Michael Burry, the hedge fund manager who made a fortune predicting the 2008 housing collapse, has turned his contrarian gaze on artificial intelligence. His latest regulatory filings reveal a striking portfolio move: short positions against some of the most prominent names riding the AI wave, including Palantir Technologies and, perhaps more provocatively, a put-option bet against Anthropic — the company behind the Claude chatbot that has become one of OpenAI’s most credible competitors.
The man who inspired “The Big Short” is, once again, betting against the crowd.
According to a Business Insider report analyzing Scion Asset Management’s 13F filing with the Securities and Exchange Commission, Burry’s fund held put options on Palantir Technologies valued at approximately $9.1 million at the end of the first quarter of 2025. Put options give the holder the right to sell shares at a specified price, a classic bearish bet that profits when a stock declines. But the Palantir position wasn’t the only AI-adjacent wager. Burry also held puts on several other technology companies that have seen their valuations swell amid the artificial intelligence frenzy that has gripped markets since late 2022.
What caught the attention of Wall Street watchers most sharply was the apparent positioning against Anthropic. The San Francisco–based AI lab, founded by former OpenAI executives Dario and Daniela Amodei, has raised billions in funding and was recently valued at $61.5 billion. It isn’t publicly traded, which makes any short-adjacent bet more complex and harder to read from a standard SEC filing. The precise structure of Burry’s Anthropic-related position remains somewhat opaque, but the signal is unmistakable: Scion sees vulnerability where the rest of the market sees inevitability.
This matters. Not because Burry is always right — he isn’t, and his track record since the mortgage trade has been decidedly mixed — but because his willingness to stake capital against AI’s most celebrated companies crystallizes a debate that has been simmering for months beneath the surface of record stock prices and breathless earnings calls.
Palantir’s stock has been one of the most remarkable performers of the AI era. Shares surged more than 340% in 2024, driven by the company’s pivot to marketing its software platform as an AI-powered analytics engine for both government and commercial clients. CEO Alex Karp has positioned the Denver-based firm as the connective tissue between large language models and real-world enterprise decision-making. The stock trades at roughly 200 times trailing earnings — a valuation that prices in years, perhaps decades, of growth that hasn’t materialized yet.
Burry, evidently, thinks the market is getting ahead of itself.
He’s not alone in his skepticism, though few have been as willing to put money behind it. A growing chorus of analysts and institutional investors have begun questioning whether AI-linked valuations have detached from any reasonable projection of future cash flows. The comparison to the dot-com bubble of the late 1990s has become a staple of financial commentary, though the analogy is imperfect. Unlike many dot-com companies, today’s AI leaders — Nvidia, Microsoft, Alphabet, and increasingly Palantir and Anthropic — generate real revenue and in many cases substantial profits. The question isn’t whether AI is real. It’s whether the prices already reflect more optimism than the technology can deliver in the near term.
Palantir’s commercial revenue growth has been impressive but uneven. The company reported a 20% year-over-year increase in U.S. commercial revenue in its most recent quarter, with government contracts still accounting for the majority of its business. Its Artificial Intelligence Platform, or AIP, has attracted significant interest from corporate clients looking to deploy large language models within their own data environments. But converting pilot programs into sustained, high-margin recurring revenue is a different challenge entirely, and Palantir’s operating margins remain thin relative to the multiples investors are paying.
And then there’s the broader question of whether Palantir’s competitive moat is as deep as its stock price implies. The company competes with cloud hyperscalers — Amazon Web Services, Microsoft Azure, Google Cloud — that have vastly larger engineering teams, deeper customer relationships, and the ability to bundle AI tools with existing infrastructure contracts. Palantir’s advantage has historically been its willingness to embed engineers directly with clients and customize deployments, a labor-intensive model that doesn’t scale as easily as a pure software business.
Burry’s bet against Anthropic raises a different set of questions. The company has positioned itself as the “safety-first” alternative to OpenAI, emphasizing research into AI alignment and responsible development. Its Claude models have earned strong reviews from developers and enterprise customers. Amazon has committed up to $4 billion in investment. Google, Salesforce Ventures, and Spark Capital are among its other backers. By any conventional measure, Anthropic is one of the most well-capitalized and technically accomplished AI startups in history.
So what’s the bear case?
Start with economics. Training and running frontier AI models is extraordinarily expensive. Anthropic’s compute costs are staggering, and the company has yet to demonstrate a clear path to profitability. Revenue has been growing — reportedly surpassing $850 million in annualized run rate — but the gap between revenue and the capital required to stay competitive at the frontier remains wide. Every new model generation demands more data, more compute, more specialized talent. The arms race shows no signs of slowing, and the companies best positioned to win it may be the ones with the deepest pockets: Google, Microsoft, and Meta, all of which are building their own frontier models in-house.
There’s also the competitive dynamic with OpenAI, which has its own set of financial pressures but benefits from first-mover advantage, a massive consumer user base through ChatGPT, and a recently restructured for-profit entity that gives it more flexibility to raise capital. Anthropic’s path to a public offering or a sustainable standalone business is far from guaranteed, and Burry may be betting that the private-market valuation has overshot what the company can realistically deliver.
The timing of these bets is worth examining. Burry has a well-documented tendency to be early — sometimes painfully so. His mortgage short in the mid-2000s cost Scion’s investors months of losses before the trade paid off spectacularly. More recently, he took bearish positions against the S&P 500 and Nasdaq in 2023 that initially moved against him as the market rallied. He later appeared to close or reduce those positions. Being right about the direction but wrong about the timing can be just as costly as being wrong altogether, particularly with options that have expiration dates.
Still, the filing offers a window into how at least one prominent investor is processing the current moment. The AI trade has been remarkably one-directional. Nvidia’s market capitalization has swelled past $3 trillion. Microsoft has poured more than $13 billion into OpenAI. Venture capital firms have flooded AI startups with funding at valuations that would have seemed absurd three years ago. Corporate earnings calls are saturated with references to AI strategy, AI integration, AI transformation — often with little specificity about actual revenue impact.
Burry seems to be asking a simple question: What if the payoff takes longer than the market expects?
It’s a question with historical resonance. The internet did transform the global economy, but investors who bought Cisco, Intel, and Sun Microsystems at their 2000 peaks waited years — in some cases, forever — to recover their money. The technology was real. The valuations were not. Whether the same dynamic is playing out with AI remains the central tension in equity markets today.
Recent trading data suggests Burry isn’t the only one hedging. Short interest in Palantir has ticked up modestly in recent weeks, though it remains well below levels that would indicate widespread bearish sentiment. Options market activity shows increased demand for downside protection on several AI-linked names, a pattern consistent with institutional investors looking to hedge concentrated long positions rather than make outright directional bets.
Palantir, for its part, has continued to announce new contracts and partnerships. The company recently expanded its work with the U.S. Department of Defense and signed several new commercial deals in the healthcare and energy sectors. Karp has been vocal about what he sees as Palantir’s unique position at the intersection of AI capability and operational deployment, arguing that most organizations don’t need better models — they need better ways to use the models that already exist.
That argument has merit. But it also describes a services-heavy business model that may not justify a valuation typically reserved for high-growth, high-margin software platforms.
Anthropic’s trajectory will be shaped in large part by forces outside its control. Regulatory developments around AI safety could either bolster the company’s positioning or create compliance burdens that favor larger incumbents. The pace of model improvement matters too — if the gap between frontier models and open-source alternatives narrows, Anthropic’s differentiation becomes harder to sustain. And the relationship with Amazon, while financially significant, introduces dependency risk. If Amazon decides to build rather than buy, Anthropic loses its most important distribution channel.
Burry’s filing doesn’t tell us his full thesis. 13F disclosures are snapshots, not narratives. They show what a fund held at a specific point in time, not why, and not whether the positions have since been adjusted. Scion’s portfolio also included long positions in several value-oriented stocks, consistent with Burry’s longstanding preference for companies trading below intrinsic value. The short bets on AI names appear to be a hedge against what he perceives as speculative excess rather than a wholesale rejection of artificial intelligence as a technology.
That distinction matters. You can believe AI will reshape industries and still believe the stocks are overpriced. Those two ideas aren’t in conflict. They’re just uncomfortable to hold simultaneously in a market that has rewarded maximalist optimism and punished caution.
Whether Burry’s latest contrarian stance proves prescient or premature, it serves as a useful corrective to the prevailing narrative. The AI boom has minted enormous wealth and attracted extraordinary talent. It has also created a market environment where questioning the consensus requires real conviction — and real capital.
Michael Burry has both. Whether he has the timing right this time is another matter entirely.
WebProNews is an iEntry Publication