Having established itself as the most valuable technology company due to overwhelming demand for its Graphics Processing Unit (GPU)—the core chip used for artificial intelligence development—Nvidia is expanding its offerings. The chip giant has now introduced its own Central Processing Unit (CPU), named Vera, which aims to challenge market leaders Intel and Advanced Micro Devices (AMD) in the rapidly growing AI server sector.
Introduction of the Vera Data Center CPU
While Nvidia initially dominated the market with GPUs for creating and deploying artificial intelligence, the company is now shipping its own CPUs. These processors could potentially be adopted by cloud service providers as alternatives to chips manufactured by Intel and AMD, initiating a new competitive phase in AI server technology.
On Tuesday, Nvidia released comprehensive details about its data center CPU, Vera. The information included specific benchmarks and architectural specifications necessary for potential customers to fully evaluate the chip. Company representatives reported that Vera units had already been delivered to major clients, including OpenAI, Anthropic, and SpaceX, in June.
Strategic Shift and Market Ambition
Nvidia has historically positioned itself as a trendsetter in the information technology industry. However, by entering the CPU space, it is re-entering competition with two highly established players: Intel and AMD, both of whom maintain deep relationships with major hyperscale cloud providers. This investment in CPUs reflects Nvidia’s strategy to achieve vertical integration across its systems, allowing the company to produce more components and technology within a single ecosystem annually.
The overarching goal is for Nvidia to sell complete computing power racks rather than merely individual chips. The firm states that this approach will enable engineers to maximize GPU performance, ensuring Nvidia’s systems remain a leading choice for advanced AI laboratories despite intense competition from AMD and custom-designed silicon.
Technical Superiority and Performance Focus
The shift in computing demand has been noted by industry experts. While early AI servers (such as those available when ChatGPT launched in 2022) paired multiple GPUs to a single CPU, the rise of “agentic AI”—which can operate autonomously with minimal human input—has restored critical attention to the CPU’s role in managing and feeding data to these agents.
“Agents have made CPUs ‘much more integral,’ Ian Buck, Nvidia’s vice president of hyperscale, stated during a presentation last week. “Particularly how fast a CPU can answer one question.”
Nvidia detailed that Vera is the first server CPU designed entirely from its core specifications, departing from previous designs that relied on off-the-shelf architectures provided by Arm. The company asserts that Vera was specifically engineered to resolve bottlenecks associated with AI agents and claims it offers 50% better performance for AI agents compared to traditional x86 chips made by Intel and AMD.
Previous CPUs for clouds from Intel and AMD focused on core count, but Nvidia’s chip and Olympus core focuses on single-core speed instead. Hannah Coutand, a Vera product marketer at Nvidia, said in a presentation that the chip focused on per-core speed, high memory bandwidth, and latency, “so that agents can return to their GPUs as quickly as possible and keep those GPUs, which are a very expensive and a highly valuable asset in the AI factory, as highly utilized as possible.”
The CPU is reported to be power-intensive, drawing between 250 watts and 450 watts. Furthermore, it supports substantial low-power memory—the same type used in laptops and phones—with capacity up to 1.5 terabytes per chip.
Market Projections and Competition
Industry analysts have noted that Nvidia may be pioneering a new category of CPUs that Intel and AMD currently lack a direct response to. Karl Freund, founder of Cambrian AI Research, stated that the Vera CPU is intended for intensive AI tasks, rather than traditional server operations like website serving.
The market potential is significant: Nvidia estimates the entire server CPU market could eventually reach $200 billion, contrasting with an earlier Bernstein estimate placing the mature server CPU market value at approximately $37 billion in 2025. Wolfe Research projected that the average selling price for a Vera chip would be around $5,000 and forecast Nvidia shipping roughly 1.3 million units this year.
Despite the strong showing from Nvidia (up 8%), AMD and Intel have shown robust performance in 2026. According to data reported on the source page, AMD’s stock was up 128%, while Intel’s stock increased by 149%.
From a competitive standpoint, Gartner analyst Kevin Knox stated that AMD is currently considered the strongest competitor for enterprise AI server CPUs. While Intel reportedly commands about 66.8% of the market and AMD holds approximately 33%, Knox noted that AMD has been successfully expanding its market share while cultivating strong ties with hyperscale providers.
Regarding availability, Nvidia confirmed it will sell Vera chips standalone, as well as in a configuration of two chips within a single server. Additionally, it can be deployed in liquid-cooled racks containing 256 Vera chips and paired directly with Nvidia’s GPUs in a system marketed as the Vera Rubin.
