Anthropic's initiative to develop its own artificial intelligence chips signifies a significant shift in its strategy amid soaring demand for computational power. The company, which has seen its annual revenue run-rate soar to $30 billion from approximately $9 billion at the end of 2025, aims to enhance its control over performance and costs while ensuring supply chain stability.
Currently dependent on hardware from major players like Google and Amazon, Anthropic's move toward custom silicon production reflects a broader trend where AI firms are striving for independence from suppliers like Nvidia. Other companies, including Meta Platforms and OpenAI, are also pursuing similar paths, with Meta investing in MTIA chips in partnership with Broadcom, and OpenAI collaborating on a $10 billion project for custom processors set for production in late 2026.
However, the journey toward custom chip development is not without its challenges. Industry experts estimate that designing 3nm chips could cost between $400 million and $600 million, while the manufacturing costs for high-end chips may exceed $3,320 per unit. Additionally, the rapid pace of technological advancement poses a risk, as custom solutions can quickly become obsolete, complicating Anthropic's efforts to establish a foothold in this competitive market.