Steve Eisman
'Big Short' Investor Explains How the AI Bubble Will Burst (October 2, 2026)
Link.
NVIDIA, the largest company in the world, has revenue growth of 110%. But 70% of NVIDIA's AR came from 5 customers. That's a lot of concentration.
Google, Meta, Amazon, and Microsoft: 70% of their AI revenue is from Anthropic and OpenAI.
50% of Oracle revenue is from OpenAI. And OpenAI loses money like crazy.
The whole AI chain between NVIDIA, hyperscalers and end users depends on the growth of Anthropic and OpenAI. OpenAI and Anthropic are the ones that are signing the huge deals with the hyperscalers.
This industry makes Eisman nervous. Suppose OpenAI fails, then the entire US would go into a recession overnight.
Eisman thought most of the OpenAI / Anthropic money was coming from VC, but it's actually coming from Microsoft, Amazon, Google, NVIDIA, and Softbank.
This industry has no moats. Google was a moat, everyone used Google. It's pretty easy to switch LLM models.
He thinks Anthropic and OpenAI are nervous. They're nervous that tokenmaxxing has ended and that there are no moats. They're also worried that Chinese open source models are taking massive market share. He thinks that OpenAI and Anthropic are hoping government regulation will create a duopoly and ban Chinese models, so they get a regulatory moat.
Some businesses don't have a moat. A grocery store doesn't have a moat because another grocery store can be opened down the street.
The hyperscalers were very profitable a few years ago and generated a lot of cash. The cash flow is now gone because they're spending so much on data centers.
People are clearly worried about Oracle and how much debt they have.
The hyperscalers are investing hundreds of billions. Expected to invest $700 billion this year on AI capex.
Elon Musk said that China has a chip problem and the US has an electrical power problem. He thinks the chip problem is easier to solve. Power is physical infrastructure and takes a long time.
He wouldn't invest in an LLM company cause there is no moat. He's more wary of hyperscalers cause they don't have cash flow anymore. He likes the companies that are getting the cash flow like NVIDIA, GE Vernova, Cisco, etc.
There will be software companies that have problems. He thinks software that's deeply embedded in enterprises is probably OK. But maybe websites for finding a hotel will lose value.
He's a "story person" when it comes to investing. He thinks companies that are cheap are generally cheap for a reason.
He likes a moat, that's why he owns Moody's and Visa.
He doesn't like management selling stock, cyclical companies.
He likes companies with a real story and growth tailwinds.
He's never owned precious metals. They just sit there and don't provide anything.
Bitcoin is worse because it trades inversely to its own thesis.
He used to invest overseas, but he doesn't think it's necessary. There are plenty of investing opportunities in the US.