AMD's 2nm EPYC rollout signals broader CPU role in AI data centers

AMD is rolling out new EPYC processors built on a 2nm manufacturing process. These advanced server chips are designed to expand the role of traditional CPUs within artificial intelligence (AI) data centers, which have historically relied more heavily on GPUs for AI workloads.

This development matters because it suggests a potential shift in how AI data centers are architected. If CPUs can take on a larger share of AI-related tasks, it could diversify the types of hardware solutions required for AI infrastructure and potentially alter current spending patterns.

The mechanism involves AMD's new EPYC CPUs offering improved performance and efficiency for certain AI workloads, making them more competitive against GPU-centric solutions. This could lead data center operators to integrate more CPUs alongside or instead of some GPUs for specific AI computations.

This move primarily impacts AMD (AMD) by potentially increasing demand for its server CPUs in the growing AI data center market. It could also affect NVIDIA (NVDA), which currently dominates the AI accelerator market with its GPUs, by introducing a stronger CPU-based alternative for some AI tasks.