• by Laurie Sullivan , 5 hours ago

Anthropic's revenue run rate has topped $30 billion -- up from $9 billion at the end of 2025, the company revealed Tuesday, along with a cybersecurity project made available to the entire software industry.

The run-rate disclosure overshadowed something much more important for advertisers -- a major advancement in cybersecurity technology, and a partnership with some of the biggest names across the technology industry to secure critical software on top of AI to build out "Project Glasswing." 

The new initiative shares knowledge with with the industry and shares assets with more than 40 organizations including Amazon Web Services, Anthropic, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks in an effort to secure the world’s most critical software.

"We formed Project Glasswing because of capabilities we’ve observed in a new frontier model trained by Anthropic that we believe could reshape cybersecurity," Anthropic wrote in a blog post, explaining that it has become possible for AI models to reach a code capability to surpass all but the most skilled humans at finding and exploiting software vulnerabilities.

Anthropic committed up to 100 million in usage credits for Mythos Preview across these efforts, as well as $4 million in direct donations to open-source security.

Claude Mythos Preview, Anthropic's cybersecurity software, has already found thousands of high-severity vulnerabilities, including some in every major operating system and web browser. 

During the past several weeks, Anthropic researchers used Claude Mythos Preview to scan critical software infrastructure. What emerged were thousands of high-severity zero-day vulnerabilities, previously unknown to the developers of the affected software.

The technology will not be made public, but the AI found vulnerabilities including a 27-year-old vulnerability in OpenBSD and a 16-year-old flaw in FFmpeg, a multiple Linux kernel that is autonomously chained together.

This is just the beginning, Anthropic wrote in the post. It added that no one organization can solve these cybersecurity problems alone, so it has partnered with other companies, including a partnership with Google and Broadcom. 

"We have signed a new agreement with Google and Broadcom for multiple gigawatts of next-generation TPU capacity that we expect to come online starting in 2027," the company disclosed in a blog post. "This significant expansion of our compute infrastructure will power our frontier Claude models and help us serve extraordinary demand from customers worldwide."

Krishna Rao, CFO of Anthropic, called the deal its "most significant compute commitment to date" and said it will allow the company to keep pace with its growth. 

The deal gives advertisers faster hardware and software support to run all types of features and advertising campaigns being built in artificial intelligence.

This massive expansion should also dramatically lower costs to run high-end AI while providing the raw power needed for instant, hyper-personalized consumer experiences.

By securing specialized hardware from Google and Broadcom, Anthropic ensures that Claude remains capable and cost-efficient. 

Demand from Anthropic's AI large language model Claude continues to climb since it announced its Series G fundraising in February, with more than 500 business customers agreeing to spend more than $1 million on an annualized basis. Today that number exceeds 1,000, and has doubled in less than two months.

While Anthropic does not provide information on the percentage gained from "brands," its growth is overwhelmingly driven by enterprise and business customers rather than individual consumers.

Anthropic will build the majority of its new compute power in the United States, making this partnership a major expansion for its November 2025 commitment to invest $50 billion in America’s computing infrastructure.

The partnership also deepens Anthropic’s existing work with Google Cloud by building on the increased TPU capacity  announced last October—as well as the relationship with Broadcom.

Anthropic also disclosed that it trains and runs Claude on Amazon Web Services’ Trainium, Google TPUs, and NVIDIA GPUs, giving it the ability to match workloads to chips best suited for them.