Analysis
OpenAI and Anthropic set new records for federal lobbying spending in the second quarter of 2026, combining for $3.17 million -- up 23% from the first quarter -- in a coordinated push that increasingly puts them on the opposite side of Washington's open-weight AI debate from most of the rest of the industry.
Anthropic's own lobbying bill nearly tripled to $1.97 million, now exceeding what Nvidia spends and closing in on Oracle's $2 million. OpenAI's spend rose almost 18% to $1.2 million. Both companies disclosed to Congress that their lobbying touches cybersecurity, copyright, cloud computing and defense procurement -- broad, uncontroversial-sounding categories that obscure a more pointed agenda underneath.
Reporting indicates the two labs are quietly lobbying regulators to restrict open-weight AI models, a position that puts them at odds with a 25-company coalition -- including Nvidia, Microsoft, Meta and a16z -- that published a joint letter this year urging Washington to avoid "premature restrictions" on open-weight releases. OpenAI and Anthropic were conspicuously absent from that letter's signatories, a gap that reads less like an oversight and more like a deliberate policy stance.
“Anthropic's own lobbying bill nearly tripled to $1.97 million, now exceeding what Nvidia spends and closing in on Oracle's $2 million.”
The motivation is not hard to trace. Both companies are heading toward blockbuster public listings -- Anthropic has confidentially filed for an IPO it hopes to complete by October at a valuation approaching $1 trillion, and OpenAI has pursued a similar confidential filing -- and both want a friendly regulatory reception in Washington ahead of the November midterms and their own market debuts. Restricting the open-weight competition that has been squeezing their API pricing power all year would be a direct commercial benefit dressed up as a safety position.
For founders building on open-weight models -- an increasingly large share of the AI application layer, given how competitive models like Moonshot's Kimi K3 have become -- this lobbying push is a regulatory risk worth tracking closely. If OpenAI and Anthropic succeed in shifting policy even modestly, the cost curve that has made open-weight infrastructure so attractive this year could tighten.
What to watch: whether the White House's forthcoming AI framework incorporates any open-weight restrictions, whether the 25-company coalition responds publicly to OpenAI and Anthropic's absence from its letter, and whether either company's IPO timeline shifts as a result of the political capital being spent now.
