Chamath Palihapitiya says the AI model business has no lasting advantage, a warning that comes as Anthropic and OpenAI both circle the public markets.
On the latest All-In podcast, the Social Capital chief said the real business model has moved to the application layer above the model and the infrastructure layer below it.
The catalyst is Moonshot AI’s Kimi K3, an open-weight Chinese model that reportedly matches frontier performance at a fraction of the cost.
Chamath says once a lab publishes its numbers, rivals match or beat them within weeks. Commoditization that used to take five to ten years (in other industries) is now taking months.
Assigning large terminal value to the model layer is “a mathematical mistake,” he added.
White House Weighs Up Action
The White House spent last week weighing restrictions on open source AI models, particularly Chinese ones, after Kimi K3 landed.
The industry pushed back fast.
Nvidia Corp. (NASDAQ:NVDA), Meta and 23 other companies put their names to an open letter defending open weights, while Google, Amazon and the closed labs kept theirs off.
A ban would arguably be the best outcome available to Anthropic and OpenAI, removing their cheapest competition by decree.
David Sacks, who stepped down as White House AI czar in March and now co-chairs the President’s Council of Advisors on Science and Technology, called it a “tragic mistake” that would backfire on the US.
Polymarket puts the odds of a US ban on an open source model this year near 14%, with a related contract on Washington removing public access to a major Chinese model at 11%.
The Question Wall Street Is Asking
Sacks said both labs are in the middle of road shows and conceded they face a legitimate question about why open source will not commoditize them.
Anthropic filed confidentially with the SEC on June 1, OpenAI followed a week later.
Polymarket prices roughly 19% odds that OpenAI IPOs by the end of the year. Anthropic is at 71% for the same timeline.
If Palihapitiya is right, the trade sits underneath the models.
He singled out Alphabet Inc. (NASDAQ:GOOGL), arguing hundreds of competing models suits a company selling the silicon and the cloud to serve all of them.
Nvidia sits in the same trade, since cheaper models may mean more inference rather than less.
Palihapitiya’s issue is narrower than it sounds.
He isn’t saying these companies won’t make money, or that their models aren’t good. He just doesn’t think the model layer is worth what it’s being priced at.
Image: Shutterstock
Login to comment