The most important question nobody's asking about AI.

Summary

Dwarkesh Patel analyzes the Department of War's supply-chain action against Anthropic and its implications for AI governance, state power, and mass surveillance. He argues that AI regulation and alignment could become tools for government control and that open-source diffusion will undermine corporate red lines. The main market implication is an unresolved risk for Big Tech defense contractors that may be forced to choose between AI providers and Pentagon work.

• The Department of War declared Anthropic a supply-chain risk over red lines on mass surveillance and autonomous weapons.
• Dwarkesh argues the government's threat to destroy a private AI business is a warning shot about state leverage over AI.
• He calculates falling AI inference costs could make nationwide CCTV mass surveillance extremely cheap within years.
• He warns that alignment and AI regulation could give future governments more control over AI labor and values.
• He expects AI technology to be multipolar, so open-source diffusion will undercut frontier-lab red lines.
• Market implication: Amazon, Nvidia, Google, and Palantir may face a future conflict between AI providers and Pentagon contracts.
• He favors regulating specific destructive AI use cases rather than giving the government broad control over AI.