Descriptions:
Episode 274 of the All-In Podcast brings in Gavin Baker of Andreessen Horowitz alongside regular hosts to cover one of the busiest weeks in recent AI industry history. The centerpiece discussion is Andrej Karpathy’s move to Anthropic, where he will lead a new pre-training team focused on recursive self-improvement — AI systems that iteratively improve themselves. Baker frames this as potentially placing Anthropic on a new Moore’s Law-style trajectory, with model quality improving parabolically as compute and self-training loops compound.
The episode also covers Anthropic’s reported EBIT-positive quarter (per the Wall Street Journal), which the panel treats as a significant moment for AI industry credibility — separating the narrative of circular funding from evidence of genuine commercial traction. Other major topics include the Nvidia stock selloff, the SpaceX and OpenAI IPO timelines, Trump’s decision to pull a prior AI executive order, and growing societal anxiety about AI-driven layoffs, sparked in part by a widely criticized internal memo from Cloudflare CEO Matthew Prince about eliminating “measurer” roles.
Baker’s investment perspective adds weight to the competitive dynamics discussion, touching on Anthropic’s hiring strategy, the recursive self-improvement thesis, and what separates durable AI moats from hype. A dense episode for anyone tracking where institutional money and top research talent are moving.
📺 Source: All-In Podcast · Published May 22, 2026
🏷️ Format: Podcast
