All-In · Sep 25, 2026 · 1h 34m · 74m saved
With Chamath Palihapitiya, Founder and CEO of Social Capital · Jason Calacanis, Angel Investor and Managing Partner at LAUNCH · David Sacks, General Partner at Craft Ventures · David Friedberg, Founder and CEO of The Production Board
Anthropic was targeting a $2 trillion valuation for an IPO planned for October. Chamath Palihapitiya said the same offering will now clear at $1 trillion or less, because buyers will demand a lower price to take on the risks the panel described.
David Sacks was the most optimistic of the four about the business. He said Anthropic and OpenAI hold a stable duopoly at the top of the market and can charge a premium for it. He also said the position is fragile.
"But these two companies are on a hamster wheel — the moment they stop being frontier, they go to zero, and the frontier is maybe only six to twelve months ahead of the commodity portion of the market."
Chamath Palihapitiya, founder and CEO of Social Capital, and his co-hosts on All-In, the week after their fifth annual All-In Summit. Palihapitiya's firm has been testing the major models on cost and quality, and he said he has lost every head-to-head recruiting contest against Anthropic. David Friedberg runs a life-sciences research company that pays for Anthropic's models. David Sacks, a general partner at Craft Ventures, and Jason Calacanis, an angel investor and managing partner at LAUNCH, round out the panel.
The full episode is covered here so you can skip it. 94 minutes of audio, 21 minutes of reading.
Here are the 18 arguments that matter.
Key Takeaways
Anthropic's IPO will clear at $1T or less instead of $2T, Palihapitiya said, and he thinks that is good for the company
Token use has flipped from 80/20 closed models to 80/20 open models in 12 weeks, in a stat Palihapitiya cited
He called it the chart that belongs in the S-1 risk factors
Sacks sees a stable frontier duopoly that goes to zero if it falls behind
He said the frontier is only 6–12 months ahead of cheaper models
Sacks said Anthropic's lobbying for tighter AI rules would slow it enough to lose its lead
AI companies are corporations with product liability, not research labs, Palihapitiya and Sacks said
Sacks said the administration will not trade a liability waiver for a 10% equity stake
Friedberg said Anthropic's San Francisco wet lab is a low-risk benchtop lab, not a pathogen lab
Personal AI agents like Meta's Muse put the App Store's 30% cut at risk, Palihapitiya said
Palihapitiya said AI is probably most, if not all, of US real growth, in his arithmetic
3.5% inflation and 5% nominal growth leave 1.5% real growth
1. Summit and the Trump Call
The four opened with their favorite moments from the fifth annual All-In Summit. Friedberg, who produced it, thanked the production team.
Sacks said the Summit calmed a national panic over AI
"There's been this scare campaign for weeks leading into the summit to convince everyone humanity was going to go extinct unless we stopped all AI development." — David Sacks
He said Microsoft's Satya Nadella and Nvidia's Jensen Huang calmed the audience, and that President Trump then called in during Huang's talk and took control of the narrative.
Friedberg said the call was not planned
"It was completely spontaneous, not some planned bit." — David Friedberg
Calacanis said Huang asked him how to put the phone on speaker, so he handed Huang his spare microphone. He joked that Huang makes the world's hardest chips to copy but could not find the speakerphone button.
