Anthropic Explores Custom AI Chips Amid Security Disputes

Anthropic is exploring the design of its own artificial intelligence chips to mitigate global hardware shortages and support the surging demand for its Claude AI models. The initiative is in its early stages, with no dedicated engineering team or final design yet in place, and development costs are estimated at $500 million. This move follows a rapid revenue increase, with the company's annualized run rate exceeding $30 billion in 2026, up from approximately $9 billion at the end of 2025.

While considering internal design, the company maintains reliance on Nvidia and utilizes cloud-based processors from Amazon and Google. Anthropic recently signed a long-term agreement with Google and Broadcom to access 3.5GW of TPU capacity by 2027, aligning with a $50 billion commitment to strengthen U.S. computing infrastructure.

Simultaneously, the company is facing significant regulatory and security challenges. The U.S. Department of the Treasury and the Federal Reserve warned bank CEOs of cyber risks associated with Anthropic's new Mythos model, which can exploit weaknesses across major operating systems and web browsers. Furthermore, the Donald Trump administration classified the company as a supply-chain concern and a national security risk after Anthropic refused to remove safety safeguards that prevent Claude from being used for autonomous weapons or surveillance. A federal appeals court recently declined to block the Pentagon's decision to maintain this security label.