Nvidia has provided restricted access to its 88-core [1] Vera CPU for a first round of Linux benchmarks during GTC 2026.
The move signals Nvidia's intent to challenge the long-standing dominance of AMD and Intel in the server market by offering a standalone processor designed for AI and analytics workloads.
Vera is a first-generation [1] custom server core. In selected tests, the processor performed comparably to the AMD EPYC and Intel Xeon lines [2]. While some reports suggest the CPU beats all competitors in tests sanctioned by Nvidia [3], other benchmark data shows it runs very close to its rivals in most general tests [2].
Nvidia plans to produce only one 88-core [4] SKU for the Vera line. Despite the limited variety of hardware, company leadership expects the business to generate billions of dollars [4] in revenue.
The company is making the Vera CPU available as a standalone offering [5]. CoreWeave will be the first customer to receive access to the new hardware [5].
The processor was showcased in San Jose, California, at the GTC 2026 developer conference. The company said it intends to use the chip to strengthen its position in the data-center market by integrating its own CPU architecture with its existing AI ecosystem [6].
“Vera is a first-generation custom server core.”
By entering the high-core-count CPU market, Nvidia is attempting to reduce the reliance of data centers on third-party silicon. If Vera can maintain parity with Intel and AMD while offering tighter integration with Nvidia GPUs, the company could capture a larger share of the total server spend, moving beyond being just a chip provider to a full-stack infrastructure vendor.
