AMD and Anthropic have announced a strategic partnership that will deploy up to 2 gigawatts of AMD Instinct MI450 Series GPUs in AMD Helios systems. Anthropic will use that enormous pool of compute to scale its AI product, Claude. The first deployment phase begins in the first half of 2027, pointing to a long-term strategy on both sides.
The deal
AMD and Anthropic have signed a multi-year agreement under which Anthropic will get up to 2 gigawatts of AMD Instinct MI450 Series GPUs, integrated into AMD Helios rack-scale systems. Those systems will pair AMD Instinct MI455X GPUs with AMD EPYC "Venice" processors and AMD Pensando networking, plus the ROCm software stack. The first gigawatt is scheduled for the first half of 2027. On top of that, AMD is committing to a strategic investment in Anthropic of up to $5 billion.
The partnership also includes engineering collaboration: the teams will use Claude to optimise workloads for AMD Instinct GPUs and speed up ROCm development. AMD also plans to roll out Claude broadly across its engineering and product teams. This deepens an existing relationship, since Anthropic already runs on AMD Instinct MI355X GPUs.
Why it matters
The partnership is a landmark for several reasons. First, it shows how fast demand for AI training and inference compute is growing — it is now measured in gigawatts. This is not simply a chip purchase but deep hardware-software integration, which is critical for squeezing out performance. For AMD it means a significantly stronger position in the AI infrastructure market; for Anthropic, the capacity it needs to compete with the other AI leaders.
Second, AMD's investment in Anthropic and the engineering collaboration show the companies are not just selling and buying hardware but building long-term strategic alliances. Using Claude to optimise ROCm and deploying Claude inside AMD underlines the mutual benefit and the push for synergy. It could yield more efficient, more specialised AI systems that other players on the market simply won't have.
What it means in practice
- Access to capacity: for developers and companies using Claude, this potentially means faster access to frontier models and higher throughput, since Anthropic will be able to scale its operations.
- Software optimised for AMD: the engineering collaboration between AMD and Anthropic could improve the ROCm software stack, making it more attractive to AI developers working on AMD GPUs.
- Hardware diversification: by running on multiple hardware platforms, Anthropic shows the value of flexibility and independence from a single supplier. That may push other AI developers to consider alternative hardware.
- Investment in AI infrastructure: AMD's large-scale investment underlines that the race for AI compute is still on and will only intensify. That creates new openings for companies building AI hardware and software.
- The future of Claude: more compute lets Anthropic keep developing Claude, potentially shipping new, more powerful versions and expanding what it can do across applications.