Summary

Burry’s Palantir put position was roughly two orders of magnitude smaller than the “$1 billion” reported on CNBC. He bought about 50,000 put options struck at $50, two years out, on a roughly $200 stock he thinks is worth $30 or less. Because the options were worth less than $2 apiece, he said the position was $10 million—not $1 billion; the press multiplied the underlying shares by the stock price. “What’s interesting is that they don’t do this for anybody else.”
His Palantir thesis: a consulting-heavy software firm put an AI cover on its applications. Stock-based compensation eats essentially all income; measure the true cost by the buybacks needed to offset dilution and deduct that from cash flow, and “historically they don’t make anything.” His tell: five billionaires out of roughly $4 billion of revenue, “the billionaires-to-revenue ratio was greater than one, and I’d never seen that before.” IBM runs a bigger, basically similar business that was growing about as fast, without a Palantir valuation.
His timing framework maps AI onto the dot-com build-out, which he calls “not really a dot-com bubble. It was a data-transmission bubble.” Cisco grew revenues 55% in 2000 and 17% in 2001 even as the NASDAQ peaked on March 10, 2000; in prior manias, the stock-market peak came “before you were even halfway done” with capital expenditures, usually before capex peaked. Net investment over nominal GDP is now at shale-revolution levels and near the dot-com peak, while a company’s market cap can rise about $3 for every $1 of announced AI capex, as with Oracle. “I thought two years would be enough.”
He shut his fund, deregistered, and now runs mostly his own money because passive ownership has changed the crash mechanics. Over 50% of money is passive and under 10% is actively managed by people thinking about stocks long term; unlike 2000, when ignored stocks rallied through the NASDAQ crash, “now I think the whole thing’s just gonna come down,” possibly in a longer bear market “more akin to 2000.” He also did not want to repeat the investor experience. His advice: buy out-of-favor health care stocks and sell anything that has risen sharply and looks overvalued.
On Berkshire’s Google buy, he’s skeptical that AI helps Google’s golden goose. Google Search worked because roughly 85% of searches are non-monetizable but cost “infinitesimal fractions of a cent”; his own AI queries “cost tens of dollars just for one inquiry.” Free-tier LLMs are already massively penetrated and “gonna be commoditized”—the money is in the developer space, not consumers who “won’t ever have to” pay.
Macro: he won’t bet against the United States’ ability to find a way through its debt problems (“waiting for Castro to die… is not a strategy”), would abolish the Fed, and calls Bitcoin at $100,000 “the tulip bulb of our time.” The math is grim—$4.5 trillion in individual taxes, $400 billion in corporate taxes, and $1 trillion in annual interest—but “it’s the United States.” The neutral rate is “probably around 4%,” so cutting now punishes savers and could steepen the curve. He’s held gold since 2005; Bitcoin is “worse than a tulip bulb” because of the criminal activity it enables.
The Big Short coda: nobody ever apologized. Investors were “generally mad at me even when things went well”; after the payout, it was “‘Ooh, we don’t wanna go through that again.’” He gave Lewis his email archive defensively—“I wanted to make sure you had everything… I didn’t do anything wrong”—and credits those emails as “the main reason I didn’t get sued.”

Deep dive

1. Lewis’s reluctant-interview setup: a 13F release made “laying low” impossible

Burry had earlier said no to the podcast, then briefly said he would like to help before reversing himself. His SEC-mandated 13F revealed big put positions against Palantir and NVIDIA, and the news “exploded… on Twitter, on CNBC” for 48 hours. Lewis’s logic for asking again: “He can’t lay low,” so “what’s the point?” of refusing.
Lewis’s primer on the original trade: credit default swaps on subprime mortgage bonds were not available, so Burry “had to help Wall Street create it for him”—like buying fire insurance on someone else’s house. Every other character in The Big Short then used the instrument he helped create.

2. What made the Big Short unique—and why his own investors never forgave him

Burry could buy relatively inexpensive insurance on incredibly illiquid bonds without owning them, when “nobody thought this could happen.” By late 2005, Goldman Sachs called asking, “What are you doing? You’re the only person we know.” Crucially, it was “the once-in-a-century opportunity to actually say, ‘I know when this is gonna happen’”—versus the 1990s bubble, where “there was no telling when that would end” and shorting was “just a high-risk endeavor.”
The human aftermath, unvarnished: investors “were generally mad at me even when things went well,” only one invested during the last year and a half, and after the win nobody ever called to apologize. “I didn’t expect it. It’s Wall Street.” He reopened in 2013 deliberately small—below the SEC investment-adviser registration threshold, with no marketing and only investors he knew—when “if I wanted to, I could have raised billions.”
His email archive, which he handed Lewis for the book, was self-protection: “those emails, I think, were the main reason I didn’t get sued by my investors… it was very clear where we stood with everybody.”

3. The Palantir short: notional distortion, stock-comp accounting, and the “luckiest companies on the planet”

The reporting mechanics Burry wants corrected: 50,000 puts struck at $50, two years out, on a roughly $200 stock he thinks “is worth $30 or less.” The options were worth less than $2 each, making the position about $10 million, but it was reported as a billion-dollar short because the press multiplied the underlying shares by the spot price. The same distortion hit his index hedges: “he’s shorting a billion and a half of the S&P 500.” Lewis compared Alex Karp’s attack on Burry with John Mack blaming short sellers during the financial crisis, adding that it is “always a really bad sign when people start going after the short sellers.”
The fundamental case: expensive-to-install applications sold with consulting; government revenue fell from a majority to “more even”; and C-suites are scrambling because they “feel under the gun to AI something.” Strip out stock-based compensation—using buybacks that offset dilution and deducting them from cash flow—and “historically they don’t make anything.” The hook that drew him in: five billionaires from roughly $4 billion of revenue, a billionaires-to-revenue ratio above one.
The framing worth keeping: “Palantir and NVIDIA are the two luckiest companies on the planet. Neither produced a product for AI.” NVIDIA got lucky twice—GPUs happened to fit crypto mining, then AI; Palantir “put an AI cover on their applications” after ChatGPT, “but that’s what every company is doing now.”

4. Timing the bubble: capex manias peak in the market before the spending peaks

The historical template: the dot-com era was “a data-transmission bubble”—fiber needed routers, and routers needed fiber. Cisco’s revenue grew 55% in 2000 and 17% in 2001 because investment continued after the market top. Burry says the investment peaked later; charting net investment—capex less depreciation—against nominal GDP produces “these nice mounds of investment manias.” In prior examples, the market peaked before capex was halfway done, and usually before capex itself peaked.
Where we are: at shale-revolution levels relative to GDP and near the dot-com peak, in the phase where announcing a dollar of AI capex can add roughly $3 of market capitalization. Oracle rose 40%, and Ellison was briefly the richest man, on bookings for a massive plan it would still have to spend to build. Burry’s hedge is deliberately qualified: “I can’t say because it hasn’t happened fully yet,” but “I thought two years would be enough.” His advice: buy out-of-favor health care stocks; sell overvalued holdings that are shooting straight up.
Why he would rather run mostly his own money now: over 50% is passive and under 10% is actively managed by people genuinely thinking long term, leaving no 2000-style pocket of ignored stocks to hide in—“the whole thing’s just gonna come down,” and being long U.S. stocks while protecting yourself “will be very hard.” He closed the fund, did not want to repeat the experience with investors, and “put on all the positions for myself right away—the same positions.”

5. Google’s golden goose, the un-shortable Treasury, and abolishing the Fed

On Berkshire buying Google: “We don’t know that that is Buffett,” and Google is merely “the value investor’s favorite in that group.” His worry is unit economics: about 85% of searches are non-monetizable and only worked at “infinitesimal fractions of a cent,” while his own AI queries cost tens of dollars each. Search is “basically all their cash flow.” Unlike the slow internet-penetration wave Amazon rode, free LLMs are already “massively penetrated”; consumers “won’t ever have to” pay, so LLMs get commoditized and the money migrates to developers.
On a U.S. debt crisis: “waiting for Castro to die. It’s not a strategy.” The arithmetic—$4.5 trillion from individuals, $400 billion from corporations, $1 trillion of annual interest, and a thinner social cushion—is “ridiculous,” but the reserve currency means “betting that they can’t find a way is not something I’d wanna do anytime soon.”
His “sick view” on Fed independence: “when Trump starts running the Fed, it might become the end of the Fed because… everybody’s gonna hate it, not just me.” He would replace it with a Treasury department making those decisions—“they’re almost the same department already”—argues the neutral rate is probably around 4%, and warns rate cuts “kill all the savers” and could steepen the curve. Refuge: gold since 2005, not Bitcoin—“It’s not worth anything… It’s the tulip bulb of our time,” and worse because of the criminal activity it enables.

Bitcoin Reclaims $90k: Have We Bottomed?How a16z Growth Invests