AMD and Anthropic announced a strategic partnership on July 22 under which Anthropic will deploy up to 2 gigawatts of AMD Instinct MI450-series GPUs to train and serve Claude — and AMD committed to a future strategic equity investment of up to $5 billion in the AI lab. AMD shares rose about 1% on the news.

The hardware

The compute arrives as AMD Helios rack-scale systems: MI455X GPUs paired with 6th-gen EPYC "Venice" CPUs, Pensando networking and the ROCm software stack. The first gigawatt is slated to come online in the first half of 2027. The deal builds on Anthropic's existing use of AMD's MI355X chips.

Supplier and shareholder

The equity commitment — up to $5 billion, which the Wall Street Journal reports is tied to deployment milestones — makes AMD a direct shareholder in one of its biggest customers. That is a cleaner alignment than AMD's OpenAI arrangement, which used warrants. The chip purchases themselves are worth tens of billions of dollars, per WSJ.

Claude inside AMD

The companies also agreed to a multi-year engineering collaboration: Anthropic will optimize Claude for Instinct GPUs and help accelerate ROCm development, while AMD adopts Claude broadly across its own engineering and product teams. "Access to compute is central to keeping Claude at the frontier," said Anthropic co-founder Tom Brown.

The compute roster grows

Anthropic now spreads its training and inference across Google, Amazon, Broadcom, SpaceX's AI unit, TeraWulf — and AMD. CEO Lisa Su said she was "thrilled to deepen our partnership with Anthropic and deploy AMD Helios at gigawatt scale." With Meta and OpenAI deals already signed, AMD's announced AI compute commitments now total roughly 14 gigawatts.