Anthropic's CEO pushes back on DC critics

For months, Silicon Valley critics painted Anthropic as the cautious, nanny-state AI lab obsessed with safety over raw power. This week, CEO Dario Amodei decided to set the record straight, publishing an extensive defense of the company’s alignment with American national security.

The timing is no accident. Amodei led by revealing that Anthropic is the fastest-growing software company in history, exploding from a $1B to a $7B run rate over just nine months. But his main goal was to squash rumors that Anthropic is politically biased or trying to hamstring startups.

Deep ties with Washington

Amodei laid out Anthropic’s surprisingly deep footprint inside the federal government. In July, the Department of War handed the company a two-year, $200M agreement to prototype frontier AI for national security.

Claude is also being offered to federal agencies for just $1 through a General Services Administration partnership, while classified versions run across sensitive intelligence networks via Palantir and Lawrence Livermore National Laboratory. Amodei even highlighted a personal meeting with Donald Trump in Pennsylvania to discuss AI energy infrastructure, alongside hiring senior former Trump officials to the company’s advisory council.

On policy, Amodei pushed back against claims that Anthropic wants to suffocate AI startups. He clarified that their support for California’s SB 53 bill included an exemption for any company making under $500M in gross revenue, specifically to keep startups safe.

He also leaned hard into geopolitical defense, noting Anthropic is the only frontier AI lab that refuses to sell services to Chinese-controlled entities, backing hawkish chip export bans even at the cost of short-term revenue. On political bias, he cited studies from the conservative Manhattan Institute and Stanford showing Claude 3.5 Sonnet was less biased than its rivals, noting further neutral gains in Sonnet 4.5 and Haiku 4.5.

Why it matters

If you are an enterprise developer building on Claude, this is Anthropic signaling total institutional permanence. By embedding itself into classified Pentagon networks and government procurement, Anthropic is insulating itself against political backlash from both sides of the aisle.

For the broader industry, this statement marks the end of AI companies pretending to be neutral global platforms. The race between OpenAI, Google, and Anthropic is no longer just about benchmarks or consumer apps. It is about becoming the critical national security infrastructure of the United States.

We think Amodei played this cleverly by rebranding Anthropic's traditional safety-first stance into hard-nosed American patriotism. It neutralizes attacks from venture capitalists who accused him of regulatory capture, while matching the Trump administration's hawkish energy policy and trade stance. The risk, however, is clear: leaning this heavily into military contracts and political alignment makes it much harder to maintain the lofty, independent public-benefit mission Anthropic originally sold to the world.