Descriptions:
Episode 282 of the All-In Podcast opens with the week’s defining story: Moonshot AI’s release of Kimi K3, a Chinese open-source model that co-hosts Chamath Palihapitiya, David Sacks, David Friedberg, and Jason Calacanis describe as directly competitive with Anthropic’s Fable and OpenAI’s GPT-5.6 at roughly 50% lower cost. The release triggered a policy response the hosts liken to the DeepSeek moment of early 2025, with the White House reported to be considering a ban on Chinese open-source AI models. Polymarket odds on such a ban in 2026 rose from 22% to 45% within days of the news.
David Sacks, who has publicly advised the Trump administration on AI policy, states on air that no ban decision has been made and that the administration favors incentivizing US open-source development over restricting Chinese contributions to the public domain. The hosts debate the logic of a potential ban, pointing out that Thinking Machines’ current best American open-source model was itself bootstrapped by distilling off Kimi K2.5 — meaning restrictions could directly harm US developers.
A substantial portion of the episode covers competitive revenue dynamics: Anthropic’s ARR grew from $10B to over $70B in the first half of 2026, while OpenAI accelerated from $33B in May to $41.3B in July. The hosts debate whether Anthropic’s calls for open-source regulation constitute regulatory capture, given that the company is simultaneously the fastest-growing tech firm in history. The episode also touches on Anthropic’s reported $1.5 billion employee payout and broader implications for AI market structure.
📺 Source: All-In Podcast · Published July 24, 2026
🏷️ Format: Podcast
