Anthropic Evaluates AMD Chips to Expand Claude Computing Capacity

Anthropic is evaluating AMD hardware as it seeks more computing capacity to train and run Claude, its family of large language models. A deal could help the AI developer diversify its chip supply and reduce its reliance on existing infrastructure, while giving Advanced Micro Devices a potentially important customer as it challenges Nvidia’s dominance in the market for AI accelerators.

Publicly available GitHub code indicates Anthropic is testing software compatibility and networking integration for AMD’s architecture, a preliminary step toward potentially buying hardware or renting AMD-based computing capacity. As of July 2026, neither company had announced a formal agreement, chip model, order volume, contract value or deployment schedule. The work remains at the technical evaluation stage, and the code does not establish that a commercial transaction will follow.