usdt

USDC

steth

HYPE

doge

usds

wbtc

wbeth

link

USD1

USDE

Gram

USDG

hbar

pyusd

avax

xaut

shib

near

paxg

WLFI

ASTER

RLUSD

ondo

usdd

MORPHO

aave

BFUSD

dexe

pepe

atom

RENDER

JITOSOL

algo

USDtb

nexo

BEAT

币安人生

RETH

LBTC

STABLE

PUMP

NIGHT

aero

tusd

METH

cake

eurc

dash

SOSO

SoSoScholar

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="78c88867-b7d1-4190-896a-fd0780f89291"><div class="bn-block" data-node-type="blockContainer" data-id="78c88867-b7d1-4190-896a-fd0780f89291"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">DeepSeek is advancing its in-house AI chip plan, with the goal of reducing reliance on external chip suppliers such as Nvidia. Taken in isolation, this is not a development significant enough to change Nvidia’s near-term fundamentals. Designing, taping out, validating and eventually deploying an in-house chip at scale is a long, difficult process with a high failure rate. Moreover, DeepSeek’s scale, financial resources and supply-chain control are not comparable to those of U.S. tech giants such as Microsoft, Meta, Google and Amazon. Therefore, interpreting this news simply as “DeepSeek replacing Nvidia” would likely overstate the short-term impact.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="80e296d7-f1b2-48a4-ad83-d7f79e622699"><div class="bn-block" data-node-type="blockContainer" data-id="80e296d7-f1b2-48a4-ad83-d7f79e622699"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The significance of DeepSeek lies in the fact that it is not an isolated case, but the latest example of a broader shift taking place across the AI supply chain. Over the past two years, the core narrative behind the AI trade has been compute scarcity: large-model companies needed to train bigger models, cloud providers needed to expand data centers, enterprise customers needed access to AI services, and almost all of that demand ultimately pointed to Nvidia’s GPUs, networking equipment and software ecosystem. Because of this supply-demand imbalance, Nvidia has not only sold more chips, but also gained exceptionally strong pricing power and margins far above those typically seen in traditional semiconductor cycles.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f428b59a-3812-4ed2-bd22-46b5d38e6a89"><div class="bn-block" data-node-type="blockContainer" data-id="f428b59a-3812-4ed2-bd22-46b5d38e6a89"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Now, as more large-model companies and cloud providers move forward with in-house AI chips, the market is starting to reassess how long this logic can continue. Microsoft, Meta, Google and Amazon have already been investing in custom chips for years. The goal is not simply to “move away from Nvidia,” but more practically to lower data-center costs, optimize specific AI workloads and reduce dependence on a single supplier. DeepSeek’s entry extends this trend from U.S. cloud providers to Chinese large-model companies, further increasing market sensitivity toward a restructuring of the AI compute supply chain.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="6012371b-8bd6-4e3e-bcfd-7230d34d1f46"><div class="bn-block" data-node-type="blockContainer" data-id="6012371b-8bd6-4e3e-bcfd-7230d34d1f46"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For Nvidia, the most important near-term variable remains demand, not substitution. As long as large-model training, inference deployment and data-center expansion continue to grow, it is difficult to see Nvidia’s orders and revenue meaningfully reversing because of one customer’s in-house chip effort. Nvidia still has clear advantages in high-end training, GPU cluster interconnects, the CUDA ecosystem, full-system solutions and software toolchains. Even if customer-built chips succeed, they usually start with specific internal use cases, such as inference, recommendation systems, search, advertising or selected model services. They are unlikely to quickly replace Nvidia’s full product stack.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b9f30e82-3772-48b5-8775-8268574f2a11"><div class="bn-block" data-node-type="blockContainer" data-id="b9f30e82-3772-48b5-8775-8268574f2a11"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The issue is that the stock market often trades not on current-quarter orders, but on the profit margins and growth curve of the next several years. Over the past two years, Nvidia benefited from a “scarce asset” valuation: customers had to buy, there were few short-term alternatives, supply could not keep up with demand, and pricing and margins naturally stayed high. Once in-house chips become an industry consensus, the market will begin to price in another possibility: even if Nvidia remains the core supplier in the AI compute market, its bargaining power may gradually decline.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9c0d6916-d636-445f-9798-c015c8eae427"><div class="bn-block" data-node-type="blockContainer" data-id="9c0d6916-d636-445f-9798-c015c8eae427"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">This pressure may not show up as revenue decline. It is more likely to appear first in margin expectations. Large cloud providers are buying Nvidia GPUs while also developing ASICs or specialized AI chips, which essentially increases their negotiating leverage. In the future, if supply is no longer extremely tight and customers’ internal chips begin to handle part of the workload, Nvidia may need to rely on more complete system solutions, more flexible pricing, stronger software lock-in and faster product iteration to maintain share. Continued revenue growth and continued margin expansion may no longer be able to coexist as easily as they did over the past two years.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="057d21c3-ee80-4b31-9c8c-bd530a00547f"><div class="bn-block" data-node-type="blockContainer" data-id="057d21c3-ee80-4b31-9c8c-bd530a00547f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">This is also why AI-related stocks have been so volatile recently. The market does not lack AI stories; what it lacks is confidence in valuations. Previously, as long as AI capital expenditure expectations kept rising, Nvidia, semiconductor equipment, advanced packaging, HBM, data-center power and server supply chains could all enjoy a broad re-rating. Now, investors are asking a different question: where will this capital expenditure eventually translate into revenue, who will capture the profits, and who will bear the costs? If large-model companies can lower inference costs through in-house chips, that may support the long-term expansion of AI applications, but it does not necessarily continue to benefit the profit margins of all hardware suppliers.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="cf3d1397-34df-4f8d-a728-1aece06f87a3"><div class="bn-block" data-node-type="blockContainer" data-id="cf3d1397-34df-4f8d-a728-1aece06f87a3"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">From a supply-chain perspective, the in-house chip trend will not simply weaken the entire AI hardware sector. Instead, it will drive further internal divergence. For Nvidia, the question is whether its high margins can be sustained. Broadcom, Marvell and other ASIC design and custom-chip supply chains may benefit from the custom silicon trend among cloud providers. TSMC, advanced packaging and HBM suppliers may still benefit from more diversified chip architectures and the associated demand for foundry and memory capacity. Server, power and cooling supply chains will depend on whether overall data-center construction continues to expand. In other words, the AI hardware trade will not end because of in-house chips, but it will shift from “buying the entire compute chain” to selecting the segments that truly benefit.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="5668cb92-71ad-422b-97d7-94863d090824"><div class="bn-block" data-node-type="blockContainer" data-id="5668cb92-71ad-422b-97d7-94863d090824"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For major customers such as Microsoft, Meta, Google and Amazon, the strategic value of in-house chips lies mainly in cost control. AI investment has entered a more realistic stage: the market no longer rewards only those who spend the most, but increasingly demands evidence of return on investment. Training large models, deploying inference services and expanding data centers all require massive capital expenditure. If AI product revenue cannot quickly cover these costs, shareholders will eventually question the return on this investment cycle. In-house chips have therefore become a tool for big tech companies to control long-term costs, as well as a constraint on the pricing power of external suppliers.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="280fda26-7cf7-4f70-9fb3-66aa84f4e70d"><div class="bn-block" data-node-type="blockContainer" data-id="280fda26-7cf7-4f70-9fb3-66aa84f4e70d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">DeepSeek’s special significance is that it reinforces the “low-cost AI” narrative. DeepSeek previously led the market to reassess the cost of model training and inference, which at the time caused a clear shock to Nvidia’s share price. Now that reports of an in-house chip plan have emerged, the market will naturally connect the two: if model capability improvements no longer depend entirely on stacking the most expensive GPUs, and if inference costs can continue to fall through model optimization and in-house chips, then profit distribution across the AI supply chain may change. This logic has limited impact on Nvidia’s near-term earnings, but it does put pressure on the upside of its valuation narrative.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="500e2bdf-a5e9-46a3-86ac-018b5255eba8"><div class="bn-block" data-node-type="blockContainer" data-id="500e2bdf-a5e9-46a3-86ac-018b5255eba8"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">That said, the in-house chip trend should not be simplistically interpreted as proof that Nvidia has reached an inflection point. Historically, custom chips developed by large technology companies have not meant that external chip suppliers immediately lose the market. More often, in-house chips take on part of the standardized, scaled and internally controllable workloads, while high-performance, general-purpose and rapidly evolving workloads still rely on leading external suppliers. As long as AI models continue to grow, inference demand continues to expand and data centers continue to scale, Nvidia will remain at the core of AI infrastructure.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c1f8d630-2800-4560-a023-1cf9cb669624"><div class="bn-block" data-node-type="blockContainer" data-id="c1f8d630-2800-4560-a023-1cf9cb669624"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The real risk lies in a change to the valuation structure. Previously, Nvidia’s valuation was supported by high growth, high visibility and high margins all at once. If the market begins to believe that high growth remains intact but margins may have peaked, the stock’s trading elasticity will decline. For a high-valuation leader, the most dangerous factor is not one customer developing an in-house chip, but more and more customers moving in the same direction. A single project may fail, but an industry trend can change how the market prices the long-term competitive landscape.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="e7444d80-2f7f-44a3-980c-16ebb4875602"><div class="bn-block" data-node-type="blockContainer" data-id="e7444d80-2f7f-44a3-980c-16ebb4875602"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Therefore, the impact of this news is not about making a simple judgment that it is “negative for Nvidia and positive for other chip stocks.” Rather, it suggests that the AI trade has entered its second phase. The first phase was about compute scarcity and supply bottlenecks, with Nvidia as the most obvious beneficiary. The second phase is about cost reduction, customer bargaining power, chip customization and margin redistribution. The market will continue to chase AI, but it will no longer be willing to assign higher valuations to all AI hardware assets unconditionally.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8ffd3764-c646-4fd2-9b58-bbccd8708841"><div class="bn-block" data-node-type="blockContainer" data-id="8ffd3764-c646-4fd2-9b58-bbccd8708841"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Going forward, the key drivers of the market will not be the progress of DeepSeek’s individual chip project, but several more important validation points: whether Nvidia’s data-center revenue continues to beat expectations; whether its margin guidance remains stable; whether cloud-provider capital expenditure continues to rise or begins to emphasize investment efficiency; whether ASIC and custom-chip orders accelerate; whether HBM, advanced packaging and foundry demand remain tight; and whether AI application revenue can justify continued expansion in compute spending. If these indicators remain strong, DeepSeek’s in-house chip effort will be more of a sentiment shock. If several of them weaken at the same time, the in-house chip wave could become a core narrative weighing on AI hardware valuations.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="48391d8f-ef65-4c06-b6d5-262dd9805473"><div class="bn-block" data-node-type="blockContainer" data-id="48391d8f-ef65-4c06-b6d5-262dd9805473"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Overall, DeepSeek’s in-house chip plan is not evidence of a reversal in Nvidia’s fundamentals, but it is a signal that AI supply-chain margins are being repriced. The AI trade is not over, but it is becoming more complex: from “who controls the scarcest GPUs” to “who can preserve profits as compute costs decline.” For Nvidia, the growth story remains intact, but the sustainability of its high margins will face more scrutiny. For the broader AI sector, this means the market is likely to move from broad-based valuation expansion into a more differentiated phase driven by earnings delivery, cost control and supply-chain positioning.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="e7373a6d-d07e-4c75-8c52-34aa61ed6d53"><div class="bn-block" data-node-type="blockContainer" data-id="e7373a6d-d07e-4c75-8c52-34aa61ed6d53"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

The Rise of In-House AI Chips: Nvidia’s Issue Is Not Orders, but Margin Expectations

自研AI芯片的崛起：英伟达的问题不在于订单，而在于利润预期

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="a531566d-d5fa-4820-88f0-e0c731f55427"><div class="bn-block" data-node-type="blockContainer" data-id="a531566d-d5fa-4820-88f0-e0c731f55427"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Strategy is undergoing a critical narrative shift.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="78e3292d-f87d-4932-aaca-1a5f68c41aee"><div class="bn-block" data-node-type="blockContainer" data-id="78e3292d-f87d-4932-aaca-1a5f68c41aee"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For the past several years, the market’s understanding of the company was relatively straightforward: it was the world’s largest corporate Bitcoin holder and the most aggressive listed-company buyer of BTC. Each time Strategy disclosed a new purchase, the Bitcoin market treated it as confirmation of structural demand. Each time MSTR traded at a premium to its Bitcoin net asset value, the company could return to capital markets, raise more capital, buy more BTC, and reinforce the same narrative.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="12b94090-3e6b-491c-bb5c-4a5f2b53f1e1"><div class="bn-block" data-node-type="blockContainer" data-id="12b94090-3e6b-491c-bb5c-4a5f2b53f1e1"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Over the past week, however, the market saw a very different side of the company. Strategy reduced its holdings by 3,588 BTC, bringing total holdings down to 843,775 BTC. Michael Saylor later said the cash was used for the second-quarter dividends on $STRF, $STRE, $STRK and $STRD, as well as the full June monthly dividend on $STRC. In other words, this sale was not simply an asset rebalancing exercise. It was a signal that Strategy’s increasingly large capital structure is beginning to feed back into its Bitcoin holdings.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="ef5491f3-7006-4849-9fbf-bf756f34284a"><div class="bn-block" data-node-type="blockContainer" data-id="ef5491f3-7006-4849-9fbf-bf756f34284a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The more important point is that this was not an isolated event. Strategy had already sold BTC once in late May. In late June, the company’s board approved a series of new measures, including a dollar reserve policy, preferred stock dividend policy, preferred and common stock repurchase authorizations, and a BTC monetization plan. Strategy used to be the most stable and symbolic BTC buyer in the market. Now, it is starting to operate with an increasingly visible cash payment schedule.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="222e79cb-3e35-4505-bc6a-2cbccb760a9e"><div class="bn-block" data-node-type="blockContainer" data-id="222e79cb-3e35-4505-bc6a-2cbccb760a9e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">This means Strategy’s business model is shifting from a one-way flywheel of continuous BTC accumulation to a balance-sheet structure in which BTC is used to support a complex capital stack.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="efa8a9a1-c0e4-434a-865f-31dc5f6bb502"><div class="bn-block" data-node-type="blockContainer" data-id="efa8a9a1-c0e4-434a-865f-31dc5f6bb502"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The old positive loop depended on three conditions: BTC price appreciation, MSTR trading at a premium to its BTC net asset value, and capital markets remaining willing to provide Strategy with low-cost funding. As long as those three conditions held, the company could issue common stock, convertible debt or preferred stock, use the proceeds to buy more BTC, and strengthen the market’s perception of MSTR as a leveraged BTC vehicle. That, in turn, supported the stock price and the company’s financing capacity.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="3f8c177e-9e89-45ee-9199-ee03e9f92f03"><div class="bn-block" data-node-type="blockContainer" data-id="3f8c177e-9e89-45ee-9199-ee03e9f92f03"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The problem is that this flywheel is highly sensitive to market conditions. When BTC falls, MSTR’s premium contracts and preferred shares come under pressure, Strategy’s cost of capital rises. At the same time, the company still needs to pay preferred dividends and debt interest on schedule, while maintaining a certain level of dollar reserves. Cash obligations do not disappear when BTC falls. They become more prominent when capital-market access narrows.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="3314fc2a-3a66-47ee-b7b6-dd2ffe22ba0a"><div class="bn-block" data-node-type="blockContainer" data-id="3314fc2a-3a66-47ee-b7b6-dd2ffe22ba0a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">That is Strategy’s current dilemma: its asset side is highly concentrated in BTC, while its liability side and quasi-liability structure are becoming increasingly institutionalized, securitized and periodic. BTC does not generate cash flow, but preferred shares and debt come with fixed payment schedules. BTC can be held for the long term, but dividends and interest must be paid on time. BTC’s balance-sheet value fluctuates sharply with the market, while the company’s cash obligations are rigid.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="def53302-e3c0-40da-814a-9e49e964fd1c"><div class="bn-block" data-node-type="blockContainer" data-id="def53302-e3c0-40da-814a-9e49e964fd1c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">As a result, the market significance of this sale goes beyond the absolute size of 3,588 BTC. Relative to Strategy’s total holdings of more than 840,000 BTC, the amount is not large. The real issue is the change in role: the market used to see Strategy as a potential buyer on BTC drawdowns. Now, it has started to become a holder that may release supply at specific points in time.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="04cd6466-d44e-4e7c-93a7-f36773957f09"><div class="bn-block" data-node-type="blockContainer" data-id="04cd6466-d44e-4e7c-93a7-f36773957f09"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The largest buyer has become a seller with a timetable. That is the core repricing point for the market.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d8bc665e-6f08-48b6-88c8-2cd1eda8f378"><div class="bn-block" data-node-type="blockContainer" data-id="d8bc665e-6f08-48b6-88c8-2cd1eda8f378"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For MSTR, the valuation framework is also changing. Previously, the stock was more easily traded as a high-beta BTC proxy. Investors were buying BTC upside, the Saylor narrative, and the company’s capital-market financing capacity. At this stage, however, MSTR can no longer be explained by BTC price alone. It now embeds BTC exposure, preferred dividend pressure, dollar reserve levels, financing access, potential dilution, buyback arrangements and expectations around BTC monetization.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="ff0dbada-63f4-4b39-a2a2-7ec97c83f291"><div class="bn-block" data-node-type="blockContainer" data-id="ff0dbada-63f4-4b39-a2a2-7ec97c83f291"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">This will make trading MSTR more complicated. When BTC rises, MSTR can still deliver stronger upside, as the market may again price in a recovery in financing capacity, a wider mNAV premium, and a possible return to the accumulation cycle. But when BTC falls, MSTR’s pressure may also be amplified, because investors will not only worry about the BTC price drawdown itself, but also whether the company needs to keep selling BTC to meet cash obligations. They may also start to price in the risk of a negative feedback loop across the capital structure.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f50e5769-c357-4533-8572-c77d44b78494"><div class="bn-block" data-node-type="blockContainer" data-id="f50e5769-c357-4533-8572-c77d44b78494"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The path of that negative feedback is not difficult to understand. A BTC decline reduces MSTR’s net asset value, causing its premium to compress or even turn into a discount. That lowers the appeal of common-stock issuance. Meanwhile, falling preferred share prices and rising yields push up the company’s cost of capital. If dollar reserves are not enough to cover future obligations, the company may need to continue selling BTC. Once the market begins to anticipate this behavior, MSTR’s discount pressure and BTC’s supply pressure may reinforce each other.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="e47d5e38-02c5-4e5a-a791-135fd4a0f12e"><div class="bn-block" data-node-type="blockContainer" data-id="e47d5e38-02c5-4e5a-a791-135fd4a0f12e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For BTC, Strategy’s sale is not yet a decisive supply shock, but it has become a sentiment and structural disturbance. A sale of 3,588 BTC is not enough to change the long-term supply-demand balance of the global BTC market, especially when spot ETFs, macro liquidity and institutional allocation remain the dominant drivers. Strategy’s weekly sale is more of a marginal variable. Still, the BTC market is highly dependent on narratives and expectations. When the most representative corporate buyer starts selling BTC to service the cost of its capital structure, the market will naturally place less faith in the idea of Strategy as a permanent source of demand.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="34bc2bd7-ad37-4079-9648-983d812e9d11"><div class="bn-block" data-node-type="blockContainer" data-id="34bc2bd7-ad37-4079-9648-983d812e9d11"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The impact on ordinary traders is direct.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="7adb0193-5705-4af1-8f1a-927939f1cd31"><div class="bn-block" data-node-type="blockContainer" data-id="7adb0193-5705-4af1-8f1a-927939f1cd31"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For BTC traders, Strategy needs to be watched as a potential source of supply during rallies rather than simply as a sentiment backstop. In the short term, each disclosure of BTC sales, dollar reserve replenishment or preferred dividend payment may weaken confidence in the durability of BTC rebounds. Over the medium to long term, however, BTC’s core pricing will still depend on macro liquidity, ETF inflows, risk appetite and on-chain supply-demand conditions. Strategy’s influence is more likely to be a localized disturbance than a trend-defining force.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="15248197-ab5c-4221-b5ac-ece4a3db7e0c"><div class="bn-block" data-node-type="blockContainer" data-id="15248197-ab5c-4221-b5ac-ece4a3db7e0c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For MSTR traders, the stock should be reclassified from a simple leveraged BTC play into a BTC asset vehicle with credit constraints. When risk appetite recovers, BTC strengthens again and MSTR’s mNAV returns to a premium, MSTR may remain one of the most elastic BTC-linked equities. But during periods of BTC weakness, tighter financing conditions and pressure on preferred shares, MSTR’s downside risk may be meaningfully higher than that of BTC itself. Trading MSTR now requires more than a view on BTC direction. It also requires a view on Strategy’s financing capacity, cash coverage and whether capital markets are still willing to support its structure.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="042dff85-ce0a-4246-a2af-a3e1725b8912"><div class="bn-block" data-node-type="blockContainer" data-id="042dff85-ce0a-4246-a2af-a3e1725b8912"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For investors focused on Strategy’s preferred shares, the analysis should be closer to credit than equity. The key question is not simply whether BTC will rise, but whether the company can keep covering dividends, whether dollar reserves are sufficient, how much cushion BTC holdings can provide in an extreme drawdown, and whether the company will use buybacks, refinancing or BTC sales to stabilize preferred share prices. The prices and yields of the preferred shares themselves will become an important dashboard for monitoring pressure inside Strategy.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="86d3038c-3b90-4803-9183-c4d3a501c3c6"><div class="bn-block" data-node-type="blockContainer" data-id="86d3038c-3b90-4803-9183-c4d3a501c3c6"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The market now needs to focus on three signals.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="59d804ce-9641-43c6-8c0c-7d68bf75ac3e"><div class="bn-block" data-node-type="blockContainer" data-id="59d804ce-9641-43c6-8c0c-7d68bf75ac3e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">First, whether Strategy continues to sell BTC, and whether the pace of sales closely corresponds to preferred dividend payments, debt interest and dollar reserve targets. If BTC sales become periodic, the market will start applying a liquidity discount to its holdings.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2089ea72-e063-4eb4-843c-d164ae738c89"><div class="bn-block" data-node-type="blockContainer" data-id="2089ea72-e063-4eb4-843c-d164ae738c89"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Second, whether MSTR’s mNAV premium can recover. As long as MSTR regains a meaningful premium, the company may be able to replace BTC sales with common-stock issuance or other financing tools, allowing the flywheel to restart. Conversely, if MSTR trades at a discount for an extended period, its financing capacity will be constrained and BTC monetization expectations will be harder to remove.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="dea3a2de-3204-4597-831d-6a8f5ce425c2"><div class="bn-block" data-node-type="blockContainer" data-id="dea3a2de-3204-4597-831d-6a8f5ce425c2"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Third, whether the preferred stock market stabilizes. The prices, yields and liquidity of STRC, STRF, STRD, STRK and STRE will directly reflect the market’s confidence in Strategy’s credit structure. If preferred shares remain under pressure, it would suggest the market is no longer treating Strategy only as a BTC long vehicle, but is reassessing its repayment capacity and capital-structure sustainability.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="958cf881-0e36-46cb-a505-4e913917db88"><div class="bn-block" data-node-type="blockContainer" data-id="958cf881-0e36-46cb-a505-4e913917db88"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Strategy remains one of the most important listed-company balance sheets in the Bitcoin market. Its BTC holdings are still enormous, its market influence remains significant, and Saylor’s narrative power has not disappeared. But from a trading perspective, Strategy has entered a more complex phase. It is both a long-term BTC holder and an issuer with scheduled cash costs. It may again become a buyer when markets recover, but it may also become a source of supply during periods of stress.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="30d56ae0-de34-4b55-889a-bbea3289de8d"><div class="bn-block" data-node-type="blockContainer" data-id="30d56ae0-de34-4b55-889a-bbea3289de8d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">This change will not immediately end MSTR’s high-beta trading profile, nor will it alone determine BTC’s medium- to long-term direction. But it will change how the market values Strategy. Investors can no longer focus only on how much BTC the company holds. They also need to look at how much cash obligation sits behind those holdings, how much preferred dividend pressure exists, how much debt interest must be paid, and how much capacity the company still has to maintain the entire capital structure without selling BTC.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9633141c-b745-483f-ac99-85d6dbb8e1a1"><div class="bn-block" data-node-type="blockContainer" data-id="9633141c-b745-483f-ac99-85d6dbb8e1a1"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Strategy’s flywheel has not stopped, but it has entered a headwind phase. For traders, the real focus is no longer simply whether Saylor keeps buying BTC. It is whether this system, built around BTC, MSTR, preferred shares and capital-market financing, can maintain balance between price volatility and scheduled cash payments.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f90d25c0-b143-4a83-8b70-ca444ade726b"><div class="bn-block" data-node-type="blockContainer" data-id="f90d25c0-b143-4a83-8b70-ca444ade726b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

Strategy’s Bitcoin Flywheel Enters a Headwind Phase

策略的 Bitcoin 飞轮进入逆风阶段

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="0b3a4b9d-d16e-4708-bfeb-5136370603d0"><div class="bn-block" data-node-type="blockContainer" data-id="0b3a4b9d-d16e-4708-bfeb-5136370603d0"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The key message from the June nonfarm payrolls report is not that the U.S. labor market has already weakened, but that the previously overheated employment narrative is starting to cool. Nonfarm payrolls rose by only 57,000, well below market expectations, while April and May were revised down by a combined 74,000, suggesting labor demand over the past two months was not as strong as the initial readings implied. The unemployment rate edged down from 4.3% to 4.2%, but the labor force participation rate fell to 61.8%, meaning a contraction in labor supply mechanically depressed the jobless rate and weakened its standalone signaling value. According to BLS data, June job gains were mainly concentrated in professional and business services, social assistance, and health care, while leisure and hospitality employment declined noticeably. Manufacturing and construction only recovered modestly. Overall, the report points to a resilient labor market that is cooling gradually, not one that is reaccelerating.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="5756541a-bc73-47f2-8098-df87ec216582"><div class="bn-block" data-node-type="blockContainer" data-id="5756541a-bc73-47f2-8098-df87ec216582"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The most direct implication for the Fed is that the report lowers the need for an immediate rate hike, while failing to fully eliminate the tail risk of another move later this year. Warsh’s message after the June FOMC meeting was fairly clear: the Fed will reduce reliance on forward guidance, limit the market’s dependence on the dot plot and policy hints, and push the reaction function back toward inflation, employment, and financial conditions themselves. More recently, at the ECB forum, he acknowledged that inflation risks had eased somewhat, but refused to rule out further rate hikes, suggesting the current policy function remains biased toward inflation control. In other words, as long as inflation does not decline consistently, the Fed has little incentive to reopen the rate-cut discussion. At the same time, softer payroll growth, downward revisions, and lower participation have also raised the bar for an immediate hike in July or September.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="ea2fa009-04f1-49b6-b198-3ff5741641e7"><div class="bn-block" data-node-type="blockContainer" data-id="ea2fa009-04f1-49b6-b198-3ff5741641e7"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">From a market-pricing perspective, traders began reducing bets on another rate hike after the payrolls release. On Polymarket, the probability of an October hike fell from 37% to 28%, while the probability of a December hike declined from 55% to 47%. This suggests the market has not fully ruled out another hike this year, but it no longer treats one as the near-term base case. The employment data carry a nuanced message: slower job growth and downward revisions reduce the need for an immediate hike, but the unemployment rate remains low and the labor market has not visibly cracked, which is also insufficient for markets to reprice toward rate cuts. Policy expectations have therefore moved into a phase of waiting for inflation confirmation.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8db50313-4170-442f-875d-15e92e11cf5d"><div class="bn-block" data-node-type="blockContainer" data-id="8db50313-4170-442f-875d-15e92e11cf5d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">As a result, whether the U.S. hikes again this year depends less on employment and more on the next two to three inflation reports. If June and July CPI and core PCE continue to ease, especially if energy shocks and AI data-center-related electricity and equipment cost pressures do not spread further, the Fed is likely to keep rates unchanged and use the previously hawkish dot plot as a communication tool to anchor inflation expectations. If inflation reaccelerates while wage growth remains resilient, October or December could again become live hike windows. The more reasonable base case for now is that the economy does not support a sustained hiking cycle this year, while the probability of a single “insurance-style” hike has declined but not disappeared. For traders, the key is not to bet on a static conclusion, but to watch whether inflation data can keep pushing market-implied hike probabilities lower.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2615d8f5-2ff8-4b65-9180-edc912b6d505"><div class="bn-block" data-node-type="blockContainer" data-id="2615d8f5-2ff8-4b65-9180-edc912b6d505"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For crypto markets, the payrolls report is modestly supportive of risk assets in the short term, because it weakens the pressure for an immediate hike without triggering a recession trade. Highly liquid assets such as BTC and ETH are most vulnerable to two macro combinations: strong employment combined with sticky inflation, which lifts real rates and the dollar; or a sharp deterioration in employment, which forces risk-asset deleveraging. Current data sit between those two extremes, helping crypto assets maintain a rebound window. But this support is more about valuations and liquidity expectations than a standalone driver of trend. If inflation continues to ease and ETF flows turn positive again, BTC could benefit further from expectations of lower real rates. If inflation rebounds and October hike odds rise again, crypto markets would likely come under pressure first, with altcoins and highly leveraged on-chain assets more vulnerable than BTC.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="93f9d254-fb1a-4229-bf85-cbd202c9dfbe"><div class="bn-block" data-node-type="blockContainer" data-id="93f9d254-fb1a-4229-bf85-cbd202c9dfbe"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">From a trading perspective, crypto investors should focus on three signals. First, whether Polymarket and rates futures continue to price down October and December hike odds. Second, whether BTC and ETH ETFs move from one-day stabilization to sustained net inflows. Third, whether the dollar index and 10-year real yields fall together. Only when these three signals align will crypto have the conditions to shift from a rebound into a broader trend recovery. If rate-hike odds merely fall temporarily while ETF flows fail to follow, BTC is more likely to remain range-bound, while higher-beta assets such as HYPE, SOL, and XRP will show stronger event-driven dispersion.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c1dd175a-a4b3-42e1-a2f0-a2b4de6d498b"><div class="bn-block" data-node-type="blockContainer" data-id="c1dd175a-a4b3-42e1-a2f0-a2b4de6d498b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For U.S. equities, weaker-than-expected payrolls provide short-term support for the AI trade. The AI sector is facing three main pressures this year: valuation, capital expenditure, and interest rates. Rising rate expectations compress long-duration growth valuations; doubts over AI capex returns amplify stock-level divergence; and slower employment could eventually weigh on enterprise IT budgets. This payrolls mix temporarily eases the first pressure, as lower discount-rate stress gives the Nasdaq and semiconductor stocks room to rebound. Softer labor data improved risk appetite and supported technology shares, which is consistent with the post-payrolls market reaction.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="1a4cef3f-8acf-4c12-9269-3270ee0fe5e8"><div class="bn-block" data-node-type="blockContainer" data-id="1a4cef3f-8acf-4c12-9269-3270ee0fe5e8"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">However, AI stocks should not be interpreted as an across-the-board beneficiary of lower hike odds. The AI trade has already moved from its first phase of aggregate compute-demand expansion into a second phase of return-on-investment verification. Hardware names such as Nvidia, Broadcom, TSMC, and Micron still benefit from cloud capex, but investors will scrutinize order sustainability, peak margins, customer concentration, and inventory risk more aggressively. Platform companies such as Microsoft, Meta, Google, and Amazon face a two-sided repricing: lower rate expectations support valuations, but the larger their AI investment becomes, the more the market will demand clear returns from advertising, cloud revenue, and enterprise AI adoption. The jobs report can relieve macro valuation pressure, but it cannot replace fundamental validation.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="52922472-14de-4929-8c68-350319df1cec"><div class="bn-block" data-node-type="blockContainer" data-id="52922472-14de-4929-8c68-350319df1cec"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Overall, the best trading framework is that risk assets can continue to repair, but investors should avoid chasing indiscriminate beta. In crypto, the preference should be for BTC, ETH, and assets with clear flow or product catalysts, while reducing exposure to highly leveraged and low-liquidity altcoins. In U.S. AI equities, the preference should be for leaders with high earnings visibility, strong order quality, and cash flows capable of supporting capex, while treating second-tier AI names that rely mainly on narrative expansion more cautiously. If inflation continues to cool, hike odds could fall further, allowing BTC, ETH, the Nasdaq, and semiconductor leaders to extend their recovery. If inflation reaccelerates, the market will quickly shift from “cooling jobs are good for risk assets” to “the Fed may be forced to tighten,” with high-valuation AI stocks and high-beta crypto assets likely to face the fastest de-risking.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="3dba389e-000c-4a5e-bdf7-775a3bc76e2e"><div class="bn-block" data-node-type="blockContainer" data-id="3dba389e-000c-4a5e-bdf7-775a3bc76e2e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The bottom line is that the June payrolls report reduced the urgency of another rate hike this year, but it did not close the hike window. From a trading perspective, it should be treated as a period of relief for risk assets, not confirmation of a full policy pivot. The next phase of market direction will depend on whether inflation can continue to decline and whether confidence in Warsh’s hawkish framework continues to erode.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="6f15d650-e6af-4813-bd97-9083aead80c8"><div class="bn-block" data-node-type="blockContainer" data-id="6f15d650-e6af-4813-bd97-9083aead80c8"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

Payrolls Cool but Do Not Crack: Fed Hike Trade Enters a Data-Validation Phase

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="0b3a4b9d-d16e-4708-bfeb-5136370603d0"><div class="bn-block" data-node-type="blockContainer" data-id="0b3a4b9d-d16e-4708-bfeb-5136370603d0"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"> 6月份非农就业报告的关键信息并非美国劳动力市场已经疲软，而是此前过热的就业形势开始降温。非农就业人数仅增加5.7万人，远低于市场预期，而4月和5月的数据合计下调了7.4万人，表明过去两个月的劳动力需求并不像最初数据所显示的那样强劲。失业率从4.3%小幅下降至4.2%，但劳动参与率降至61.8%，这意味着劳动力供给的萎缩直接压低了失业率，削弱了其独立的信号价值。根据美国劳工统计局（BLS）的数据，6月份新增就业岗位主要集中在专业和商业服务、社会援助以及医疗保健行业，而休闲和酒店行业的就业人数则明显下降。制造业和建筑业的复苏幅度也十分有限。总体而言，该报告表明劳动力市场具有韧性，但正在逐步降温，而非重新加速增长。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="5756541a-bc73-47f2-8098-df87ec216582"><div class="bn-block" data-node-type="blockContainer" data-id="5756541a-bc73-47f2-8098-df87ec216582"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">对美联储而言，最直接的影响是这份报告降低了立即加息的必要性，但同时也未能完全消除今年晚些时候再次加息的尾部风险。沃什在6月联邦公开市场委员会（FOMC）会议后传递的信息相当明确：美联储将减少对前瞻性指引的依赖，限制市场对点阵图和政策暗示的依赖，并将政策反应机制重新转向通胀、就业和金融状况本身。最近，在欧洲央行论坛上，他承认通胀风险有所缓解，但拒绝排除进一步加息的可能性，暗示当前的政策功能仍然偏向于控制通胀。换句话说，只要通胀没有持续下降，美联储就没有多少动力重新开启降息讨论。与此同时，非农就业增长放缓、预期值下调以及参保率下降也提高了7月或9月立即加息的可能性。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="ea2fa009-04f1-49b6-b198-3ff5741641e7"><div class="bn-block" data-node-type="blockContainer" data-id="ea2fa009-04f1-49b6-b198-3ff5741641e7"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">从市场定价的角度来看，非农就业数据公布后，交易员开始减少对再次加息的预期。在Polymarket平台上，10月份加息的概率从37%降至28%，12月份加息的概率从55%降至47%。这表明市场并未完全排除今年再次加息的可能性，但已不再将其视为近期加息的基本预期。就业数据传递出的信息较为复杂：就业增长放缓和预期值下调降低了立即加息的必要性，但失业率仍然很低，劳动力市场尚未出现明显疲软迹象，这也不足以促使市场重新调整定价，转向降息预期。因此，政策预期已进入等待通胀确认的阶段。 </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8db50313-4170-442f-875d-15e92e11cf5d"><div class="bn-block" data-node-type="blockContainer" data-id="8db50313-4170-442f-875d-15e92e11cf5d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">因此，美国今年是否再次加息，与其说是取决于就业情况，不如说是取决于未来两到三个月的通胀报告。如果6月和7月的CPI和核心PCE继续放缓，尤其是在能源冲击和人工智能数据中心相关的电力及设备成本压力没有进一步蔓延的情况下，美联储可能会维持利率不变，并继续使用此前鹰派的点阵图作为沟通工具来锚定通胀预期。如果通胀再次加速，而工资增长依然强劲，那么10月或12月可能再次成为加息窗口。目前更合理的假设是，今年的经济状况不足以支撑持续的加息周期，而单次“保险式”加息的可能性虽然降低，但并未完全消失。对于交易员而言，关键不在于押注于一个固定的结论，而在于观察通胀数据能否继续压低市场隐含的加息概率。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2615d8f5-2ff8-4b65-9180-edc912b6d505"><div class="bn-block" data-node-type="blockContainer" data-id="2615d8f5-2ff8-4b65-9180-edc912b6d505"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">对于加密货币市场而言，就业报告短期内对风险资产起到一定的支撑作用，因为它削弱了短期内加息的压力，同时又不会引发经济衰退交易。比特币和以太坊等高流动性资产最容易受到两种宏观经济组合的影响：一是强劲的就业形势伴随顽固的通胀，这将推高实际利率和美元；二是就业形势急剧恶化，这将迫使风险资产去杠杆化。目前的数据介于这两种极端情况之间，有助于加密资产维持反弹窗口。但这种支撑更多地源于估值和流动性预期，而非趋势的独立驱动因素。如果通胀持续放缓，ETF资金流动再次转正，比特币可能会进一步受益于实际利率走低的预期。如果通胀反弹，10月份加息的可能性再次上升，加密货币市场可能会率先承压，其中山寨币和高杠杆链上资产比比特币更容易受到冲击。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="93f9d254-fb1a-4229-bf85-cbd202c9dfbe"><div class="bn-block" data-node-type="blockContainer" data-id="93f9d254-fb1a-4229-bf85-cbd202c9dfbe"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">从交易角度来看，加密货币投资者应关注三个信号。首先，Polymarket 和利率期货是否会继续压低 10 月和 12 月加息的概率。其次，BTC 和 ETH ETF 是否会从一日企稳转为持续净流入。第三，美元指数和 10 年期实际收益率是否会同步下跌。只有当这三个信号一致时，加密货币才具备从反弹转向更广泛的趋势复苏的条件。如果加息概率只是暂时下降，而 ETF 资金流入未能跟进，那么 BTC 更有可能维持区间震荡，而 HYPE、SOL 和 XRP 等高贝塔系数资产则会表现出更强烈的事件驱动型波动。 </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c1dd175a-a4b3-42e1-a2f0-a2b4de6d498b"><div class="bn-block" data-node-type="blockContainer" data-id="c1dd175a-a4b3-42e1-a2f0-a2b4de6d498b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">对于美国股市而言，弱于预期的非农就业数据为人工智能板块提供了短期支撑。人工智能板块今年面临三大压力：估值、资本支出和利率。利率预期上升会压缩长期增长型股票的估值；对人工智能资本支出回报的担忧会加剧个股层面的分化；而就业放缓最终可能会对企业IT预算造成压力。此次非农就业数据暂时缓解了第一个压力，因为较低的贴现率压力为纳斯达克指数和半导体股票提供了反弹空间。疲软的劳动力市场数据提振了风险偏好，支撑了科技股，这与非农就业数据公布后的市场反应一致。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="1a4cef3f-8acf-4c12-9269-3270ee0fe5e8"><div class="bn-block" data-node-type="blockContainer" data-id="1a4cef3f-8acf-4c12-9269-3270ee0fe5e8"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">然而，不应将人工智能股票视为加息概率降低的普遍受益者。人工智能交易已从计算需求总体扩张的第一阶段进入投资回报验证的第二阶段。英伟达、博通、台积电和美光等硬件公司仍受益于云计算资本支出，但投资者将更加严格地审查订单可持续性、利润率峰值、客户集中度和库存风险。微软、Meta、谷歌和亚马逊等平台公司面临着双重定价：较低的加息预期支撑了估值，但随着其人工智能投资规模的扩大，市场将更加关注广告、云收入和企业人工智能应用带来的明确回报。就业报告可以缓解宏观估值压力，但无法取代基本面验证。 </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="52922472-14de-4929-8c68-350319df1cec"><div class="bn-block" data-node-type="blockContainer" data-id="52922472-14de-4929-8c68-350319df1cec"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">总体而言，最佳交易框架是风险资产可能持续修复，但投资者应避免盲目追逐高贝塔系数。在加密货币领域，应优先选择比特币 (BTC)、以太坊 (ETH) 以及具有明确现金流或产品催化剂的资产，同时减少对高杠杆、低流动性山寨币的投资。在美国人工智能股票领域，应优先选择盈利可见性高、订单质量强、现金流足以支撑资本支出的领头羊公司，同时对主要依赖市场叙事扩张的二线人工智能公司保持谨慎。如果通胀持续降温，加息的可能性可能会进一步降低，从而使比特币、以太坊、纳斯达克指数和半导体领头羊公司得以延续复苏势头。如果通胀再次加速，市场将迅速从“就业降温有利于风险资产”转变为“美联储可能被迫收紧货币政策”，届时高估值人工智能股票和高贝塔系数加密资产的风险降低速度可能会最快。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="3dba389e-000c-4a5e-bdf7-775a3bc76e2e"><div class="bn-block" data-node-type="blockContainer" data-id="3dba389e-000c-4a5e-bdf7-775a3bc76e2e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">总而言之，6月份的非农就业报告降低了今年再次加息的紧迫性，但并未关闭加息窗口。从交易角度来看，这应被视为风险资产的一段喘息期，而非政策全面转向的确认。市场下一阶段的走向将取决于通胀能否继续下降，以及市场对沃什鹰派政策框架的信心是否会持续减弱。 </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="6f15d650-e6af-4813-bd97-9083aead80c8"><div class="bn-block" data-node-type="blockContainer" data-id="6f15d650-e6af-4813-bd97-9083aead80c8"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

薪资数据降温但未失速：美联储加息交易进入数据验证阶段

Memory Stocks Suddenly Slump: What Did Meta’s Compute-Selling Plan Really Hit?

内存股突然暴跌：Meta 的计算销售计划到底触及了什么？

How Will Tens of Billions Actually Flow Into SpaceX After It Crashes the Index Party — and Will the Stock Really Get Ripped Higher?

SpaceX 在“指数派对”崩盘后，数千亿美元资金究竟将如何流入？其股价真的会大幅飙升吗？

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="666b3214-d9f9-43d1-96ec-bf229e943aa4"><div class="bn-block" data-node-type="blockContainer" data-id="666b3214-d9f9-43d1-96ec-bf229e943aa4"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">On June 23 local time, Ethereum co-founder Vitalik Buterin confirmed on X that the Ethereum Foundation (EF) will cut its 2026 budget by about 40%. On the same day, the foundation confirmed layoffs of 54 people, roughly 20% of its total headcount. This marks the ninth senior-level departure from the foundation since January. Former co-executive director Tomasz Stańczak stepped down in February, while another co-executive director, Hsiao-Wei Wang, resigned this month after taking leave. On the surface, this looks like a round of crisis-driven layoffs. But viewed together with the foundation’s Treasury Management Policy published last year, it looks more like a pre-planned balance sheet restructuring.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2f87d890-c87f-4364-af30-aa7569b75a90"><div class="bn-block" data-node-type="blockContainer" data-id="2f87d890-c87f-4364-af30-aa7569b75a90"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The picture becomes much clearer if EF is viewed as an endowment managing crypto-native assets. In the past, the foundation consumed about 15% of its remaining funds each year. Such a drawdown rate would be considered aggressive in the framework of university endowments or sovereign funds. Mature perpetual funds typically keep their annualized spending rate in the 4%-5% range to ensure capital can operate across multiple cycles. Buterin’s target is to reduce that ratio to around 5% before 2030. This is not merely cost-cutting. It is an attempt to transform EF from a “consumption-based” project-funding institution into one that can exist indefinitely — in essence, a redefinition of its duration risk.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="607c6e3f-8539-46df-82f1-787bc6fd943c"><div class="bn-block" data-node-type="blockContainer" data-id="607c6e3f-8539-46df-82f1-787bc6fd943c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The spending contraction is taking place across several fronts. The first is a paradigm shift in security strategy. Ethereum previously relied on multi-client redundancy to mitigate risk: as long as no single client’s failure rate exceeded 33%, the network could continue producing blocks and reaching finality. The foundation is now moving toward greater reliance on AI-assisted formal verification, shifting clients from “mutual backups” to “specialized division of labor.” In essence, it is using automated verification to replace human redundancy and lower maintenance costs. Second, PSE, the team focused on zero-knowledge proof research under Privacy and Scaling Explorations, has been dissolved as an independent unit. That means this budget is moving from pure “exploration” toward “implementation” directly embedded in the protocol layer and access layer — closer to the corporate logic of shifting basic research budgets into product engineering. Third, Devcon continues to shrink in scale. Buterin has begun personally funding “beyond Ethereum” projects that were previously led by the foundation, while institutional partnerships have also narrowed toward a small number of replicable CROPS-friendly pilots — censorship resistance, resilience, openness, privacy and security. This is a classic case of SG&amp;A deleveraging.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a4331d22-aa2e-4609-be09-f97a1937ff68"><div class="bn-block" data-node-type="blockContainer" data-id="a4331d22-aa2e-4609-be09-f97a1937ff68"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The restructuring is happening against an unfriendly market backdrop. ETH is hovering around $1,600, down more than 60% from its August 2025 high of nearly $4,956. Its performance has clearly lagged most major crypto assets and exceeded Bitcoin’s roughly 48% drawdown over the same period. More importantly, the Client Incentive Program, which expired in April, had previously provided annual funding to independent client teams such as Geth, Erigon and Lighthouse. Former EF contributor Trenton Van Epps previously estimated that maintaining the existing core development ecosystem requires about $30 million a year, while EF’s budget contraction makes it difficult for the foundation itself to fully cover that gap. The market response came almost simultaneously: on June 22, Ethlabs officially launched, led by several former senior EF researchers and funded by BitMine, SharpLink and Ethereum co-founder Joseph Lubin. The project focuses on institutional-grade settlement, native asset issuance and cross-chain infrastructure. This creates a structural complement to EF’s contraction: as the foundation steps back from parts of its “growth investment” role, private capital and ecosystem organizations are beginning to absorb the R&amp;D and funding functions being released. The dynamic resembles a mature company spinning off high-risk growth businesses into independently financed platforms while returning its own focus to conservative balance sheet management.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="ce4cc347-6ed6-4962-88e1-34daeb5cadcf"><div class="bn-block" data-node-type="blockContainer" data-id="ce4cc347-6ed6-4962-88e1-34daeb5cadcf"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Looking further ahead, Buterin’s endgame can be understood as a “soft lean-out.” Strawmap, part of Ethereum’s protocol roadmap, covers multiple core areas including consensus, proof systems, privacy and account models. It is viewed as an important upgrade package for Ethereum’s next protocol phase after the PoW era and The Merge. Once this roadmap is gradually completed, the foundation may significantly raise the threshold for launching new features. Its governance style could also shift away from high-intensity, broad-based R&amp;D expansion toward a model that places greater emphasis on security, convergence and long-term maintenance. In other words, Ethereum’s development curve is moving away from that of a growth-stage technology company with sustained high R&amp;D spending, and gradually toward a governance path closer to that of a sovereign monetary network.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="be043b6f-12ad-464b-8397-7e6883548619"><div class="bn-block" data-node-type="blockContainer" data-id="be043b6f-12ad-464b-8397-7e6883548619"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Solana co-founder Anatoly Yakovenko interpreted the move as positive, arguing that budget constraints will force focus, and that a smaller, leaner foundation can respond and correct mistakes faster. But the risks are equally real: the talent outflow has not yet stopped, there is still no clear solution to the client incentive funding gap, and whether new organizations such as Ethlabs can shoulder the R&amp;D responsibilities previously carried by EF remains an untested assumption. Ultimately, this “balance sheet reduction” is not testing how much money the foundation still has. It is testing whether the entire ecosystem can complete a transition from dependence on capital injections from a single foundation to a diversified supply of market capital — and that is the real variable that will determine Ethereum’s next narrative and valuation.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="69dca9f2-fd27-4690-bf5d-6300d6ede13d"><div class="bn-block" data-node-type="blockContainer" data-id="69dca9f2-fd27-4690-bf5d-6300d6ede13d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

Ethereum Foundation “Balance Sheet Reduction”: Restructuring from 15% to 5%

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="666b3214-d9f9-43d1-96ec-bf229e943aa4"><div class="bn-block" data-node-type="blockContainer" data-id="666b3214-d9f9-43d1-96ec-bf229e943aa4"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">当地时间6月23日，Ethereum联合创始人Vitalik Buterin在X上确认，Ethereum基金会（EF）将削减其2026年预算约40%。同日，该基金会确认裁员54人，约占其总人数的20%。这是自1月以来该基金会第九位高级别离职人员。前联合执行董事Tomasz Stańczak于2月辞职，而另一位联合执行董事Hsiao-Wei Wang在本月请假后辞职。表面上看，这似乎是一轮危机驱动的裁员。但结合基金会去年发布的国库管理政策来看，这更像是一次有计划的资产负债表重组。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2f87d890-c87f-4364-af30-aa7569b75a90"><div class="bn-block" data-node-type="blockContainer" data-id="2f87d890-c87f-4364-af30-aa7569b75a90"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">如果将EF视为一个管理加密原生资产的捐赠基金，那么情况就清晰多了。过去，该基金会每年消耗其剩余资金的约15%。在大学捐赠基金或主权基金的框架下，这种提款率被认为是激进的。成熟的永久基金通常将其年化支出率保持在4%-5%的范围内，以确保资本能够跨越多个周期运行。Buterin的目标是在2030年前将该比率降至5%左右。这不仅仅是削减成本。这是试图将EF从一个“消费型”项目资助机构转变为一个可以无限期存在的机构——本质上是重新定义其期限风险。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="607c6e3f-8539-46df-82f1-787bc6fd943c"><div class="bn-block" data-node-type="blockContainer" data-id="607c6e3f-8539-46df-82f1-787bc6fd943c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">支出收缩正在多个方面进行。首先是安全策略的范式转变。Ethereum此前依赖多客户端冗余来降低风险：只要没有单个客户端的故障率超过33%，网络就可以继续生成区块并达到最终性。该基金会现在正转向更多地依赖AI辅助的形式化验证，将客户端从“互为备份”转变为“专业分工”。本质上，它正在使用自动化验证来取代人工冗余并降低维护成本。其次，专注于隐私和扩展性探索（PSE）下的零知识证明研究团队已被解散为一个独立的单位。这意味着预算正从纯粹的“探索”转向直接嵌入协议层和访问层的“实现”——更接近于将基础研究预算转移到产品工程的企业逻辑。第三，Devcon的规模持续缩小。Buterin已开始亲自资助以前由基金会领导的“超越Ethereum”的项目，同时机构合作伙伴关系也缩小到少数可复制的CROPS友好型试点——抗审查性、弹性、开放性、隐私和安全性。这是销售、一般和行政（SG&amp;A）去杠杆化的典型案例。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a4331d22-aa2e-4609-be09-f97a1937ff68"><div class="bn-block" data-node-type="blockContainer" data-id="a4331d22-aa2e-4609-be09-f97a1937ff68"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">重组发生在不利的市场背景下。ETH徘徊在1,600美元左右，较2025年8月近4,956美元的峰值下跌了60%以上。其表现明显落后于大多数主要加密资产，并且超过了Bitcoin同期约48%的跌幅。更重要的是，于4月到期的客户激励计划（Client Incentive Program）此前曾为Geth、Erigon和Lighthouse等独立客户团队提供年度资助。前EF贡献者Trenton Van Epps此前估计，维持现有的核心开发生态系统每年需要约3000万美元，而EF的预算缩减使得基金会本身难以完全弥补这一缺口。市场反应几乎是同时发生的：6月22日，Ethlabs正式启动，由几位前EF高级研究员领导，并由BitMine、SharpLink和Ethereum联合创始人Joseph Lubin资助。该项目专注于机构级结算、原生资产发行和跨链基础设施。这为EF的收缩创造了结构性互补：随着基金会从部分“增长投资”角色中退出，私人资本和生态系统组织开始吸收被释放的研发和资助职能。这种动态类似于一家成熟公司将高风险增长业务剥离给独立融资平台，同时将其自身重点转回保守的资产负债表管理。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="ce4cc347-6ed6-4962-88e1-34daeb5cadcf"><div class="bn-block" data-node-type="blockContainer" data-id="ce4cc347-6ed6-4962-88e1-34daeb5cadcf"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">展望未来，Buterin的最终目标可以被理解为一种“软退出”。Strawmap是Ethereum协议路线图的一部分，涵盖了共识、证明系统、隐私和账户模型等多个核心领域。它被认为是PoW时代和The Merge之后Ethereum下一协议阶段的重要升级包。一旦该路线图逐步完成，基金会可能会显著提高推出新功能的门槛。其治理风格也可能从高强度、广泛的研发扩张转向一种更侧重于安全、融合和长期维护的模式。换句话说，Ethereum的发展曲线正从一个持续高研发支出的成长型科技公司，逐渐转向更接近主权货币网络的治理路径。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="be043b6f-12ad-464b-8397-7e6883548619"><div class="bn-block" data-node-type="blockContainer" data-id="be043b6f-12ad-464b-8397-7e6883548619"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Solana联合创始人Anatoly Yakovenko将此举解读为积极的，认为预算限制将迫使重点突出，而一个更小、更精简的基金会可以更快地做出反应和纠正错误。但风险同样真实：人才流失尚未停止，客户激励资金缺口仍无明确解决方案，而像Ethlabs这样的新组织能否承担EF以前承担的研发责任仍是一个未经检验的假设。最终，这种“资产负债表缩减”并非在测试基金会还剩下多少钱。它是在测试整个生态系统是否能完成从依赖单一基金会的资本注入到多元化市场资本供应的转变——而这才是决定Ethereum下一个叙事和估值的真正变量。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="69dca9f2-fd27-4690-bf5d-6300d6ede13d"><div class="bn-block" data-node-type="blockContainer" data-id="69dca9f2-fd27-4690-bf5d-6300d6ede13d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

以太坊基金会“资产负债表缩减”：从15%重组至5%

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="5990badb-11df-4a1f-9faa-ba15cfc2a5cf"><div class="bn-block" data-node-type="blockContainer" data-id="5990badb-11df-4a1f-9faa-ba15cfc2a5cf"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">On June 23, South Korea’s capital market delivered a rare circuit-breaker event, reminding global investors to reassess the more fragile side of the AI trade. The KOSPI triggered a circuit breaker during the session, halting trading for 20 minutes. Selling pressure showed little sign of easing after trading resumed, and the index ultimately closed down 9.99% at 8,203.84, marking its steepest one-day decline since March. Samsung Electronics and SK Hynix, the two heavyweight constituents, both fell more than 12%, becoming the main drag on the benchmark. Just one trading day earlier, the KOSPI had broken above the 9,100 level, with market sentiment still elevated and the optimism driven by semiconductor leaders yet to fade.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f140b5c0-8f3c-4d45-9f83-bdf64ea112e6"><div class="bn-block" data-node-type="blockContainer" data-id="f140b5c0-8f3c-4d45-9f83-bdf64ea112e6"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">This was not an isolated event. Since June, South Korea’s market has repeatedly triggered circuit breakers or sell-side Sidecar mechanisms. The underlying cause has been broadly similar: the semiconductor sector carries excessive index weight, meaning any marginal change in expectations around major customers’ capital expenditure, HBM procurement schedules or AI-chip production plans can be rapidly amplified through heavyweight stocks into the broader index. Previous corrections were linked to Broadcom’s weaker-than-expected guidance, shifts in U.S. rate expectations and concerns over AI-chip demand. The June 23 selloff, meanwhile, reflected a market reassessment of the Vera Rubin platform’s production timetable, the HBM4 supply path and the certainty of SK Hynix’s capacity expansion. The structural features of South Korea’s semiconductor supply chain were laid bare: the Korean equity market is highly dependent on Samsung Electronics and SK Hynix, the country’s two domestic memory-chip giants, while the valuation and earnings expectations of both companies are largely tied to the AI capex cycle of Nvidia and North American cloud providers. Bargaining power across the supply chain, order visibility from key customers and the pace of technology migration together mean that Korean equities often swing more sharply than short-term changes in fundamentals alone would imply.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="212b8543-8d7c-4f7a-9a37-ecfc7d25ad1e"><div class="bn-block" data-node-type="blockContainer" data-id="212b8543-8d7c-4f7a-9a37-ecfc7d25ad1e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The force behind this amplification is South Korea’s distinctive retail-investor structure. The Korean market has long been characterized by mass retail participation, with tens of millions of individual investors acting as a major source of market liquidity. Data from Korea’s securities registration institutions show annual securities transaction tax revenue of about 6.6 trillion won, roughly 75% of which comes from retail investors. This structure magnifies upside in bull markets and downside in corrections. From mid-May to early June, forced liquidations in the Korean market rose significantly, with some statistical measures showing margin-call liquidations reaching hundreds of billions of won within just several days. More importantly, the clearing of leveraged positions has not meaningfully cooled retail sentiment. The market has continued to show a pattern of “buying more as prices fall” and using borrowed money to enter the market: some investors were forced out after margin calls, while new leveraged funds kept flowing in to chase rebounds. When pension funds and other institutional investors turned net sellers for a period, the marginal bid previously supported by retail and leveraged capital quickly weakened. Goldman Sachs and other institutions later noted that the recent rally in Korean equities had been driven more by technical and liquidity factors than by fresh fundamental buying.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="95893f29-feb0-4628-9d49-ec1748214fdd"><div class="bn-block" data-node-type="blockContainer" data-id="95893f29-feb0-4628-9d49-ec1748214fdd"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Political uncertainty has further added to market fragility. After taking office, the Lee Jae-myung administration proposed a tax-reform package that included lowering the threshold for individual shareholders subject to capital-gains tax from 5 billion won to 1 billion won, raising the securities transaction tax from 0.15% to 0.2%, and increasing the top corporate tax rate from 24% to 25%. The proposal triggered petitions from retail investors and political controversy, and Korean equities at one point sold off sharply. The government subsequently signaled some willingness to soften certain details and left the issue for further discussion in the National Assembly. Yet the unresolved policy status itself remains a source of risk. For the market, expectations that major shareholders may reduce holdings before year-end to avoid tax liabilities, combined with the heavy dependence of securities transaction tax revenue on active retail trading, mean that any progress on tax reform could become a catalyst for renewed selling.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="38256fc4-4a90-4929-aa6b-f6e1ee58a969"><div class="bn-block" data-node-type="blockContainer" data-id="38256fc4-4a90-4929-aa6b-f6e1ee58a969"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">From a global perspective, South Korea’s violent market swings are a microcosm of the AI capex cycle entering a phase of expectation adjustment. U.S. semiconductor and AI-related assets have also come under pressure, while technology shares in Japan, Taiwan and other Asian markets have seen varying degrees of declines, highlighting the strong cross-market linkage among Asia-Pacific tech stocks. Even though Nvidia’s management has remained optimistic about AI demand, market sentiment has not fully recovered. Some investors have begun to reassess whether current valuations have excessively priced in the long-term benefits of AI infrastructure buildout. Over the past year, the core logic of the AI trade rested on sustained growth in compute demand, upward revisions to cloud-provider capex and a high-margin cycle for memory chips. The market is now demanding greater order visibility, clearer cash-flow realization and a more reasonable valuation cushion.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="1467f70c-8a34-4879-8554-abf5e888ba95"><div class="bn-block" data-node-type="blockContainer" data-id="1467f70c-8a34-4879-8554-abf5e888ba95"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Overall, the current turbulence in South Korea’s market reflects the fragility of its liquidity structure and policy expectations more than a fundamental deterioration in the semiconductor industry. The industrial logic of tight HBM supply and demand still has near-term support, and the recovery in the memory cycle has not been clearly invalidated. However, highly leveraged retail investors, a dual-heavyweight index structure and unresolved tax reform together have created a market ecosystem that is highly sensitive to external shocks. Before the AI memory supercycle is decisively disproven, Korean equities are likely to retain their high-beta characteristics. For global investors, South Korea is increasingly becoming a key barometer for tracking crowding in the AI trade, valuation elasticity in technology stocks and the risks embedded in retail leverage.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="e4b89bd9-f109-47b0-80b6-535a9901c5c3"><div class="bn-block" data-node-type="blockContainer" data-id="e4b89bd9-f109-47b0-80b6-535a9901c5c3"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

Behind South Korea’s Stock-Market Circuit Breaker: A Confluence of AI Trading, Semiconductor Concentration and Retail Leverage

韩国股市熔断机制背后：AI交易、半导体集中化和散户杠杆的汇合

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="a8e4752f-6093-4b00-a990-143452fb7584"><div class="bn-block" data-node-type="blockContainer" data-id="a8e4752f-6093-4b00-a990-143452fb7584"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The recent release of GLM-5.2 by Zhipu has brought the market’s attention back to a core question: does improving AI model capability necessarily require hardware demand, training costs, and capital expenditure to rise in a linear fashion?</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c043d1dc-7281-40e4-98d4-4243be8d66e0"><div class="bn-block" data-node-type="blockContainer" data-id="c043d1dc-7281-40e4-98d4-4243be8d66e0"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Based on publicly available information, the significance of GLM-5.2 is not merely that it ranks highly on benchmarks. More importantly, it demonstrates performance close to frontier closed-source models in long-horizon tasks, coding, and agentic workflows. Zhipu official materials show that GLM-5.2 supports a 1M-token context window and is designed for long-horizon tasks. Its architectural improvements can reduce per-token FLOPs by 2.9x at a 1M context length, while speculative decoding further improves inference efficiency. Zhipu also says GLM-5.2 scored 81.0 on Terminal-Bench 2.1, close to Claude Opus 4.8’s 85.0 and ahead of Gemini 3.1 Pro.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b7a85c20-71b4-4139-9ebd-74f9329400db"><div class="bn-block" data-node-type="blockContainer" data-id="b7a85c20-71b4-4139-9ebd-74f9329400db"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">These developments are easily simplified by the market into narratives such as “open-source models are crushing closed-source models” or “AI no longer needs chips.” A more objective conclusion is this: <strong>hardware demand will not disappear, but its structure will change; closed-source model companies will face greater pressure on capital expenditure; and the semiconductor industry’s valuation framework will shift from simply betting on training-scale expansion to assessing competition across training, inference, interconnects, memory, and energy efficiency.</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d7e83e2d-99e0-41e2-8575-0d5390d85f52"><div class="bn-block" data-node-type="blockContainer" data-id="d7e83e2d-99e0-41e2-8575-0d5390d85f52"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">First, the impact on hardware demand is not that companies will “buy fewer GPUs,” but that lower unit intelligence costs will reactivate demand elasticity. When DeepSeek R1 emerged in early 2025, Nvidia’s shares fell by roughly 17% in a single day, wiping out nearly $600 billion in market value. At the time, investors worried that if a Chinese team could build near-frontier models at much lower cost, the need for U.S. tech giants to keep piling into compute would be questioned. But the industry’s development since then suggests that low-cost models have not ended compute demand. Instead, they have enabled more developers, enterprises, and agent applications to enter real usage scenarios. The cheaper models become, the more frequently they are called, and the easier it is for inference-side demand to expand.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c6bc7365-0442-425b-b0a1-abee9c0bb403"><div class="bn-block" data-node-type="blockContainer" data-id="c6bc7365-0442-425b-b0a1-abee9c0bb403"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Microsoft’s reported consideration of using Azure-hosted DeepSeek in Copilot Cowork reflects exactly this logic. According to Axios, Microsoft is shifting Copilot Cowork toward usage-based pricing and exploring fine-tuned DeepSeek V4 or other open-source models as lower-cost alternatives to OpenAI and Anthropic models. The reason is that agentic tools repeatedly call models while completing tasks, causing AI bills to rise quickly. This shows that large enterprise customers care not only about the “most powerful model,” but also about the “unit task cost at an acceptable level of capability.” In the future, enterprise AI platforms will likely adopt multi-model routing: the strongest closed-source models for complex tasks, and cheaper open-source or Chinese models for routine coding, documentation, customer service, and data-processing tasks.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="05c5b8dd-18e2-4490-b4aa-5ad4af0b5ff5"><div class="bn-block" data-node-type="blockContainer" data-id="05c5b8dd-18e2-4490-b4aa-5ad4af0b5ff5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Second, the capital investment logic of closed-source model companies will come under stronger constraints. The previous commercial model for closed-source AI rested on three assumptions: capability leadership, user dependence on APIs, and pricing high enough to cover massive training and inference infrastructure spending. But open-weight models such as GLM-5.2 are compressing that premium. Third-party platforms show that GLM-5.2 is priced on OpenRouter at about $1 per million input tokens and $4 per million output tokens, while supporting a 1M context window. OfficeChai, citing Artificial Analysis data, said GLM-5.2 ranked fourth overall on the Intelligence Index v4.1, first among open-weight models, and sat on the cost-intelligence Pareto frontier. Even if these rankings do not fully represent real enterprise scenarios, they are already enough to change customers’ bargaining power.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f88a60e7-265c-4045-875c-ec7fdefdf444"><div class="bn-block" data-node-type="blockContainer" data-id="f88a60e7-265c-4045-875c-ec7fdefdf444"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For OpenAI, Anthropic, Google DeepMind, Meta, and other leading AI labs, the implication is that closed-source models will still require continued investment in larger-scale training, because frontier capability, reliability, safety alignment, and ecosystem distribution remain valuable. However, it will be harder for them to earn excess gross margins through a single high-priced API model. Capital expenditure will increasingly need to emphasize “effective capex”: higher data-center utilization, lower inference costs, stronger caching, model distillation, mixture-of-experts architectures, dedicated inference chips, and model combinations tailored to specific industry scenarios, rather than simply expanding GPU clusters.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4c9be4ab-823d-44a3-bad0-5db67992cc86"><div class="bn-block" data-node-type="blockContainer" data-id="4c9be4ab-823d-44a3-bad0-5db67992cc86"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Third, for the semiconductor industry, the impact is one of differentiation rather than broad-based bearishness. On the training side, if algorithmic efficiency continues to improve, the incremental GPU demand required to train each frontier model may be lower than the market’s most aggressive expectations, weakening the “infinite GPU buildout” narrative. But on the inference side, agents, long context windows, multimodality, and enterprise automation will significantly increase token consumption. This is especially true for long tasks such as code generation, software engineering, research assistance, and data analysis, which consume large amounts of output tokens and reasoning tokens. OfficeChai’s data also noted that GLM-5.2’s output-token consumption per task is not low in high-capability settings, suggesting that “cheaper models” do not necessarily mean “lower total bills.”</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="11f66652-1dd5-496b-a342-89b3d02d41c5"><div class="bn-block" data-node-type="blockContainer" data-id="11f66652-1dd5-496b-a342-89b3d02d41c5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">As a result, opportunities across the semiconductor supply chain will spread beyond GPUs alone into more specialized areas: high-bandwidth memory, advanced packaging, networking interconnects, inference accelerators, low-power edge AI, domestic compute adaptation, and cloud scheduling software. Nvidia will still benefit from the largest-scale training and high-end inference clusters, but investors will pay closer attention to its pricing power, the sustainability of customer capex, and whether it can maintain margins in an environment of “cheaper models but higher usage.” AMD, ASIC vendors, cloud providers’ self-developed chips, and Chinese AI chipmakers may gain more opportunities in replacement and tiered deployment.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="11c57f81-35b0-420d-97e9-a127136a1894"><div class="bn-block" data-node-type="blockContainer" data-id="11c57f81-35b0-420d-97e9-a127136a1894"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">In summary, GLM-5.2 and Microsoft’s consideration of DeepSeek are not proof that AI hardware demand has peaked. Rather, they show that the AI industry is entering a new stage of <strong>unit intelligence cost competition</strong>. In the short term, this will pressure pricing and gross-margin expectations for closed-source models and trigger valuation volatility in semiconductors. In the long term, cheaper models will expand AI application penetration and shift compute demand from a small number of giant training clusters toward broader enterprise inference, agent workflows, and industry automation. What truly needs to be revalued is not whether AI needs compute, but who can convert compute into monetizable intelligent productivity at the lowest cost.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="16fb6ef6-e93c-43e5-bc8c-8a1b5e3a50ec"><div class="bn-block" data-node-type="blockContainer" data-id="16fb6ef6-e93c-43e5-bc8c-8a1b5e3a50ec"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

GLM-5.2 and “Cheap Compute”: The AI Supply Chain Is Being Repriced

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="a8e4752f-6093-4b00-a990-143452fb7584"><div class="bn-block" data-node-type="blockContainer" data-id="a8e4752f-6093-4b00-a990-143452fb7584"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">智谱近期发布的GLM-5.2将市场的目光拉回了一个核心问题：提升AI模型能力是否必然要求硬件需求、训练成本和资本支出呈线性增长？</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c043d1dc-7281-40e4-98d4-4243be8d66e0"><div class="bn-block" data-node-type="blockContainer" data-id="c043d1dc-7281-40e4-98d4-4243be8d66e0"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">根据公开信息，GLM-5.2的意义不仅在于其在基准测试中排名靠前。更重要的是，它在长上下文任务、编码和代理工作流方面展现出了接近前沿闭源模型的性能。智谱官方资料显示，GLM-5.2支持1M token的上下文窗口，专为长上下文任务设计。其架构改进可在1M上下文长度下将每token的FLOPs降低2.9倍，而投机解码则进一步提高了推理效率。智谱还表示，GLM-5.2在Terminal-Bench 2.1上得分为81.0，接近Claude Opus的85.0，并优于Gemini 3.1 Pro。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b7a85c20-71b4-4139-9ebd-74f9329400db"><div class="bn-block" data-node-type="blockContainer" data-id="b7a85c20-71b4-4139-9ebd-74f9329400db"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">市场很容易将这些进展简化为“开源模型正在碾压闭源模型”或“AI不再需要芯片”之类的叙事。一个更客观的结论是：<strong>硬件需求不会消失，但其结构将发生变化；闭源模型公司将面临更大的资本支出压力；半导体行业的估值框架将从单纯押注训练规模扩张转向评估训练、推理、互连、内存和能效方面的竞争。</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d7e83e2d-99e0-41e2-8575-0d5390d85f52"><div class="bn-block" data-node-type="blockContainer" data-id="d7e83e2d-99e0-41e2-8575-0d5390d85f52"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">首先，对硬件需求的影响并非公司将“减少GPU购买量”，而是较低的单位智能成本将重新激活需求弹性。当DeepSeek R1于2025年初出现时，Nvidia的股价单日下跌约17%，市值蒸发近6000亿美元。当时，投资者担心如果中国团队能以低得多的成本构建接近前沿的模型，美国科技巨头持续投入计算的需求将受到质疑。但此后的行业发展表明，低成本模型并未终结计算需求。相反，它们使更多的开发者、企业和代理应用得以进入实际使用场景。模型越便宜，调用频率越高，推理端需求越容易扩展。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c6bc7365-0442-425b-b0a1-abee9c0bb403"><div class="bn-block" data-node-type="blockContainer" data-id="c6bc7365-0442-425b-b0a1-abee9c0bb403"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">微软考虑在Copilot Cowork中使用Azure托管的DeepSeek，正是反映了这一逻辑。据Axios报道，微软正将Copilot Cowork转向按使用量付费的模式，并探索使用微调的DeepSeek V4或其他开源模型作为OpenAI和Anthropic模型的低成本替代方案。原因是代理工具在完成任务时会反复调用模型，导致AI账单迅速上涨。这表明大型企业客户不仅关心“最强大的模型”，也关心“在可接受能力水平下的单位任务成本”。未来，企业AI平台可能会采用多模型路由：最强大的闭源模型用于复杂任务，而更便宜的开源或中国模型用于常规编码、文档、客户服务和数据处理任务。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="05c5b8dd-18e2-4490-b4aa-5ad4af0b5ff5"><div class="bn-block" data-node-type="blockContainer" data-id="05c5b8dd-18e2-4490-b4aa-5ad4af0b5ff5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">第二，闭源模型公司的资本投资逻辑将受到更强的制约。此前闭源AI的商业模式基于三个假设：能力领先、用户依赖API，以及足够高的定价以覆盖巨大的训练和推理基础设施支出。但GLM-5.2等开放权重模型正在压缩这种溢价。第三方平台显示，GLM-5.2在OpenRouter上的定价约为每百万输入token 1美元，每百万输出token 4美元，同时支持1M上下文窗口。OfficeChai援引Artificial Analysis的数据称，GLM-5.2在Intelligence Index v4.1中总体排名第四，在开放权重模型中排名第一，并处于成本-智能帕累托前沿。即使这些排名不能完全代表真实的企业场景，它们也足以改变客户的议价能力。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f88a60e7-265c-4045-875c-ec7fdefdf444"><div class="bn-block" data-node-type="blockContainer" data-id="f88a60e7-265c-4045-875c-ec7fdefdf444"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">对于OpenAI、Anthropic、Google DeepMind、Meta等领先的AI实验室而言，这意味着闭源模型仍需要持续投入更大规模的训练，因为前沿能力、可靠性、安全对齐和生态系统分发仍然有价值。然而，它们将更难通过单一的高价API模式赚取超额毛利。资本支出将越来越需要强调“有效资本支出”：提高数据中心利用率、降低推理成本、加强缓存、模型蒸馏、混合专家架构、专用推理芯片以及针对特定行业场景定制的模型组合，而不是简单地扩大GPU集群。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4c9be4ab-823d-44a3-bad0-5db67992cc86"><div class="bn-block" data-node-type="blockContainer" data-id="4c9be4ab-823d-44a3-bad0-5db67992cc86"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">第三，对半导体行业而言，影响是分化而非普遍的看跌。在训练方面，如果算法效率持续提高，训练每个前沿模型所需的增量GPU需求可能低于市场最激进的预期，削弱了“无限GPU建设”的叙事。但在推理方面，代理、长上下文窗口、多模态和企业自动化将显著增加token消耗。特别是对于代码生成、软件工程、研究辅助和数据分析等长任务，它们会消耗大量的输出token和推理token。OfficeChai的数据也指出，在高性能场景下，GLM-5.2每任务的输出token消耗并不低，这表明“更便宜的模型”不一定意味着“总账单更低”。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="11f66652-1dd5-496b-a342-89b3d02d41c5"><div class="bn-block" data-node-type="blockContainer" data-id="11f66652-1dd5-496b-a342-89b3d02d41c5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">因此，半导体供应链的机会将从单纯的GPU扩展到更专业的领域：高带宽内存、先进封装、网络互连、推理加速器、低功耗边缘AI、国内计算适配以及云调度软件。Nvidia仍将受益于最大规模的训练和高端推理集群，但投资者将更密切地关注其定价能力、客户资本支出的可持续性，以及在“模型更便宜但使用量更高”的环境中维持利润率的能力。AMD、ASIC供应商、云服务商自研芯片以及中国AI芯片制造商在替代和分级部署方面可能会获得更多机会。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="11c57f81-35b0-420d-97e9-a127136a1894"><div class="bn-block" data-node-type="blockContainer" data-id="11c57f81-35b0-420d-97e9-a127136a1894"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">总而言之，GLM-5.2和微软对DeepSeek的考虑并非AI硬件需求已达顶峰的证据。相反，它们表明AI行业正进入一个全新的<strong>单位智能成本竞争</strong>阶段。短期内，这将给闭源模型的定价和毛利率预期带来压力，并引发半导体行业的估值波动。长期来看，更便宜的模型将扩大AI应用的渗透率，并将计算需求从少数巨型训练集群转移到更广泛的企业推理、代理工作流和行业自动化。真正需要重新评估的不是AI是否需要计算，而是谁能以最低的成本将计算转化为可盈利的智能生产力。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="16fb6ef6-e93c-43e5-bc8c-8a1b5e3a50ec"><div class="bn-block" data-node-type="blockContainer" data-id="16fb6ef6-e93c-43e5-bc8c-8a1b5e3a50ec"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

GLM-5.2 与“廉价算力”：AI 供应链正在重新定价

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="b935c88e-a7a7-48e1-a04f-9e99228e2a7b"><div class="bn-block" data-node-type="blockContainer" data-id="b935c88e-a7a7-48e1-a04f-9e99228e2a7b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">On June 12, 2026, SpaceX went public on Nasdaq. SoDEX listed $SPCX the same day — you can trade it directly here: <a target="_blank" rel="noopener noreferrer nofollow" href="https://sodex.com/join/JELLYZ">https://sodex.com/join/JELLYZ</a></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9c24eb84-d77f-4727-b059-c55f32450ecd"><div class="bn-block" data-node-type="blockContainer" data-id="9c24eb84-d77f-4727-b059-c55f32450ecd"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Priced at $135 per share, the stock jumped 29% after opening, briefly pushing SpaceX’s market value above $2 trillion and making it the world’s sixth-largest listed company.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0799e981-df10-4a4a-bed1-07f0234ff457"><div class="bn-block" data-node-type="blockContainer" data-id="0799e981-df10-4a4a-bed1-07f0234ff457"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a7f775fd-27ec-41f2-bba2-4188e3ede68e"><div class="bn-block" data-node-type="blockContainer" data-id="a7f775fd-27ec-41f2-bba2-4188e3ede68e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>Why is SpaceX valued so highly? What Is the Market Pricing In?</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="6fa44436-3ed1-4595-8de8-1d0372f153b4"><div class="bn-block" data-node-type="blockContainer" data-id="6fa44436-3ed1-4595-8de8-1d0372f153b4"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The key is that SpaceX is no longer just a rocket company.&nbsp;</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="18cc8dff-d09c-4407-a20a-83ad2a06651e"><div class="bn-block" data-node-type="blockContainer" data-id="18cc8dff-d09c-4407-a20a-83ad2a06651e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Falcon launches and government contracts make it a high-barrier aerospace business; Starlink turns it into a global communications platform; and xAI, Grok, Colossus and orbital data centers add an AI infrastructure story.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="e3911289-34b9-4b9a-90e3-c3b52863c004"><div class="bn-block" data-node-type="blockContainer" data-id="e3911289-34b9-4b9a-90e3-c3b52863c004"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Its business can therefore be divided into three segments: Space, Connectivity and AI.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="015f33a6-0d55-4cde-bab5-194da3e688dc"><div class="bn-block" data-node-type="blockContainer" data-id="015f33a6-0d55-4cde-bab5-194da3e688dc"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9fbfd07b-f603-4588-a0ba-0ff12beae8b1"><div class="bn-block" data-node-type="blockContainer" data-id="9fbfd07b-f603-4588-a0ba-0ff12beae8b1"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>Space — the foundation.&nbsp;</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="01d2ecaa-7ab5-4eb6-84dc-92a160b6bd64"><div class="bn-block" data-node-type="blockContainer" data-id="01d2ecaa-7ab5-4eb6-84dc-92a160b6bd64"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Reusable, low-cost access to orbit is SpaceX’s core strategic moat. It determines the deployment cost of Starlink and provides the foundation for Starship, orbital data centers and future Mars missions.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2e7c7b5a-7fba-4460-a854-90c02da320d2"><div class="bn-block" data-node-type="blockContainer" data-id="2e7c7b5a-7fba-4460-a854-90c02da320d2"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">In 2025, the Space segment generated $4.1 billion in revenue(~22% of Spacex), with $653 million in adjusted EBITDA.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4110d3f5-3888-40cc-9912-015bf5a48a33"><div class="bn-block" data-node-type="blockContainer" data-id="4110d3f5-3888-40cc-9912-015bf5a48a33"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="902af5ea-fa8e-43a8-aee9-94be971a7b71"><div class="bn-block" data-node-type="blockContainer" data-id="902af5ea-fa8e-43a8-aee9-94be971a7b71"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>Connectivity — the cash cow.&nbsp;</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="17254e99-a328-4ccc-8578-5a9d0168f771"><div class="bn-block" data-node-type="blockContainer" data-id="17254e99-a328-4ccc-8578-5a9d0168f771"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Mostly Starlink. 2025 revenue $11.387B (~61% of group), up 50% YoY, with $4.4B operating profit. Users passed 10.3 million in Q1, double a year earlier, with no rival at comparable scale; institutions put its addressable market as high as $1.6 trillion.&nbsp;</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a26ead41-ef98-445e-ae05-3b04664d42ff"><div class="bn-block" data-node-type="blockContainer" data-id="a26ead41-ef98-445e-ae05-3b04664d42ff"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Aviation, maritime, government and mobile are markets terrestrial networks can't reach, and those customers pay far more than households — supporting long-term ARPU and margins.&nbsp;</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="e7af1aca-8ccc-4da9-8d38-34c4b05bb515"><div class="bn-block" data-node-type="blockContainer" data-id="e7af1aca-8ccc-4da9-8d38-34c4b05bb515"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">This high-growth, high-margin business forms the foundation of SpaceX’s current valuation.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="dad0b050-d26d-432b-82cf-d5f36b058676"><div class="bn-block" data-node-type="blockContainer" data-id="dad0b050-d26d-432b-82cf-d5f36b058676"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="1a881e14-3acd-4481-bd85-14b95bb615c9"><div class="bn-block" data-node-type="blockContainer" data-id="1a881e14-3acd-4481-bd85-14b95bb615c9"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>AI — the biggest story.&nbsp;</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="fbb00763-54e0-456d-8365-7e7766841a94"><div class="bn-block" data-node-type="blockContainer" data-id="fbb00763-54e0-456d-8365-7e7766841a94"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The AI segment includes X, Grok, Colossus and potential orbital compute infrastructure.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d6f58eb1-f6c6-403f-81ed-fdb8d2cc30f9"><div class="bn-block" data-node-type="blockContainer" data-id="d6f58eb1-f6c6-403f-81ed-fdb8d2cc30f9"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">It remains highly capital-intensive, losing $6.3 billion in 2025, while first-quarter 2026 capital expenditure reached $7.7 billion. However, based on announced contracts, I estimate external compute revenue could reach around $11.5 billion in 2026 and $26 billion in 2027.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="aba7e7cf-2627-4783-85bc-49bf26cff325"><div class="bn-block" data-node-type="blockContainer" data-id="aba7e7cf-2627-4783-85bc-49bf26cff325"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="52813b6a-5d2b-45dd-aad0-15a6c8b7ca52"><div class="bn-block" data-node-type="blockContainer" data-id="52813b6a-5d2b-45dd-aad0-15a6c8b7ca52"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">So the valuation stacks in three layers:</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4b9b41ef-d413-42fa-b3da-1636fae78d8e"><div class="bn-block" data-node-type="blockContainer" data-id="4b9b41ef-d413-42fa-b3da-1636fae78d8e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">- Cash flow today — Starlink, launches, government contracts — sets the floor.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="012088dd-c677-4aba-b141-1107f80b562c"><div class="bn-block" data-node-type="blockContainer" data-id="012088dd-c677-4aba-b141-1107f80b562c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">- Commercializing growth — Direct to Cell, enterprise connectivity, military satellites, Starship — small now, but could expand the market sharply.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0dd79c5c-4b5f-40cd-8525-59fe44bb5308"><div class="bn-block" data-node-type="blockContainer" data-id="0dd79c5c-4b5f-40cd-8525-59fe44bb5308"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">- Long-term options — xAI, orbital data centers, the Moon, Mars — unquantifiable today, yet already partly priced in.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="dee42730-5cb0-4230-a353-0ba4b9e9dd18"><div class="bn-block" data-node-type="blockContainer" data-id="dee42730-5cb0-4230-a353-0ba4b9e9dd18"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="04b12754-082f-48da-b193-d2758a3da6f1"><div class="bn-block" data-node-type="blockContainer" data-id="04b12754-082f-48da-b193-d2758a3da6f1"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>In one line: Connectivity makes money, AI burns money, and Space builds the foundation.</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a3e547d6-afc8-4ebd-bb92-a751e3f44f11"><div class="bn-block" data-node-type="blockContainer" data-id="a3e547d6-afc8-4ebd-bb92-a751e3f44f11"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The math: nearly $18.7B in 2025 revenue against a ~$4.9B net loss. At ~$2.1 trillion, that's over 110x 2025 sales — or about 57x on projected 2026 revenue of $36.6B.&nbsp;</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="5900538f-de39-4597-8f48-9f83019ed9e2"><div class="bn-block" data-node-type="blockContainer" data-id="5900538f-de39-4597-8f48-9f83019ed9e2"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="712aab1b-6bcb-4111-83fc-4a576e8c785b"><div class="bn-block" data-node-type="blockContainer" data-id="712aab1b-6bcb-4111-83fc-4a576e8c785b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Goldman Sachs and Morgan Stanley go further: nearly $160B revenue by 2028, over $470B by 2030, and possibly $3.4 trillion by 2040 — assuming Musk's team keeps making the right calls for 14 years through every uncertainty in AI, space, regulation and competition.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d726afe1-03cc-4d8b-a25b-16424e6ad356"><div class="bn-block" data-node-type="blockContainer" data-id="d726afe1-03cc-4d8b-a25b-16424e6ad356"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f204ca11-6f80-400d-8b94-2c4226ba130c"><div class="bn-block" data-node-type="blockContainer" data-id="f204ca11-6f80-400d-8b94-2c4226ba130c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Ultimately, SpaceX's valuation is a probability-weighted bet on future scenarios. Starlink is the base, Starship the growth optionality, AI and space-based data centers the upside. Execution risk decides how much of it becomes real.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f6cf0626-4ec8-48bb-aa83-9d5141a10acb"><div class="bn-block" data-node-type="blockContainer" data-id="f6cf0626-4ec8-48bb-aa83-9d5141a10acb"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="612d2853-7042-47cb-a020-a04ee299f27d"><div class="bn-block" data-node-type="blockContainer" data-id="612d2853-7042-47cb-a020-a04ee299f27d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>So the real question: are you willing to bet on Elon Musk?</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2b6b012b-5f69-46a7-8e44-f595018952a4"><div class="bn-block" data-node-type="blockContainer" data-id="2b6b012b-5f69-46a7-8e44-f595018952a4"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

SpaceX at $2 Trillion: What Is the Market Pricing In?

SpaceX 达 2 万亿美元：市场在定价什么？

Daily Roast

Comparison of xAI, OpenAI, and Anthropic

xAI、OpenAI 和 Anthropic 的比较

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="c74fd4d9-128a-4474-8d61-3259c6b0a0cc"><div class="bn-block" data-node-type="blockContainer" data-id="c74fd4d9-128a-4474-8d61-3259c6b0a0cc"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Oracle reported earnings today. I opened its chart on SoDEX — the stock was down nearly 12% in pre-market trading.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="58447a45-3102-4c6a-b878-3fb2c057d1c9"><div class="bn-block" data-node-type="blockContainer" data-id="58447a45-3102-4c6a-b878-3fb2c057d1c9"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><a target="_blank" rel="noopener noreferrer nofollow" href="https://sodex.com/join/JELLYZ">https://sodex.com/join/JELLYZ</a><br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4641a6bd-664e-4ae0-86f4-6986d842d451"><div class="bn-block" data-node-type="blockContainer" data-id="4641a6bd-664e-4ae0-86f4-6986d842d451"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">This is not an isolated case. Over the past week, three key companies across the AI infrastructure chain reported earnings, and the market reaction was surprisingly consistent:<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9f89eee3-8773-4a99-a259-97e5fdbddb0a"><div class="bn-block" data-node-type="blockContainer" data-id="9f89eee3-8773-4a99-a259-97e5fdbddb0a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">- June 3: Broadcom reported FY2026 Q2. Results beat — the stock sold off sharply.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="89fd1d8c-6e64-42c9-9613-262350eec544"><div class="bn-block" data-node-type="blockContainer" data-id="89fd1d8c-6e64-42c9-9613-262350eec544"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">- June 4: Ciena reported FY2026 Q2. Results beat — the stock also fell sharply.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f19463ef-c7ff-4be9-bf1f-2019d50a86d8"><div class="bn-block" data-node-type="blockContainer" data-id="f19463ef-c7ff-4be9-bf1f-2019d50a86d8"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">- June 10, after the close: Oracle reported FY2026 Q4. Results beat — the stock dropped anyway.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="713f6d4b-b192-42f4-b19b-707f4e3d221b"><div class="bn-block" data-node-type="blockContainer" data-id="713f6d4b-b192-42f4-b19b-707f4e3d221b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Three earnings reports. None of them were bad. Three stock reactions. None of them went up.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="06581b80-9f50-4399-96e2-037d0aa08431"><div class="bn-block" data-node-type="blockContainer" data-id="06581b80-9f50-4399-96e2-037d0aa08431"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">So what went wrong?<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="151d910a-4148-4e7c-853e-d2bd3fcbcc80"><div class="bn-block" data-node-type="blockContainer" data-id="151d910a-4148-4e7c-853e-d2bd3fcbcc80"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">First, Oracle: strong demand, but heavy financing pressure.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0a2ce375-e93d-460f-a016-159bd7da382d"><div class="bn-block" data-node-type="blockContainer" data-id="0a2ce375-e93d-460f-a016-159bd7da382d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The numbers were solid: Q4 revenue of $19.18B, up 21% YoY; adjusted EPS of $2.11; OCI cloud infrastructure revenue up 93% YoY.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b6a79829-7a31-4faf-85d9-0d86780373ee"><div class="bn-block" data-node-type="blockContainer" data-id="b6a79829-7a31-4faf-85d9-0d86780373ee"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The problem is not demand. The problem is investment. In FY2026, Oracle's CapEx reached $55.66B while free cash flow was -$23.69B. FY2027 CapEx could reach as high as $95B, funded by continued debt and equity financing for AI data center construction.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="81cad8f4-9d25-4e72-ad59-f40b60670952"><div class="bn-block" data-node-type="blockContainer" data-id="81cad8f4-9d25-4e72-ad59-f40b60670952"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Oracle's central tension is clear: AI cloud orders are strong, but fulfilling them requires massive upfront spending on data centers, GPUs, networking, power, and land. The market isn't worried about whether demand exists — it's worried about whether these orders can generate a high enough return on capital, and when free cash flow will turn positive again.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4fdd773f-bb01-4a2e-b62c-6fbb8d981059"><div class="bn-block" data-node-type="blockContainer" data-id="4fdd773f-bb01-4a2e-b62c-6fbb8d981059"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Then Broadcom: the business is strong, but expectations are too high.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="19a03b4e-794e-44d3-a445-8c9c213b12cc"><div class="bn-block" data-node-type="blockContainer" data-id="19a03b4e-794e-44d3-a445-8c9c213b12cc"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Revenue was $22.19B, up 48% YoY; AI semiconductor revenue was $10.8B, up 143% YoY. Very strong — yet the stock still fell.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4a782c61-5c05-4093-a43f-905ae9a1368b"><div class="bn-block" data-node-type="blockContainer" data-id="4a782c61-5c05-4093-a43f-905ae9a1368b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Expectations for core AI suppliers have become extreme. Broadcom guided Q3 AI chip revenue to around $16B — strong, but not enough for the market's more aggressive hopes. Broadcom didn't fall because AI ASIC and networking demand is weak. It fell because the stock had already priced in too much of the future.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c9ff802e-60b7-40f2-8dcd-a4f676cba3b4"><div class="bn-block" data-node-type="blockContainer" data-id="c9ff802e-60b7-40f2-8dcd-a4f676cba3b4"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Finally, Ciena: revenue beat, but the trade was too crowded.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d3c54b7f-3f51-45f2-88bb-d550b86246b5"><div class="bn-block" data-node-type="blockContainer" data-id="d3c54b7f-3f51-45f2-88bb-d550b86246b5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Revenue was $1.57B, up 40% YoY; adjusted EPS was $1.64, up 290%; full-year revenue guidance was raised to around $6.3B. Again, not a bad report.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b404dbe9-0675-4de6-bd0f-8daf020b12b5"><div class="bn-block" data-node-type="blockContainer" data-id="b404dbe9-0675-4de6-bd0f-8daf020b12b5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Ciena's problem: the market had long been trading it as a core beneficiary of AI optical networking — AI data centers need higher-bandwidth, lower-latency optical connections, and Ciena sits directly in that part of the chain. But after a sharp year-to-date rally, the bar for another positive surprise was simply too high.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0ea9e794-40de-4341-a13d-403aa293a24f"><div class="bn-block" data-node-type="blockContainer" data-id="0ea9e794-40de-4341-a13d-403aa293a24f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Three companies, one signal.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="7b9db97e-71f5-46b4-9784-0aeb63dc30dd"><div class="bn-block" data-node-type="blockContainer" data-id="7b9db97e-71f5-46b4-9784-0aeb63dc30dd"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Oracle provides cloud and databases. Broadcom provides ASICs, AI networking chips, and infrastructure software. Ciena provides optical networking and data center interconnects. Placed inside the AI infrastructure chain, they are links in the same chain:<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="e4786c49-3afc-4c39-8f4b-d77050bb3792"><div class="bn-block" data-node-type="blockContainer" data-id="e4786c49-3afc-4c39-8f4b-d77050bb3792"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">AI data center construction → AI ASICs / networking chips → optical networking / interconnects → cloud revenue and compute monetization<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="53d617bf-dfc6-48e4-9664-f9ba23288cb4"><div class="bn-block" data-node-type="blockContainer" data-id="53d617bf-dfc6-48e4-9664-f9ba23288cb4"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Look at all three together and the signal is clear: AI infrastructure trading is moving from phase one to phase two.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a790c193-f62a-443f-8926-6fdf7a4fb38c"><div class="bn-block" data-node-type="blockContainer" data-id="a790c193-f62a-443f-8926-6fdf7a4fb38c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">In phase one, the market bought the narrative: who has AI orders, who is in the chain, who benefits from data center expansion.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="39059fea-db1e-4921-b10a-8178a00480bc"><div class="bn-block" data-node-type="blockContainer" data-id="39059fea-db1e-4921-b10a-8178a00480bc"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">In phase two, the market buys verification: can orders turn into revenue, revenue into profit, profit into free cash flow? Can CapEx generate enough ROIC? Is valuation already stretched? Can guidance keep moving higher?<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b3a24943-5f4f-4c50-92d0-d0f865808f19"><div class="bn-block" data-node-type="blockContainer" data-id="b3a24943-5f4f-4c50-92d0-d0f865808f19"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The selloffs across all three companies show that the market will keep trading AI — but it will no longer blindly reward every AI infrastructure company.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a6dcffc5-a5b5-4931-a399-a59ef8cbdc69"><div class="bn-block" data-node-type="blockContainer" data-id="a6dcffc5-a5b5-4931-a399-a59ef8cbdc69"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Going forward, the real upside may belong to two types of companies: the surest recipients of AI CapEx dollars — GPU, ASIC, HBM, networking, and power chain companies — and the operators that can prove CapEx returns by turning orders into revenue, profit, and free cash flow.<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="6d19a3c6-3836-4fb3-ac56-4fb60a8cde48"><div class="bn-block" data-node-type="blockContainer" data-id="6d19a3c6-3836-4fb3-ac56-4fb60a8cde48"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">If you also follow U.S. AI stocks, you can view and trade related U.S. equity contracts on SoDEX, including $$ORCL, $$MU, and other AI infrastructure names.</p></div><div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="de5a0dd8-ebd9-4c91-9f64-3f02911ada0d"><div class="bn-block" data-node-type="blockContainer" data-id="de5a0dd8-ebd9-4c91-9f64-3f02911ada0d"><div class="bn-block-content" data-content-type="image" data-url="https://static.sosovalue.com/2026/06/11/9434de2d-bd4a-4be4-9340-5834ec98f328.jpg" data-file-block=""><div class="bn-file-block-content-wrapper"><div class="bn-visual-media-wrapper"><img class="bn-visual-media" src="https://static.sosovalue.com/2026/06/11/9434de2d-bd4a-4be4-9340-5834ec98f328.jpg" alt="BlockNote image"></div></div></div></div></div></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4c03b9e4-6913-44cb-986e-4065d1253e8f"><div class="bn-block" data-node-type="blockContainer" data-id="4c03b9e4-6913-44cb-986e-4065d1253e8f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

Three companies, one signal.

三家公司，一个信号。

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="2d2fdbea-1810-45a9-bf59-c4e0fa0a397b"><div class="bn-block" data-node-type="blockContainer" data-id="2d2fdbea-1810-45a9-bf59-c4e0fa0a397b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Over the past few days, U.S. equities have received two signals of sharply different nature: SpaceX is about to complete the largest IPO in history, while Bank of America (BofA) has issued one of its most serious reduction warnings to clients in recent years. Viewed on the same timeline, the two events together form a judgment call about the market cycle.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9f929f90-c850-4488-a16a-fbc31b4568f1"><div class="bn-block" data-node-type="blockContainer" data-id="9f929f90-c850-4488-a16a-fbc31b4568f1"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>SpaceX Listing: A High-Stakes Bet Priced at Elevated Levels</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d114435c-62df-496a-836d-4e2fc018fe4c"><div class="bn-block" data-node-type="blockContainer" data-id="d114435c-62df-496a-836d-4e2fc018fe4c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">SpaceX plans to sell more than 555 million shares at an offering price of $135 per share, aiming to raise at least $75 billion, with the company valued at more than $1.75 trillion. If the overallotment option is exercised, the total fundraising size could reach $85.7 billion. If completed successfully, it would surpass the record set by Saudi Aramco in 2019 and become the world’s largest IPO.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="80aae89e-3046-4f4a-a3a9-397a7fd3b92f"><div class="bn-block" data-node-type="blockContainer" data-id="80aae89e-3046-4f4a-a3a9-397a7fd3b92f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">However, there is a clear gap between the grand narrative on paper and the company’s financial reality. The S-1 prospectus shows that the company generated consolidated revenue of $18.674 billion in 2025, with an operating loss of $2.589 billion and adjusted EBITDA of $6.584 billion. At the business-segment level, Starlink’s connectivity business contributed $11.4 billion, or roughly 61% of total revenue, and was the only consistently profitable business. The AI segment, xAI, was the main source of losses. Full-year 2025 GAAP net loss was $4.9 billion, reflecting the company’s highly capital-intensive investment in Starship construction, satellite network expansion, and AI computing infrastructure.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9e1ea882-7465-4fdb-9ece-bc0454f17063"><div class="bn-block" data-node-type="blockContainer" data-id="9e1ea882-7465-4fdb-9ece-bc0454f17063"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Notably, as much as 30% of the offering has been reserved for retail investors, who could receive more than $20 billion worth of shares. That proportion is three times or more the typical 5% to 10% retail allocation in conventional IPOs. This structure reflects Musk’s bet on retail-investor loyalty, but it also means that FOMO sentiment in the early stage of trading could be systematically amplified.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="430c82cd-ce3b-4777-8e88-da258be52a92"><div class="bn-block" data-node-type="blockContainer" data-id="430c82cd-ce3b-4777-8e88-da258be52a92"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Historical experience shows that QQQ’s performance around major technology IPOs has been unstable. Facebook, Snowflake, and Coinbase mostly recorded positive returns in the 20 trading days after listing, while Uber and Arm showed clear divergence and volatility during similar windows. The conclusion is that a giant IPO is both an amplifier of market sentiment and a real drain on liquidity.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="531bb5d1-2677-436e-b84c-2afe0c7a3fff"><div class="bn-block" data-node-type="blockContainer" data-id="531bb5d1-2677-436e-b84c-2afe0c7a3fff"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>BofA’s Warning: 70% of Bear-Market Signals Have Flashed</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="5431e9cb-34ab-49d1-a8cf-7a53d56c98f1"><div class="bn-block" data-node-type="blockContainer" data-id="5431e9cb-34ab-49d1-a8cf-7a53d56c98f1"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">In contrast to the IPO boom, one of Wall Street’s most significant risk warnings has emerged. Bank of America’s U.S. equity strategist Savita Subramanian wrote in a June 5 report: “Of the bear-market precursor signals we track, 70% have been triggered, consistent with the average level seen before previous market peaks — additional caution is warranted at this point.”</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="00b98827-c143-486d-9471-463f997dc279"><div class="bn-block" data-node-type="blockContainer" data-id="00b98827-c143-486d-9471-463f997dc279"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Among the 10 bear-market signals, seven were triggered in May, up from five in April and four in March. Seven is also the average number of signals triggered before bear markets since 1990. Based on this, BofA has set its year-end target for the S&amp;P 500 at 7,100, implying about 6% downside from the current level of around 7,400.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f5a42da8-414f-4f2b-baa9-ab8434a5ee2d"><div class="bn-block" data-node-type="blockContainer" data-id="f5a42da8-414f-4f2b-baa9-ab8434a5ee2d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The S&amp;P 500 appears “statistically expensive” on 17 of 20 valuation metrics, with some measures even exceeding levels seen during the dot-com bubble. Within the technology sector, dispersion has widened: the median performance gap between the best- and worst-performing quintiles of stocks has expanded to its highest level since February 2000. At the same time, capital expenditure by hyperscale technology companies is expected to approach 100% of operating cash flow this year. This means the monetization efficiency of AI investment is facing a serious test.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="fdde1261-b37b-46f0-8e09-759385b4af9e"><div class="bn-block" data-node-type="blockContainer" data-id="fdde1261-b37b-46f0-8e09-759385b4af9e"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>Two Questions, One Structural Judgment</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="ceb686a2-b77b-4537-a66e-1e7bd63e8c01"><div class="bn-block" data-node-type="blockContainer" data-id="ceb686a2-b77b-4537-a66e-1e7bd63e8c01"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Faced with the above information, the market has two very different interpretive frameworks: one is that AI does indeed contain bubble-like elements; the other is that BofA is simply calling the top early. Frankly, these two possibilities are not mutually exclusive. Historically, classic “top calls” have sometimes occurred while bull markets were still in full swing, and at other times have proved highly accurate. The difference this time is that the infrastructure investment supporting valuations is real and substantial, but the monetization path remains under verification. Market valuations are above historical medians across multiple dimensions, but not all sectors are at the same risk level.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4de62e1b-5092-4d76-b80f-b570c5ef3cdb"><div class="bn-block" data-node-type="blockContainer" data-id="4de62e1b-5092-4d76-b80f-b570c5ef3cdb"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The core question has always been whether prices have already incorporated enough future growth expectations. When SpaceX comes to market at roughly 94 times price-to-sales (P/S), and is currently ineligible for inclusion in the S&amp;P 500 due to ongoing losses, the market’s willingness to accept a “story premium” will become an important test. The outcome will directly affect the pricing and market reception of subsequent super IPOs such as OpenAI and Anthropic. Mitrade</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="fb8e14df-a11b-4003-95ba-7dd742a3002d"><div class="bn-block" data-node-type="blockContainer" data-id="fb8e14df-a11b-4003-95ba-7dd742a3002d"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>What This Means for Investors</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="aa12498f-0b1c-4e88-b2fe-b7b8317f15cf"><div class="bn-block" data-node-type="blockContainer" data-id="aa12498f-0b1c-4e88-b2fe-b7b8317f15cf"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The simultaneous emergence of SpaceX’s listing and BofA’s warning creates a fairly typical stress test for market sentiment: extremely optimistic narratives coexisting with systemic valuation pressure. For investors holding technology exposure, the question at this point is less about whether to “keep buying or exit entirely,” and more about whether the risk-reward profile of current positions still matches their original investment logic.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="64309b09-e4f1-4794-81e2-d911fe561135"><div class="bn-block" data-node-type="blockContainer" data-id="64309b09-e4f1-4794-81e2-d911fe561135"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Three questions are worth asking: Does the core thesis of the portfolio depend on the rapid monetization of AI? Have existing unrealized gains already priced in a substantial degree of optimism? If the market falls 10% to 15% over the next six months, is there enough capital and psychological capacity to wait for a recovery?</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b2be63c4-a0f9-4eb9-bcd2-83ee7f3dd6fe"><div class="bn-block" data-node-type="blockContainer" data-id="b2be63c4-a0f9-4eb9-bcd2-83ee7f3dd6fe"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Historical patterns around large IPOs show that their listing windows usually appear when market liquidity is relatively abundant and risk appetite is elevated. This phenomenon itself can sometimes serve as a structural annotation to overheated market sentiment. Ultimately, the real test for investors lies in maintaining independent judgment — neither being swept up by the narrative enthusiasm created by a giant IPO, nor treating institutional warnings as a signal for panic selling. Instead, both should be used as reference tools to calibrate the market’s current coordinates, allowing investors to make clearer asset-allocation decisions based on their own portfolio structure and risk tolerance.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="13b52c74-f432-4bf5-b22d-453fd0374963"><div class="bn-block" data-node-type="blockContainer" data-id="13b52c74-f432-4bf5-b22d-453fd0374963"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

How Should Investors Understand Current Risks in U.S. Equities?

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="2d2fdbea-1810-45a9-bf59-c4e0fa0a397b"><div class="bn-block" data-node-type="blockContainer" data-id="2d2fdbea-1810-45a9-bf59-c4e0fa0a397b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">在过去的几天里，美国股市收到了两个性质截然不同的信号：SpaceX 即将完成史上最大规模的 IPO，而美国银行（BofA）则向客户发出了近年来最严重的减持预警。将这两个事件放在同一时间线上来看，它们共同构成了一个关于市场周期的判断。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9f929f90-c850-4488-a16a-fbc31b4568f1"><div class="bn-block" data-node-type="blockContainer" data-id="9f929f90-c850-4488-a16a-fbc31b4568f1"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>SpaceX 上市：高风险高估值的豪赌</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d114435c-62df-496a-836d-4e2fc018fe4c"><div class="bn-block" data-node-type="blockContainer" data-id="d114435c-62df-496a-836d-4e2fc018fe4c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">SpaceX 计划以每股 135 美元的价格出售超过 5.55 亿股股票，目标是筹集至少 750 亿美元，公司估值超过 1.75 万亿美元。如果行使超额配售权，总募资规模可能达到 857 亿美元。若成功完成，将超越沙特阿美在 2019 年创下的纪录，成为全球规模最大的 IPO。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="80aae89e-3046-4f4a-a3a9-397a7fd3b92f"><div class="bn-block" data-node-type="blockContainer" data-id="80aae89e-3046-4f4a-a3a9-397a7fd3b92f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">然而，纸面上的宏大叙事与公司的财务现实之间存在明显差距。S-1 招股说明书显示，该公司在 2025 年实现合并收入 186.74 亿美元，营业亏损 25.89 亿美元，调整后 EBITDA 为 65.84 亿美元。在业务板块层面，Starlink 的连接业务贡献了 114 亿美元，约占总收入的 61%，是唯一持续盈利的业务。AI 部门 xAI 是亏损的主要来源。2025 年全年 GAAP 净亏损为 49 亿美元，反映了公司在星舰建造、卫星网络扩张和 AI 计算基础设施方面的巨额资本密集型投资。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9e1ea882-7465-4fdb-9ece-bc0454f17063"><div class="bn-block" data-node-type="blockContainer" data-id="9e1ea882-7465-4fdb-9ece-bc0454f17063"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">值得注意的是，高达 30% 的发行份额已预留给散户投资者，他们可能获得价值超过 200 亿美元的股票。这一比例是传统 IPO 中通常 5% 至 10% 散户配额的三倍或更多。这种结构反映了马斯克对散户投资者忠诚度的押注，但也意味着交易早期可能出现 FOMO 情绪的系统性放大。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="430c82cd-ce3b-4777-8e88-da258be52a92"><div class="bn-block" data-node-type="blockContainer" data-id="430c82cd-ce3b-4777-8e88-da258be52a92"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">历史经验表明，在大型科技 IPO 期间，QQQ 的表现并不稳定。Facebook、Snowflake 和 Coinbase 在上市后 20 个交易日内大多录得正回报，而 Uber 和 Arm 在类似窗口期则表现出明显的分化和波动性。结论是，巨型 IPO 既是市场情绪的放大器，也是流动性的真正消耗者。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="531bb5d1-2677-436e-b84c-2afe0c7a3fff"><div class="bn-block" data-node-type="blockContainer" data-id="531bb5d1-2677-436e-b84c-2afe0c7a3fff"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>BofA 预警：70% 的熊市信号已触发</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="5431e9cb-34ab-49d1-a8cf-7a53d56c98f1"><div class="bn-block" data-node-type="blockContainer" data-id="5431e9cb-34ab-49d1-a8cf-7a53d56c98f1"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">与 IPO 热潮形成对比的是，华尔街发出了最重大的风险预警之一。美国银行美国股票策略师 Savita Subramanian 在 6 月 5 日的报告中写道：“在我们追踪的熊市前兆信号中，已有 70% 被触发，这与此前市场顶部出现前的平均水平一致——目前需要额外的谨慎。”</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="00b98827-c143-486d-9471-463f997dc279"><div class="bn-block" data-node-type="blockContainer" data-id="00b98827-c143-486d-9471-463f997dc279"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">在 10 个熊市信号中，7 个在 5 月份被触发，高于 4 月份的 5 个和 3 月份的 4 个。自 1990 年以来，7 个也是熊市出现前触发信号的平均数量。基于此，BofA 将标普 500 指数的目标价设定在年底 7,100 点，意味着较目前约 7,400 点的水平有约 6% 的下跌空间。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f5a42da8-414f-4f2b-baa9-ab8434a5ee2d"><div class="bn-block" data-node-type="blockContainer" data-id="f5a42da8-414f-4f2b-baa9-ab8434a5ee2d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">在 20 个估值指标中，标普 500 指数有 17 个指标显示“统计上昂贵”，其中一些指标甚至超过了互联网泡沫时期的水平。在科技行业内部，分化正在扩大：股票表现最好和最差的五分之一股票之间的中位数表现差距已扩大到 2000 年 2 月以来的最高水平。与此同时，超大规模科技公司的资本支出预计将接近其运营现金流的 100%。这意味着 AI 投资的货币化效率正面临严峻考验。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="fdde1261-b37b-46f0-8e09-759385b4af9e"><div class="bn-block" data-node-type="blockContainer" data-id="fdde1261-b37b-46f0-8e09-759385b4af9e"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>两个问题，一个结构性判断</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="ceb686a2-b77b-4537-a66e-1e7bd63e8c01"><div class="bn-block" data-node-type="blockContainer" data-id="ceb686a2-b77b-4537-a66e-1e7bd63e8c01"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">面对以上信息，市场有两种截然不同的解读框架：一种是 AI 确实包含泡沫成分；另一种是 BofA 只是过早地看顶。坦率地说，这两种可能性并非相互排斥。历史上，经典的“顶部预测”有时发生在牛市仍在如火如荼进行之时，有时则被证明非常准确。这次的不同之处在于，支撑估值的ا基础设施投资是真实且实质性的，但货币化路径仍有待验证。市场估值在多个维度上都高于历史中位数，但并非所有板块的风险水平都相同。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4de62e1b-5092-4d76-b80f-b570c5ef3cdb"><div class="bn-block" data-node-type="blockContainer" data-id="4de62e1b-5092-4d76-b80f-b570c5ef3cdb"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">核心问题始终在于价格是否已经包含了足够的未来增长预期。当 SpaceX 以约 94 倍的市销率（P/S）进入市场，并且由于持续亏损而暂时不符合纳入标普 500 指数的资格时，市场接受“故事溢价”的意愿将成为一项重要考验。其结果将直接影响 OpenAI 和 Anthropic 等后续超级 IPO 的定价和市场接受度。Mitrade</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="fb8e14df-a11b-4003-95ba-7dd742a3002d"><div class="bn-block" data-node-type="blockContainer" data-id="fb8e14df-a11b-4003-95ba-7dd742a3002d"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>这对投资者意味着什么</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="aa12498f-0b1c-4e88-b2fe-b7b8317f15cf"><div class="bn-block" data-node-type="blockContainer" data-id="aa12498f-0b1c-4e88-b2fe-b7b8317f15cf"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">SpaceX 上市和 BofA 预警的同时出现，为市场情绪构成了一个相当典型的压力测试：极端乐观的叙事与系统性的估值压力并存。对于持有科技敞口的投资者来说，此时的问题与其说是“继续买入还是全部退出”，不如说是当前头寸的风险回报特征是否仍然符合其最初的投资逻辑。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="64309b09-e4f1-4794-81e2-d911fe561135"><div class="bn-block" data-node-type="blockContainer" data-id="64309b09-e4f1-4794-81e2-d911fe561135"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">有三个问题值得思考：投资组合的核心论点是否依赖于 AI 的快速货币化？现有的未实现收益是否已经消化了相当程度的乐观情绪？如果未来六个月市场下跌 10% 至 15%，是否有足够的资本和心理承受能力来等待复苏？</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b2be63c4-a0f9-4eb9-bcd2-83ee7f3dd6fe"><div class="bn-block" data-node-type="blockContainer" data-id="b2be63c4-a0f9-4eb9-bcd2-83ee7f3dd6fe"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">大型 IPO 周围的历史模式表明，它们的上市窗口通常出现在市场流动性相对充裕、风险偏好较高的时候。这种现象本身有时可以作为市场情绪过热的结构性注脚。最终，对投资者而言，真正的考验在于保持独立判断——既不被巨型 IPO 制造的叙事热情所裹挟，也不将机构的警告视为恐慌性抛售的信号。相反，两者都应作为参考工具，用于校准市场当前的位置坐标，从而使投资者能够根据自身的投资组合结构和风险承受能力，做出更清晰的资产配置决策。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="13b52c74-f432-4bf5-b22d-453fd0374963"><div class="bn-block" data-node-type="blockContainer" data-id="13b52c74-f432-4bf5-b22d-453fd0374963"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

投资者应如何理解当前美股的风险？

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="c8c0a711-56b3-4696-96bd-5645df09856a"><div class="bn-block" data-node-type="blockContainer" data-id="c8c0a711-56b3-4696-96bd-5645df09856a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>Tech stocks have pulled back recently, and I'm considering trading </strong>$USTECH100<strong> on </strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a2cc973c-4d1b-4c52-ae5b-84564ac098f8"><div class="bn-block" data-node-type="blockContainer" data-id="a2cc973c-4d1b-4c52-ae5b-84564ac098f8"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/sodex_official">@sodex_official</a></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="659485e2-a0db-45d5-905c-14e8df6b7a43"><div class="bn-block" data-node-type="blockContainer" data-id="659485e2-a0db-45d5-905c-14e8df6b7a43"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">  <a target="_blank" rel="noopener noreferrer nofollow" href="https://sodex.com/join/JELLYZ">https://sodex.com/join/JELLYZ</a>  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8dc63ac3-aca2-4cc4-b1d2-02df440bdba2"><div class="bn-block" data-node-type="blockContainer" data-id="8dc63ac3-aca2-4cc4-b1d2-02df440bdba2"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Since early June, the Nasdaq-100 has already pulled back more than 5% from its high.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9127bcaf-341e-4a88-ab27-e024824ed739"><div class="bn-block" data-node-type="blockContainer" data-id="9127bcaf-341e-4a88-ab27-e024824ed739"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">My current view: this correction is mainly driven by a crowded AI trade unwinding, plus an upward repricing of rate expectations.From a historical perspective, this is still a small-scale pullback.But the real question is:  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="45869ad7-6554-42dc-861b-82154a7d8604"><div class="bn-block" data-node-type="blockContainer" data-id="45869ad7-6554-42dc-861b-82154a7d8604"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>Will it turn into a deeper decline?</strong>  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="44788a12-8e4e-4af2-a461-09badd789ddd"><div class="bn-block" data-node-type="blockContainer" data-id="44788a12-8e4e-4af2-a461-09badd789ddd"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="04cd0a1c-878b-4cb1-ad86-3c807ce60b81"><div class="bn-block" data-node-type="blockContainer" data-id="04cd0a1c-878b-4cb1-ad86-3c807ce60b81"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The next few days will be key. We need to watch macro data and earnings expectations closely to judge whether this is becoming a real risk, or a chance to buy the dip.  So before taking the trade, I pulled 10 years of Nasdaq-100 Index data, from June 10, 2016 to June 9, 2026, to answer one question:  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="1d040d4e-eb8b-40ac-bfa5-cba023bb12f3"><div class="bn-block" data-node-type="blockContainer" data-id="1d040d4e-eb8b-40ac-bfa5-cba023bb12f3"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>What are the drawdown patterns of the Nasdaq-100?</strong>  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="cc70fbc5-93c9-4efe-8b92-bff98ca226bf"><div class="bn-block" data-node-type="blockContainer" data-id="cc70fbc5-93c9-4efe-8b92-bff98ca226bf"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f3b14232-5521-4ec4-9fca-744ccc568fba"><div class="bn-block" data-node-type="blockContainer" data-id="f3b14232-5521-4ec4-9fca-744ccc568fba"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Here is the key takeaway first.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="3c4df29d-8165-42b9-8175-f26363f5b2e8"><div class="bn-block" data-node-type="blockContainer" data-id="3c4df29d-8165-42b9-8175-f26363f5b2e8"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">A 5% pullback is usually about sentiment and positioning. </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="bd6ba9fc-b6dd-4a96-81ac-2daafb9fa36f"><div class="bn-block" data-node-type="blockContainer" data-id="bd6ba9fc-b6dd-4a96-81ac-2daafb9fa36f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">A 10% correction means we need to watch rates, earnings, and the semiconductor chain closely. </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="25bde6a8-ede3-4b56-8830-3aeb33764bf0"><div class="bn-block" data-node-type="blockContainer" data-id="25bde6a8-ede3-4b56-8830-3aeb33764bf0"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">A 20%+ drawdown usually means a major new variable has hit the market, and investors are repricing the entire macro and valuation framework.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="857dfbb5-34f1-43c6-b6e4-91dea0361bda"><div class="bn-block" data-node-type="blockContainer" data-id="857dfbb5-34f1-43c6-b6e4-91dea0361bda"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a30b2c16-fb47-4eec-a0c2-ee7692e4b417"><div class="bn-block" data-node-type="blockContainer" data-id="a30b2c16-fb47-4eec-a0c2-ee7692e4b417"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Now let's look at the data.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="3a660703-0118-44ce-a313-565f6038c7bc"><div class="bn-block" data-node-type="blockContainer" data-id="3a660703-0118-44ce-a313-565f6038c7bc"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Over the past 10 years, the Nasdaq-100 has seen: 20 drawdowns of more than 5%; </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="5c9262d6-4364-400c-b0b4-ac396d029755"><div class="bn-block" data-node-type="blockContainer" data-id="5c9262d6-4364-400c-b0b4-ac396d029755"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">11 drawdowns of more than 10%; 4 major drawdowns of more than 20%.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="cf13d11f-3a31-4f23-a6d2-ee09f75f566f"><div class="bn-block" data-node-type="blockContainer" data-id="cf13d11f-3a31-4f23-a6d2-ee09f75f566f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">In other words, 5% pullbacks are common, 10% corrections come almost every one to two years, and 20%+ drawdowns usually mean the market is repricing interest rates, liquidity, policy risk, or tech earnings.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="66766339-823e-4bbe-b507-cfa6a8c56724"><div class="bn-block" data-node-type="blockContainer" data-id="66766339-823e-4bbe-b507-cfa6a8c56724"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b3b0028a-b935-482b-8217-a598ee3c0243"><div class="bn-block" data-node-type="blockContainer" data-id="b3b0028a-b935-482b-8217-a598ee3c0243"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Specifically, the four 20%+ drawdowns of the past decade were:  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="95e779b5-3805-4029-8d4e-2178c1f1e3b0"><div class="bn-block" data-node-type="blockContainer" data-id="95e779b5-3805-4029-8d4e-2178c1f1e3b0"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>1. Q4 2018: max drawdown about 22.99%.</strong> The pressure came from Fed rate hikes, balance-sheet tightening, and trade tensions. The Nasdaq-100 fell from around 7,660 on August 29 to around 5,899 on December 24, 2018, and did not reclaim its high until April 2019. The essence was simple: once the rate and liquidity backdrop shifted, growth valuations had to compress.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="6efa9ae1-af62-4f88-9824-57530131f26b"><div class="bn-block" data-node-type="blockContainer" data-id="6efa9ae1-af62-4f88-9824-57530131f26b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>2. The 2020 COVID shock: max drawdown about 28.03%.</strong> The Nasdaq-100 fell from around 9,719 on February 19 to around 6,994 on March 20, 2020 — a nearly 30% drop in just 22 trading days. This was not a collapse in tech fundamentals. It was a liquidity shock across global risk assets. After the Fed moved quickly to inject liquidity, the Nasdaq-100 reclaimed its high in just 55 trading days. The lesson: the Nasdaq-100 falls fast, but when liquidity turns, it can recover just as fast.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="80c18701-fac5-4ddb-a55b-df2a5da6692a"><div class="bn-block" data-node-type="blockContainer" data-id="80c18701-fac5-4ddb-a55b-df2a5da6692a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>3. November 2021 to late 2022: max drawdown about 35.56%.</strong> This was the most important Nasdaq-100 bear market of the decade. The cause was not one company blowing up. It was persistent inflation, aggressive Fed tightening, surging real rates, and a full repricing of tech-growth valuations. The Nasdaq-100 fell from around 16,573 on November 19, 2021 to around 10,679 on December 28, 2022 — 281 trading days to the low, then another 243 trading days to reclaim the high. With this kind of drawdown, you cannot call a bottom just because “it has fallen a lot.” The underlying rate cycle had changed.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d7c0279a-6b66-465d-bac1-0527835bb967"><div class="bn-block" data-node-type="blockContainer" data-id="d7c0279a-6b66-465d-bac1-0527835bb967"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>4. The tariff-war shock in H1 2025: max drawdown about 22.93%.</strong> This was the fourth 20%+ drawdown of the past decade. The Nasdaq-100 fell from around 22,176 on February 19 to around 17,090 on April 8, 2025, before reclaiming its high on June 24.  The triggers were tariff escalation, trade-policy uncertainty, and growth concerns.For high-valuation tech, tariffs are not just a trade issue. They feed into costs, profit expectations, inflation, and ultimately  H1 2025 was not ordinary volatility. It was a classic policy-shock drawdown.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9e3d2195-fbc6-497f-a7ac-454d0b43bbfb"><div class="bn-block" data-node-type="blockContainer" data-id="9e3d2195-fbc6-497f-a7ac-454d0b43bbfb"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d10329e3-7023-4187-8d17-009d81d908cb"><div class="bn-block" data-node-type="blockContainer" data-id="d10329e3-7023-4187-8d17-009d81d908cb"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Beyond these four, the Nasdaq-100 also had several mid-sized corrections of 10–20%:  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="df284a20-dab0-4910-b7e4-fbd17742e884"><div class="bn-block" data-node-type="blockContainer" data-id="df284a20-dab0-4910-b7e4-fbd17742e884"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">May–June 2019: about 11.06%, on trade tensions. </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="63d8f571-63f0-48bb-9bbc-2e3631975548"><div class="bn-block" data-node-type="blockContainer" data-id="63d8f571-63f0-48bb-9bbc-2e3631975548"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">September 2020: about 12.78%, as tech cooled after its strong post-COVID rally. </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="88f6bef4-1c7f-4018-88e5-4902b3fdd8c7"><div class="bn-block" data-node-type="blockContainer" data-id="88f6bef4-1c7f-4018-88e5-4902b3fdd8c7"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">February–March 2021: about 10.93%, on rising Treasury yields. </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="da6f1b6b-ec9c-4557-91db-b602f68cdb7c"><div class="bn-block" data-node-type="blockContainer" data-id="da6f1b6b-ec9c-4557-91db-b602f68cdb7c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">July–August 2024: about 13.58%, as the AI and semiconductor trade got too crowded and the market digested earlier gains. </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="28228bb1-e028-426a-a076-a07e34974066"><div class="bn-block" data-node-type="blockContainer" data-id="28228bb1-e028-426a-a076-a07e34974066"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">January–March 2026: about 12.12%, mainly on the U.S.–Iran conflict.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a37c0483-8034-4129-9cc4-6172b85f65c3"><div class="bn-block" data-node-type="blockContainer" data-id="a37c0483-8034-4129-9cc4-6172b85f65c3"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Clearly, 10% corrections are not rare for the Nasdaq-100. They are periodic stress tests within a strong long-term uptrend.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4bfbf65e-e7dc-4b1a-bf6b-345b92a12c2a"><div class="bn-block" data-node-type="blockContainer" data-id="4bfbf65e-e7dc-4b1a-bf6b-345b92a12c2a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="e93c4afd-f81c-44f4-a67e-42b9e27fd029"><div class="bn-block" data-node-type="blockContainer" data-id="e93c4afd-f81c-44f4-a67e-42b9e27fd029"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">So the key is not simply asking how much the index has fallen.  The real question is: <strong>what kind of drawdown is this?</strong> If it is just sentiment and positioning, the dip may be a setup.  If rates, earnings, and semiconductors all start breaking down together, the risk is much bigger.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="748cedad-bc05-47b3-be5d-630b38ac45fb"><div class="bn-block" data-node-type="blockContainer" data-id="748cedad-bc05-47b3-be5d-630b38ac45fb"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">That is why I am watching the next few sessions closely before making the trade. </p></div><div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="e4cc775c-4b87-4216-8632-232ef9cd989a"><div class="bn-block" data-node-type="blockContainer" data-id="e4cc775c-4b87-4216-8632-232ef9cd989a"><div class="bn-block-content" data-content-type="image" data-url="https://static.sosovalue.com/2026/06/10/ffa0cc61-9f13-43bb-8b38-37548c8317f3.png" data-file-block=""><div class="bn-file-block-content-wrapper"><div class="bn-visual-media-wrapper"><img class="bn-visual-media" src="https://static.sosovalue.com/2026/06/10/ffa0cc61-9f13-43bb-8b38-37548c8317f3.png" alt="BlockNote image"></div></div></div></div></div></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a1da9b4f-93ba-4451-8cec-dcbeada1641f"><div class="bn-block" data-node-type="blockContainer" data-id="a1da9b4f-93ba-4451-8cec-dcbeada1641f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="edc787a2-d720-4bfb-a936-b9c20f3c2a6b"><div class="bn-block" data-node-type="blockContainer" data-id="edc787a2-d720-4bfb-a936-b9c20f3c2a6b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"> <a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/hashtag/AI?src=hashtag_click">#AI</a> <a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/hashtag/sodex?src=hashtag_click">#sodex</a></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="73b849c9-954a-432b-a9e3-6fad8705a37b"><div class="bn-block" data-node-type="blockContainer" data-id="73b849c9-954a-432b-a9e3-6fad8705a37b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

What are the drawdown patterns of the Nasdaq-100?

纳斯达克100指数的下跌模式是什么？

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="a047cf32-4878-4457-b64c-9f70a936c6b0"><div class="bn-block" data-node-type="blockContainer" data-id="a047cf32-4878-4457-b64c-9f70a936c6b0"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">AI capital expenditure refers to the fixed-asset investments cloud giants like Amazon, Microsoft, Google and Meta — plus some vertical AI players — make in GPU clusters, data centers, networking, storage and power infrastructure. In essence, it's the starting point of the entire AI supply chain.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="45317ca1-0c07-4961-9c01-3944085049ac"><div class="bn-block" data-node-type="blockContainer" data-id="45317ca1-0c07-4961-9c01-3944085049ac"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2a6a4be2-1323-4ea0-aa57-084b8131a833"><div class="bn-block" data-node-type="blockContainer" data-id="2a6a4be2-1323-4ea0-aa57-084b8131a833"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Once hyperscalers raise CapEx, capital flows down the chain:  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d402c6a7-77c2-4d3f-9cca-db8a53730050"><div class="bn-block" data-node-type="blockContainer" data-id="d402c6a7-77c2-4d3f-9cca-db8a53730050"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>GPU / ASIC → HBM → AI servers → networking equipment → data centers → power infrastructure → cloud revenue</strong>  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="895d4feb-ab39-402c-b9c5-1e92f962f362"><div class="bn-block" data-node-type="blockContainer" data-id="895d4feb-ab39-402c-b9c5-1e92f962f362"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">That's why the pace of AI CapEx directly reflects compute demand, supply-chain orders, application innovation and the ability of AI products to scale.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="415fd4a2-0488-4f00-8ec1-da7cb999b43b"><div class="bn-block" data-node-type="blockContainer" data-id="415fd4a2-0488-4f00-8ec1-da7cb999b43b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2996452b-36e9-48c6-86b9-524dbc197178"><div class="bn-block" data-node-type="blockContainer" data-id="2996452b-36e9-48c6-86b9-524dbc197178"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Looking at the data, the CapEx cycle of the four major cloud giants (Amazon, Microsoft, Google, Meta) splits into four stages:  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c2d3a689-75fd-4a47-8284-0a6e5c0759b7"><div class="bn-block" data-node-type="blockContainer" data-id="c2d3a689-75fd-4a47-8284-0a6e5c0759b7"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>2011–2023:</strong> Traditional cloud expansion. CapEx was driven by enterprise cloud migration, SaaS, video, advertising, e-commerce and storage.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f69c1139-5c8c-4ffd-ad4e-910e8f3d91fc"><div class="bn-block" data-node-type="blockContainer" data-id="f69c1139-5c8c-4ffd-ad4e-910e8f3d91fc"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>2024: </strong>An extraordinary acceleration begins. After being down 2.5% YoY in 2023, combined CapEx jumped 54.8% to $228.4 billion. Post-ChatGPT and GPT-4, AI infrastructure had firmly entered the tech giants' budgets.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="1768e6f0-e051-4f44-a65d-97428424a403"><div class="bn-block" data-node-type="blockContainer" data-id="1768e6f0-e051-4f44-a65d-97428424a403"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>2025: </strong>The arms race confirmed. CapEx grew a further 64.6% to $376 billion — proof that 2024 wasn't a one-off rebound, but the start of sustained expansion in AI compute demand.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0f0516d7-3a20-447e-8b03-35d0935307fb"><div class="bn-block" data-node-type="blockContainer" data-id="0f0516d7-3a20-447e-8b03-35d0935307fb"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>2026:</strong> Explosive growth. On current guidance, combined CapEx could reach $710 billion, up nearly 89% YoY. This is no longer an extension of the cloud cycle — it's a massive buildout as tech giants race to secure the next generation of compute.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="bab55998-f6a2-4176-905d-507ae7dc2f2e"><div class="bn-block" data-node-type="blockContainer" data-id="bab55998-f6a2-4176-905d-507ae7dc2f2e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="37c44bc7-4b67-462a-81e0-649f1534d87c"><div class="bn-block" data-node-type="blockContainer" data-id="37c44bc7-4b67-462a-81e0-649f1534d87c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>And the expansion is far from over. </strong>As free cash flow gets consumed by CapEx, the giants are leaning more on external financing: Alphabet recently moved forward with an equity raise of around $80–85 billion, and Meta is exploring more options to fund its data center buildout.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4b5d13bd-51f4-457a-a005-2709e3604b35"><div class="bn-block" data-node-type="blockContainer" data-id="4b5d13bd-51f4-457a-a005-2709e3604b35"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8a827e19-d9fb-41cb-9ff3-0d5f578ff9e0"><div class="bn-block" data-node-type="blockContainer" data-id="8a827e19-d9fb-41cb-9ff3-0d5f578ff9e0"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>So where is the money going?</strong>  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c180e4cd-6c91-48b7-8f10-a1f5e48c29ca"><div class="bn-block" data-node-type="blockContainer" data-id="c180e4cd-6c91-48b7-8f10-a1f5e48c29ca"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>1. AI chips and accelerators </strong>— Nvidia and AMD GPUs, Google TPUs, Amazon Trainium, Microsoft Maia. The most visible part of the spend.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="22d94173-f5d7-4957-b113-be73139f7a0a"><div class="bn-block" data-node-type="blockContainer" data-id="22d94173-f5d7-4957-b113-be73139f7a0a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>2. HBM, DRAM and enterprise SSDs </strong>— model parameters, training data and inference cache all need high-speed memory; the stronger the GPU, the greater the demand.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b5ff419d-0fdf-4cb5-aba6-72b004ee6f17"><div class="bn-block" data-node-type="blockContainer" data-id="b5ff419d-0fdf-4cb5-aba6-72b004ee6f17"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>3. AI servers and rack-scale systems </strong>— hyperscalers buy full servers and, increasingly, rack-scale systems like GB200 and GB300, not individual GPUs.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="dec9822c-e47e-46ff-8b34-92a3feff43d9"><div class="bn-block" data-node-type="blockContainer" data-id="dec9822c-e47e-46ff-8b34-92a3feff43d9"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>4. Networking and optical modules </strong>— training spans thousands of GPUs, so switches, NICs, optical modules and interconnects become critical.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="df9e419a-67ea-409c-a626-9cdc015bcb07"><div class="bn-block" data-node-type="blockContainer" data-id="df9e419a-67ea-409c-a626-9cdc015bcb07"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>5. Data centers and power </strong>— land, buildings, liquid cooling, transformers, grid connections and long-term power agreements, all built for high power density.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8b3bf2a5-0f85-4858-bb72-73c71afef8fb"><div class="bn-block" data-node-type="blockContainer" data-id="8b3bf2a5-0f85-4858-bb72-73c71afef8fb"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="93e18bdc-917d-4a45-8071-c5eed06704e1"><div class="bn-block" data-node-type="blockContainer" data-id="93e18bdc-917d-4a45-8071-c5eed06704e1"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">So AI CapEx isn't just about buying GPUs — it's about building an entire "compute factory." That's why the AI trade has widened from Nvidia to HBM, memory, servers, optical modules, data centers, power equipment and liquid cooling.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2509240d-de23-49ce-ae00-3a7910b43172"><div class="bn-block" data-node-type="blockContainer" data-id="2509240d-de23-49ce-ae00-3a7910b43172"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For investors, the real question isn't how much the giants spend — it's whether that spending converts into large, sustainable AI revenue. Short term, CapEx means supply-chain orders; medium term, cloud compute capacity; long term, the winner won't be whoever spends the most, but whoever turns each dollar of CapEx into the most revenue and profit.  This cycle may look like a model race on the surface. Underneath, it's a race for compute, power, memory and data center capacity.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="fa950fe9-790b-4de4-b7b1-5bc1f5fa3f6f"><div class="bn-block" data-node-type="blockContainer" data-id="fa950fe9-790b-4de4-b7b1-5bc1f5fa3f6f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="33d06726-8163-4929-96df-93c3573f01e8"><div class="bn-block" data-node-type="blockContainer" data-id="33d06726-8163-4929-96df-93c3573f01e8"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Therefore, the Big Four cloud giants — along with the core suppliers capturing the largest share of AI CapEx — are the <strong>key players in this AI infrastructure cycle.</strong>   <strong>If you want to invest in this theme, you can trade them on </strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="cc2136d5-92c5-44cb-b0a7-3c84ca529205"><div class="bn-block" data-node-type="blockContainer" data-id="cc2136d5-92c5-44cb-b0a7-3c84ca529205"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/sodex_official">@sodex_official</a> such  as <a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/search?q=%24GOOGL&amp;src=cashtag_click">$GOOGL</a>, <a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/search?q=%24MSFT&amp;src=cashtag_click">$MSFT</a>, <a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/search?q=%24MU&amp;src=cashtag_click">$MU</a> and <a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/search?q=%24SNDK&amp;src=cashtag_click">$SNDK</a>. <a target="_blank" rel="noopener noreferrer nofollow" href="https://sodex.com/join/JELLYZ">https://sodex.com/join/JELLYZ</a></p></div><div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="c390c10a-d43e-4de1-8eb8-c095f01da84c"><div class="bn-block" data-node-type="blockContainer" data-id="c390c10a-d43e-4de1-8eb8-c095f01da84c"><div class="bn-block-content" data-content-type="image" data-url="https://static.sosovalue.com/2026/06/09/9c4cbc7f-b138-4ced-972b-f013dee29c4d.jpg" data-file-block=""><div class="bn-file-block-content-wrapper"><div class="bn-visual-media-wrapper"><img class="bn-visual-media" src="https://static.sosovalue.com/2026/06/09/9c4cbc7f-b138-4ced-972b-f013dee29c4d.jpg" alt="BlockNote image"></div></div></div></div></div></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="7675a5cb-cc8b-4fc9-8e22-33d2f11bfe8a"><div class="bn-block" data-node-type="blockContainer" data-id="7675a5cb-cc8b-4fc9-8e22-33d2f11bfe8a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

Why should AI stock investors pay attention to AI CapEx?

为什么AI股票投资者应关注AI资本支出（CapEx）？

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="e1c5c878-cf32-48dd-919e-ba527503d02c"><div class="bn-block" data-node-type="blockContainer" data-id="e1c5c878-cf32-48dd-919e-ba527503d02c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>1. Fear of weaker DRAM demand from customers like Nvidia.</strong>  <br><br>Nvidia's Rubin NVL72 is reportedly set to cut total DRAM capacity per rack from an originally planned 55TB to 28TB, while memory per cabinet may drop from 192GB to 96GB. The market has read this as a sign of weaker memory demand, triggering fears that the expected "volume and price upcycle" in memory may fail to materialize.  <br><br><strong>But in our view, the market has overreacted. </strong>The reason for this adjustment is not weak demand — it's supply bottlenecks.  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2be529b6-1127-4861-95a8-532f9f9b1a92"><div class="bn-block" data-node-type="blockContainer" data-id="2be529b6-1127-4861-95a8-532f9f9b1a92"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div><div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="bf312965-a681-45b3-83ad-050577c06fb7"><div class="bn-block" data-node-type="blockContainer" data-id="bf312965-a681-45b3-83ad-050577c06fb7"><div class="bn-block-content" data-content-type="image" data-url="https://static.sosovalue.com/2026/06/08/10422a55-98c2-4bc9-9068-9d08960b69c0.png" data-file-block=""><div class="bn-file-block-content-wrapper"><div class="bn-visual-media-wrapper"><img class="bn-visual-media" src="https://static.sosovalue.com/2026/06/08/10422a55-98c2-4bc9-9068-9d08960b69c0.png" alt="BlockNote image"></div></div></div></div></div></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a0b28ae7-4b18-467f-ba0f-228772f8bf14"><div class="bn-block" data-node-type="blockContainer" data-id="a0b28ae7-4b18-467f-ba0f-228772f8bf14"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><br><br>First, due to tight allocation and extremely low packaging yields, SK Hynix and Samsung cannot supply enough 192GB memory modules to meet demand. To ensure every Rubin GPU has sufficient DRAM, Nvidia is proactively managing supply-chain risk by lowering memory per cabinet.  <br><br>Second, Nvidia's original reference specification was already very high, and the new architecture is highly flexible. Even with a reduced factory-shipped configuration, end users can still provision additional memory capacity through cloud providers if needed.  So this adjustment does not materially change the reality of tight supply or the growth trajectory of memory demand.  <br><br><strong>2. Macro pressure: Fed / BoJ rate concerns and a heavy data calendar.</strong>  <br>Concerns are rising that the Fed may stay restrictive for longer and that the Bank of Japan may keep hiking. The remarks and policy stance of the new Fed Chair, Kevin Warsh, are worth watching closely.   <br><br>The upcoming U.S. CPI release on June 10 and the FOMC meeting on June 16–17 could further fuel risk-off sentiment.  <br><br><strong>3. A big prior rally, high expectations, and a market waiting for fresh catalysts.</strong>  <br>Memory stocks delivered a blistering rally in April and May. The market now needs fresh industry-specific catalysts to support further upside. The sharp sell-offs in Broadcom and Ciena after strong earnings reports are a case in point — which makes Micron's earnings on June 24 worth watching closely.  <br><br>In summary, <strong>June is a macro-driven month, but the fundamentals of SK Hynix and Samsung remain solid. </strong>  <br><br>Unless clear reversal signals emerge, investors can keep buying the dip in <a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/search?q=%24SKHXUSD&amp;src=cashtag_click">$SKHXUSD</a> and <a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/search?q=%24SAMSUNGUSD&amp;src=cashtag_click">$SAMSUNGUSD</a> on  </p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="e26a8e73-2674-44f5-992d-7b251a9322f7"><div class="bn-block" data-node-type="blockContainer" data-id="e26a8e73-2674-44f5-992d-7b251a9322f7"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/sodex_official">@sodex_official</a> and wait for macro risks to ease. <a target="_blank" rel="noopener noreferrer nofollow" href="https://sodex.com/join/JELLYZ">https://sodex.com/join/JELLYZ</a>  <a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/hashtag/SoDEX?src=hashtag_click">#SoDEX</a> <a target="_blank" rel="noopener noreferrer nofollow" href="https://x.com/hashtag/SoSoValue?src=hashtag_click">#SoSoValue</a></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c71350e6-cf48-40df-9592-edde2987f9ac"><div class="bn-block" data-node-type="blockContainer" data-id="c71350e6-cf48-40df-9592-edde2987f9ac"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

Memory stocks like SK Hynix and Samsung have recently pulled back more than 15%. What is the market worried about?

像 SK Hynix 和 Samsung 这样的内存股最近下跌了超过 15%。市场担心什么？

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="499e7221-8a78-4e07-9f2e-685ef4ebae25"><div class="bn-block" data-node-type="blockContainer" data-id="499e7221-8a78-4e07-9f2e-685ef4ebae25"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Coinbase recently announced that the first U.S. mortgage involving Bitcoin as part of the collateral structure and supported within the Fannie Mae system has been successfully issued. The loan was originated and serviced by Better, while Coinbase provided the digital-asset custody and collateral infrastructure. The borrower was able to use BTC or USDC held in a Coinbase account as collateral to finance the down payment for a home purchase, without first having to sell the digital assets.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f8e4d38d-9dd2-45cd-8ced-dcea40d961a0"><div class="bn-block" data-node-type="blockContainer" data-id="f8e4d38d-9dd2-45cd-8ced-dcea40d961a0"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">One fact needs to be clarified at the outset: this is not “Fannie Mae directly accepting Bitcoin as collateral for a home loan,” nor is it “using Bitcoin directly to pay for a house.” More precisely, it is a two-layer loan structure. The first layer is a standard Fannie Mae-conforming residential mortgage. The second layer is a loan backed by BTC or USDC and used to fund part of the down payment. In other words, the crypto-asset risk is placed in the second-lien loan and custody structure, rather than directly entering Fannie Mae’s core credit balance sheet.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="580dca01-63af-45a6-a271-6198f16e1ac5"><div class="bn-block" data-node-type="blockContainer" data-id="580dca01-63af-45a6-a271-6198f16e1ac5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The reason this development is worth watching is not the size of the first loan itself, but the fact that it advances the long-running narrative of whether crypto assets can enter the traditional financial collateral system into an executable policy and product prototype.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8b5978ba-cb78-4ee7-a6ef-5f546ccc47ef"><div class="bn-block" data-node-type="blockContainer" data-id="8b5978ba-cb78-4ee7-a6ef-5f546ccc47ef"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>A regulatory anchor has emerged.</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0c8bec29-0743-450a-a1b2-bfdad73b1125"><div class="bn-block" data-node-type="blockContainer" data-id="0c8bec29-0743-450a-a1b2-bfdad73b1125"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">In the past, crypto assets were largely viewed within traditional credit systems as peripheral assets: highly volatile, difficult to value, and difficult to regulate. What makes this product important is that it is not a purely on-chain lending product, nor an isolated experiment by a small private lender. Instead, it has been embedded into the “conforming mortgage” framework at the core of the U.S. housing finance system. The involvement of the Fannie Mae system suggests that regulators and housing-finance infrastructure are beginning to explore how to build an auditable bridge among valuation haircuts, custody, risk management, asset verification, and loan compliance.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f19316c1-0e1b-43bf-96ea-228a68b42fb3"><div class="bn-block" data-node-type="blockContainer" data-id="f19316c1-0e1b-43bf-96ea-228a68b42fb3"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The real historical significance lies here: this does not prove that Bitcoin has already become a mainstream form of collateral. Rather, it shows that policymakers are beginning to open a pathway for digital assets to become assets that can be identified, haircut, custodied, and incorporated into risk models.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d4b0a10d-20b8-4109-9293-634eb802a1cd"><div class="bn-block" data-node-type="blockContainer" data-id="d4b0a10d-20b8-4109-9293-634eb802a1cd"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>The customer logic is very clear.</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f8dd6f8d-8fec-46a9-a3cb-f40106ec7f08"><div class="bn-block" data-node-type="blockContainer" data-id="f8dd6f8d-8fec-46a9-a3cb-f40106ec7f08"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">This product is not a mass-market financial tool for all first-time homebuyers. It is designed for a specific group of borrowers: those who hold substantial BTC or USDC, do not want to sell those assets, but need cash liquidity for a home down payment.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="7b42c910-799c-4b20-878c-537bd1b1cb18"><div class="bn-block" data-node-type="blockContainer" data-id="7b42c910-799c-4b20-878c-537bd1b1cb18"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">These users typically share three characteristics. First, their asset structure is tilted toward digital assets, meaning they may have significant paper wealth but limited traditional cash reserves. Second, they are long-term bullish on BTC and do not want to reduce their exposure simply to buy a home. Third, they are able to tolerate overcollateralization, asset lock-up, and additional interest costs. In other words, this is more of a balance-sheet conversion tool for “crypto-rich, cash-light” users than a form of inclusive finance that meaningfully lowers the barrier to homeownership for ordinary buyers.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="781fb428-045d-4cd2-863c-b5e148df262f"><div class="bn-block" data-node-type="blockContainer" data-id="781fb428-045d-4cd2-863c-b5e148df262f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">From a financial-product perspective, it addresses an asset-liquidity mismatch: the user has assets, but those assets are not held in bank deposits, brokerage accounts, or other traditional collateralizable forms. The innovation lies in converting digital assets, through compliant custody and haircut-based valuation, into collateral that can support down-payment financing.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4d845de8-1657-4e1d-8b0d-234a7065a06a"><div class="bn-block" data-node-type="blockContainer" data-id="4d845de8-1657-4e1d-8b0d-234a7065a06a"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>Tax motivation is a core driver.</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b30020dd-f113-4413-9ac1-b18d35678802"><div class="bn-block" data-node-type="blockContainer" data-id="b30020dd-f113-4413-9ac1-b18d35678802"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">For long-term BTC holders, selling assets to fund a down payment may trigger capital-gains taxes while also reducing their exposure to future upside. The appeal of the collateralized-loan structure is that users can obtain dollar liquidity for a home down payment without selling their BTC.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="ef70a6d3-7cce-47ff-97b4-14fec6b9cd0e"><div class="bn-block" data-node-type="blockContainer" data-id="ef70a6d3-7cce-47ff-97b4-14fec6b9cd0e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">This is essentially a crypto version of a long-standing financing strategy used by wealthy asset holders: do not sell the asset; borrow against its liquidity. Stocks, bonds, and real estate have long been used as collateral for financing. BTC and USDC are now being tested within a similar framework. For crypto-asset holders, this means digital assets may gradually become more than trading instruments; they may become part of financing, credit, and asset-allocation systems.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="97a44744-f6ac-4dbe-b3f2-609b7fb52b63"><div class="bn-block" data-node-type="blockContainer" data-id="97a44744-f6ac-4dbe-b3f2-609b7fb52b63"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">That said, this development should be viewed rationally.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="05e0b804-d769-4a0b-88c5-708819e2a5f7"><div class="bn-block" data-node-type="blockContainer" data-id="05e0b804-d769-4a0b-88c5-708819e2a5f7"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The first loan amount is very small, and the transaction remains essentially a proof of concept. Its narrative value is far greater than its near-term business value. A single case does not prove that the product has scalable risk-management capacity, nor does it mean that banks, GSEs, investors, and regulators have fully accepted crypto assets as routine collateral within the mortgage system.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9b21b708-6f73-41ee-8610-9dc85660a63e"><div class="bn-block" data-node-type="blockContainer" data-id="9b21b708-6f73-41ee-8610-9dc85660a63e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Moreover, the applicable user base is inherently limited. BTC collateral requires a significant haircut and overcollateralization. Borrowers must also accept asset lock-up, limited ability to trade freely, and the cost of servicing two layers of loans. For ordinary homebuyers, this may not necessarily be more attractive than increasing the cash down payment, selling part of their assets, or using traditional financing options.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9620adde-cc8a-4df3-983a-5700d154a80a"><div class="bn-block" data-node-type="blockContainer" data-id="9620adde-cc8a-4df3-983a-5700d154a80a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">In addition, the product does not eliminate BTC volatility. It merely isolates that volatility risk through haircuts, custody arrangements, loan structuring, and default-resolution mechanisms. As long as the underlying asset remains highly volatile, financial institutions will have to use more conservative loan-to-value ratios, stricter custody rules, and more complex compliance procedures to hedge the risk.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8919a91a-12b9-4fd3-b65b-6e7ded82d908"><div class="bn-block" data-node-type="blockContainer" data-id="8919a91a-12b9-4fd3-b65b-6e7ded82d908"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Therefore, this development should be assessed on two levels. From a business-scale perspective, it remains very small and should not be overinterpreted in the short term as the start of a broad crypto-mortgage boom. But from a policy and financial-infrastructure perspective, it is an important signal: crypto assets are taking a step from being merely “tradable assets” toward becoming “collateral assets that can be incorporated into the risk-management framework of traditional finance.”</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="602bb810-1244-418d-85cb-c4151646205c"><div class="bn-block" data-node-type="blockContainer" data-id="602bb810-1244-418d-85cb-c4151646205c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">What deserves attention is not how much revenue the first loan can generate, but whether Fannie Mae, Freddie Mac, the FHFA, lenders, and custodians will eventually develop more standardized rules: which digital assets can be recognized, how they should be valued, what haircut ratios should apply, how custody requirements should be set, how assets should be handled in the event of default, and whether these assets can be more broadly incorporated into personal asset verification and credit assessment.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="3ac454c4-0273-4c38-b106-a76e1a3efe3e"><div class="bn-block" data-node-type="blockContainer" data-id="3ac454c4-0273-4c38-b106-a76e1a3efe3e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">If these questions are gradually institutionalized, the narrative of crypto assets entering the traditional financial collateral system will move from “story” to “rule.” And in financial history, what changes the market is often not the first transaction itself, but the regulatory and institutional entry point confirmed behind that first transaction.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9e39564b-7e94-4e20-9c3b-363c9de79d1e"><div class="bn-block" data-node-type="blockContainer" data-id="9e39564b-7e94-4e20-9c3b-363c9de79d1e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

Bitcoin Finds a Home— America's First Fannie Mae-Backed Crypto Mortgage

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="499e7221-8a78-4e07-9f2e-685ef4ebae25"><div class="bn-block" data-node-type="blockContainer" data-id="499e7221-8a78-4e07-9f2e-685ef4ebae25"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Coinbase 最近宣布，在美国房利美（Fannie Mae）系统内成功发行了首笔涉及 Bitcoin 作为部分抵押结构贷款的抵押贷款。该笔贷款由 Better 提供发起和管理服务，Coinbase 则负责数字资产托管和抵押基础设施。借款人能够使用存放在 Coinbase 账户中的 BTC 或 USDC 作为抵押，为购房首付融资，而无需先出售数字资产。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f8e4d38d-9dd2-45cd-8ced-dcea40d961a0"><div class="bn-block" data-node-type="blockContainer" data-id="f8e4d38d-9dd2-45cd-8ced-dcea40d961a0"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">首先需要澄清一点：这并非“房利美直接接受 Bitcoin 作为房屋贷款抵押”，也不是“直接使用 Bitcoin 支付房屋”。更准确地说，这是一种双层贷款结构。第一层是符合房利美标准的常规住宅抵押贷款。第二层是抵押品为 BTC 或 USDC 的贷款，用于资助首付的一部分。换句话说，加密资产的风险被置于第二顺位贷款和托管结构中，而不是直接进入房利美的核心信用资产负债表。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="580dca01-63af-45a6-a271-6198f16e1ac5"><div class="bn-block" data-node-type="blockContainer" data-id="580dca01-63af-45a6-a271-6198f16e1ac5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">这一发展值得关注的原因并非首笔贷款本身的规模，而是它将关于加密资产能否进入传统金融抵押品体系的长期叙事，推进到了一个可执行的政策和产品原型。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8b5978ba-cb78-4ee7-a6ef-5f546ccc47ef"><div class="bn-block" data-node-type="blockContainer" data-id="8b5978ba-cb78-4ee7-a6ef-5f546ccc47ef"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>监管的锚点已出现。</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0c8bec29-0743-450a-a1b2-bfdad73b1125"><div class="bn-block" data-node-type="blockContainer" data-id="0c8bec29-0743-450a-a1b2-bfdad73b1125"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">过去，在传统信贷体系中，加密资产在很大程度上被视为边缘资产：波动性高、估值困难、监管困难。这款产品之所以重要，是因为它并非纯粹的链上借贷产品，也不是小型私人贷方孤立的实验。相反，它已被嵌入到美国住房金融体系核心的“符合性抵押贷款”框架中。房利美系统的介入表明，监管机构和住房金融基础设施正开始探索如何建立一个可审计的桥梁，连接估值折扣、托管、风险管理、资产验证和贷款合规。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f19316c1-0e1b-43bf-96ea-228a68b42fb3"><div class="bn-block" data-node-type="blockContainer" data-id="f19316c1-0e1b-43bf-96ea-228a68b42fb3"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">真正的历史意义在于此：这并非证明 Bitcoin 已成为主流抵押品形式。相反，它表明政策制定者正开始为数字资产成为可识别、可打折、可托管并纳入风险模型的资产开辟一条道路。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d4b0a10d-20b8-4109-9293-634eb802a1cd"><div class="bn-block" data-node-type="blockContainer" data-id="d4b0a10d-20b8-4109-9293-634eb802a1cd"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>客户逻辑非常清晰。</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f8dd6f8d-8fec-46a9-a3cb-f40106ec7f08"><div class="bn-block" data-node-type="blockContainer" data-id="f8dd6f8d-8fec-46a9-a3cb-f40106ec7f08"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">该产品并非面向所有首次购房者的主流金融工具。它针对的是特定借款人群体：那些持有大量 BTC 或 USDC，不愿出售这些资产，但又需要现金流动性来支付购房首付的人。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="7b42c910-799c-4b20-878c-537bd1b1cb18"><div class="bn-block" data-node-type="blockContainer" data-id="7b42c910-799c-4b20-878c-537bd1b1cb18"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">这些用户通常具有三个特征。首先，他们的资产结构偏向数字资产，这意味着他们可能拥有可观的账面财富，但现金储备有限。其次，他们长期看好 BTC，不愿为了购房而减少其敞口。第三，他们能够容忍超额抵押、资产锁定以及额外的利息成本。换句话说，这更像是一种为“加密资产富裕、现金匮乏”用户提供的资产负债表转换工具，而不是一种真正降低普通购房者置业门槛的普惠金融形式。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="781fb428-045d-4cd2-863c-b5e148df262f"><div class="bn-block" data-node-type="blockContainer" data-id="781fb428-045d-4cd2-863c-b5e148df262f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">从金融产品角度来看，它解决了资产流动性错配的问题：用户拥有资产，但这些资产并未存放在银行存款、经纪账户或其他可传统抵押的形式中。其创新之处在于，通过合规的托管和基于折扣的估值，将数字资产转化为可支持首付融资的抵押品。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="4d845de8-1657-4e1d-8b0d-234a7065a06a"><div class="bn-block" data-node-type="blockContainer" data-id="4d845de8-1657-4e1d-8b0d-234a7065a06a"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>税务动机是核心驱动力。</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b30020dd-f113-4413-9ac1-b18d35678802"><div class="bn-block" data-node-type="blockContainer" data-id="b30020dd-f113-4413-9ac1-b18d35678802"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">对于长期持有 BTC 的人来说，出售资产来支付首付款可能会触发资本利得税，同时也会减少他们未来获得收益的可能性。抵押贷款结构的好处在于，用户可以在不出售其 BTC 的情况下获得用于购房首付的美元流动性。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="ef70a6d3-7cce-47ff-97b4-14fec6b9cd0e"><div class="bn-block" data-node-type="blockContainer" data-id="ef70a6d3-7cce-47ff-97b4-14fec6b9cd0e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">这本质上是富裕资产持有者长期以来使用的融资策略的加密版本：不卖出资产；而是以其流动性进行借贷。股票、债券和房地产长期以来一直被用作融资抵押品。现在，BTC 和 USDC 正在类似的框架下进行测试。对于加密资产持有者来说，这意味着数字资产可能逐渐超越交易工具的范畴；它们可能成为融资、信贷和资产配置系统的一部分。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="97a44744-f6ac-4dbe-b3f2-609b7fb52b63"><div class="bn-block" data-node-type="blockContainer" data-id="97a44744-f6ac-4dbe-b3f2-609b7fb52b63"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">尽管如此，这一发展仍应理性看待。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="05e0b804-d769-4a0b-88c5-708819e2a5f7"><div class="bn-block" data-node-type="blockContainer" data-id="05e0b804-d769-4a0b-88c5-708819e2a5f7"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">首笔贷款金额非常小，交易本质上仍是概念验证。其叙事价值远大于短期商业价值。单一案例并不能证明该产品具有可扩展的风险管理能力，也不能说明银行、政府支持企业（GSEs）、投资者和监管机构已完全接受加密资产作为抵押品纳入抵押贷款系统的常规操作。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9b21b708-6f73-41ee-8610-9dc85660a63e"><div class="bn-block" data-node-type="blockContainer" data-id="9b21b708-6f73-41ee-8610-9dc85660a63e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">此外，适用用户群体本身就有限。BTC 抵押需要显著的折扣和超额抵押。借款人还必须接受资产锁定、有限的自由交易能力以及偿还两层贷款的成本。对于普通购房者来说，这可能并不比增加现金首付、出售部分资产或使用传统融资方案更具吸引力。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9620adde-cc8a-4df3-983a-5700d154a80a"><div class="bn-block" data-node-type="blockContainer" data-id="9620adde-cc8a-4df3-983a-5700d154a80a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">此外，该产品并未消除 BTC 的波动性。它只是通过折扣、托管安排、贷款结构和违约解决机制来隔离这种波动性风险。只要标的资产保持高度波动，金融机构就必须使用更保守的贷款价值比、更严格的托管规则和更复杂的合规程序来对冲风险。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8919a91a-12b9-4fd3-b65b-6e7ded82d908"><div class="bn-block" data-node-type="blockContainer" data-id="8919a91a-12b9-4fd3-b65b-6e7ded82d908"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">因此，这一发展应从两个层面进行评估。从业务规模角度看，它仍然很小，短期内不应过度解读为广泛的加密抵押贷款热潮的开端。但从政策和金融基础设施角度看，它是一个重要信号：加密资产正从仅仅是“可交易资产”迈向“可纳入传统金融风险管理框架的抵押资产”。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="602bb810-1244-418d-85cb-c4151646205c"><div class="bn-block" data-node-type="blockContainer" data-id="602bb810-1244-418d-85cb-c4151646205c"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">值得关注的不是首笔贷款能产生多少收入，而是房利美、房地美（Freddie Mac）、联邦住房金融局（FHFA）、贷方和托管方最终是否会制定更标准化的规则：哪些数字资产可以被认可，应如何估值，应适用何种折扣比例，如何设定托管要求，在发生违约时如何处理资产，以及这些资产是否可以更广泛地纳入个人资产验证和信用评估中。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="3ac454c4-0273-4c38-b106-a76e1a3efe3e"><div class="bn-block" data-node-type="blockContainer" data-id="3ac454c4-0273-4c38-b106-a76e1a3efe3e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">如果这些问题逐渐制度化，加密资产进入传统金融抵押品体系的叙事将从“故事”变为“规则”。而在金融史上，改变市场的往往不是第一笔交易本身，而是这第一笔交易背后确认的监管和制度入口。</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9e39564b-7e94-4e20-9c3b-363c9de79d1e"><div class="bn-block" data-node-type="blockContainer" data-id="9e39564b-7e94-4e20-9c3b-363c9de79d1e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

Bitcoin 找到了归宿——美国首个由房利美（Fannie Mae）支持的加密抵押贷款

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="335f7ff3-0486-4368-8039-2f21d234adc5"><div class="bn-block" data-node-type="blockContainer" data-id="335f7ff3-0486-4368-8039-2f21d234adc5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">No single event can explain a market rout. But as June began, crypto sold off across the board: Bitcoin fell below $66,000, ETH dropped under $1,900, and spot Bitcoin ETFs recorded 12 consecutive trading days of net outflows, with cumulative redemptions over the past three weeks approaching $4 billion. Strategy, which once loudly proclaimed it would “never sell,” sold 32 BTC — a negligible amount in dollar terms, but enough to signal that an unprecedented capital migration has quietly begun. The crypto market is now doing what it always does: digesting a liquidity revolution taking place in U.S. equities.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="331f295d-5825-49f7-9ba6-99a924418e23"><div class="bn-block" data-node-type="blockContainer" data-id="331f295d-5825-49f7-9ba6-99a924418e23"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Based on the 24-hour low of $65,426, Bitcoin has retraced nearly 50% from its October 2025 all-time high of $126,200. Using the Jan. 1, 2026 closing price of $88,839, BTC is down about 25% year to date. Over the past 24 hours, crypto derivatives markets saw roughly $744 million in liquidations. Ether also fell below $1,900, while the total cryptocurrency market capitalization stood at about $2.38 trillion, down roughly 5% over 24 hours.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="042c5b04-2b66-4abf-a9ec-098a0ef6ad66"><div class="bn-block" data-node-type="blockContainer" data-id="042c5b04-2b66-4abf-a9ec-098a0ef6ad66"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>Institutions Are Voting With Their Feet</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="fb61b412-62a8-4006-aca5-2e4884e6fae4"><div class="bn-block" data-node-type="blockContainer" data-id="fb61b412-62a8-4006-aca5-2e4884e6fae4"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">ETF flows are currently the clearest pressure gauge. According to SoSoValue data, U.S. spot Bitcoin ETFs have posted 12 consecutive trading days of net outflows, with cumulative redemptions of about $3.97 billion, marking the longest outflow streak since the products launched. Spot Ethereum ETFs have also remained under pressure, with 16 consecutive trading days of net outflows and cumulative redemptions of about $847 million. While the absolute amount is smaller than Bitcoin’s, Ethereum is in a more fragile position.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="680adaf6-4905-4b29-af5a-36d5d65c54d0"><div class="bn-block" data-node-type="blockContainer" data-id="680adaf6-4905-4b29-af5a-36d5d65c54d0"><div class="bn-block-content" data-content-type="image" data-url="https://static.sosovalue.com/2026/06/03/thumbnails/ecc4903a-4944-4208-9396-cbaf66f99b20.png" data-file-block=""><div class="bn-file-block-content-wrapper"><div class="bn-visual-media-wrapper"><img class="bn-visual-media" src="https://static.sosovalue.com/2026/06/03/thumbnails/ecc4903a-4944-4208-9396-cbaf66f99b20.png" alt="BlockNote image"></div></div></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a561753e-2ada-4b58-9822-56b274afedfd"><div class="bn-block" data-node-type="blockContainer" data-id="a561753e-2ada-4b58-9822-56b274afedfd"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Bitcoin still has at least the support of the “digital gold” narrative and corporate treasury demand. ETH, by contrast, is facing ETF outflows, weak on-chain fees, slowing ecosystem growth, and competition from alternative chains all at once. Ether’s break below $1,900 is not merely a technical breakdown; it reflects the market’s reduced willingness to price in an “Ethereum revival.”</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="bb2f5bd3-8c8e-4b0d-b230-094081058d84"><div class="bn-block" data-node-type="blockContainer" data-id="bb2f5bd3-8c8e-4b0d-b230-094081058d84"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">ETFs were once viewed as a stable channel for institutional capital to enter crypto assets. Recent data, however, show that part of that capital was never a long-term conviction allocation, but simply one component of a broader risk-asset portfolio. When prices break key moving averages and risk appetite rotates toward stronger themes, ETF flows can amplify downside volatility.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8729f1ed-3d98-4806-80e5-50e88224d051"><div class="bn-block" data-node-type="blockContainer" data-id="8729f1ed-3d98-4806-80e5-50e88224d051"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">At the same time, on-chain data are flashing clear signs of cooling. Glassnode’s latest figures show Bitcoin’s realized-cap growth has collapsed 57% to near zero, while only 59.8% of BTC supply remains in profit, down from 61.5% a week earlier. New capital inflows have nearly stalled, leaving the market increasingly dependent on existing liquidity.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b5f247a7-086c-498a-9cd5-d9cc59b2d888"><div class="bn-block" data-node-type="blockContainer" data-id="b5f247a7-086c-498a-9cd5-d9cc59b2d888"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>When the Strongest Believers Start Selling, How Should the Market Price Conviction?</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="bf403c49-7533-4982-97a2-f8dcbbdf19bd"><div class="bn-block" data-node-type="blockContainer" data-id="bf403c49-7533-4982-97a2-f8dcbbdf19bd"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">On June 1, a Form 8-K from Strategy made the crypto market even quieter. The company sold 32 Bitcoin between May 26 and May 31, raising roughly $2.5 million, with the proceeds earmarked for STRC preferred-stock dividends. The 32 BTC represented less than 0.004% of its 843,706-BTC holdings. The amount was almost negligible, and the transaction itself was not enough to change Bitcoin’s supply-demand balance. But its symbolic meaning far exceeded the numbers.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c09dad51-1c92-4a25-bacd-4a6289824b65"><div class="bn-block" data-node-type="blockContainer" data-id="c09dad51-1c92-4a25-bacd-4a6289824b65"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Strategy’s sale further intensified scrutiny of the crypto-treasury model. Because the proceeds were used for preferred-stock distributions, the transaction underscored that once companies rely on equity, preferred shares, convertible debt and other financing tools to accumulate Bitcoin over the long term, balance-sheet management inevitably becomes a market focus. Investors will naturally arrive at a deeper question: when Bitcoin’s appreciation can no longer sustainably cover the cost of the capital structure, so-called Bitcoin treasury companies may shift from “levered accumulation machines” into financial entities that need cash-flow management. Once the narrative turns from offense to defense, the valuation premium will be reassessed.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="43e8a08b-9aa6-4979-a4ee-9a30f796281e"><div class="bn-block" data-node-type="blockContainer" data-id="43e8a08b-9aa6-4979-a4ee-9a30f796281e"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>Where Has the Liquidity Gone?</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="7ae1a284-9e75-4fa2-9d66-226d7a30d3e5"><div class="bn-block" data-node-type="blockContainer" data-id="7ae1a284-9e75-4fa2-9d66-226d7a30d3e5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">To understand crypto’s liquidity drain, one must look to another arena.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="e0f14fd8-5035-4834-8410-926e34b98d79"><div class="bn-block" data-node-type="blockContainer" data-id="e0f14fd8-5035-4834-8410-926e34b98d79"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">SEC filings and media reports show that SpaceX has filed an S-1, with the market expecting its IPO to raise as much as $60 billion to $80 billion and the company potentially listing on Nasdaq as early as June 12. Anthropic has also reportedly submitted a confidential IPO filing to the SEC. Following its latest funding round, the company is valued at about $965 billion, with annualized revenue of roughly $47 billion. OpenAI, meanwhile, remains in the reported IPO-preparation stage and has not publicly filed a prospectus.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="65083e94-e85b-420a-8367-eecd6bdca1da"><div class="bn-block" data-node-type="blockContainer" data-id="65083e94-e85b-420a-8367-eecd6bdca1da"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">At the same time, Alphabet has announced plans to raise as much as $80 billion through equity financing to support AI infrastructure and compute expansion. The combination of this financing and potential IPOs is likely to absorb part of the risk budget of institutional and high-net-worth investors. Against the backdrop of continued capital chasing AI-related stocks, crypto assets’ relative appeal is declining.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="b0781b75-e1ee-4462-99c1-3ac147c1837e"><div class="bn-block" data-node-type="blockContainer" data-id="b0781b75-e1ee-4462-99c1-3ac147c1837e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Moreover, the ETF structure lowered the threshold for entry, but it also reduced the friction of exit. This double-edged sword becomes especially evident when liquidity is being drawn toward competing opportunities. For institutions holding both crypto assets and technology stocks, the combination of more visible AI cash flows and scarcity premiums around major IPOs is becoming a powerful driver of portfolio rebalancing.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="254fae34-8422-4f91-9461-f79129e80832"><div class="bn-block" data-node-type="blockContainer" data-id="254fae34-8422-4f91-9461-f79129e80832"><div class="bn-block-content" data-content-type="heading"><h3 class="bn-inline-content"><strong>Crypto Is Being Actively Repriced by Capital</strong></h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="6b9ec249-c600-45dd-8d3a-d950a455f251"><div class="bn-block" data-node-type="blockContainer" data-id="6b9ec249-c600-45dd-8d3a-d950a455f251"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The current crypto downturn is different in nature from the systemic deleveraging of 2022.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f7f36ce7-4bfb-425d-a816-db8fab03c4c9"><div class="bn-block" data-node-type="blockContainer" data-id="f7f36ce7-4bfb-425d-a816-db8fab03c4c9"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">In 2026, AI has become the dominant narrative in equity markets. Hedge funds and asset managers are pouring capital into AI-related semiconductors, cloud computing, and infrastructure plays. Strategy’s symbolic breach, persistent ETF outflows, and the breakdown of key levels in both Bitcoin and Ethereum all point to the same conclusion: in the face of the first super IPO cycle of the AI era, crypto is undergoing a passive valuation reset rather than a simple cyclical replay.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0e42f407-f873-4d58-b22d-4a138bd8a06d"><div class="bn-block" data-node-type="blockContainer" data-id="0e42f407-f873-4d58-b22d-4a138bd8a06d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The current decline is therefore better understood as a stress test of funding conditions and allocation logic. If ETF outflows slow, derivatives leverage is flushed out, and macro rate expectations ease, Bitcoin could still form a low-level recovery. But if capital continues flowing into AI stocks and large IPOs, rebounds in Bitcoin and Ether will depend more heavily on support from existing investors, and market volatility is likely to remain elevated.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="86010a91-db26-4056-8abe-4aabcc7431be"><div class="bn-block" data-node-type="blockContainer" data-id="86010a91-db26-4056-8abe-4aabcc7431be"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

When Crypto Loses Its Last Liquidity Fantasy

当加密货币失去其最后的流动性幻想

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="8629febf-3c60-4daa-8371-0c79a62a0449"><div class="bn-block" data-node-type="blockContainer" data-id="8629febf-3c60-4daa-8371-0c79a62a0449"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Leveraging AI and with the right infrastructure, a one-person team can build an on-chain finance business empire.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="34339e5e-e6da-4038-ad03-18e8185a60f5"><div class="bn-block" data-node-type="blockContainer" data-id="34339e5e-e6da-4038-ad03-18e8185a60f5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Now is your chance to start building with SoSoValue.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="a5bf7ab6-a8df-4dc8-84a8-4a73d3d432de"><div class="bn-block" data-node-type="blockContainer" data-id="a5bf7ab6-a8df-4dc8-84a8-4a73d3d432de"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Build Your One-Person On-Chain Finance Business with SoSoValue</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="6cc42492-fc27-4f54-aa64-1864c1b54ff3"><div class="bn-block" data-node-type="blockContainer" data-id="6cc42492-fc27-4f54-aa64-1864c1b54ff3"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"> Be your own boss and build your agentic on-chain finance business, powered by SoSoValue’s one-stop infrastructure spanning financial news, data, index tools, trading API, and high-performance order book.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2e776dfa-dea3-4fd0-88e8-1d602965e93b"><div class="bn-block" data-node-type="blockContainer" data-id="2e776dfa-dea3-4fd0-88e8-1d602965e93b"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The Buildathon page is now live, and builder registration is open.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="473852a8-e76e-435a-907b-5424ff07f7c8"><div class="bn-block" data-node-type="blockContainer" data-id="473852a8-e76e-435a-907b-5424ff07f7c8"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Prize Pool: 10,000 USDC</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="1d125f2e-92d0-4343-902b-268d239ca41a"><div class="bn-block" data-node-type="blockContainer" data-id="1d125f2e-92d0-4343-902b-268d239ca41a"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">Wave 1: 3,000 USDC</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8447ce09-b0c9-4026-9ac4-e2028e84c3f9"><div class="bn-block" data-node-type="blockContainer" data-id="8447ce09-b0c9-4026-9ac4-e2028e84c3f9"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">Wave 2: 3,000 USDC</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="8a913574-3424-450f-99bf-315329be7a03"><div class="bn-block" data-node-type="blockContainer" data-id="8a913574-3424-450f-99bf-315329be7a03"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">Wave 3: 4,000 USDC</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="c1be53e3-11db-4491-86b8-2fe9bcd8083f"><div class="bn-block" data-node-type="blockContainer" data-id="c1be53e3-11db-4491-86b8-2fe9bcd8083f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">From early concept to working prototype to final product, this is your opportunity to turn a bold idea into something real.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="198d3555-9dba-417d-9daf-64e7fce38052"><div class="bn-block" data-node-type="blockContainer" data-id="198d3555-9dba-417d-9daf-64e7fce38052"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Join here: <a target="_blank" rel="noopener noreferrer nofollow" href="https://app.akindo.io/wave-hacks/JBEQXgN4Zi2jA3wA">https://app.akindo.io/wave-hacks/JBEQXgN4Zi2jA3wA</a></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="f223d20f-c19a-47e0-bb03-09b09e53e427"><div class="bn-block" data-node-type="blockContainer" data-id="f223d20f-c19a-47e0-bb03-09b09e53e427"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Want to brainstorm with other builders before the Buildathon officially starts?  Join the Discord Buildathon Channel: <a target="_blank" rel="noopener noreferrer nofollow" href="https://discord.gg/HQuGhhkhUW">https://discord.gg/HQuGhhkhUW</a></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d4e2d8e8-4b29-4809-9a37-6cf7c0b324aa"><div class="bn-block" data-node-type="blockContainer" data-id="d4e2d8e8-4b29-4809-9a37-6cf7c0b324aa"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Your idea starts now.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="42a5c0da-1ea5-4318-9eff-27abaf267465"><div class="bn-block" data-node-type="blockContainer" data-id="42a5c0da-1ea5-4318-9eff-27abaf267465"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"> #SoSoValue #Buildathon #AI #Agentic #OnePerson #Web3 #Crypto #Builder #BuildInPublic</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="dba73c3c-0e3f-4846-b19e-7667573b1f6d"><div class="bn-block" data-node-type="blockContainer" data-id="dba73c3c-0e3f-4846-b19e-7667573b1f6d"><div class="bn-block-content" data-content-type="image" data-url="https://static.sosovalue.com/2026/04/19/8da6a46a-4f8b-4c83-ae66-fea6346295c6.png" data-file-block=""><div class="bn-file-block-content-wrapper"><div class="bn-visual-media-wrapper"><img class="bn-visual-media" src="https://static.sosovalue.com/2026/04/19/8da6a46a-4f8b-4c83-ae66-fea6346295c6.png" alt="BlockNote image"></div></div></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="44aad451-ab22-478e-9f61-0037f18b8fb9"><div class="bn-block" data-node-type="blockContainer" data-id="44aad451-ab22-478e-9f61-0037f18b8fb9"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

Build Your One-Person On-Chain Finance Business with SoSoValue

使用 SoSoValue 构建您的个人链上金融业务

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="4fd4a121-e33c-43ff-af2a-e9464fcba299"><div class="bn-block" data-node-type="blockContainer" data-id="4fd4a121-e33c-43ff-af2a-e9464fcba299"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The results for <strong>#SoSoScholar 2025</strong> are officially out.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2d2d676a-60df-4f3d-983f-a322cb55333e"><div class="bn-block" data-node-type="blockContainer" data-id="2d2d676a-60df-4f3d-983f-a322cb55333e"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Huge respect to the <strong>40 researchers</strong> who stood out this year. With the amount of noise in crypto right now, solid research matters more than ever — and these folks delivered real signal.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0cfa128d-c38c-4469-b661-9e1257234874"><div class="bn-block" data-node-type="blockContainer" data-id="0cfa128d-c38c-4469-b661-9e1257234874"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"><strong>Some quick stats:</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="78e4eff1-a581-4c98-a026-66336f43c3cc"><div class="bn-block" data-node-type="blockContainer" data-id="78e4eff1-a581-4c98-a026-66336f43c3cc"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">📊 <strong>4,279 submissions</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="360d95e2-1c1a-48fe-bb24-60acc7bcefc7"><div class="bn-block" data-node-type="blockContainer" data-id="360d95e2-1c1a-48fe-bb24-60acc7bcefc7"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">🌍 <strong>15+ regions</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="cc05c441-8269-443d-a3a8-ec92057a92b6"><div class="bn-block" data-node-type="blockContainer" data-id="cc05c441-8269-443d-a3a8-ec92057a92b6"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">💰 <strong>$33,000 distributed</strong> (upgraded from the original $30K pool)</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="e220f698-d3b2-46cf-803c-dbfc75e875a5"><div class="bn-block" data-node-type="blockContainer" data-id="e220f698-d3b2-46cf-803c-dbfc75e875a5"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">If you’re curious what actually won — or want to read through the research itself — the full leaderboard and submissions are here:<br>👉 <a target="_blank" rel="noopener noreferrer nofollow" href="https://scholar2025.sosovalue.com/">https://scholar2025.sosovalue.com/</a><br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="349c1b25-e949-41d7-b175-642997216995"><div class="bn-block" data-node-type="blockContainer" data-id="349c1b25-e949-41d7-b175-642997216995"><div class="bn-block-content" data-content-type="heading" data-level="3"><h3 class="bn-inline-content">From Research to Earn: Why We’re Building SoDEX</h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="bc9847d3-f07f-45cf-a2dc-6029f4ef36e2"><div class="bn-block" data-node-type="blockContainer" data-id="bc9847d3-f07f-45cf-a2dc-6029f4ef36e2"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The competition is only part of the picture.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="be37ec94-7c7c-4144-addf-05748eaff1f4"><div class="bn-block" data-node-type="blockContainer" data-id="be37ec94-7c7c-4144-addf-05748eaff1f4"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">We’re using SoSoScholar as the foundation for something bigger: <strong>SoDEX</strong>, which we see as a “New Ark” for crypto research and value creation.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="9950c4a0-97fd-4aff-8d2c-78c4f92a382f"><div class="bn-block" data-node-type="blockContainer" data-id="9950c4a0-97fd-4aff-8d2c-78c4f92a382f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Instead of the usual trading-fee rebate or influencer-driven models, the goal is to align <strong>investors, researchers, and the platform</strong> in a way that’s actually sustainable.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="92099b29-76a3-48e9-9c27-66191ff0271d"><div class="bn-block" data-node-type="blockContainer" data-id="92099b29-76a3-48e9-9c27-66191ff0271d"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">The guiding idea is simple:<br><strong>“Love one another inclusively, benefit one another mutually.”</strong></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="bb76b2a6-cbd2-4715-90d7-e199db5c3517"><div class="bn-block" data-node-type="blockContainer" data-id="bb76b2a6-cbd2-4715-90d7-e199db5c3517"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Practically, that means enabling researchers (“Lookouts”) to turn insight into real, professional-grade assets:</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="41502bb7-6756-4bf6-8b76-7ace9fd84a45"><div class="bn-block" data-node-type="blockContainer" data-id="41502bb7-6756-4bf6-8b76-7ace9fd84a45"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">💰 <strong>Direct incentives</strong>: $SoPoints &amp; $SOSO rewards for delivering alpha</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="53dde29b-7427-4f63-bc4d-14ab1b6b91b6"><div class="bn-block" data-node-type="blockContainer" data-id="53dde29b-7427-4f63-bc4d-14ab1b6b91b6"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">🛠️ <strong>Index Builder</strong>: build custom strategies using SoSoValue data</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="33715d29-b359-43ff-8cc1-0347b197755e"><div class="bn-block" data-node-type="blockContainer" data-id="33715d29-b359-43ff-8cc1-0347b197755e"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">📊 <strong>Monetization</strong>: offer Pro-level research and services directly to global investors<br></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="cf0a5b5a-3468-4963-a72a-579016150ff8"><div class="bn-block" data-node-type="blockContainer" data-id="cf0a5b5a-3468-4963-a72a-579016150ff8"><div class="bn-block-content" data-content-type="heading" data-level="3"><h3 class="bn-inline-content">What Comes Next</h3></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="aa79c262-22f0-442e-ac14-307f2cc2c187"><div class="bn-block" data-node-type="blockContainer" data-id="aa79c262-22f0-442e-ac14-307f2cc2c187"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Our long-term mission is filtering noise and uncovering <strong>true value</strong> — and building infrastructure where research creates real economic impact instead of zero-sum games.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="68eae852-0a15-4e5e-bf88-0ae111eedc1a"><div class="bn-block" data-node-type="blockContainer" data-id="68eae852-0a15-4e5e-bf88-0ae111eedc1a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">If you’re a <strong>researcher, analyst, or builder</strong> interested in shaping the SoDEX &amp; ValueChain era, we’d love to collaborate.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="19575b31-6936-4e4a-a4f5-9ba412e2cde7"><div class="bn-block" data-node-type="blockContainer" data-id="19575b31-6936-4e4a-a4f5-9ba412e2cde7"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Join the voyage here:<br>👉 <a target="_blank" rel="noopener noreferrer nofollow" href="https://forms.gle/pjEBkTG6MnUfXRJ58">https://forms.gle/pjEBkTG6MnUfXRJ58</a><br><br><br>Happy to answer questions or discuss how this model works in practice.<br>👉<a target="_blank" rel="noopener noreferrer nofollow" href="https://discord.com/invite/sodex">https://discord.com/invite/sodex</a></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0e29c16e-40a1-4a1c-a350-238892d4f5e1"><div class="bn-block" data-node-type="blockContainer" data-id="0e29c16e-40a1-4a1c-a350-238892d4f5e1"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

Announcement: SoSoScholar2025 Final Results & Research Leaderboard Now Live

公告：SoSoScholar2025 最终结果和研究排行榜现已上线

From Casino to the New Ark,  To Those Who Still Survive in Crypto,  with the Launch of SoDEX

致这个周期的幸存者：从赌场回到新方舟—— 写在 SoDEX 上线之际

<div class="bn-block-group" data-node-type="blockGroup"><div class="bn-block-outer" data-node-type="blockOuter" data-id="a6801dfd-b202-4dfc-9f32-4e4e6ff19669"><div class="bn-block" data-node-type="blockContainer" data-id="a6801dfd-b202-4dfc-9f32-4e4e6ff19669"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Change is knocking—whether we like it or not. 🚪</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="32fec682-830d-4afe-b572-38a8864077af"><div class="bn-block" data-node-type="blockContainer" data-id="32fec682-830d-4afe-b572-38a8864077af"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">So here’s the scoop: a major U.S. banking trade group is pushing senators to <strong>ban stablecoin issuers (and their partners) from paying interest</strong>. Their argument? Crypto platforms weren’t built to fix gaps in the traditional banking system—and stablecoins don’t come with <strong>FDIC insurance</strong> like regular bank deposits do.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0d1f784c-356d-4408-b3a5-ad422caaf56a"><div class="bn-block" data-node-type="blockContainer" data-id="0d1f784c-356d-4408-b3a5-ad422caaf56a"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Translation in plain English:<br>Banks are saying, <em>“If it looks like a savings account and pays interest, it should play by bank rules.”</em><br>Crypto folks are saying, <em>“Innovation doesn’t wait for permission.”</em></p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="630dbb19-e38b-497b-9c59-721eedc3f82f"><div class="bn-block" data-node-type="blockContainer" data-id="630dbb19-e38b-497b-9c59-721eedc3f82f"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">And that’s where the tension lives.</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="d95794bd-78a3-4bde-b965-5afa246d2424"><div class="bn-block" data-node-type="blockContainer" data-id="d95794bd-78a3-4bde-b965-5afa246d2424"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">Like it or not, <strong>the financial world is changing</strong>. New tools show up, old rules get questioned, and lawmakers try to catch up—sometimes by pumping the brakes. ⛔️</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="94f4ee23-608b-4da5-99ce-1156587bdc51"><div class="bn-block" data-node-type="blockContainer" data-id="94f4ee23-608b-4da5-99ce-1156587bdc51"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">So what do <em>you</em> think?</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="44378d21-03ee-4292-9e3d-bacd37aa416f"><div class="bn-block" data-node-type="blockContainer" data-id="44378d21-03ee-4292-9e3d-bacd37aa416f"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">Should stablecoins be treated like banks if they act like banks?</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="2957be13-1e0b-4070-a56e-46f2bbf981d8"><div class="bn-block" data-node-type="blockContainer" data-id="2957be13-1e0b-4070-a56e-46f2bbf981d8"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">Is banning interest protection… or resistance to change?</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="5b4b51be-f8e6-4a1e-a115-7154743869ae"><div class="bn-block" data-node-type="blockContainer" data-id="5b4b51be-f8e6-4a1e-a115-7154743869ae"><div class="bn-block-content" data-content-type="bulletListItem"><p class="bn-inline-content">Can innovation and regulation actually coexist?</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="530543de-cb7c-4ef4-a92d-bd959efc25b6"><div class="bn-block" data-node-type="blockContainer" data-id="530543de-cb7c-4ef4-a92d-bd959efc25b6"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content">One thing’s clear: change is bound to happen—even when it makes people uncomfortable. The real question is <strong>how we shape it</strong>. 💬</p></div></div></div><div class="bn-block-outer" data-node-type="blockOuter" data-id="0f41923b-6a28-45a4-8206-f3e3da680fd1"><div class="bn-block" data-node-type="blockContainer" data-id="0f41923b-6a28-45a4-8206-f3e3da680fd1"><div class="bn-block-content" data-content-type="paragraph"><p class="bn-inline-content"></p></div></div></div></div>

MAG7

Change Doesn’t Ask for Permission

改变无需请求许可