Anthropic Explores Developing In-House AI Chips Amid Growing Demand
SAN FRANCISCO, April 9 — Anthropic is exploring the possibility of designing its own artificial intelligence chips as demand for advanced computing power continues to surge, according to sources familiar with the matter.
The initiative remains in its early stages, and the company has not yet committed to a formal chip development program. Sources said Anthropic may ultimately decide to continue relying on external suppliers rather than pursue in-house design, with no dedicated team or finalized chip architecture currently in place.
Rising Demand Driving Strategic Considerations
The move comes as demand for Anthropic’s AI models, including its Claude chatbot, accelerates rapidly. The company recently reported that its annualized revenue run rate has exceeded $30 billion in 2026, a sharp increase from approximately $9 billion at the end of 2025.
To support its growing infrastructure needs, Anthropic currently relies on a mix of hardware, including tensor processing units developed by Google and chips supplied by Amazon.
Earlier this week, the company also entered into a long-term agreement with Google and Broadcom, which is involved in designing advanced AI chips. The partnership forms part of a broader commitment to invest $50 billion in strengthening computing infrastructure in the United States.
Industry-Wide Shift Toward Custom Silicon
Anthropic’s exploration reflects a broader trend across the technology sector, where companies are increasingly considering custom chip development to reduce reliance on external suppliers and optimize performance for AI workloads.
Major players such as Meta and OpenAI are also pursuing similar strategies, seeking greater control over hardware tailored to their specific AI models and applications.
However, designing advanced AI chips is both complex and costly. Industry estimates suggest that developing a single chip can require investments of up to $500 million, factoring in engineering talent and manufacturing processes.
