Anthropic CEO says AI centralizes by nature and open models just shift power to whoever owns the chips

Details

The dispute started when investor Gavin Baker claimed on the All-In Podcast that Anthropic's CEO had privately suggested the company could one day be the only private firm left in the world. Anthropic researcher Sholto Douglas denied this, arguing the AI market is already the most competitive in the world and that economic concentration is exactly what Anthropic opposes.\n\nBaker framed two basic policy positions: either AI is too dangerous to spread widely — concentrate it in a few companies — or too dangerous to concentrate — spread it as broadly as possible. Yann LeCun came down firmly for openness, comparing AI to the printing press and internet: tools that amplify human intelligence, where societies need diversity of systems with different values.\n\nAmodei entered the thread to reject the binary framing. He argues that regulation doesn't automatically mean regulatory capture — it can also constrain corporate power, including Anthropic's own. On open models specifically, he says they shift market power toward the players with the most compute to run them, not toward end users or smaller actors.\n\nFormer White House adviser David Sacks countered that Anthropic deliberately hires ex-officials to shape regulation in its favor, and that an AI approval agency would weaken the US against China. The debate remains unresolved and directly frames what AI policy will look like in 2027.