AMD invests up to $5 billion in AI firm Anthropic

Semiconductor giant AMD has announced a significant strategic investment of up to $5 billion in artificial intelligence company Anthropic. This collaboration is designed to bolster next-generation AI infrastructure and deepen the integration of AMD's high-performance computing solutions within Anthropic's operations.

As part of the agreement, Anthropic will adopt AMD's comprehensive AI hardware system, Helios. This system integrates AMD's Instinct MI455X Graphics Processing Units (GPUs) and EPYC 'Venice' Central Processing Units (CPUs), among other components. AMD states that the Helios system is engineered for advanced AI training and inference, offering the performance, efficiency, and scalability required for the most demanding AI workloads.

“

The collaboration brings together Anthropic’s leadership in cutting-edge AI with AMD’s high-performance computing power, and we expect it will accelerate the adoption of AI at scale and establish Helios as a key platform for the next generation of AI infrastructure.

Furthermore, the two companies will collaborate to leverage Anthropic's AI model, Claude, to optimize workloads for AMD's Instinct GPUs. This partnership also aims to accelerate the development of AMD's ROCm software platform. AMD plans to extensively utilize Claude across its engineering and product development teams, underscoring the strategic importance of this investment.

AMD CEO Lisa Su highlighted that the collaboration merges Anthropic's expertise in cutting-edge AI with AMD's high-performance computing power. She expressed confidence that the partnership will expedite the widespread adoption of AI and establish Helios as a pivotal platform for future AI infrastructure. Tom Brown, co-founder and chief of computing at Anthropic, noted that the agreement secures necessary capacity and optimization for training and running Claude, emphasizing the benefit of running workloads on diverse hardware.

“

The fact that we are able to run the system on a diverse range of hardware allows us to assign the right workloads to the right hardware.