Burry Is Wrong Again?
Why We’re Not Joining His Shorts on Nvidia and Palantir
Burry, the 2008 Hero… Now Fighting AI
Michael Burry earned his fame by calling the 2008 subprime crisis. That’s not up for debate. But in 2025–2026 he has picked a new villain: the “artificial intelligence bubble,” with short bets against Nvidia and Palantir that he has laid out in his Substack, Cassandra Unchained.
At Primus Pilus Capital, we respect his courage and independent thinking. It takes guts to short the market’s favorite names. But looking at today’s numbers, we believe this time he is reading the timing, the companies, and the data wrong. Again.
“This Is Another 2000”: Burry’s Story
Burry’s story is powerful and easy to remember:
Companies are allegedly overspending on data centers and AI chips out of fear of missing out, not because of genuine demand.The accounting on those assets is supposedly too generous, making reported earnings look better than they really are.
Nvidia is cast as the “new Cisco” of a bubble, and Palantir as a cult stock supported more by narrative than by business substance.
It’s a narrative that resonates with anyone who remembers 1999. But just because a story sounds familiar doesn’t make it true. The key difference between then and now is that today, AI is already deeply embedded in real business processes and real P&L, not just in pitch decks.
What We See at Primus Pilus Capital
Our starting point with Nvidia and Palantir is not “Is Burry right?” but something colder and simpler:
What are we paying today for these companies’ growth, and what are we getting in return?
That’s where we bring in a tool we like a lot: PEGY. It’s not magic, but it forces two very concrete questions:
Does the speed of earnings growth justify the multiple?
Are we paying a high price for something very good, or a crazy price for a promise that may never show up?
With that lens, the picture we see is quite different from Burry’s doomsday setup.
Nvidia: Expensive, Yes. But Not Absurd
For Nvidia, the essential facts are:
It trades at a high P/E.
Its earnings are growing at an extraordinary pace.
Its margins and returns on capital are in a different league compared with an average company.
When you cross price, growth, and profitability, you end up with a PEGY close to 1, which, in plain English, means:
It’s not cheap.
But it’s also not the cartoonish, “totally detached from reality” bubble that only lives off expectations.
Put simply: paying a high multiple for a company that prints cash, dominates a de facto standard, and grows earnings at high double‑digit rates is not the same as paying insane multiples for businesses with barely any revenue, like many .com names in 2000.
If Nvidia’s growth slows someday, the market will adjust. But that’s very different from justifying a big structural short here and now. The current numbers don’t support that kind of apocalypse trade.
Palantir: Lofty Expectations, Real Business
Palantir is a different story. Here, the market definitely fell in love hard:
The stock price bakes in many years of strong growth.
The P/E is very demanding.
When we run PEGY, the number shoots above 4, which tells us something important:
The market is paying not just for today’s business, but for a very optimistic vision of tomorrow’s business.
Still, that doesn’t automatically make it “smoke and mirrors.” Palantir:
Is growing revenue.
Is improving margins and profitability.
And is shifting from “that weird government contractor” to a software and AI platform used by private companies to solve very concrete problems.
At these prices, for us it’s a high‑expectation bet, not a zero in waiting. The risk is overpaying for something that does exist, not discovering there was nothing there.
The PEGY Block, at a Glance
Here’s how we frame it at Primus Pilus Capital, without unnecessary jargon:
Nvidia (NVDA)
High valuation.
Earnings growing very fast.
PEGY close to 1 → expensive, but consistent with what it delivers.
Palantir (PLTR)
Very demanding valuation.
Strong growth and improving margins.
PEGY clearly above 1 → expectation risk, not a hollow model.
With this snapshot, calling both names a “total bubble” is more slogan than analysis.
Burry’s Real Problem: Timing and Crash Obsession
The most worrying thing about Burry’s thesis is not that he sees excesses (they always exist), but how he chooses to play them: big and against the grain.
In the last few years, we’ve already seen him:
Go heavily short indices while the market kept grinding higher.
Blast out a broad “sell” call that he later had to walk back publicly when equities did the exact opposite.
And now, bet aggressively against Nvidia and Palantir precisely while their businesses show traction, renewed contracts, and customers paying for what they use.
Yes, at some point the AI frenzy will cool down. They all do. But there is a crucial gap between 2000 and 2026: back then, many business models were still hypothetical; today, a big chunk of AI is already embedded into tools we use daily and that companies monetize in very concrete ways.
Burry is essentially betting that all of this is an accounting illusion. Our numbers tell a different story.