Nvidia targets Intel and AMD in server CPUs

Nvidia dominates advanced GPUs used for AI workloads, with market share often projected at more than 90% for AI model training. Estimates for its share of AI inference applications often range between 60% and 75%, but the company has historically had a smaller presence in CPUs used for some inference tasks and broader server workloads.

That position is changing as more AI processing shifts to CPUs and operational AI applications expand. Nvidia disclosed in its fiscal Q2 2027 report that its Grace CPU family had generated $5 billion in sales over the trailing 12-month period. The company expects data center CPU revenue to more than double from this fiscal year to the next fiscal year.

Nvidia has entered full production of its Vera CPUs, which are designed to pair with Vera Rubin GPUs and also sell as stand-alone products. The company says Vera delivers performance and efficiency advantages over other data center CPUs in key areas, creating a potential challenge for Intel and AMD as Nvidia seeks a larger role across data center compute hardware.