AMD has entered an agreement to invest up to $5 billion in Anthropic, which includes a commitment to deploy two gigawatts of AI infrastructure using AMD hardware.
Key facts
- •AMD’s investment of up to $5 billion is contingent upon Anthropic meeting deployment milestones.
- •The agreement includes the deployment of two gigawatts of capacity using AMD’s Instinct MI450-series accelerators.
- •The first gigawatt of capacity is expected to be deployed in the first half of 2027.
- •AMD will integrate Claude into its own engineering and product-development teams as part of the partnership.
- •Anthropic currently secures capacity through multiple providers, including Amazon, Google, and SpaceX.
AMD has announced a strategic agreement to invest up to $5 billion in AI developer Anthropic. The deal involves a large-scale hardware order, with Anthropic planning to deploy two gigawatts of capacity using AMD's Instinct MI450-series accelerators. Deployment of the first gigawatt is scheduled to begin in the first half of 2027.
By the numbers
Infrastructure and Hardware Details
Anthropic will utilize AMD’s Helios platform, which integrates MI455X GPUs, EPYC “Venice” processors, Pensando networking, and ROCm software. This deployment is intended to support the training and serving of Anthropic’s Claude AI models. The agreement also establishes a multi-year engineering program where the companies will use Claude to optimize workloads for AMD accelerators and further develop the ROCm software platform.
Strategic Investment Structure
The $5 billion investment is tied to specific deployment milestones, though the companies have not disclosed the exact conditions or the valuation of the equity. Unlike previous agreements AMD reached with OpenAI and Meta, which provided those companies with rights to acquire shares based on performance targets, this deal involves a direct equity commitment from AMD. The transaction combines this financial arrangement with a separate hardware procurement agreement.
Broader AI Capacity Strategy
Anthropic continues to diversify its hardware usage, maintaining partnerships with Amazon, Google, and others. The company currently utilizes Nvidia GPUs, Amazon Trainium processors, and Google tensor processing units. Anthropic’s compute strategy includes direct deployments, cloud agreements, and leased data-center facilities, including capacity at SpaceX’s Colossus 1 site.
Advertisement
This article was independently rewritten by ManyPress editorial AI from reporting originally published by Artificial Intelligence News.
