METAL for iPhone

Read AI news in the METAL app.

Download METAL and discover fresh AI stories every day.

Download on the App Store

For iPhone · Free download

Search for METAL AI Magazine in the App Store on your iPhone.

METAL

Anthropic Co-Founder Tom Brown's Washington Role

Anthropic co-founder Tom Brown has emerged as the negotiator who ended June's White House standoff over a model shutdown and struck a $1.25 billion-a-month compute deal with Musk. The Pentagon's security-risk designation remains in place.

Anthropic Co-Founder Tom Brown's Washington Role

Image: METAL

Summary

  • Anthropic co-founder and chief compute officer Tom Brown reportedly ended a June standoff with the White House over the shutdown of two models in two and a half weeks.
  • Brown personally pitched Musk in late March, and in early May Anthropic announced it would rent SpaceX compute for $1.25 billion a month.
  • A Republican, Brown holds regular talks with the commerce secretary and the White House science office chief, but the Pentagon's security-risk designation of Anthropic is still in effect.

Tom Brown, a co-founder of AI company Anthropic, has stepped forward as the company's negotiator with the Trump administration and Elon Musk. According to reports, Brown, Anthropic's chief compute officer (CCO), ended a two-and-a-half-week standoff in June after the White House forced Anthropic to shut off all access to two of its models over security concerns, and in May he closed a deal to rent SpaceX's computing capacity for $1.25 billion a month. A Republican, he is handling the political and business alliances the company needs as it prepares for an initial public offering that could value it at more than $2 trillion.

During the June shutdown, CEO Dario Amodei sent Brown to Washington after his own efforts fell short. Brown ended the standoff by assuring Commerce Department, Pentagon and White House officials that Anthropic understood their security concerns and had upgraded its safeguards. According to a person familiar with the matter, Brown and Amodei spoke by phone every day during that period, and Brown taking charge of the Washington negotiations let Amodei focus on customer relationships and other CEO duties.

Brown, who stands 6 feet 6 inches tall, posed for a photo last week at a White House event celebrating America's AI advances, shoulder to shoulder with technology CEOs and President Trump, who has previously criticized Anthropic as woke. Amodei also attended the same lunch, making Anthropic one of two companies with two executives present. Some observers see Brown's improved relationship with the administration as one factor that helped Amodei secure his first lengthy face-to-face meeting with Trump at a White House dinner in late September, after which he drew praise from the president.

Brown's Washington push extends to where he lives. Earlier this year he bought a house in Washington with his wife, Michelle Valentine, a startup co-founder and former venture investor. People familiar with the matter said the couple have described themselves in private conversations as allies of the Republican party who want to build relationships in Washington. Brown has supported the expansion of data centers and nuclear energy that the White House is pushing. At a G-20 tech summit last month, he said he really loved a Truth Social post by Trump saying communities that don't embrace data centers risk becoming backward and poor. "He was pointing out that the data centers are just an enormous source of prosperity," Brown said. METAL previously reported on the optimism inside the G20 venue in Chapel Hill and the data center protests outside its walls.

Views in Washington have shifted too. "He is increasingly playing the dual role of founder and diplomat," said Joseph Hoefer, chief AI officer at lobbying firm Monument Advocacy, which represents technology companies but not Anthropic.

Brown grew up in the San Francisco Bay Area, studied engineering at the Massachusetts Institute of Technology (MIT), and worked at several startups and the incubator Y Combinator. After reading Nick Bostrom's book Superintelligence, he taught himself AI through online courses and math textbooks, and offered to mop the floors at OpenAI to get in. OpenAI co-founder Greg Brockman hired him as one of the startup's first 20 employees. He was later laid off from OpenAI and spent a short time at Google DeepMind, before Amodei brought him back to OpenAI in 2018 to lead engineering work on GPT-3. In early 2021 he founded Anthropic together with Amodei and other OpenAI leaders.

His connection to Musk goes back to 2016. According to reports, Musk, then OpenAI's main funder, savaged a demonstration by a renowned researcher during one visit but liked Brown's tool that let an AI play the video game StarCraft. A decade later, that encounter led to a computing deal between the two men's companies. After Anthropic cut off xAI's access to Claude Code, Musk spent months attacking Anthropic on X as "smug, sanctimonious and hypocritical" and nicknamed it Misanthropic. In late March, Brown drove to SpaceX's offices in Hawthorne, California, and told Musk that Anthropic was not as woke as he thought, while acknowledging that some of its past models had been. In early May, the two companies announced that Anthropic would rent excess data center capacity from SpaceX, which had acquired xAI, for $1.25 billion a month.

Securing compute was Brown's job from the start. Unlike OpenAI, which relied mainly on Nvidia chips through Microsoft, Anthropic spread its suppliers across several companies, and when demand rose after the launch of Claude in early 2023, Brown led deals with Google and Amazon. As its models gained traction with businesses, compute fell short and customers faced usage limits and outages, which set the stage for the SpaceX deal. In recent weeks Anthropic has also signed agreements worth tens of billions of dollars with AMD and three smaller companies.

Only one side of the relationship with the administration has thawed. Anthropic clashed with the Defense Department earlier this year over guardrails for military use of AI, and the Pentagon's security-risk designation remains in place. The Pentagon has barred all use of Anthropic's tools and told contractors to stop using them, a move that has been upheld by one court and struck down by another. METAL previously reported that a U.S. appeals court upheld the Pentagon's exclusion of Anthropic. Brown speaks regularly with Emil Michael, undersecretary of defense for research and engineering, and in recent weeks has discussed Anthropic's models and AI safety with Commerce Secretary Howard Lutnick and Michael Kratsios, head of the White House Office of Science and Technology Policy.

On stage at the G20, when Lutnick asked what lies ahead for AI, Brown said he expects major progress on curing cancer and on the economy within five years, echoing Amodei's claims. "Looking back we'll be like, 'Wow, back in 2026 we thought stuff was going fast and we thought we were seeing this, but actually we underestimated how impactful this would be and what the progress would be,'" Brown said. The full report METAL reviewed says Brown was seen during the G20 talking casually with Lutnick and officials from other countries at the Carolina Inn on the University of North Carolina's campus.

Seen through a legal lens, the core of this arrangement is the separation of negotiating channels. Amodei handles risk warnings and customers while Brown handles regulators and rivals, and once June showed that the administration can halt access to models outright, the government channel becomes a tool for managing business continuity. With the Pentagon designation and conflicting court rulings still standing, the next test is whether Brown's diplomacy can change the company's legal status as well.

Comments