Wall Street just threw the biggest coming-out party in stock market history. Anthropic priced its IPO at $83 a share on November 18, above its planned range, and raised more than $81 billion ahead of its Nasdaq debut under the ticker ANT, according to CNBC. For AI bulls, it looked like fresh proof that the boom still has plenty of fuel left.
Michael Burry is not buying it.
According to StockTwits, the “Big Short” investor told readers of his Substack that he would pass on the offering and skip the shares once they start trading, adding that the deal “is pumping futures for now.”
His bigger worry lies beneath the celebration, and it has as much to do with Nvidia as with Anthropic.
Michael Burry may let the AI rally run before shorting it
Michael Burry said he would look to short any rally in AI stocks, but he is in no hurry to do it. He pointed out that bear market rallies averaged about 22% in both the 1930s and the 2000s. A hint that he may let prices climb for a while before he leans in.
Also read: Wall Street is watching a troubling move in Treasury yields
He has already been laying the groundwork. Burry doubled his put position on the Nasdaq-100 in a November 12 Substack post, using contracts that expire in early 2027 with a strike roughly 20% below where the index traded when he bought. The index would need a deep slide for those puts to pay off at expiration.
Burry sees the recent bounce as a trap. He compared the Nasdaq-100’s October 6 high to the March 2000 peak that came right before the dot-com crash, and he bought back the slice of his VanEck Semiconductor ETF puts that he had trimmed on November 5, according to TheStreet. He expects a significant correction in chip and tech stocks before mid-2027.
That patience lines up with how he has talked about Nvidia lately. In a November 10 post, Burry said he was “not tempted as a long-term hold from here,” but he suspected the market would soon see the stock as “a heck of a trade.” In other words, he can see a bounce coming without believing in the story behind it.
Why smaller AI Models worry Burry
The sharpest part of Burry’s latest reader Q&A came from someone who runs a law firm. Burry agreed that AI tools capable of reviewing years of filings are on the way, but he argued that businesses could run smaller models on their own machines using their own records, and that compression could fit those models onto smaller devices.
That shift runs straight counter to the way today’s AI giants make money. “This works against the Anthropic model and the Nvidia model,” Burry wrote, adding that brute-force use of resources is not the way forward, as reported by Stocktwist. He also pointed to rising inference costs in China as a sign of how expensive running AI in the U.S. could become.
The stakes are enormous for Anthropic. Its offering valued the company at about $1.5 trillion, a price that rests on businesses continuing to pay for access to its large models. If firms can achieve good-enough results with compact models run in-house, that growth path becomes much harder to defend.
Nvidia faces a similar test. The chipmaker’s quarterly revenue more than doubled from a year earlier to $115.8 billion in its first report since Jensen Huang resigned as CEO on November 12. Those sales lean heavily on massive data centers, which is exactly the kind of spending Burry thinks smaller, local models could eventually chip away at.
Tencent’s cash burn kept Burry on the sidelines
Burry’s doubts about AI spending reach well beyond the U.S. Tencent made up just over 2% of his portfolio earlier this month, but the company’s negative quarterly free cash flow of 69.1 billion yuan kept him from adding shares.
The spending behind that cash drain is hard to ignore. Tencent’s capital expenditure jumped more than sixfold from a year earlier as it expanded its computing capacity, while third-quarter revenue grew 8% but still missed market expectations, as reported by the South China Morning Post.
For Burry, the problem is less about Tencent itself and more about the cycle it represents. He said he was unsure when hyperscaler spending would stop or slow if the AI bulls turn out to be right. That kind of uncertainty makes it tough to value companies pouring cash into data centers ahead of proven demand.
More Wall Street:
- Arista Networks stock gets fresh backing from Jim Cramer and Wall Street
- Oracle stock sinks as Wall Street confronts a $20 billion AI problem
- Wall Street just got a candid look at who is winning the AI race
The same race shows up on Nvidia’s side of the ledger. The chipmaker’s outlook for its current quarter assumes no data center compute revenue from China, according to Fox Business. That leaves its forecast resting on customers outside China, the very buyers Burry believes could slow their orders.
Private credit buyers may be early
Burry also weighed in on private credit, one of the market’s sore spots this year. He said he had heard capital was already moving into depressed private credit, but he warned that the trend could become a self-fulfilling prophecy and that today’s buyers may prove early.
The strain in that market has been building for months. In April, Blue Owl capped withdrawals from two of its largest private credit funds after investors sought to withdraw roughly $5.4 billion, citing concerns about AI disruption at software companies.
Burry’s point is that the real test may still be ahead. He has argued that leverage across the AI system runs as high as 76 times and that new investments could turn tentative or end in write-offs. If the bust proves large enough at the data center level, he believes today’s bargain hunters in private credit could simply be overwhelmed.
Related: Michael Burry’s latest move puts the Nasdaq 100 under scrutiny


















