Anthropic Explores Custom AI Chip Development as Revenue Surpasses $30B Run Rate

Anthropic is exploring designing its own AI chips as its Claude revenue run rate has surged from approximately $9 billion in late 2025 to over $30 billion, though the effort remains at an early stage with no committed design or dedicated team. The San Francisco-based company currently runs Claude on tensor processing units from Google and Broadcom, Amazon's custom chips, and Nvidia hardware, and signed a long-term deal for 3.5 gigawatts of TPU capacity starting in 2027. Custom chip development would cost roughly $500 million and reduce dependence on third-party suppliers, following similar moves by Meta and OpenAI. Despite explosive revenue growth and a $50 billion US infrastructure commitment, Anthropic remains unprofitable.

Anthropic's exploration of custom AI chip development

• ▪Anthropic is exploring the possibility of designing its own AI chips
• ▪Anthropic has not committed to a specific chip design
• ▪Anthropic may decide to continue purchasing chips from third parties rather than building its own
• ▪Anthropic's AI chip development effort is at an early stage
• ▪The development cost of an advanced AI chip is roughly $500 million according to industry sources
• ▪Anthropic has not assembled a dedicated team for the AI chip project

Anthropic's revenue growth and existing chip partnerships

• ▪Anthropic currently runs Claude on Amazon's custom chips
• ▪Anthropic currently runs Claude across tensor processing units designed by Alphabet's Google in partnership with Broadcom
• ▪Anthropic's annualized revenue run rate has surpassed $30 billion
• ▪Anthropic was consuming roughly one gigawatt of compute capacity earlier in 2026
• ▪Anthropic committed to invest $50 billion in US computing infrastructure in November 2025
• ▪Anthropic's annualized revenue was approximately $9 billion at the end of 2025
• ▪Anthropic signed a long-term deal with Google and Broadcom for approximately 3.5 gigawatts of TPU-based compute capacity starting in 2027

Industry trend toward proprietary AI silicon

• ▪A custom AI silicon market is emerging as an alternative to Nvidia's general-purpose GPUs

Perspective of Anthropic

• ▪Anthropic's revenue run rate growth from $9 billion to over $30 billion creates financial capacity for custom chip investment
• ▪Anthropic's exploration of custom AI chips represents a strategic shift toward greater hardware independence

Perspective of Nvidia

• ▪The emerging custom AI silicon market challenges Nvidia's general-purpose GPU dominance in AI workloads

Perspective of Google and Broadcom

• ▪Google and Broadcom's 3.5 gigawatt TPU deal with Anthropic starting in 2027 represents a significant near-term revenue commitment

Perspective of AI industry competitors

• ▪OpenAI's work on custom silicon with Broadcom demonstrates the strategic importance of proprietary chip development for AI leaders
• ▪Meta's development of custom AI training chips reflects industry-wide efforts to reduce reliance on third-party hardware suppliers