Anthropic considers building its own AI chips
What's the story
Anthropic, a leading artificial intelligence (AI) lab, is said to be considering the development of its own AI chips. The move comes as the company and others in the industry face a shortage of these critical components. The plans are still in their early stages and there's no guarantee that Anthropic will actually go through with them.
Chip strategy
Current chip usage and partnerships
Anthropic currently uses a variety of chips, including Google's tensor processing units (TPUs) and Amazon's chips, to develop its AI software and chatbot, Claude. The company recently signed a long-term deal with Google and Broadcom for TPUs. Anthropic has seen a massive surge in demand for its AI tools. The company's run-rate revenue has now crossed $30 billion, up from about $9 billion at the end of 2025.
Chip development
Cost and challenges of designing AI chips
Designing an advanced AI chip can cost around half a billion dollars, industry sources say. This is because companies have to hire skilled engineers and invest in ensuring that the manufacturing process is defect-free. Anthropic's talks reflect a broader trend among tech giants like Meta and OpenAI, all of whom are racing to develop in-house AI chips.