Neither Anthropic, Open AI nor SpaceX in its current form is a good candidate for an IPO at valuations that all but guarantee hundreds of billions of dollars in "passive capital" aka pension funds and ETFs will have to buy in.
SpaceX might have been a candidate on its own core business (aka: launching spacecraft and Starlink), but ever since the weird side deals with all of the other companies in the Muskverse (Twitter, xAI, Tesla) it is far too contaminated.
Sooner or later the AI bubble will burst - and assuming that the pension funds and ETFs buy in as projected, they stand to lose a lot of money that will make Covid's first lockdown + dotcom + 2007 Lehman combined look pale.
Short dot.com stocks (55% return), short subprime mortgages (Massive return), long Gamestop 2019, short ARKK 2021, The shorts on Palantir and NVDA are probably still running (PLTR 25% in profit, NVDA 20% loss).
Burry is well aware of this, he has written about how passive investing is contributing to this problem.
It has been broken since ~2008 (ZIRPs, etc.) and has really gone off the rails since BTC and memestocks have taken off. Now everything's a memestock. It's all vibes-based.
The market can remain irrational longer than you can remain solvent.
A long-term principle that I think does still apply.
“In the short run, the market is a voting machine, but in the long run, it is a weighing machine”
- benjamin graham
The problem is that every company is valued by the market as if it could capture all the value of the new technology, but obviously, not all of them will. This is what happened during the first Web bubble. All the big names were bad bets. Everyone lost a lot of money. Many companies folded immediately. Others, like Yahoo, bled out eventually. Amazon took 10 years to recover. MSFT took 15. And the biggest winner, Google, didn't IPO until well after the bubble popped.
So the question isn't "is Anthropic worth $1T"? The question is "is AI worth Anthropic's $1T + OpenAI's $1T + SpaceX/xAI's $1.8T altogether?" Maybe, but probably not. That implies you are betting on a horse, not investing.
Which one will be the winner in 10 years? Are you confident making that pick? Is it even one of these companies or some other Google yet to be formed or will rise up from China or whatever?
That's the paradox. AI can still change the world and create a ton of value, yet none of these companies might be a good investment at these prices. Or one of them could be a great investment, and the others terrible investments.
But Micheals arguments are valid. There could be competition, or even local models, thus indeed becoming 'commoditized'.
Because I used frontier models this weekend (I had 78% of my assigned tokens for this month left, I wanted to burn them before June 1st, ended up with 24% left), and tbh, I don't see much of the improvement compared to the models I use day-to-day. I'd rather pay less for a slightly worse model. Stacktrace analysis (or any bug analysis really) is where LLMs have the most success rate imho, and free models are good enough since last year. As for coding/architecture tasks, frontier models seems to hallucinate less, but I wonder if it's the guardrails or the he model themselves.
We all know the market can stay irrational much longer than you can stay solvent if you bet against it. If you watched "The Long Short" (excellent movie btw.) you know how close Michael burry came to capitulation before his subprime bet paid off. He seems to have a tendency to be too early with his predictions, even with his genius GameStop investment. So while he may be right again fundamentally, his timing may be completely off and those companies could be "worth" significantly more than a trillion dollars, at least temporarily, in stock valuations.
My personal prediction is this: The hype will go on longer than people think, just like with the New Economy. There is this quote from market analyst Larry Wachtel in 1999 who said: "Everybody's happy, everybody's making money - something's wrong here"[1]. Ironically, even Wachtel eventually succumbed to FOMO, capitulated, went in late, and lost a lot of money[2]. I am trying to not make his mistake, but it will be tempting to do so, I am sure about that.
I need to figure out the next one.
1. Do an IPO.
2. Sell a small amount (5%) to price insensitive Elon Musk-fans at a $2 trillion valuation
3. Get in all the indexes, because you are huge.
4. Unlock more stock, which the index funds have to purchase
[1] https://www.bloomberg.com/opinion/newsletters/2026-06-01/the...
My impressions is that US investors also have a special love for the S&P500 and could likely benefit from a non-US bias.
Seems equivalent to removing road safety rules for the least-tested, most-powerful new vehicles only.
Whoever you are Michael Burry, you don't know shit about the implications.
According to your logic, it should have a market cap of $2.6 trillion.
Conservative is to look at P/E, which is 10 for Mercedes.
Anthropic isn't even a growth stock, since it has already been force fed to everyone with one of the largest marketing and coercion campaigns in history.
[1] https://www.sec.gov/Archives/edgar/data/1326801/000132680124...
beernet•1h ago
They will be right eventually and inevitably. Until then, it's funny watching them build a personal "brand" just to say "I told you so" when the market drops in X years.
epolanski•1h ago
The same way semiconductor, internet or railroad companies were not great investments regardless of how important the technology was going to be. It's still a financial investment and it's only going to pay off if bought at the right price, not at crazy multiples.
I will also add: if all your moat is your latest model, you're as good as your latest model and can be easily dethroned.
Strong moats are monopoly-like concessions (Verisign), exclusive technological edge (ASML), brands (Coca Cola), etc.
alexpotato•42m ago
Paul Graham doesn't think so
https://paulgraham.com/brandage.html
hparadiz•40m ago
epolanski•34m ago
There's no moat in LLMs when you're as good as your latest model.
Companies out there aren't in the business of throwing money down the drain.
Take DS4, you can use Deepseek APIs directly with Claude Code, and you're unlikely to notice a difference for the overwhelming majority of your use cases. But your bills run in few $ per day. I'm talking 2 magnitudes less.
hparadiz•11m ago
potatototoo99•33m ago
mschuster91•40m ago
Agreed with the exception of Verisign. Many a "security" company went bust like DigiNotar after mishaps or hacks. Being a globally trusted root CA or DNS operator is a strong moat - but also an incredibly brittle one.
And brands... brands aren't as safe as we thought either, as "store brands"/"private labels" are taking up more and more market share [1].
[1] https://www.nbcnews.com/business/consumer/shoppers-are-tradi...
monooso•50m ago