Michael Burry, the investor made famous by The Big Short, has escalated his warnings about the artificial intelligence boom, saying this time that a severe market downturn that prevents OpenAI and Anthropic from going public would be beneficial to humanity.

“For the benefit of humanity, the markets should tank hard and prevent the OpenAI and Anthropic IPOs,” Burry wrote Tuesday on X.

Burry went further in replies to his post, saying that the two AI companies could “suck up” and ultimately destroy “TRILLIONS of dollars of capital,” adding that the financial damage would be “the least of the damage they do.”

He also responded “Along those lines” to a user who joked about crashing the market so that “Skynet cant IPO,” invoking the fictional AI system from the Terminator films.

The comments extend a bearish argument Burry has been developing around the economics of frontier AI. His concern is not simply that AI stocks have become expensive. He argues that the industry is committing enormous amounts of capital to models, chips and data centers before the returns from that spending have been established.

That thesis puts the planned public-market debuts of OpenAI and Anthropic at the center of the debate.

Both companies require enormous amounts of capital to develop more capable models and secure the computing infrastructure needed to operate them. An IPO would provide access to a much broader pool of investors and potentially give both companies additional financing capacity as their infrastructure requirements expand.

OpenAI CEO Sam Altman said earlier this month that taking the company public in 2026 would be “ill-advised” amid concerns surrounding AI safety. Anthropic, meanwhile, is expected to pursue an IPO after the November midterm elections, according to a prospectus reviewed by Reuters.

Burry’s latest comments follow an earlier post this week in which he said he was “more confident than ever” that his bearish AI thesis would play out over the next year. He previously identified 2028 as his base-case timeline.

He said the change in his timeline prompted him to adjust his positions against several companies and indexes, including Nvidia, Palantir, Micron, Oracle and the Nasdaq 100.

At the heart of Burry’s position is the scale of investment required to build the AI ecosystem.

Technology companies and their infrastructure partners have committed enormous sums to data centers, processors, networking equipment and electricity capacity. Burry has argued that much of that expansion is increasingly supported by debt and complex financing arrangements.

“If the spending stops or slows, it all comes apart,” he wrote earlier this week.

His argument rests on the assumption that AI infrastructure spending must continue at exceptionally high levels to support current valuations and growth expectations. If companies eventually reduce capital expenditure, the consequences could extend beyond individual AI developers to chipmakers, cloud providers, data-center operators and lenders that have financed the buildout.

Higher interest rates would add another pressure point by increasing the cost of financing the infrastructure boom.

If that happens, there will be a more consequential risk than a conventional correction in AI stocks. A sharp decline in valuations could make it more difficult for private AI companies to raise capital on favorable terms, while a broader market selloff could reduce the appetite of public investors to finance companies with enormous infrastructure requirements and uncertain long-term profitability.

Burry has previously pointed to debt-fueled spending on chips and data centers as a key vulnerability. He has also questioned whether accounting practices, stock-based compensation and financing arrangements across the AI industry may make underlying economics look stronger than they are.

Those arguments remain Burry’s investment thesis rather than evidence that OpenAI or Anthropic will fail or that an AI market collapse is imminent.

The broader market has so far provided little confirmation of an imminent breakdown. AI-related companies have continued to attract capital and have been among the major drivers of equity-market gains this year. That resilience creates a significant test for Burry’s thesis. The AI industry has continued to raise money, expand computing capacity and generate growing demand even as investors debate whether the resulting valuations can ultimately be supported by cash flows.

Therefore, the IPO question carries implications beyond simply accessing public capital for OpenAI and Anthropic. A listing would expose their financial performance, infrastructure commitments and customer economics to a much broader investor base at a time when the sustainability of AI spending is already under intense scrutiny.

Burry’s position is that a market collapse before those offerings reach investors would prevent additional capital from flowing into what he considers an unsustainable investment cycle. His warning is therefore less a prediction about the mechanics of an IPO than an argument about whether the financial system should continue supplying capital to frontier AI at its current scale.

While the market continues to finance the expansion, Burry is betting that the cost of that expansion will eventually become too large to ignore.

The U.S. Federal Reserve on Wednesday finalized changes to its annual stress tests for large banks, moving to make the closely watched examinations more transparent and reduce sharp year-to-year swings in the capital requirements imposed on major lenders.

The changes largely mirror proposals released by the central bank a year ago after years of complaints from the banking industry that the stress tests relied on opaque models and subjective judgments. The Fed’s revised framework seeks to give banks and investors greater visibility into how the tests are designed while preserving the role of the examinations in determining how much capital banks need to withstand severe economic losses.

The stress tests are a central component of U.S. bank regulation. Each year, the Fed subjects large lenders to hypothetical downturns involving scenarios such as recessions, market shocks, falling asset prices, and rising credit losses. The results help determine a bank’s “stress capital buffer,” an additional capital requirement that sits on top of regulatory minimums.

Under the new framework, the Fed will seek public feedback before making major changes to the models used in the tests or to the hypothetical economic scenarios used to assess banks. That marks a significant change in the way the central bank approaches a process that has historically given banks limited visibility into how the models and scenarios evolve. Greater consultation could make it easier for banks, investors and other market participants to understand why capital requirements change from year to year.

The Fed is also changing the way it calculates the stress capital buffer. Rather than basing the requirement on a single year’s stress-test result, the central bank will use the average of a bank’s two most recent results.

The objective is to reduce volatility in capital requirements without materially changing the overall amount of capital held by the banking system.

The Fed estimates that the changes will cut year-over-year volatility in capital requirements by about 50%, while leaving aggregate bank capital levels broadly unchanged.

The overhaul is not primarily a move to reduce the amount of capital banks collectively need to hold. Instead, it is designed to make the capital framework more stable and predictable, reducing the possibility that a particularly severe stress scenario in one year could suddenly force a bank to hold substantially more capital.

For banks, greater predictability could make capital planning easier. Lenders use their expected capital requirements when deciding how much money they can return to shareholders through dividends and stock buybacks, as well as how much they can deploy into loans and other investments.

Large swings in stress capital buffers can therefore affect decisions well beyond the stress test itself.

The changes also address one of the banking industry’s longstanding objections to the Fed’s process: uncertainty over the models used to translate hypothetical economic shocks into estimated losses.

Because banks cannot fully anticipate how the Fed’s models will behave under different scenarios, even lenders with similar balance sheets can face different capital outcomes. Requiring public feedback on major model and scenario changes gives the industry and other participants a formal opportunity to scrutinize those assumptions before they influence capital requirements.

The Fed’s decision comes as regulators continue to face pressure over how bank capital rules should evolve. Higher capital requirements can provide a larger cushion against losses, but they can also affect the amount of money banks have available for lending and shareholder distributions.

The stress tests have become necessary because they effectively link a bank’s capital planning to the regulator’s assessment of its resilience under hypothetical conditions. A methodology that changes sharply from year to year can yield uncertainty even when a bank’s underlying financial position has not changed by the same magnitude.

The two-year averaging mechanism addresses part of that problem by smoothing the impact of unusually high or low test results. It also means that a single year’s deterioration in a bank’s stress-test performance will have a more gradual effect on its capital requirement, while an improvement will similarly take longer to flow through completely. That could make the system less sensitive to individual annual results.

For investors, the revised framework may make comparisons across years easier because capital requirements should be less susceptible to sudden changes driven by the testing methodology itself.

The Fed’s estimate that aggregate capital levels will not materially change also signals that the central bank is seeking greater stability without fundamentally altering the amount of loss-absorbing capacity in the banking system.

The changes are therefore less about lowering the regulatory capital burden than changing how that burden is calculated and communicated. The central bank’s decision also places greater emphasis on transparency. By inviting public feedback on major changes to stress-test models and economic scenarios, the Fed is effectively acknowledging that the credibility of the exercise depends not only on the results but also on confidence in the methodology behind them.

For the largest U.S. banks, the practical effect will be a stress-testing system in which capital requirements are expected to move more gradually and with greater visibility into the assumptions driving those changes.

The Fed will retain the stress tests as a key safeguard against losses at major lenders, while attempting to address concerns that unpredictable methodology changes can themselves create unnecessary volatility in banks’ capital planning.

China has warned the European Union that it will take a “resolute response” to protect Chinese industries if Brussels moves ahead with new restrictions on Chinese companies or products, raising the risk of another escalation in trade tensions between the two economies.

China’s Ministry of Commerce said the reported EU initiative amounted to a “typical protectionist and unilateralist measure” that could disrupt the stability of China-EU trade and broader industrial supply chains.

The warning came after reports that several EU member states are preparing a joint document calling on the European Commission to accelerate the development of a new instrument for responding to what the bloc considers unfair trade practices, particularly those involving China.

The reported proposal would give the EU a mechanism resembling the United States’ Section 301 trade authority, which Washington uses to investigate foreign trade practices and impose measures in response to what it determines to be unfair conduct.

Rhodium Group senior adviser Noah Barkin reported that France and Germany are finalizing a joint paper calling for Brussels to fast-track such an instrument. According to Barkin, the document is intended to give the European Commission greater powers to restrict China’s access to the European market.

Beijing’s response indicates that the proposal is being viewed not simply as another European trade-defense measure but as a potential expansion of the EU’s ability to impose restrictions on Chinese companies.

“If the European side engages in talks and consultations with China, while increasingly pressuring China, it will seriously undermine mutual trust, interfere with the overall process of consultation,” China’s Commerce Ministry said.

The ministry said China would defend the legitimate interests of its industries if the EU introduced discriminatory measures.

The timing of the warning is significant because Beijing and Brussels have been attempting to stabilize their trade relationship through a new consultation process.

China and the EU held their first Trade and Investment Consultations in Brussels in June, agreeing to work on four areas: trade and investment balancing, export controls, intellectual property rights and reform of the World Trade Organization. The two sides also agreed to hold another ministerial-level meeting in autumn.

Chinese officials subsequently said that more than 20 consultations had been held by the respective working teams as they sought to address trade concerns. Beijing has repeatedly called for negotiations based on what it describes as mutual respect and balanced treatment.

The reported Franco-German initiative introduces a harder-edged element into those discussions. Rather than relying only on existing WTO procedures and individual trade-defense investigations, the proposed mechanism could give Brussels a broader instrument for responding to foreign trade practices.

For China, the concern is that such a framework could eventually be applied across multiple industries and become a permanent part of the EU’s trade policy toward Chinese companies.

Beijing has already signaled that it has options for responding. State-backed Global Times reported that China’s potential “strong policy toolbox” could include anti-discrimination investigations, security probes into industrial and supply chains, and scrutiny of the effects of foreign subsidies. The newspaper cited an unnamed Chinese observer for the possible measures.

That means a European trade measure could generate pressure beyond tariffs. Investigations into European companies operating in China, supply chains and foreign subsidies could create additional costs for businesses on both sides.

The dispute comes as European governments increasingly grapple with the scale of China’s industrial capacity and the competitive pressure Chinese companies are creating in sectors ranging from electric vehicles and renewable energy to chemicals, telecommunications and other advanced manufacturing industries.

Germany’s powerful industrial sector has itself been debating how aggressively Europe should respond. The German Federation of Industries recently called for stronger tools against unfair competition and greater supply-chain diversification, while cautioning against broad protectionism.

The European debate is not just about restricting Chinese imports, as it also concerns whether existing EU trade instruments are sufficient to address state subsidies, excess industrial capacity and the growing presence of Chinese companies in European markets.

That creates a difficult balance for Brussels. European governments want to protect domestic industries and reduce vulnerabilities while maintaining access to Chinese markets and avoiding a broad trade confrontation.

The same tension is visible in telecommunications. The EU is considering giving operators more time to remove equipment from high-risk suppliers such as Huawei after companies warned that replacing the equipment could cost as much as €40 billion. The issue illustrates the economic cost that can accompany efforts to reduce dependence on Chinese technology.

The stakes are equally substantial for China. The EU remains an important trading partner, while Chinese manufacturers now depend on overseas markets as domestic competition intensifies in several industries.

A widening cycle of trade restrictions and retaliation could therefore hit companies on both sides at a time when Beijing and Brussels are formally trying to rebalance their relationship through negotiations.

The immediate issue is still only a reported policy proposal rather than an EU-wide restriction already in force. The Franco-German paper is expected to call for the Commission to develop the instrument, after which member states and EU institutions would still have to determine its scope and implementation. Barkin described the proposal as potentially significant because it would signal a stronger common position from Europe’s two largest economies toward Beijing.

China’s warning nevertheless shows that Beijing is already treating the proposed mechanism as a potential escalation.

The central risk for European and Chinese businesses is that a tool originally designed to strengthen Europe’s negotiating position could become the foundation for a broader cycle of reciprocal investigations and market restrictions. With formal China-EU consultations still underway, the dispute now adds another test of whether both sides can pursue tougher trade policies without allowing their economic relationship to slide into a wider confrontation.