AMD has landed one of the largest AI infrastructure deals announced this year, securing an agreement to supply Anthropic with AI servers reportedly worth tens of billions of dollars while planning to invest up to $5 billion in the Claude developer.
Under the multiyear partnership, Anthropic will deploy up to 2 gigawatts of AMD Instinct MI450 Series GPUs in AMD’s Helios rack-scale systems, with the first gigawatt expected to come online in the first half of 2027. Reuters reported the server agreement is worth tens of billions of dollars, although neither company disclosed a financial value.
The partnership extends beyond chip sales. The companies will work together over multiple years to optimize Claude for AMD Instinct GPUs and speed development of AMD’s ROCm software platform. AMD also plans to deploy Claude across its engineering and product development teams.
Anthropic already runs workloads on Nvidia GPUs, Google’s TPUs, and Amazon’s Trainium chips. Adding AMD expands its hardware mix as demand for Claude continues to grow. According to Reuters, Anthropic will use some of the new systems in its own data centers while accessing additional capacity through cloud providers, helping it secure the computing power needed to train and run future AI models.
A bigger challenge to Nvidia
The deal strengthens AMD’s position in the AI infrastructure market, where Nvidia remains the dominant supplier. Nvidia still controls the vast majority of the AI accelerator market, but AMD has steadily won larger commitments from frontier AI developers.
Industry analysts see the agreement as another sign that major AI companies are looking to diversify their hardware suppliers rather than relying on a single vendor. Reuters cited Emarketer analyst Jacob Bourne, who said the win reinforces AMD’s position as Nvidia’s closest competitor in AI accelerators.
The agreement also follows AMD’s previously announced multiyear partnership with OpenAI, underscoring the company’s strategy of securing large AI customers through long-term infrastructure commitments.
The larger picture
The AMD-Anthropic agreement highlights how AI companies are increasingly signing long-term infrastructure deals years before the hardware is deployed. Building AI systems at the gigawatt scale requires early planning, significant financing, and partnerships across chipmakers, cloud providers, and data center operators.
For AMD, landing another hyperscale AI customer adds momentum to its effort to chip away at Nvidia’s dominance. For Anthropic, the agreement reduces the risk that limited computing capacity could slow the growth of Claude as enterprise adoption continues to rise.
The strategy also comes with tradeoffs. Massive AI infrastructure projects require billions of dollars in capital, depend on future data center construction, and face growing scrutiny over power consumption and community concerns surrounding large-scale data centers. Those factors could influence how quickly these ambitious deployment plans become reality.
Whether all of the planned capacity comes online on schedule will depend on chip production, data center construction, power availability, and continued demand for AI services. But the agreement underscores a broader shift across the industry: securing long-term computing capacity is becoming just as important as building the next generation of AI models.
Other News: SpaceX may soon be doing more than launching rockets. The company is reportedly in talks to supply AI computing power to the Pentagon.
