AI Infrastructure 4 min read

AMD Lands 2 GW Anthropic Deployment

AMD and Anthropic unveiled a sweeping strategic partnership centered on AI infrastructure, engineering collaboration, and capital investment. Anthropic plans to deploy up to 2 gigawatts of AMD Instinct MI450 Series GPUs in AMD Helios rack-scale systems, with the first 1 GW deployment scheduled to begin in the first half of 2027. AMD also committed to make a strategic equity investment of up to $5 billion in Anthropic, marking one of the company’s largest AI ecosystem investments to date.

The deployment will use AMD Helios rack-scale systems featuring AMD Instinct MI455X GPUs, AMD EPYC “Venice” CPUs, AMD Pensando networking, and the ROCm software platform. AMD said the installation builds on Anthropic’s existing deployment of Instinct MI355X accelerators and is intended to support both large-scale AI training and inference for Claude. The announcement represents one of the largest publicly disclosed AI infrastructure commitments based on AMD hardware and further strengthens AMD’s position as an alternative supplier to NVIDIA for hyperscale AI deployments.

Beyond hardware, the companies will establish a multi-year engineering collaboration focused on software optimization. Anthropic will use Claude to optimize AI workloads running on AMD Instinct GPUs and accelerate development of AMD’s ROCm software platform. AMD also plans to broadly deploy Claude across its engineering and product development organizations to support software development and product design.

• Anthropic plans to deploy up to 2 GW of AMD Instinct MI450 Series GPUs.
• First 1 GW deployment is scheduled for the first half of 2027.
• Systems will use Helios rack-scale architecture with MI455X GPUs, EPYC Venice CPUs, Pensando networking, and ROCm software.
• Deployment expands Anthropic’s existing use of AMD Instinct MI355X GPUs.
• AMD and Anthropic will jointly optimize Claude workloads for AMD hardware and accelerate ROCm development.
• AMD will deploy Claude broadly across its engineering and product development teams.
• AMD committed to make a strategic equity investment of up to $5 billion in Anthropic.

“We are thrilled to deepen our partnership with Anthropic and deploy AMD Helios at gigawatt scale,” said Dr. Lisa Su, chair and CEO of AMD. “This collaboration brings together Anthropic’s leadership in frontier AI with the full strength of AMD high-performance computing. Together, we will accelerate AI adoption at scale and establish Helios as a major platform for the next generation of AI infrastructure.”

🌐 Analysis: This announcement significantly expands AMD’s AI infrastructure strategy beyond silicon into long-term ecosystem partnerships. The combination of a multi-gigawatt hardware commitment, software co-development around ROCm, and a multibillion-dollar equity investment demonstrates AMD’s effort to secure sustained AI accelerator demand while strengthening its software ecosystem. The deal also reflects a broader industry trend in which frontier AI model developers increasingly diversify compute suppliers rather than relying on a single accelerator vendor.

Anthropic Infrastructure Announcements 2026
Major compute capacity, silicon and AI data center agreements announced through July 22, 2026
DATE / PARTNERSCAPACITY & PLATFORMINFRASTRUCTURE COMMITMENT
July 22, 2026
AMD + Anthropic
Up to 2 GW
MI450 Series GPUs
Helios rack-scale systems
The first 1 GW deployment will begin in the first half of 2027 using MI455X GPUs, EPYC Venice CPUs, Pensando networking and ROCm software. AMD also committed to invest up to $5 billion in Anthropic.
May 6, 2026
SpaceXAI + Anthropic
300+ MW
220,000+ NVIDIA GPUs
Colossus 1
Anthropic secured all available compute capacity at the Colossus 1 data center for Claude training and inference. The companies also outlined exploratory work on potential multi-gigawatt orbital AI compute.
April 20, 2026
Amazon AWS + Anthropic
Up to 5 GW
Trainium2–Trainium4
Graviton CPUs
Anthropic committed more than $100 billion over 10 years to AWS infrastructure. Nearly 1 GW of Trainium2 and Trainium3 capacity is scheduled to operate by the end of 2026, with additional inference capacity planned across Asia and Europe.
April 6, 2026
Google + Broadcom + Anthropic
5 GW
Next-generation Google TPUs
Broadcom silicon
The companies expanded Anthropic’s access to Google’s next-generation TPU infrastructure, with capacity expected to begin operating in 2027. Most of the new infrastructure will be located in the United States.
2026 Capacity Program
Fluidstack + Anthropic
$50B Program
U.S. AI data centers
NVIDIA GPU infrastructure
Anthropic’s previously announced U.S. infrastructure program calls for Fluidstack-developed data center sites to begin entering service during 2026. The program targets approximately 800 permanent jobs and 2,400 construction jobs.
Strategic Capacity
Microsoft + NVIDIA + Anthropic
$30B Azure Capacity
NVIDIA GPUs
Microsoft Azure
The multiyear partnership provides Anthropic with NVIDIA-based Azure capacity for Claude training and deployment. It complements the company’s AWS Trainium, Google TPU, SpaceXAI and AMD infrastructure agreements.
Infrastructure strategy: Anthropic is distributing Claude workloads across AMD Instinct GPUs, NVIDIA GPUs, AWS Trainium and Google TPUs while securing capacity from hyperscalers, neocloud operators and dedicated AI data center platforms.

Share this article

Help others discover this reporting.

Explore More