Nvidia management told analysts at a closed-door briefing on Tuesday that its next-generation Vera server CPU will be offered as a standalone product, not only bundled with its Blackwell and Rubin GPU platforms, and that the company estimates the resulting opportunity at roughly $200 billion in annual addressable revenue. The disclosure came less than a week after Nvidia posted a record $58.3 billion in net profit for the February-April quarter.

Chief financial officer Colette Kress, speaking at the briefing, said Nvidia anticipated about $20 billion of Vera CPU revenue in calendar 2027. “Vera CPU opens a brand-new $200 billion TAM for Nvidia, a market we have never addressed before,” she said. The company has so far sold CPUs only inside its Grace-Blackwell super-chip assemblies, where the Grace CPU sits adjacent to a Hopper or Blackwell GPU on a coherent fabric.

For Intel and AMD, the entry into discrete server CPUs is the most direct strategic threat Nvidia has yet posed. Intel’s Xeon line and AMD’s EPYC line have, between them, controlled essentially the entire merchant server-CPU market for a decade. AMD reported $5.8 billion in data-centre revenue for the first quarter and has been gaining server-CPU share against Intel — momentum the new Vera positioning is likely to interrupt.

Markets read the news as confirming Nvidia’s ability to extend AI-cycle margins beyond the GPU itself. Nvidia closed 3.1 percent higher on Tuesday at a fresh record. Intel slipped 4.2 percent; AMD held roughly flat, with several analysts pointing to its broader product diversification as insulating it from immediate Vera competition.

Nvidia’s strategic logic for going standalone has been telegraphed for several quarters. The company’s software stack — CUDA, cuDNN, the Nvidia Inference Microservices — increasingly treats CPU and GPU as a coherent compute fabric, and Nvidia executives have argued that customers prefer to source the entire pipeline from a single vendor for both procurement efficiency and integration support.

Several large hyperscalers, including Microsoft Azure, AWS and CoreWeave, are already prepared to deploy Vera-based systems through their existing supply pipelines. Oracle Cloud Infrastructure, the smallest of the top-tier hyperscalers, has said it would treat the standalone Vera offering as a preferred path for its agentic AI workloads.

Wall Street’s reaction was uniformly positive. Morgan Stanley raised its 12-month Nvidia price target by 9 percent; Bernstein flagged the announcement as “the most consequential expansion of TAM Nvidia has ever signalled.”