AI powerhouse Anthropic is approaching an $800 billion valuation despite Pentagon restrictions, with government agencies finding workarounds to access its technology. The staggering figure represents more than double the company's $380 billion valuation from just months ago, as secondary markets now price Anthropic at $688 billion - a 75% surge in three months. This explosive growth follows the success of Claude Code and a revenue run rate that has skyrocketed from $9 billion to $30 billion annually, with over 1,000 enterprise customers now spending over $1 million each per year.

Venture firms push Anthropic toward an $800 billion price tag

The new valuation talk is hitting as Anthropic gets even more attention for a new model called Claude Mythos. The model came out last week and drew attention inside security circles because it can uncover serious software flaws that human researchers had not found.

Even with President Donald Trump’s ban on federal use of Anthropic technology, officials across Washington are still dealing with the company.

Staff from at least two large federal agencies recently contacted Anthropic about using Mythos in cyber defense work, according to a former senior U.S. technology official with direct knowledge of the talks.

The Commerce Department’s Center for AI Standards and Innovation is also actively testing Mythos, according to a Politico report that cites four people familiar with the matter.

Those people allegedly included one current cybersecurity official, one former cybersecurity official, a former Trump administration official, and a former senior national security official. The center evaluates U.S. and foreign AI models for risks and opportunities.

On Capitol Hill, staff on at least three congressional committees have either held or requested briefings from Anthropic over the past week to learn more about Mythos and its cyber scanning abilities, according to three congressional aides working on AI policy.

Federal agencies keep testing Anthropic despite Trump’s Pentagon ban

The clash started in late February, when Trump and Defense Secretary Pete Hegseth told federal agencies to stop using Anthropic’s technology after CEO Dario Amodei opposed letting the Pentagon use its models for autonomous lethal attacks or mass surveillance against Americans. Last month, Hegseth formally labeled Anthropic a supply chain risk.

That move was unusual for a U.S. company and effectively blocks its AI models from use on Defense Department contracts. Even so, parts of the federal government appear to be moving around that order as interest in Mythos grows.

The same tension showed up on Wall Street. On Tuesday, JPMorgan Chase CEO Jamie Dimon said AI tools may help defend companies someday, but right now, they are opening more weak spots. He said JPMorgan is testing Anthropic’s Mythos preview as part of the bank’s push to use AI without giving attackers an edge. “AI’s made it worse, it’s made it harder,” Jamie said on the bank’s earnings call. “It does create additional vulnerabilities, and maybe down the road, better ways to strengthen yourself too.”

Asked later about Mythos, Jamie pointed to Anthropic’s warning that the model had already found thousands of vulnerabilities in corporate software. “It shows a lot more vulnerabilities need to be fixed,” he said.

Jamie added that JPMorgan, the world’s largest bank by market value, spends heavily on cybersecurity, keeps top experts on staff, and stays in constant contact with government agencies. Still, he said, banks remain connected to exchanges and other outside systems that add more layers of risk.

If you're reading this, you’re already ahead. Stay there with our newsletter.