# AI infrastructure stocks — X 热门讨论 (2026-09-26 06:24 UTC)
## @HeroOkMan (OkMan) · 09-25 14:07 · ♥30 ↻2 💬4 Michael Burry spent two years watching his own investors send him hostile letters, demand their money back, and threaten lawsuits over a bet he refused to unwind. In November 2025, the same man closed his hedge fund again, this time after warning that artificial intelligence stocks were the next bubble.
Burry was a former neurologist, diagnosed with Asperger's as an adult, running a small hedge fund called Scion Capital out of California. Around 2005, reading mortgage bond prospectuses line by line, a habit almost no one else on Wall Street bothered with, he became convinced the American housing market was built on loans that couldn't possibly be repaid. He wanted to bet against it, but in 2005 there was no simple way to short a mortgage bond. So he invented one, persuading Goldman Sachs and other banks to sell him credit default swaps, insurance-like contracts that would pay out if subprime mortgage bonds failed, something nobody had really built a retail-sized market for before.
For nearly two years, the trade did nothing but bleed money. Mark-to-market losses piled up every quarter while the underlying mortgages hadn't actually defaulted yet. Investors who had trusted him with their capital called him reckless. Some tried to force him to exit. Burry refused, spending those years re-reading his own analysis, checking for the mistake he was sure everyone else could see and he couldn't.
There was no mistake. By 2007 the subprime bonds started failing exactly as his spreadsheets predicted. Scion's position exploded in value. Between November 2000 and June 2008, the fund returned 489%. Burry personally made roughly $100 million. His investors, the same ones who had written him angry letters two years earlier, made an estimated $700 million.
He closed Scion in 2008, calling the years of isolation and investor hostility not worth repeating.
Seventeen years later, in an October 27, 2025 letter to investors, Burry wrote that his estimation of value in securities "is not now, and has not been for some time, in sync with the markets," and wound down his fund again, this time after placing roughly $1.1 billion in bearish options positions against Palantir and Nvidia. His reasoning echoed 2005 almost exactly: he argues that massive capital flows into passive index funds and AI infrastructure spending are distorting prices the same way CDO purchases once distorted subprime mortgage bonds, without anyone checking the underlying value.
As he put it himself: sometimes we see bubbles. Sometimes there's something to do about it. Sometimes the only winning move is not to play. > 引用 @HeroOkMan: HOW $100 BECAME $1,340,000: THE FIVE MECHANISMS MONEY HAS EVER BEEN MADE THROUGH https://x.com/HeroOkMan/status/2103486447244214364