Anthropic’s $40 Million Bet: Regulation Is Now Part of the Product Strategy Anthropic is not waiting for Washington to write the rules; it is financing the political machinery deciding which rules survive. The company added a second $20 million contribution to Public First Action, taking its 2026 commitment to $40 million, while the safeguards group says it has raised more than $80 million. Anthropic says its money is restricted to public education and policy work rather than direct electioneering, but issue advertising can still reshape a race. The rival rulebooks Leading the Future—backed by figures including OpenAI president Greg Brockman and Andreessen Horowitz co-founder Marc Andreessen—has reported raising about $75 million, with more pledged. Public First is pushing safeguards and developer transparency, while Anthropic’s framework calls for independent evaluations, enforceable safety practices and government power to block deployments carrying catastrophic risk. Leading the Future wants the lighter-touch version: fewer barriers, faster development and rules closer to the White House’s industry-friendly posture. Both networks have supported candidates from both parties; bipartisan money does not make this neutral, it makes the campaign durable. This is policy capex My market read is that political spending has become another layer of AI capex. Compute determines who can train a model, but regulation can determine who may deploy it, sell it into regulated markets and absorb the compliance burden. Harder rules could slow the sector, yet high fixed compliance costs could also entrench the best-capitalized labs and hyperscalers. That is why Anthropic can be sincere about safety and self-interested at the same time; investors should judge the mechanism, not demand pure motives. For MSFT, AMZN, GOOGL and META, the read-through is not a one-day move, but margins, liability, government access and competitive concentration. November is the next test These groups have already spent millions across primaries and still have firepower for November, turning races into proxy battles over AI governance. The Alex Bores contest showed the ambiguity: Public First backed him, Leading the Future opposed him, and he lost to another candidate who also favored regulation. Bottom line: the AI war is no longer model versus model; it is coalition versus coalition, and the faction that writes the rules may capture more value than the faction that wins the next benchmark.