Advanced Micro Devices has secured one of its largest AI infrastructure agreements to date, signing a deal in which Anthropic will deploy up to 2 gigawatts of AMD AI infrastructure while AMD commits to invest as much as $5 billion in the AI company.
The agreement strengthens AMD’s challenge to NVIDIA in the market for AI accelerators and gives Anthropic access to the computing resources it needs as demand for its AI models continues to grow. The transaction is valued at tens of billions of dollars through long-term infrastructure deployments.
Anthropic plans to begin deploying the first gigawatt of AMD’s Instinct MI450 Series GPUs during the first half of 2027. The systems will run in AMD’s Helios rack-scale platform, combining Instinct MI455X GPUs with EPYC Venice processors, Pensando networking technology and ROCm software.
The AI company will split its deployment between infrastructure it owns directly and capacity leased through cloud and neocloud providers. AMD and Anthropic are jointly selecting data center locations that will host the new systems.
“The structure of this deal tells you more than the size. To win OpenAI and Meta, AMD had to hand each of them warrants for 160 million shares, essentially paying for those commitments with equity dilution. With Anthropic, the money flows the other direction,” Brendan Burke, Research Director at the Futurum Group, told Techstrong Semi.
“AMD is investing up to $5 billion in its customer, with no warrants disclosed. That’s the posture of a supplier with leverage, not a challenger buying its way onto the roadmap. Anthropic becomes the third frontier lab to commit gigawatt-scale capacity to Instinct, and it’s the first deal where AMD didn’t have to give up a piece of itself to get it.”
Anthropic AI to Support AMD Performance
The companies will also collaborate on software optimization. Anthropic’s Claude models will be used to improve performance across AMD’s Instinct GPU platform and accelerate development of ROCm, the company’s open software stack for AI workloads. AMD also plans to deploy Claude across its engineering and product development organizations.
Anthropic already operates workloads across infrastructure supplied by Google, Amazon and NVIDIA. The AMD deployment expands that mix as leading AI developers seek additional suppliers amid sustained demand for advanced AI chips.
The agreement also advances AMD’s strategy of expanding its presence in the AI accelerator market. The company has recently announced AI infrastructure partnerships with OpenAI and Meta as it works to gain share in a market dominated by NVIDIA.
AMD is reportedly discussing additional financial support that could include backing future Anthropic data center lease commitments, a structure that has become more common as AI developers seek to finance massive infrastructure buildouts.
For Anthropic, the partnership adds another major infrastructure agreement during a year of rapid expansion. The company has previously announced compute agreements with Amazon, Google and SpaceX as it scales capacity for Claude and Claude Code.
Anthropic has said surging customer demand has periodically strained its infrastructure and affected service availability during peak usage. The company also reported that annualized revenue run rate reached $47 billion earlier this year and recently completed a funding round valuing the business at $965 billion as it prepares for a potential IPO.
