I started buying gold after Silicon Valley Bank collapsed on March 10, 2023. The FDIC later confirmed that SVB was closed by California regulators on March 10, 2023, and announced on March 13 that it would protect all depositors; the FDIC’s deposit insurance reform report that same year also placed this event in the context of re-examining the U.S. deposit insurance system. [G1][G2] Since then, gold has tripled.

Gold is about to return to its essence and once again challenge the authority of the dollar.

Today, June 12, 2026, SpaceX is listed on Nasdaq. [X5] The IPO that tests the market’s temperature has arrived.

Stagflation

The first time I realized something was off was when a glaring detail appeared in the Q3 2025 earnings reports of major U.S. banks. After the pandemic, global economic growth was like a capsized boat in a ditch—half dead.

Note: The World Bank said that 2020–2024 would be the weakest five-year period for global growth in 30 years; the IMF also described the world economy at the time as “still limping along.” [WG1][WG2]

That day, several major U.S. banks released their third-quarter earnings. On the surface, the market heard a very pretty story: the U.S. consumer remained “resilient,” bank performance was still strong, and card spending was still growing. Reuters later reported that Wall Street’s big banks emphasized consumer resilience in their Q3 earnings, with customer activity remaining solid, including increased credit and debit card spending, and credit trends still very good. [C1] That same day, bank stocks were also traded as good news: Reuters’ headline was “Wall Street ends mixed; banks rally on upbeat results,” with Wells Fargo rising on better-than-expected Q3 profits, Citigroup also rising, and the S&P 500 bank index strengthening. [C2]

But what really struck me as odd was that credit cards were not a marginal item in the earnings reports; they were directly written into the banks’ performance structures.

JPMorgan’s Q3 2025 earnings report showed that its debit and credit card sales volume increased 9% year-over-year; Card Services & Auto net revenue reached $7.2 billion, up 12% year-over-year, driven in part by higher net interest income in Card Services from increased credit card revolving balances. [C3] Wells Fargo’s Q3 materials also showed that card fees increased by $127 million, up 12% year-over-year, due to higher merchant processing card fees and increased consumer credit card activity; its Credit Card business grew 13% year-over-year, including higher loan balances and higher card fees. [C4] Citigroup’s Q3 report similarly stated that U.S. Personal Banking net income grew 64% year-over-year, Branded Cards revenue reached $3.0 billion, up 8% year-over-year, driven by higher loan spreads, higher interest-earning balances, and higher interchange. [C5]

This is where I first saw “stagflationization” happening.

The “consumer resilience” in bank earnings may not be spending power in ordinary people’s lives, but rather revolving borrowing capacity.

Ordinary people continuing to swipe cards does not necessarily represent prosperity. It is more likely that cash flow has been squeezed tighter by living costs, and they are using high-interest credit cards to roll over debt: using this card to maintain daily life, and using the next paycheck, another card, balance transfers, cash advances, or minimum payments to pay off the previous card, rolling today’s inflation and high-rate pressure into the future.

The so-called resilient consumer may just be revolving debt that hasn’t yet blown up.

This is not an emotional judgment. Subsequent data began to confirm this direction. The Philadelphia Fed’s large bank credit card data later showed that in Q3 2025, credit card balances and purchase volumes at large banks continued to rise year-over-year, with the average purchase volume of borrowers with credit scores below 660 growing most noticeably; at the same time, revolving balances hit new highs, and the average purchase APR on general-purpose credit cards remained near 24.5%, close to historical highs. [C6] The New York Fed subsequently showed that by Q4 2025, U.S. household debt rose to $18.8 trillion, with credit card balances rising to a high of about $1.28 trillion; by Q1 2026, credit card balances, though seasonally lower, were still $70 billion higher than a year earlier, and the share of credit card balances flowing into serious delinquency was 7.10%, up from 7.04% a year earlier. [C7][C8]

This is not a simple story of “consumers are strong,” but more like “consumers haven’t broken yet, but are already using high-interest debt to stay alive.”

If we pull the lens from bank earnings to the retail end, this “resilient consumer” story looks even less like broad prosperity and more like a K-shaped consumer society: the upper arm is still spending, while the lower arm has already begun to deform. Walmart’s data is very telling. When Reuters reported in 2024 that Walmart raised its full-year outlook, it mentioned that Walmart’s market share gains spanned all income levels but were primarily driven by higher-income households. [K1] By Q3 2025, Walmart’s own earnings materials put it more bluntly: Walmart U.S. comparable sales grew 4.5%, e-commerce grew 28%, market share gains spanned all income segments but were led by upper-income households; management also said on the earnings call that the U.S. business saw strength across all income levels, especially higher-income households, with growth mainly driven by middle- and upper-income households. [K2]

This is not prosperity in the traditional sense; it is high-income consumers also starting to seek discounts. Channels that used to belong only to low-income households as “money-saving channels” are now absorbing middle- and upper-income traffic. Costco is the same thread. Costco’s Q4 fiscal 2025 net sales grew 8.0% year-over-year to $84.4 billion, and full-year net sales grew 8.1% to $269.9 billion; Reuters’ coverage of that earnings report said that U.S. consumers, under pressure from inflation and tariffs, were flocking to membership-based warehouse retailers like Costco in search of cheap necessities, and Costco was also attracting price-sensitive consumers through its Kirkland private label, low-priced eggs and butter, and extended gas station hours. [K3] So, rather than saying the U.S. consumer is broadly strong, it is more accurate to say that some high-end consumption and traditional supermarket traffic are beginning to migrate to “low-price but decent” channels like Walmart and Costco.

The low-income end is a different picture. By October 2025, the threat of a U.S. government shutdown jeopardized SNAP food stamp distributions; Reuters reported that U.S. food retailers and food companies were preparing for a November sales decline, and if federal food aid were interrupted, it could cause a roughly $8 billion grocery revenue gap, with Walmart being the largest recipient of SNAP grocery spend at about 26.1%. The same report also cited research saying that if benefits were delayed, Walmart, Dollar General, and Dollar Tree’s Q4 sales could decline by less than 1% year-over-year, depending on the duration of the shutdown. [K4]

This is not pure extrapolation. Earlier, in 2014, U.S. food stamp cuts had already left a precedent for Walmart: Reuters reported at the time that severe weather and reduced U.S. food stamp benefits dragged down Walmart’s fiscal Q4 comparable sales, forcing the company to lower its quarterly and full-year earnings forecasts; Walmart’s official earnings report subsequently showed that Walmart U.S. Q4 comparable sales fell 0.4%, traffic fell 1.7%, and the company’s Q4 profit fell about 21% year-over-year. It needs to be written precisely: that time it was not a comprehensive year-over-year decline in Walmart’s total revenue—Walmart U.S. Q4 net sales still grew 2.4%—what was really hit were the comparable sales, traffic, and profit metrics. [K5][K6]

This is the ugliest part of a K-shaped economy: banks see credit cards still being swiped and bad debts not yet exploding; the stock market trades on big bank profits and consumer resilience; the retail end sees high-income groups sinking into discount channels, while low-income groups rely on food stamps to maintain basic consumption. On the surface, consumption has not collapsed, but structurally it has already split. The so-called “resilient” is not broad strength, but the U.S. economy still managing to piece together a decent aggregate number through upper-tier consumption, discount migration, credit expansion, and government transfer payments.

And this is precisely the most insidious part of stagflationization. It does not necessarily erupt in the form of a financial crisis, nor does it necessarily cause the job market to suddenly collapse. It is more like a slow cash-flow drain: inflation pushes up food, gasoline, rent, and daily consumption; high interest rates push up credit card and debt-rolling costs; wage growth cannot keep up with price increases. Macro data still looks unbroken, but the household sector is already paying today’s prices with future cash flow.

In Q1 2026, U.S. real GDP grew at an annualized rate of only 1.6%, but over the same period the PCE price index rose at an annualized 4.5%, and core PCE rose 4.4%; by May, CPI was up 4.2% year-over-year, energy was up 23.5%, and gasoline was up 40.5%. [S1][S2] Employment data remained stable on the surface: in May, the U.S. unemployment rate held at 4.3%, and nonfarm payrolls increased by 172,000; but average hourly earnings grew 3.4% year-over-year, already lagging the 4.2% CPI, and the number of long-term unemployed reached 2.0 million, an increase of 524,000 from a year earlier. [S2][S3]

Forecasts and survey data are also converging in the same direction. The Philadelphia Fed’s Q2 2026 Survey of Professional Forecasters has already placed weaker growth, higher inflation, and higher recession probabilities in subsequent quarters into its forecast path. [S4] The University of Michigan’s preliminary May 2026 consumer survey showed that consumer sentiment remained near 2022 lows, current conditions fell about 9%, consumers continued to be hit by high prices, gasoline prices, and tariff pressures, and the one-year inflation expectation remained at 4.5%; more importantly, groups with little or no stock assets felt the pressure of high prices significantly more heavily. [S5] Reuters, citing the June Beige Book, reported that U.S. economic activity and inflation both rose in recent weeks, retail visits declined, credit card use increased, demand for necessities was stronger, and middle-income households were stretching every dollar further; the same report also mentioned that 9 out of 12 Fed districts mentioned data center construction. [S6] Another Reuters economist survey also showed that, against the backdrop of persistent war-driven inflation, expectations for rate cuts this year have further faded, and the Fed is likely to continue holding rates. [S7]

This is what I mean by stagflationary pressure: growth is not truly expanding at high speed, but living costs are pressing down on ordinary people again; employment has not collapsed across the board, but real purchasing power is being eroded; consumption has not stopped, but an increasing share of consumption may be sustained by credit cards and debt rolling.

My original judgment was that capital did not want to directly trigger a financial crisis, but rather wanted to avoid a financial crisis erupting in an uncontrollable way. It was more like trying to manually engineer a controlled stagflation: letting inflation slowly steal ordinary people’s purchasing power, letting high interest rates suppress demand and valuations, and letting asset prices reprice over a long cycle. That way, the crisis would not arrive as a one-day crash, but would be broken up into years of costs—gasoline, food, rent, credit card interest, job anxiety, and real wage declines—paid by all ordinary people together.

But what I did not expect was that in the end it would go to an even more direct step.

If it were only stagflation, capital could at least pretend it was a macro cycle; but when AI companies, cloud providers, chipmakers, data center contractors, and giant platforms like SpaceX begin to concentrate on selling “the future” to the public market, the matter becomes blunt. They are not waiting for the bubble to digest naturally; they are, while the bubble still has narrative power, while retail investors still want to believe in the future, while index funds and pensions still must allocate to growth assets, distributing the last round of risk as equally as possible.

This is what I call “harvesting the whole society.”

Not because AI has no future. On the contrary, AI is definitely the future. The problem is that the future itself cannot prove that all of today’s valuations are reasonable, nor can it guarantee that those who sell the future today will bear the same kind of risk as those who take over the future tomorrow.

Technological revolution and financial bubbles have never been mutually exclusive. Classic studies on financial instability, manias-panics-crashes, irrational exuberance, and IPO cycles repeatedly illustrate the same thing: when financing, narrative, leverage, and exit windows stack together, genuine technological progress will also be processed by the capital market into a bubble structure. [B1][B2][B3][B4] Railroads, electricity, the internet, and new energy have all proven: technologies that truly change the world often first create a bubble that destroys the latecomers. The place where capital is most adept at substitution is packaging “the direction is correct” as “the price is reasonable,” packaging “the technology will exist” as “the current company will win,” and packaging “the industry will grow” as “the people who take over today will make money.”

AI is the same.

Even crueler is that the companies that look most like winners today may not be the ones that truly remain after the bubble. Technology cycles never guarantee that first movers survive to the end, nor do they guarantee that the companies doing best now will still do best in the next paradigm. Today’s lead may just be a phase lead manufactured by financing windows, GPU supply, brand credibility, and distribution channels; when cost structures, hardware forms, algorithm paradigms, or regulatory environments change, today’s winners can also become tomorrow’s old assets.

OpenAI Infrastructure

OpenAI was the first place I felt something was off. As an AI giant, it should be the one closest to the future, but its actions looked less like a company that had already found a profitable endgame and more like a capital node that had to bind everyone into the same growth curve.

The Stargate project stated that over the next four years it would invest $500 billion to build new AI infrastructure for OpenAI in the United States; participants included OpenAI, SoftBank, Oracle, MGX, with technology partners including Oracle and NVIDIA. [O1][O2] Subsequently, OpenAI reached an AI infrastructure agreement with CoreWeave worth up to about $11.9 billion and obtained equity in CoreWeave; Reuters also reported that the deal occurred on the eve of CoreWeave’s IPO. [O3][O4] Later came Oracle’s roughly $300 billion computing procurement agreement, AMD’s 6GW GPU deployment, and Broadcom’s 10GW OpenAI-designed AI accelerators. [O5][O8][O9][O10]

These numbers are too big; they are no longer ordinary business expansion but the financial structure itself.

If OpenAI already had a clear, stable, self-financing path to profitability, these agreements could be understood as capacity expansion. But reality is more complicated. Reuters, citing WSJ reports, said that as OpenAI sprinted toward an IPO, its revenue and user targets fell short of expectations, and the outside world worried whether it could support the massive data center spending. [O11] HSBC analysts also estimated that OpenAI might need about $207 billion in new funding to meet its data center spending commitments through 2030, and might still be unprofitable by 2030; this is only an analyst estimate, not a company-confirmed fact, but it is enough to show why the market began to worry about its funding gap. [O12]

This is no ordinary commercialization. It is more like binding cloud providers, chipmakers, data center financiers, the U.S. policy narrative, and public market investors all to the same future story.

OpenAI may not be the most evil one. On the contrary, it may be the person at this table who first realized the bubble was inevitable. Sam Altman knew from the beginning that, whether it was OpenAI or not, AI would be pushed into a bubble. Given that, the most rational move was not to exit, but to turn oneself into the center of the bubble, binding the entire AI industry, the U.S. government, cloud providers, chipmakers, and capital markets.

It thought the U.S. government would bail out the market, and perhaps eventually it will. But politics is not a stabilizer; politics itself can also become a transaction. Especially when Trump once again ties tariffs, industrial policy, capital market sentiment, and personal political narrative together, the market is no longer just a market; it becomes a K-line that can be drawn together by policy, public opinion, and liquidity.

This is the sharper point I wanted to make in my original text: when a capitalist-style president regains the ability to directly influence asset prices, many traditional “policy objectives” will be reordered by asset prices. Tariffs can be a negotiating tool, regulation can be a political tool, AI can be a national competition narrative, and the stock market itself can become a dashboard to display governance capability. Whether the government will bail out the market is no longer just an economic question, but a question of political price.

NVIDIA Circular Investment

Originally, every party at the table was gambling, betting that they were not the loser. So everyone went all out, until everyone discovered they were all exhausted. Profits had been compressed to the limit, while the AI economy was still circulating internally. Model companies need computing power, cloud providers need contracts, chipmakers need orders, data centers need financing, electricity and land need policy, and capital markets need growth stories. Every node is proving the valuation of another node, and every node needs another node to keep believing.

This is the most dangerous part of so-called circular trading.

It is not necessarily financial fraud, nor necessarily a conspiracy, but it will entangle real demand and capital demand together, letting investment, revenue, equity, contracts, and market confidence feed each other. NVIDIA invests in OpenAI, OpenAI purchases NVIDIA systems; OpenAI signs huge computing contracts with cloud providers, and cloud providers use those contracts to finance, go public, and expand, which in turn supports OpenAI’s computing narrative. NVIDIA’s announcement of up to $100 billion investment in OpenAI and its cooperation with OpenAI to deploy 10GW of NVIDIA systems is precisely the most representative node in this kind of structure. [O6][O7][O14]

This structure looks like a synergistic ecosystem in an upcycle, but in a downcycle it will turn into a chain of cross-collateralization.

NVIDIA is actually the most special. Unlike other companies that can complete a cash-out exit through IPOs, secondary market share sales, or equity financing, as the hardest shovel-seller in the entire AI bubble, its cash-out method has always been selling hardware: selling GPUs, selling systems, selling networking, selling entire data center infrastructure. Every move it makes will be analyzed by the market, because it can no longer claim to be just a bystander; its customers, investment targets, revenue expectations, and the entire AI infrastructure bubble are tied together. It is not that it cannot make money, but that it makes so much money that it can only keep selling hardware, praying to become a second IBM—an infrastructure company that still exists after the bubble passes—rather than a hardware cyclical stock that gets revalued when the bubble bursts.

Flash Memory Surge

The part about flash memory that truly made me feel something was off was not the fact that “AI needs storage” itself. Of course AI needs storage. Data centers need SSDs. Long context, RAG, vector databases, and KV cache will of course increase read/write pressure. The question is: LLMs have been hot for so long—was flash memory free before? Why was it precisely at the end of 2025, when AI FOMO began to be repeatedly questioned over circular deals, data center capex, and profitability, that memory / NAND / enterprise SSD suddenly got pushed into the role of a new scarce asset?

This is the suspicious part.

The timeline is very subtle. In October 2025, Bloomberg had already begun using circular deals to describe the AI transaction network among OpenAI, NVIDIA, AMD, Oracle, and others, saying these intertwined transactions were raising market concerns over whether the AI boom was being sustained by the deal structure itself. [O14] Around the same period, Reuters also began putting the AI bubble into market discussion. In November, NVIDIA’s earnings temporarily calmed the market, but Reuters’ headline already directly used the phrase “AI bubble jitters”; by December, AI capex from Oracle, Broadcom, and Meta continued to make the market worry again about overheated valuations. [F11][F12]

In other words, the market atmosphere at the end of 2025 was not simply “AI will rise forever.” Quite the opposite: the AI narrative was beginning to show fatigue. Investors started asking whether model companies could actually make money, whether cloud providers’ data center spending was too large, and whether NVIDIA’s growth had already been pushed too high by circular transactions and capex expectations.

Right in this window, flash memory and memory were pushed to center stage.

Reuters had already written the first clue on October 20: the AI boom was driving up prices of the “less trendy” memory. The mechanism was not that LLMs directly consumed all NAND, but that the AI chip race pushed chipmakers to shift capacity, investment, and priority toward HBM, server memory, and high-end AI-related products, which in turn squeezed the supply of traditional memory used in smartphones, computers, and servers. [F13] On November 14, Reuters reported again that Samsung had raised some memory chip prices by as much as 60% from September levels, and some customers had begun panic buying. [F14]

By December 3, Reuters directly framed this as a global supply chain crisis: AI and consumer electronics companies were being forced to compete for a shrinking supply of memory chips, with prices soaring; tech giants such as Microsoft, Google, and ByteDance were all scrambling for supply from Micron, Samsung, and SK hynix. Reuters also explained the mechanism clearly: appetite for high-end AI chips driven by NVIDIA and giants such as Google, Microsoft, and Alibaba meant chipmakers still could not produce enough high-end semiconductors; but their tilt toward AI-related products then choked the traditional memory needed by smartphones, PCs, and consumer electronics. [F15]

TrendForce’s data quantified this process. In Q3 2025, enterprise SSD prices and shipments both rose, and the combined revenue of the top five enterprise SSD brands grew 28% quarter-over-quarter to about $6.54 billion, a new high for the year; TrendForce also said the market had tilted in favor of sellers and expected Q4 enterprise SSD contract prices to rise by more than another 25%. [F16] This was not an ordinary recovery, but the reappearance of a seller’s market.

Then the price increases began to lose control. TrendForce forecast in January that NAND Flash demand was splitting into consumer and AI segments, with enterprise SSD becoming the largest segment; under NAND suppliers’ disciplined capacity management and server demand crowding out other applications, contract prices for all NAND Flash products were expected to rise 33–38% in Q1 2026, with client SSD rising at least 40%. [F3] By Q1 2026, TrendForce said CSPs needed high-speed data transmission and massive storage capacity to build AI server infrastructure; structural HDD shortages also pushed orders toward QLC enterprise SSDs; as a result, combined revenue of the top five NAND Flash suppliers jumped 83.7% quarter-over-quarter, surpassing $38.9 billion. [F4] Another June report showed that rapid adoption of AI Agent services and CSP procurement drove enterprise SSD revenue up 86.1% quarter-over-quarter in Q1, exceeding $18.46 billion; supplier inventories fell to historical lows, output lagged order growth, vendors raised prices aggressively amid tight supply, and enterprise SSD contract prices rose about 80% in Q1. [F5]

So the question is not “is there demand.” Of course there is demand. The question is that this demand narrative appeared too timely, too neatly, and too perfectly suited to keeping the AI bubble alive.

When the market began questioning whether AI infrastructure chains such as OpenAI, Oracle, CoreWeave, and NVIDIA had enough real returns, a new explanation immediately appeared: it is not that AI demand is weak, but that the bottleneck has not yet been solved; it is not that the GPU story is finished, but that memory is also insufficient; it is not that model companies are burning too much money, but that the data pipeline, enterprise SSD, KV cache, and long-context reasoning all need to be rebuilt. NVIDIA was also pushing along the same line: in 2025 it launched AI Data Platform, putting storage vendors such as DDN, Dell, HPE, IBM, NetApp, Pure Storage, VAST Data, and WEKA into the AI enterprise infrastructure narrative; in 2026, BlueField-4 STX and CMX further tied storage, DPUs, GPU memory extension, KV cache, and agentic AI long-context reasoning together. [F8][F9][F10]

I cannot say this already proves legal conspiracy. Public materials contain no price agreement, no antitrust investigation material, and no internal emails that can prove Samsung, SK hynix, Micron, Kioxia, SanDisk, and NVIDIA privately conspired to manipulate prices.

But public materials are enough to prove another thing: at the end of 2025, when AI FOMO was beginning to tire and AI bubble doubts were rising, the memory / NAND / enterprise SSD narrative was suddenly repackaged from a cyclical hardware recovery into an AI infrastructure bottleneck. The timing was too coincidental, the price increases too large, and the supporting reports too neat, making it look less like pure natural demand release and more like a narrative life-extension jointly completed by the industrial chain.

It may not be a conspiracy, but it accomplished the function a conspiracy would need to accomplish: manufacture scarcity, manufacture panic buying, manufacture long-term supply lockups, manufacture the FOMO that “even storage is not enough,” and continue the sentence that the IPO year of 2026 needed most—AI infrastructure is still far from enough, so everything connected to AI infrastructure deserves to be repriced.

LLM Hardware Cost-Effectiveness

There is also a technical judgment here that cannot be covered up by the capital narrative: what may fail first on the LLM path is not necessarily the technology itself, but hardware cost-effectiveness.

I believe scaling laws still exist. Larger models, larger data, longer training, more complex inference may indeed continue to bring capability improvements. But the problem is, if every capability improvement requires larger clusters, higher inference costs, more complex engineering scheduling, and higher electricity and cooling costs, then the bottleneck is no longer “can it be done,” but “is anyone willing to pay for this cost long-term.”

When a model must run across dozens or even more devices, it may certainly be stronger, but it also begins to move away from the economics of ordinary software. What is truly fatal is not that it is unintelligent, but that it is too expensive, too slow, too heavy, and too dependent on hardware cycles. API price wars will compress profits, inference costs will swallow cash flow, enterprise deployments will lengthen payment collection cycles, and open-source models will continuously lower the chargeable boundary. The technology path may not fail, but commercial cost-effectiveness will fail first.

So I say, the fundamental failure of LLMs is not technology, but a cost-effectiveness mismatch on hardware. When better hardware appears, we may certainly have stronger models; but by that day, the algorithm paradigm itself may have already changed. History often goes like this: the first-generation technology path proves the direction, and the second-generation hardware and algorithms truly harvest the world. Those who build it first today may not live to the end; those who do it best today may not continue to do best in the next stage.

Companies like Oracle and CoreWeave are more like middle layers. Oracle can continue to hold up with its balance sheet, cloud contracts, and capital market narrative; neocloud companies like CoreWeave are more fragile. Their value comes from GPUs, data centers, contracts, and financing ability, but if the market begins to doubt the payment ability of model companies, or doubts that the growth rate of AI demand can cover capital expenditures, then these companies will turn from “core AI infrastructure assets” into “high-leverage cyclical assets.”

The biggest problem for this type of company is not that they have no assets, but that their assets are too heavy and their credibility too light. GPUs, server rooms, power contracts, and long-term computing contracts all look like moats in an upcycle; but once the market is no longer willing to value GPUs and long-term computing contracts as growth stocks, they will quickly turn from “future infrastructure” into “hardware contractors weighed down by debt, depreciation, and utilization rates.” Giants still have cash flow, ecosystems, and policy relationships to tell stories; neoclouds’ stories are narrower: if model company demand is discounted, they will be crushed into the road by the hardware cycle like a steamroller, becoming the sunk cost of the next round of infrastructure.

Executives Eager to Cash Out

The most glaring thing about CoreWeave is not just its dependence on GPUs, data centers, massive debt, and a few large customer contracts, but the cash-out pace of insiders.

According to a Bloomberg report in June 2026, CoreWeave’s stock price had more than doubled since its IPO in March 2025; but over the same period, company executives had sold more than $2.3 billion in personal holdings. Bloomberg, citing insider trading data from Washington Service, said the main sellers were CoreWeave’s three co-founders Michael Intrator, Brannin McBee, and Brian Venturo. [CW1] Subsequent market reports further indicated that these transactions were executed through prearranged trading plans; among them, Brian Venturo sold more than $1.1 billion after the lock-up period ended, and the three founders still held about 18% of the company’s shares after the sales, with Intrator remaining the largest shareholder at about 10.4%. [CW2]

This does not mean CoreWeave necessarily has problems, nor that these transactions are illegal. Prearranged trading plans are a common share-sale arrangement used by executives of U.S. listed companies; without further confirmation from individual SEC filings, I will not write it as a verified 10b5-1 fact. But when a company has just been pushed onto the public market by the AI infrastructure narrative, its stock price has skyrocketed, its capital expenditures are huge, its debt pressure is heavy, and its customer concentration is extremely high, the founding team and executives cashing out a cumulative total of over $2 billion is itself a very strong signal: the primary market and insiders have already turned the AI infrastructure story into cash, while public market investors are beginning to bear valuation, financing, and cycle risks.

CoreWeave’s IPO documents can also explain why this matters. Its S-1/A shows that in the IPO, the company issued 47,178,660 shares of Class A common stock, and selling shareholders issued 1,821,340 shares; the company would not receive any proceeds from the shares sold by selling shareholders. At the same time, after the IPO, the three co-founders Michael Intrator, Brian Venturo, and Brannin McBee still held approximately 37.0%, 23.2%, and 18.7% of voting power respectively, totaling about 79.0%. [CW3] In other words, CoreWeave is not a company that has been fully publicized in the ordinary sense: control remains highly concentrated, but market risk has been made public.

This is the core structure of the entire AI bubble: control is in the hands of insiders, risk is in the hands of the public market.

IPO

Google / Alphabet cannot stay out of it either. It is of course a genuinely profitable company, with search, advertising, cloud, and Gemini. But when Alphabet announced a massive equity financing to expand AI infrastructure and compute, its signaling significance was still very strong. [A1][A2] In the same phase, OpenAI also officially submitted a confidential draft S-1, putting the future listing option on the table. [I1] Even the most cash-flow-rich giant and the most core model company are beginning to connect AI infrastructure, the IPO path, and the capital market more directly, which shows that AI is no longer an internal R&D project, but an infrastructure war that needs to be jointly borne by the capital market.

The truly dangerous signal is not that one company is financing, but that everyone is financing almost simultaneously. Every company is afraid of being the last one to stand up and ask for money, because the last financier is most easily identified by the market as “already unable to hold on.” So the rush itself becomes a run: Google first launches equity financing, OpenAI quickly puts out a confidential S-1, Anthropic also queues up for an IPO, and SpaceX packages xAI into the public market. Every company says it is financing for the future, but from the market structure, it looks more like everyone sees that the number of lifeboats is limited and suddenly starts running.

Anthropic is the same. On one hand, it represents the AI safety narrative; on the other, it submits a confidential draft S-1, preparing for an IPO. [H1] This does not prove Anthropic is hypocritical, nor can it prove it is deliberately harvesting. But it does show that discourses like safety, ethics, alignment, pause, and frontier risk have already coexisted with the capital exit cycle. When the Anthropic Institute discusses recursive self-improvement, it proposes the need to establish verifiable mechanisms to slow down or temporarily pause frontier AI development; this may be sincere in terms of technical governance. But when it appears in the same period as an IPO, what the market sees is not just safety, but “the safety narrative is also going public.” [H2]

The problem is not that these safety discussions are necessarily fake, but that they can also become the emotional packaging most easily understood by ordinary investors before an IPO: the more you emphasize “this thing is too powerful and must be regulated,” the more you reinforce “this thing represents the future”; the more you talk about frontier risk, the more you tell the market “we stand at the frontier.” This is the most subtle part of the safety narrative: it can be sincere governance language, and at the same time it can become fear marketing for the capital market. Whether you believe its motives or not, the market will translate it into valuation language.

SpaceX pushed this to another level.

If you only look at the name, SpaceX looks like an aerospace company going public; but if you look at its own S-1, the intent is almost written on the page. In its prospectus, SpaceX said it identified “the largest actionable TAM in human history”: the quantifiable total addressable market is $28.5 trillion, of which Space is only $370 billion, Connectivity is about $1.6 trillion, but AI is $26.5 trillion, about 93%. [X1] Bloomberg also directly summarized this as: SpaceX is marketing a $26.5 trillion AI market story to IPO investors, not a simple rocket or satellite internet story. [X7]

Note, TAM is not a revenue forecast, nor is it money the company has already received; it is just the “addressable market” that the company itself has circled. But precisely because it is not financial reality but a valuation narrative, it reveals even more clearly what this IPO is really selling. SpaceX’s own TAM table has already written the answer: what the public market is truly being asked to believe in is not the $370 billion Space, nor the $1.6 trillion Connectivity, but the $26.5 trillion AI. Even Reuters Breakingviews called such market claims “planet-scale absurdity.” [X8]

The S-1’s operating data further breaks this structure apart: in 2025, the Space segment had revenue of about $4.086 billion and an operating loss of $657 million; the Connectivity / Starlink segment had revenue of about $11.387 billion, operating profit of $4.423 billion, and Segment Adjusted EBITDA of about $7.168 billion; the AI segment had revenue of about $3.201 billion, but an operating loss of $6.355 billion, and Segment Adjusted EBITDA of -$1.237 billion. [X1] Even more glaring is capital expenditure: in 2025, AI segment capex was about $12.727 billion, and in Q1 2026, AI segment capex reached another $7.723 billion. [X1]

This is the key point: Starlink is cash flow and profit; rockets, Starshield, and defense communications are credibility; xAI / Grok / AI compute are losses and capital expenditures, but also the largest source of valuation imagination in the entire prospectus. The SpaceX IPO is not simply selling rockets and Starlink to the market; it is using assets that truly have business intersections, cash flow, and engineering credibility—rockets, Starlink, Starshield, U.S. defense and communication networks—as a shell for the losses of xAI/Grok, the capital expenditures of AI compute, and the platform narrative of X.

So this paragraph cannot be written as “SpaceX becomes the parent entity of Musk’s platform.” That would be too diffuse, and would instead help smooth out its story. A more accurate way to put it is: Musk is using the shell of SpaceX, the cash flow of Starlink, the credibility of defense communications, and the index mechanism of Nasdaq to package xAI into the public market. Rockets are responsible for making the story credible, Starlink is responsible for making the financial statements still look decent, the index mechanism is responsible for bringing in passive funds, and xAI is responsible for pushing the valuation space from a few hundred billion dollars to the mythical scale of a “$26.5 trillion AI TAM.”

This is what I think looks most like “deliberately selling off xAI.” There is almost no sufficiently natural business intersection between xAI and SpaceX. The so-called space data center can be told as a long-term sci-fi narrative; the S-1 indeed writes “deploy orbital AI compute at scale” and “AI infrastructure in space” into the growth story. [X1] But in terms of real-world costs, launch, maintenance, heat dissipation, bandwidth, latency, reliability, and operations are all worse than ground-based data centers. As long as data centers can still be built on Earth, moving computing power into orbit is not a reasonable business synergy, but more like adding a mythical interface to the IPO story that “only SpaceX can tell.”

Without this packaging, pushing a high-loss, high-capex, unproven-business-model xAI alone onto the public market would face much greater resistance; but once packed into SpaceX, it is no longer just xAI, it becomes a combined commodity of “rockets + Starlink + defense communications + Musk + Nasdaq index funds + $26.5 trillion AI TAM.”

So 2026 naturally becomes the harvest year of the last dance.

This last dance most resembles a financing run. In a normal cycle, companies finance according to business maturity; at the end of a bubble, companies finance according to “who hasn’t yet managed to sell their story.” Everyone is afraid of becoming the clown who hasn’t raised money before the deadline, because that means the market has already started asking: why do you need money now? Why has everyone else already sold the future, while you are still standing at the door? So the rush itself exposes fear, and fear drives more rushing.

The FOMO sentiment of 2025 was visible to everyone. When OpenClaw exploded in popularity, there were scenes of nearly a thousand developers and AI enthusiasts lining up for installation outside Tencent’s building in Shenzhen. Sina Technology reported that on March 6, 2026, nearly a thousand people lined up outside Tencent’s building, and Tencent Cloud engineers completed free cloud installations of OpenClaw for users; because the local deployment environment for OpenClaw was complex, platforms like Xiaohongshu already had door-to-door installation services, with one-time installation fees in Shenzhen, Guangzhou, Hangzhou, Chongqing, and other places ranging from 300 to 1,000 yuan; platforms like Xianyu also had tutorials, one-click installation, and remote assistance services, with prices ranging from tens of yuan to nearly 200 yuan. [OC1] 21st Century Business Herald also reported that nearly a thousand people lined up outside Tencent’s Shenzhen office area, and domestic platforms like Xiaohongshu and Xianyu priced door-to-door installation mostly at 300–800 yuan, with 500 yuan per session becoming the mainstream price, while remote installation also cost 50–100 yuan, and some online stores had sold over 1,000 orders. [OC2]

This is not just a tool fad; it is the industrialization of FOMO. OpenClaw requires calling large model capabilities and can be deployed locally or in the cloud, involving model APIs, cloud computing, servers, plugins, and configuration costs. Sina Technology mentioned that Tencent Lighthouse Cloud launched a one-click deployment template for OpenClaw at the end of January, and Baidu AI Cloud launched an ultra-simple deployment plan for OpenClaw in February. [OC1] Cailian Press further reported that domestic cloud providers including Tencent Cloud, Alibaba Cloud, China Mobile Cloud, China Telecom Cloud, JD Cloud, Volcano Engine, and Baidu AI Cloud all connected to OpenClaw; cloud providers, through one-click deployment, permission encapsulation, security hardening, and other methods, turned open-source projects into sellable SaaS services, essentially using computing power, storage, networking, and engineering capabilities to lower user barriers and seek commercial monetization loops. [OC3]

At the same time, cloud providers and model companies also simultaneously launched various Coding Plans. Tencent Cloud developer articles said that Agent workflows like OpenClaw would consume tokens at high frequency, with a moderately complex task potentially triggering dozens of model calls; major domestic cloud providers and model companies intensively launched Coding Plan subscription packages between late 2025 and March 2026, replacing per-token billing with fixed monthly fees, including Alibaba Cloud Bailian, Volcano Ark, Tencent Cloud, Zhipu, Kimi, MiniMax, and others. [OC4] Sohu tech articles also summarized this phenomenon as the OpenClaw “national shrimp-raising” era, with various cloud and model providers launching Coding Plans, exchanging fixed monthly fees for model call quotas, attempting to solve the token cost anxiety of heavy OpenClaw users. [OC5]

But I believe this is precisely the most typical part of a bubble: at the time they looked like entrances to a new era, like AI workflows everyone must have; but once the craze recedes, many things will turn into negative assets. Subscriptions, quotas, computing power, plugins, workflows, training costs, team migration costs—all will become sunk costs after the bubble ebbs. BBC Chinese later wrote in a report that a reversal from “raising lobsters” to “uninstalling lobsters” was emerging: some users thought OpenClaw had high technical barriers, was expensive, and had little practical use, and began paying to seek uninstallation; experts also reminded that OpenClaw is currently “not very practical,” with costs and risks disproportionate. [OC6] People’s Daily Online Shenzhen also reminded that while OpenClaw represents a new trend in AI Agents, the deployment threshold, long-term token costs, hardware costs, operation and maintenance costs, and security risks are not low, and ordinary users should not blindly follow the trend. [OC7]

So this is not as simple as “a tool got popular.” It is more like a rehearsal of the AI bubble at the ordinary person level: first use a new entrance to create anxiety, then use free installation to lower the barrier, then use cloud deployment, Coding Plans, plugin markets, and workflow migration to lock users into the ecosystem. While the bubble is still there, these costs are explained as “learning the future”; after the bubble tide goes out, they become bills one by one.

The stock market is the same. As prices are chased higher, everyone gets tempted. Fund managers cannot miss AI, pensions cannot miss AI, indexes cannot be without AI, and retail investors cannot watch other people make money. So everyone knows it is expensive, but everyone fears getting off too early. The strongest moment of a bubble is not when nobody knows it is a bubble, but when everyone thinks they can leave before the music stops.

This year, all model vendors have understood: they need to sell models, directly sell the concept itself, sell the credibility of Gemini, Claude, ChatGPT, and Grok in their hands, exchange it for money, and survive the difficult era ahead.

This sentence sounds harsh, but it may be the essence of the capital market. The models themselves may not immediately make money. API price wars will compress profits, open-source models will lower the moat, enterprise deployment will lengthen sales cycles, and inference costs will keep consuming cash. But the idea that “models represent the future” itself can be securitized. As long as the market is still willing to pay for the future, model companies can sell the future to the market.

OpenAI is selling the credibility of ChatGPT, Anthropic is selling the credibility of Claude and safety, Google is selling the credibility of Gemini, and this time SpaceX looks more like it is using rockets, Starlink, and Musk’s personal credibility to provide credit backing for xAI / Grok / X.

What ordinary people buy is the future others have already priced.

Index Funds

The more important part of the SpaceX IPO is the index mechanism.

Reuters reported in May that SpaceX had accelerated its IPO timetable and was targeting Nasdaq; on June 11, Reuters reported that SpaceX priced this record IPO at $135 per share, raising about $75 billion; on June 12, Reuters confirmed that SpaceX began trading on Nasdaq, closing its first day at a valuation of about $2.1 trillion. [X2][X5][X6] Nasdaq’s official FAQ makes the mechanism clearer: starting May 1, 2026, eligible newly listed Nasdaq companies ranked in the top 40 can be added to the Nasdaq-100 after 15 trading days, subject to a 3x float cap; Nasdaq also adjusted the market-cap ranking method so that listed shares and unlisted shares can both be used for eligibility / ranking market capitalization, while unlisted shares are used for eligibility and ranking but not for actual weight calculation. [X3]

This means SpaceX is not a normal case of “after listing, let the market slowly choose.” If the rule conditions are met, it can quickly trigger mechanical buying pressure from index funds after listing.

Active investors can at least choose whether to buy or not; but ETFs, mutual funds, derivatives, and benchmark strategies tracking the Nasdaq-100 must rebalance by weight once the rules add SpaceX. Bloomberg described this risk as a “feedback loop,” citing Intropic estimates that passive investors could hold about 30% of the free float after fast-entry; this is not a confirmed fact that has already happened, but a possible mechanical buying pressure created by the overlap of index rules, low float, and a mega-cap valuation narrative. [X4]

This is what I mean by “the whole society bears it together.”

Ordinary people may not have actively chosen xAI, may not have studied Grok’s losses, and may not understand Starship’s capex; but as long as they hold Nasdaq-100 products, technology index funds, pensions, passive products, or indirectly allocate to U.S. growth assets through asset-management accounts, they may become passive buyers of this IPO through the index mechanism.

This is not a story of everyone actively placing bets in a free market, but a story of financial infrastructure distributing risk to everyone.

Banks first earn revenue from credit card revolving borrowing, investment banks then earn fees from IPOs and refinancings, index companies set inclusion rules, ETFs and pensions execute rebalancing, and ordinary people finally enter the table together through wages, savings, retirement money, fund accounts, and passive index products. Each layer says it is only executing rules, but when the rules are connected, they form an extremely efficient risk-transfer machine.

Here, the role of banks and investment banks is especially important. They do not need to sincerely believe these assets can win long term, nor do they need to bear the last baton like ordinary investors. They are more like sellers: design the transaction, underwrite the issuance, arrange liquidity, write research reports, organize roadshows, collect fees, and then hand the final risk to public markets and passive funds. Saying they “aid the tiger” does not mean every banker is subjectively evil; it means that in this structure, banks naturally stand in the position of sellers and dealers. They sell tickets to enter the future, earn the ticket-handling fee, and those left in the venue waiting for the music to stop are the ticket buyers.

Prediction

Not investment advice, but a record of my personal positioning and judgment for this cycle.

I have now moved all assets except gold back into cash.

If my expectation is correct, when the whole world finally confirms that a financial crisis has already happened, we will first see a short, violent, almost indiscriminate decline across all asset categories caused by a liquidity crisis. Stocks will fall, bonds will fall, cryptocurrencies will fall, and even gold may fall first. Because in a true liquidity crisis, the market first sells not the worst assets, but the assets that can still be sold.

Then gold will rebound.

At that time I will sell all my gold, then buy U.S. stocks and cryptocurrency in stages. The only cryptocurrency will be BTC.

For me, BTC is not because I want to bet on “the next 100x coin,” nor because it will definitely not fall in the early stage of a crisis. On the contrary, in the most violent moment of a liquidity crisis, BTC may also be sold off together. But BTC’s meaning is not short-term price, but that at its birth it already wrote the annotation for this round of financial crisis.

The Bitcoin white paper proposed in 2008 a peer-to-peer electronic cash system without financial-institution intermediaries. [BT1] The line written into Bitcoin’s genesis block was even more direct:

The Times 03/Jan/2009 Chancellor on brink of second bailout for banks

This is not decoration; it is the product manual. [BT2] When the financial system socializes bad debts, institutionalizes bailouts, and concentrates monetary power inside sovereigns, central banks, and banking systems, Bitcoin itself is the antithesis of that structure. BTC is Satoshi Nakamoto’s annotation written in advance for this financial crisis.

I will not understand it as an ordinary tech stock, nor as an ordinary risk asset. What truly crosses a financial crisis is not something with fancier narratives, higher leverage, and more fragile exchange liquidity, but something that, after global credit is repriced, can still be understood as a non-sovereign reserve asset. For me, that can only be Bitcoin.

The cruelest part of this risk transfer is that it will pull already injured people back onto the table. Capital first lays off employees; employees hold severance and savings, and cannot quickly find better jobs. Meanwhile, the market is still rising, AI stocks are still telling the future story, indexes are still making new highs, and social feeds and media are still saying “missing AI means missing the era.”

Prices are also rising, and they watch their limited money shrink and depreciate. So unemployed people, anxious people, and people whose cash flow has worsened may instead use their last savings to chase the companies that just laid them off, or will soon continue replacing them through AI.

This is what I call the second harvest: the first time, inflation, high interest rates, and layoffs drain ordinary people’s cash flow; the second time, rising asset prices suck their remaining severance and savings back into the market. The financial system is most cruel not when it makes people despair, but when it shows people hope at the moment they feel least secure.

From the perspective of programmers’ employment environment, AI demand after the crisis will again explode, just as after the internet bubble the internet companies and software demand that truly changed the world continued to grow. AI will not disappear because the bubble bursts. Quite the opposite: after the bubble bursts, AI demand that truly improves efficiency, lowers costs, and reshapes workflows may remain more clearly.

But I will not look forward to it, because that is too far away.

There will also be core assets that do not disappear.

In this crisis, which company will come out unscathed? Apple. It loses too slowly, and wins too slowly. I only hope that in the end it does not become a company with nothing left but cash flow.

Conclusion

This does not mean AI will fail. AI may continue to progress, models may continue to become stronger, and Agents may truly change software, office work, scientific research, education, finance, and the military. The future direction can be correct, but that does not prevent today’s capital structure from being dangerous.

The internet was the future, but that did not mean every internet company in 2000 deserved to survive. Railroads were the future, but that did not mean every railroad stock was worth its bubble-era price. Electric vehicles are the future, but that does not mean every high-valuation new-energy company can cross the cycle. AI is the future, but that does not mean all AI companies, AI infrastructure, AI compute contracts, AI IPOs, and AI index weights today deserve to be bought by society as a whole.

So I am not writing an “AI is useless” essay.

I am writing this: when AI turns from a technology narrative into a capital exit, when compute contracts turn from business expansion into mutual balance-sheet support, when IPOs turn from a company-growth milestone into a risk-transfer mechanism, when stagflationary pressure squeezes ordinary people’s cash flow tighter, yet the market still pushes them toward “future assets,” this is no longer merely technology investing.

It is redistribution.

A financial crisis is originally a way of rebalancing. Every crisis destroys a rotten old order, and may also let the public benefit again after rebalancing. But this time, capital seems unwilling to let the crisis happen in the traditional form. It first tried to use stagflation to spread out the cost, breaking the crash into slow bleeding in daily life; when that was still not enough to digest AI infrastructure and valuation pressure, it packaged the future for listing and sold unrealized growth to public markets and index funds.

So in the end, the financial crisis did not disappear. It was only delayed, split apart, packaged, and resold.

But capital cannot maintain order through packaging forever. Capital will eventually move toward disorder, because every participant can rationally choose to run first: companies finance first, insiders sell first, banks collect underwriting fees first, funds rebalance first, retail investors chase first, and finally everyone hopes someone else will bear the liquidity risk for them. When everyone thinks this way, market order no longer comes from consensus; it can only come from force.

Final order can only come from power intervention: central bank liquidity, government rescue, regulatory suspension, trading-rule adjustment, fiscal backstop, even political orders. But for the capital market, power usually does not intervene when the bubble is most comfortable; it intervenes only when it has no choice. And that moment of no choice is often already the eve of complete financial collapse. In other words, the beginning of forced restoration of order by power is often the beginning of the financial crisis being officially acknowledged.

The cruelest part of a bubble is not that it deceives everyone, but that it tells partial truths. AI really is the future. Compute really is important. Models really will change the world. The United States really will treat AI as the core of national competition. Precisely because all of these are true, the bubble is harder to see through. Lies are easy to refute; a half-true future is hardest to reject.

This is the last dance.

The music is still playing, the lights are still on, investment banks are still smiling, founders are still speaking, big tech is still raising next year’s capex guidance, models are still updating, GPUs are still shipping, data centers are still breaking ground, SpaceX is still launching rockets, and Nasdaq is still adjusting rules. Everyone knows where the door is, but everyone hopes they are not the last one to leave.

And ordinary people stand at the door, holding money diluted by inflation, being told:

This is the future.

---

References / Sources

Opening Context, Gold, Bank Crisis, Bitcoin

[G1] Federal Deposit Insurance Corporation. (2023, March 13). FDIC Acts to Protect All Depositors of the former Silicon Valley Bank, Santa Clara, California. FDIC.
https://www.fdic.gov/news/press-releases/2023/pr23019.html

[G2] Federal Deposit Insurance Corporation. (2023). Options for Deposit Insurance Reform — Section 2: Introduction and Background. FDIC.
https://www.fdic.gov/analysis/options-deposit-insurance-reforms/report/options-deposit-insurance-reform-section-2.pdf

[BT1] Nakamoto, S. (2008). Bitcoin: A Peer-to-Peer Electronic Cash System. Bitcoin.org.
https://bitcoin.org/en/bitcoin-paper

[BT2] Bitcoin Core. (n.d.). chainparams.cpp — Genesis block timestamp string. GitHub.
https://github.com/bitcoin/bitcoin/blob/master/src/kernel/chainparams.cpp https://raw.githubusercontent.com/bitcoin/bitcoin/master/src/kernel/chainparams.cpp

Post-Pandemic Global Growth and the U.S. K-Shaped Consumer Split

[WG1] World Bank. (2024, January 9). Global Economy Set for Weakest Half-Decade Performance in 30 Years.
https://www.worldbank.org/en/news/press-release/2024/01/09/global-economic-prospects-january-2024-press-release

[WG2] Gourinchas, P.-O. / IMF Blog. (2023, October 10). Resilient Global Economy Still Limping Along, With Growing Divergences.
https://www.imf.org/en/blogs/articles/2023/10/10/resilient-global-economy-still-limping-along-with-growing-divergences

[K1] Reuters. (2024, November 19). Walmart raises annual forecasts again, signals holiday shopping resilience.
https://www.reuters.com/business/retail-consumer/walmart-raises-annual-forecasts-betting-strong-holiday-shopping-2024-11-19

[K2] Walmart. (2025, November 20). Q3 FY26 Earnings Release / Earnings Presentation / Earnings Call Transcript.
https://corporate.walmart.com/news/2025/11/20/walmart-releases-q3-fy26-earnings

[K3] Costco Wholesale Corporation. (2025, September 25). Costco Wholesale Corporation Reports Fourth Quarter and Fiscal Year 2025 Operating Results; Reuters. (2025, September 25). Costco tops quarterly estimates as Americans seek bargains amid inflation.
https://investor.costco.com/news/news-details/2025/Costco-Wholesale-Corporation-Reports-Fourth-Quarter-and-Fiscal-Year-2025-Operating-Results/default.aspx
https://www.reuters.com/business/retail-consumer/costco-beats-quarterly-revenue-estimates-strong-demand-cheaper-essentials-2025-09-25

[K4] Reuters. (2025, October 31). US grocers brace for sales dip as food aid set to lapse.
https://www.reuters.com/business/retail-consumer/us-grocers-brace-sales-dip-food-aid-set-lapse-2025-10-31

[K5] Reuters via Yahoo Finance. (2014, January 31). Wal-Mart cuts outlook due to store closings, Sam’s Club weakness.
https://finance.yahoo.com/news/wal-mart-cuts-outlook-due-140937005.html

[K6] Walmart. (2014, February 20). Walmart reports Q4 underlying EPS of $1.60, Fiscal 2014 underlying EPS of $5.11.
https://corporate.walmart.com/news/2014/02/20/walmart-reports-q4-underlying-eps-of-1-60-fiscal-2014-underlying-eps-of-5-11

Credit Cards, Bank Earnings, and Household Debt

[C1] Reuters. (2025, October 15). Wall Street top banks highlight consumer resilience.
https://www.reuters.com/business/finance/wall-street-top-banks-highlight-consumer-resilience-2025-10-15

[C2] Reuters / The Straits Times. (2025, October 15). Wall Street ends mixed; banks rally on upbeat results.
https://www.straitstimes.com/business/companies-markets/wall-street-ends-mixed-banks-rally-on-upbeat-results

[C3] JPMorgan Chase & Co. (2025, October 14). 3Q25 Earnings Press Release.
https://www.jpmorganchase.com/content/dam/jpmc/jpmorgan-chase-and-co/investor-relations/documents/quarterly-earnings/2025/3rd-quarter/f24f154c-2653-4da3-b1b0-185586086d50.pdf https://www.sec.gov/Archives/edgar/data/19617/000162828025044845/a3q25erfexhibit991narrative.htm

[C4] Wells Fargo & Company. (2025, October 14). 3Q25 Financial Results / Presentation.
https://www.wellsfargo.com/assets/pdf/about/investor-relations/earnings/third-quarter-2025-presentation.pdf

[C5] Citigroup Inc. (2025, October 14). Third Quarter 2025 Results and Key Metrics.
https://www.citigroup.com/rcs/citigpa/storage/public/Earnings/Q32025/2025prqtr3rslt.pdf

[C6] Federal Reserve Bank of Philadelphia. (2026, January 21). Large Bank Credit Card and Mortgage Data 2025 Q3 Narrative.
https://www.philadelphiafed.org/surveys-and-data/2025-q3-large-bank

[C7] Federal Reserve Bank of New York. (2026, February 10). Household Debt Balances Grow Modestly; Early Delinquencies Level Out for Non-Housing Debts.
https://www.newyorkfed.org/newsevents/news/research/2026/20260210

[C8] Federal Reserve Bank of New York. (2026, May 12). Household Debt Balances Rise Slightly as Delinquency Transition Rates Hold Steady.
https://www.newyorkfed.org/newsevents/news/research/2026/20260512

U.S. Stagflationary Pressure

[S1] U.S. Bureau of Economic Analysis. (2026, May 28). GDP (Second Estimate) and Corporate Profits, 1st Quarter 2026.
https://www.bea.gov/news/2026/gdp-second-estimate-and-corporate-profits-1st-quarter-2026

[S2] U.S. Bureau of Labor Statistics. (2026, June 10). Consumer Price Index Summary — May 2026.
https://www.bls.gov/news.release/pdf/cpi.pdf

[S3] U.S. Bureau of Labor Statistics. (2026, June 5). Employment Situation Summary — May 2026.
https://www.bls.gov/news.release/pdf/empsit.pdf

[S4] Federal Reserve Bank of Philadelphia. (2026, May 15). Second Quarter 2026 Survey of Professional Forecasters.
https://www.philadelphiafed.org/surveys-and-data/real-time-data-research/spf-q2-2026

[S5] University of Michigan Surveys of Consumers. (2026, May 8). Preliminary Results from the May 2026 Survey.
https://data.sca.isr.umich.edu/fetchdoc.php?docid=81123

[S6] Reuters. (2026, June 3). US economic activity, inflation both up in recent weeks, Fed survey shows.
https://www.reuters.com/business/us-economic-activity-inflation-both-up-recent-weeks-fed-survey-shows-2026-06-03

[S7] Reuters. (2026, June 9). Fed to hold rates this year, cut calls fade as war inflation persists, economists say.
https://www.reuters.com/business/fed-hold-rates-this-year-cut-calls-fade-war-inflation-persists-economists-say-2026-06-09

OpenClaw, Coding Plans, and AI Agent FOMO

[OC1] Sina Technology. (2026, March 6). Nearly a Thousand People Queue Downstairs at Tencent to Install It: What Are Users Doing with OpenClaw?
https://finance.sina.com.cn/tech/roll/2026-03-06/doc-inhqaham2427583.shtml

[OC2] 21jingji / EqualOcean. (2026, March 9). A Thousand People Queue to Install Top AI App OpenClaw: Has Tencent Made Sky-High Installation Fees Go Cold?
https://www.21jingji.com/article/20260309/herald/763cb0f461b6fbd9d89bc83b046c4ee6.html

[OC3] Cailian Press. (2026, March 9). The “Lobster” Is on the Table, Listed Companies Rush to “Raise” It! OpenClaw Ignites the Tech Circle.
https://www.cls.cn/detail/2306947

[OC4] Tencent Cloud Developer Community. (2026). The Most Cost-Effective Way to Buy Tokens for OpenClaw: Understand Each Platform’s Coding Plan in One Article.
https://cloud.tencent.com/developer/article/2654022

[OC5] Sohu. (2026). OpenClaw’s “Everyone Raises Shrimp” Era: Which Coding Plan Fits You? Details of Mainstream Domestic Coding Plan Packages.
https://www.sohu.com/a/997473785_616364

[OC6] BBC News Chinese. (2026). OpenClaw and the AI Agent Boom Explained: From “Raising Lobsters” to “Uninstalling Lobsters”.
https://www.bbc.com/zhongwen/articles/c93wvdn91kxo/simp

[OC7] People’s Daily Online Shenzhen. (2026, March 9). Shenzhen Web Affairs | Beware of “Catching Shrimp”! Do Not Blindly Follow the Trend.
http://sz.people.com.cn/BIG5/n2/2026/0309/c202846-41519152.html

OpenAI, AI Infrastructure, and Circular Deals

[O1] OpenAI. (2025, January 21). Announcing The Stargate Project.
https://openai.com/index/announcing-the-stargate-project

[O2] Reuters. (2025, January 21). Trump announces private-sector AI infrastructure investment.
https://www.reuters.com/technology/artificial-intelligence/trump-announce-private-sector-ai-infrastructure-investment-cbs-reports-2025-01-21

[O3] CoreWeave. (2025). CoreWeave Announces Agreement with OpenAI to Deliver AI Infrastructure.
https://www.coreweave.com/news/coreweave-announces-agreement-with-openai-to-deliver-ai-infrastructure

[O4] Reuters. (2025, March 10). CoreWeave strikes $12 billion contract with OpenAI ahead of IPO, sources say.
https://www.reuters.com/technology/artificial-intelligence/coreweave-strikes-12-billion-contract-with-openai-ahead-ipo-sources-say-2025-03-10

[O5] Reuters. (2025, September 10). OpenAI, Oracle sign $300 billion computing deal, WSJ reports.
https://www.reuters.com/technology/openai-oracle-sign-300-billion-computing-deal-wsj-reports-2025-09-10

[O6] Reuters. (2025, September 22). Nvidia to invest up to $100 billion in OpenAI, linking two artificial intelligence titans.
https://www.reuters.com/business/nvidia-invest-100-billion-openai-2025-09-22

[O7] OpenAI. (2025, September 22). OpenAI and NVIDIA announce strategic partnership to deploy 10 gigawatts of NVIDIA systems.
https://openai.com/index/openai-nvidia-systems-partnership

[O8] AMD. (2025, October 6). AMD and OpenAI Announce Strategic Partnership to Deploy 6 Gigawatts of AMD GPUs.
https://www.amd.com/en/newsroom/press-releases/2025-10-6-amd-and-openai-announce-strategic-partnership-to-d.html

[O9] Reuters. (2025, October 13). OpenAI taps Broadcom to build its first AI processor in latest chip deal.
https://www.reuters.com/business/openai-taps-broadcom-build-its-first-ai-processor-latest-chip-deal-2025-10-13

[O10] OpenAI. (2025, October 13). OpenAI and Broadcom announce strategic collaboration to deploy 10 gigawatts of OpenAI-designed AI accelerators.
https://openai.com/index/openai-and-broadcom-announce-strategic-collaboration

[O11] Reuters. (2026, April 28). OpenAI falls short of revenue and user targets as it races toward IPO, WSJ reports.
https://www.reuters.com/business/openai-falls-short-revenue-user-targets-it-races-toward-ipo-wsj-reports-2026-04-28

[O12] Data Center Dynamics. (2026, May 21). OpenAI must find $207bn to meet AI data center spending commitments - HSBC.
https://www.datacenterdynamics.com/en/news/openai-must-find-207bn-to-meet-ai-data-center-spending-commitments-hsbc

[O14] Bloomberg. (2025, October 7). OpenAI’s Nvidia, AMD deals boost $1 trillion AI boom with circular deals.
https://www.bloomberg.com/news/features/2025-10-07/openai-s-nvidia-amd-deals-boost-1-trillion-ai-boom-with-circular-deals https://www.bloomberg.com/news/videos/2025-10-08/openai-nvidia-fuel-1t-ai-market-with-circular-deals-video

Flash Memory, Enterprise SSD, and AI Storage Bottlenecks

[F3] TrendForce. (2026, January 5). Memory Makers Prioritize Server Applications, Driving Across-the-Board Price Increases in 1Q26, Says TrendForce.
https://www.trendforce.com/presscenter/news/20260105-12860.html

[F4] TrendForce. (2026, May 25). Combined Revenue of Top Five Global NAND Flash Suppliers Rose by 83.7% QoQ for 1Q26 as Supply Shortages Drove Price Hikes.
https://www.trendforce.com/presscenter/news/20260525-13058.html

[F5] TrendForce. (2026, June 11). AI Agent Boom Triggers Enterprise SSD Supply Crunch; Top Five Enterprise SSD Brands Post Record US$18.46 Billion Revenue in 1Q26, Says TrendForce.
https://www.trendforce.com/presscenter/news/20260611-13092.html

[F8] NVIDIA. (2025, March 18). NVIDIA and Storage Industry Leaders Unveil New Class of Enterprise Infrastructure for the Age of AI.
https://nvidianews.nvidia.com/news/nvidia-and-storage-industry-leaders-unveil-new-class-of-enterprise-infrastructure-for-the-age-of-ai

[F9] NVIDIA. (2026, March 16). NVIDIA Launches BlueField-4 STX Storage Architecture With Broad Industry Adoption.
http://nvidianews.nvidia.com/news/nvidia-launches-bluefield-4-stx-storage-architecture-with-broad-industry-adoption

[F10] NVIDIA. (2026). NVIDIA CMX Context Memory Storage Platform.
https://www.nvidia.com/en-us/data-center/ai-storage/cmx

[F11] Reuters. (2025, November 19). Nvidia’s strong forecast calms AI bubble jitters, for now.
https://www.reuters.com/world/china/ai-leader-nvidia-forecasts-fourth-quarter-revenue-above-estimates-2025-11-19

[F12] Reuters. (2025, December 12). Oracle-Broadcom one-two punch hits AI trade, but investor sentiment remains bullish.
https://www.reuters.com/business/finance/oracles-stumble-hits-ai-trade-many-remain-bullish-2025-12-12

[F13] Reuters. (2025, October 20). Chip crunch: how the AI boom is stoking prices of less trendy memory.
https://www.reuters.com/world/china/chip-crunch-how-ai-boom-is-stoking-prices-less-trendy-memory-2025-10-20

[F14] Reuters. (2025, November 14). Samsung hikes memory chip prices by up to 60% as shortage worsens, sources say.
https://www.reuters.com/world/china/samsung-hikes-memory-chip-prices-by-up-60-shortage-worsens-sources-say-2025-11-14

[F15] Reuters. (2025, December 3). The AI frenzy is driving a memory chip supply crisis.
https://www.reuters.com/world/china/ai-frenzy-is-driving-new-global-supply-chain-crisis-2025-12-03

[F16] TrendForce. (2025, December 5). Enterprise SSD Prices and Shipments Surge in 3Q25, Industry Revenue Climbs 28%, Says TrendForce.
https://www.trendforce.com/presscenter/news/20251205-12819.html

CoreWeave

[CW1] Bloomberg. (2026, June 9). CoreWeave Founders Have Dumped $2.3 Billion in Stock Since IPO.
https://www.bloomberg.com/news/articles/2026-06-09/coreweave-founders-have-dumped-2-3-billion-in-stock-since-ipo

[CW2] Yahoo Finance / GuruFocus. (2026). CoreWeave Founders Dump Billions Amid Explosive Stock Rally.
https://finance.yahoo.com/markets/stocks/articles/coreweave-founders-dump-billions-amid-195320772.html

[CW3] CoreWeave, Inc. (2025, March 20). Form S-1/A Registration Statement. U.S. Securities and Exchange Commission.
https://www.sec.gov/Archives/edgar/data/1769628/000119312525058309/d899798ds1a.htm

SpaceX, xAI, and the Nasdaq Index Mechanism

[X1] Space Exploration Technologies Corp. (2026). Form S-1 Registration Statement. U.S. Securities and Exchange Commission.
https://www.sec.gov/Archives/edgar/data/1181412/000162828026036936/spaceexplorationtechnologi.htm

[X2] Reuters. (2026, May 15). Exclusive: SpaceX accelerates IPO timeline, targets June 12 listing on Nasdaq, sources say.
https://www.reuters.com/world/spacex-accelerates-ipo-timeline-targets-june-11-pricing-nasdaq-2026-05-15

[X3] Nasdaq Global Indexes. (2026, May). Nasdaq-100 Index® Methodology Changes Frequently Asked Questions.
https://indexes.nasdaqomx.com/docs/2026_May_NDX_Changes_FAQ.pdf

[X4] Bloomberg. (2026, June 9). SpaceX IPO Risks Feedback Loop as Index Funds Eye 30% of Float.
https://www.bloomberg.com/news/articles/2026-06-09/spacex-ipo-risks-feedback-loop-as-index-funds-eye-30-of-float

[X5] Reuters. (2026, June 12). SpaceX shares surge after world’s biggest IPO, sending its value above $2 trillion.
https://www.reuters.com/world/us/live-elon-musks-spacex-set-stock-market-trading-after-worlds-biggest-ipo-2026-06-12

[X6] Reuters. (2026, June 11). Musk’s SpaceX prices record $75 billion IPO at $135 a share.
https://www.reuters.com/world/musks-spacex-prices-record-75-billion-ipo-135-share-2026-06-11

[X7] Bloomberg. (2026, May 21). SpaceX Joins Battle for Control of $26.5 Trillion AI Market.
https://www.bloomberg.com/news/articles/2026-05-21/spacex-challenges-ai-rivals-for-control-of-26-5-trillion-market

[X8] Reuters Breakingviews. (2026, April 24). SpaceX’s market claims are planet-scale absurdity.
https://www.reuters.com/commentary/breakingviews/spacexs-market-claims-are-planet-scale-absurdity-2026-04-24

Alphabet, Anthropic, and the IPO Wave

[A1] Alphabet. (2026, June 1). Alphabet Announces Proposed $80 Billion Equity Capital Raise to Expand AI Infrastructure and Compute.
https://abc.xyz/investor/news/news-details/2026/Alphabet-Announces-Proposed-80-Billion-Equity-Capital-Raise-to-Expand-AI-Infrastructure-and-Compute-2026-b0myAMewCa/default.aspx

[A2] Alphabet. (2026, June 3). Alphabet Announces Upsize and Pricing of $84.75 Billion Equity Capital Raise to Expand AI Infrastructure and Compute.
https://abc.xyz/investor/news/news-details/2026/Alphabet-Announces-Upsize-and-Pricing-of-84-75-Billion-Equity-Capital-Raise-to-Expand-AI-Infrastructure—and-Compute-2026-QzN3D9yMAj/default.aspx

[H1] Anthropic. (2026, June 1). Anthropic confidentially submits draft S-1 to the SEC.
https://www.anthropic.com/news/confidential-draft-s1-sec

[H2] Anthropic Institute. (2026). When AI builds itself.
https://www.anthropic.com/institute/recursive-self-improvement

[I1] OpenAI. (2026, June 8). Confidential submission of draft S-1 to the SEC.
https://openai.com/index/openai-submits-confidential-s-1

Financial Bubble and IPO Cycle Theory

[B1] Minsky, H. P. (1986). Stabilizing an Unstable Economy. Yale University Press.

[B2] Kindleberger, C. P., & Aliber, R. Z. (2011). Manias, Panics, and Crashes: A History of Financial Crises (6th ed.). Palgrave Macmillan.

[B3] Shiller, R. J. (2015). Irrational Exuberance. Princeton University Press.

[B4] Ritter, J. R., & Welch, I. (2002). A Review of IPO Activity, Pricing, and Allocations. The Journal of Finance, 57(4), 1795–1828.

---