All-In with Chamath, Jason, Sacks & Friedberg

AI Sovereignty Wars, Palantir-Nvidia Deal, SCOTUS Birthright Ruling, Newsom's CA Budget Lie

July 3, 2026

AI Summary

5 min read

The Case for AI Sovereignty: Why Enterprises Are Rejecting Frontier Labs

The Palantir-Nvidia partnership announced this week marks a turning point in the enterprise AI market. Palantir will use Nvidia's open-source Nemotron models to build a "sovereign AI operating system" for U.S. government agencies, who will own the hardware, data, and model weights themselves. The deal crystallizes a growing revolt against the dominant frontier AI labs—and the hosts argue it signals a fundamental shift in how enterprises will approach AI.

The Trust Problem with Frontier Labs

The central argument of the episode is that enterprises are waking up to a dangerous dynamic: by feeding their proprietary data to frontier model providers like Anthropic and OpenAI, they are effectively handing over their competitive advantage to companies that may eventually compete with them. Alex Karp of Palantir made this case bluntly on CNBC, asking whether the U.S. government should really "outsource the battlefield of this country to the consensus view in Silicon Valley."

Continue reading the full summary in the app — free to try.

Read Full Summary →

Free • No credit card required

Listen to Audio Summary Open in App

Never miss an episode of All-In with Chamath, Jason, Sacks & Friedberg

Get every new episode summarized in your inbox — free, ~5 minutes to read.

No spam. Unsubscribe anytime.

What you'll learn

1 (00:26) **Palantir-Nvidia Sovereign AI Partnership** - Palantir and Nvidia announce a partnership to build a custom, open-source AI model for the U.S. government, with Palantir calling it a "sovereign AI operating system."
2 (04:24) **Sachs: The Enterprise AI Safety Argument** - David Sachs reframes Carp's "crash out" as a brilliant insight on enterprise AI safety, where safety means retaining control over compute, models, data, and proprietary knowledge.
3 (09:29) **Chamath's Enterprise Data on Open Source vs. Frontier Models** - Chamath presents real-world data from 8090, showing a 16.4x cost reduction when using an open-source model with their harness compared to using Claude alone.
4 (15:20) **Friedberg: Life Sciences Companies Rejecting Frontier Labs** - Dave Friedberg confirms that life sciences companies are refusing to share proprietary data with Anthropic, recognizing it would commoditize their core asset.
5 (19:30) **Warning to Founders: Don't Trust the Platform** - The hosts warn founders against taking "free tokens" from Sam Altman or partnering with frontier labs, comparing it to the Microsoft and Facebook platform traps.
6 (25:17) **Nvidia's Open Source Play and the Three-Layer Stack** - The hosts discuss Nvidia's NemoTron open-source model and why Nvidia and Palantir are natural partners against the emerging Anthropic-OpenAI duopoly at the model layer.
7 (30:28) **The Deflationary Cost of Tokens and the Return to On-Prem** - The hosts predict token costs will drop 90% per year, leading to a "buying frenzy" in enterprise hardware and a shift back to on-premise inference.