What Peter Thiel's Nvidia Exit Really Means!
And why this time might actually be different from every other bubble warning you've ignored
Peter Thiel made $55 million when eBay bought PayPal in 2002. The first thing he did with that money was turn $500,000 of it into a 10% stake in a college kid’s website that let you rate how hot your classmates were.
That investment was worth north of two billion dollars. Facebook turned out to be more than a hot-or-not app for Harvard students.
So when Thiel’s hedge fund quietly dumped its entire $100 million stake in Nvidia right before earnings, my first thought wasn’t “he’s finally lost his shit.” My first thought was “what does he know that I don’t?”
Because, love him or hate him, Thiel launched PayPal during the late 1990s internet boom, when it felt like there was an open frontier or gold rush. He lived through the collapse. He watched companies with zero revenue and astronomical valuations evaporate overnight. And unlike most people who survived that bloodbath, he actually learned something from it.
Now he’s doing it again.
Thiel sold all 537,742 shares of his Nvidia holdings during the third quarter, joining SoftBank’s $5.8 billion exit and Michael Burry’s aggressive short position. Three legendary investors with completely different styles and philosophies all heading for the exit at roughly the same time.
When sharks swim away from the beach, you don’t need to know why. You just get out of the water.
Microsoft, Meta, Tesla, Amazon, and Google invested about $560 billion in AI infrastructure over the last two years, but brought in just $35 billion in AI-related revenue combined, according to Fortune.
Stop and think about that ratio for a second. These aren’t desperate startups burning through venture capital. These are the most sophisticated tech companies on the planet, run by some of the smartest operators in history.
And they’re spending sixteen dollars for every one dollar of AI revenue they generate.
You know what else had math that didn’t work?
Pets.com spent $300 on marketing to acquire customers who bought $20 worth of dog food. They went public anyway. Nine months later, they were liquidating their sock puppet mascot on eBay. The difference is that Pets.com was spending millions. These guys are spending hundreds of billions.
In what universe does that math work? In the universe where AI eventually transforms everything, and those early investments pay off 50x. Except that’s the exact same pitch we heard in 1999 about the internet. And you know what? The internet did transform everything. Amazon survived. Google thrived. Microsoft adapted.
But that didn’t stop the Nasdaq from dropping 80% when reality caught up with expectations.
Nvidia just reported $57 billion in quarterly revenue, absolutely crushing estimates. The company prints money. Its margins are obscene. By every traditional metric, this looks like the opposite of a bubble stock. But an MIT study found that 95% of AI pilot projects fail to yield meaningful results, despite more than $40 billion in generative AI investment.
That’s a “we have no idea how to make money with this yet” problem.
Here’s where it gets super interesting. Michael Burry, the guy who predicted the housing crisis by actually reading mortgage-backed security documents that nobody else bothered to read, recently broke a two-year silence to point something out.
Burry’s done this sorta thing before. In 2005, he was reading actual mortgage documents, the stuff buried in the footnotes that made lawyers fall asleep. He found adjustable-rate mortgages being sold to people who couldn’t afford the initial rate, let alone the adjusted one. Everyone said he was loco. Then he made $700 million when the whole thing imploded. Now he’s reading depreciation schedules. Maybe we should, too.
He estimates Big Tech will understate depreciation by $176 billion between 2026 and 2028, artificially inflating reported profits by more than 20% at some companies, according to Fortune.
The trick is simple: companies are buying Nvidia chips that become obsolete in two to three years, but they’re depreciating them over five or six years. Meta recently extended the estimated useful lives of certain servers and network assets from four or five years to 5.5 years.
It’s accounting magic.
Completely legal. And it makes current earnings look fantastic while pushing costs into the future. Think of it like putting your rent on a credit card. Your checking account looks great right now, but eventually that bill comes due with interest.
The 1999 Comparison Everyone’s Making Wrong
People love comparing the current moment to the dot-com boom, and the truth is both yes and no in ways that should make any investor a little uneasy.
The four biggest companies pouring money into AI infrastructure trade at around 26 times forward earnings. That doesn’t sound too bad until you remember that at the height of the dot-com frenzy, the market was trading at about 70 times earnings. That comparison might feel comforting, but we’re still measuring ourselves against the most overpriced period in modern market history.
It’s like bragging that you’re only half as drunk as you were that time you woke up in someone else’s Camry, or was it a Kia?
Anyhoo…the real difference is this: The late-90s tech darlings were raking in almost no real revenue. Today’s AI giants are minting money. The difference is the speed and sheer scale of spending.
AI investment has grown by less than half a percent of US GDP since 2022, compared to 1.2 percent during the dot-com run. But the raw dollar amounts today are enormous because the companies writing the checks are far larger than the ones we were dealing with back then. When giants spend, the numbers get big quickly.
What Thiel Knows That You Don’t
Thiel told the Times a while back that AI sits somewhere between “complete nothingburger” and “total societal transformation.” From anyone else, that’d be hedging.
Coming from the guy who co-founded PayPal and Palantir, invested in Facebook at the seed stage, and basically wrote the playbook on contrarian tech investing, that’s pattern recognition. You know who else is good at pattern recognition?
And here’s what he did after dumping Nvidia: He increased positions in Microsoft and Apple while slashing Tesla holdings by 76%. He’s still in tech. Just tech that makes money ten different ways instead of one. Those companies have diversified revenue streams and don’t need AI to save the quarter or justify their valuation at every earnings call.
That’s the tell.
He’s pointing out that the price expectations are floating higher than the actual ground they’re standing on. And markets have a habit of reintroducing gravity when they feel like it.
The Fire Exit Strategy
I’ve watched a few bubbles up close. One thing is consistent: they don’t burst because everyone wakes up one morning and realizes they’re complete idiots. People know they’re idiots. Heck….they’ve known for months.
Bubbles pop when some external thing happens and forces everyone to act on what they already believed but were ignoring.
In 2000, it was interest rates. In 2008, it was the subprime mortgages that finally defaulted. In 2022, it was inflation forcing the Fed’s hand. We don’t know what triggers the next correction.
But when three of the smartest investors on the planet all quietly head for the exit within weeks of each other, asking “why?” seems like a pretty reasonable response.
Look, I’m not telling you to sell everything and hide under your desk. I’m telling you that when the people who’ve made billions by being right about these inflection points start moving, it’s worth paying attention to where they’re moving to, not just what they’re leaving behind.
Because the last person to leave a party always gets stuck with the bill. And this party has been going for a while now.
Thank you so much for taking the time.
WorkmanShit is a reader-supported publication.
To support my whisky habit, consider becoming a free or paid subscriber.
No cash? No problem!
Smashing that ❤️ button or sharing this post keeps the wheels on this greasy hamster wheel, too.
You've got this one exactly right Neela! The bubble will burst and it will burst soon.
I recently gave a business talk about how, yes, there is an AI bubble, but AI trends are real and will change society. I used two of the same three companies you did to say just because a bubble bursts, it doesn’t prevent some companies from changing society anyway. I used Amazon, Apple, and Microsoft.
The best advice I read recently was not to exit the market. As long as the music plays, they will keep on dancing. But the advice was to sell your more speculative investments and raise your cash percentage a bit. Keep the rock solid investments which, like Apple in 2000-2001 did fall with everything else. But the rock solid ones bounce right back, while the ones playing with accounting fiction struggle to recover. If you have a bit more cash, you are more cushioned in a crash.
Finally, just because a crash may be coming, it can take longer than you think is rational. Irrational market behavior is a feature, not a bug.