Nvidia targets server CPUs with Vera, raising new competitive pressure on Intel and AMD

Nvidia is indicating that it wants a bigger share of the data center market beyond GPUs, and that ambition is landing squarely on the shoulders of Intel and AMD. A recent report highlighted Nvidia’s “Vera CPU” effort as part of a broader strategy to compete in server and other compute workloads that have long been dominated by Intel’s x86 processors and AMD’s similar ecosystem.

The new CPU initiative is wrapped into Nvidia’s “Vera Rubin platform,” which the company has been described as positioning for “agentic AI” workloads. In one industry summary of Nvidia’s updates, the Vera Rubin platform was said to include the Vera CPU, along with Nvidia’s BlueField-4 STX accelerated storage infrastructure. Those components matter because modern AI deployments typically rely on more than raw compute, with networking and storage increasingly treated as part of the same performance system.

According to commentary citing Nvidia’s expectations, the company is targeting meaningful CPU revenue within a near-term fiscal window. The figure repeated in the coverage is $20 billion in CPU revenue for the current fiscal year. Because the number is attributed to a discussion rather than presented as a directly quoted, filed forecast in the materials reviewed here, it should be treated as forward-looking commentary rather than a confirmed earnings line item.

The report also frames Nvidia’s CPU bet as a direct challenge to the so-called “duopoly” that Intel and AMD have maintained for server central processing units. Intel and AMD together account for most of the x86 server CPU market, the coverage notes, with decades of software compatibility and ecosystem lock-in. Nvidia’s counter-argument is effectively integration, bundling, and platform-level performance rather than attempting to displace x86 one-for-one.

At the center of the CPU narrative is the Vera CPU itself, described as an ARM-based processor created in partnership with MediaTek. Nvidia’s approach, as characterized in the coverage, is to marry an ARM-style CPU with Nvidia GPU technology inside a larger platform. For Nvidia, the strategic logic is that AI systems are increasingly sold as tightly tuned compute stacks, where the total system performance can outweigh any single component’s market share.

The competitive implications for Intel and AMD are less about immediate replacement of their server platforms and more about where new demand could flow if buyers prioritize end-to-end platforms. If Nvidia can deliver compelling performance-per-watt, lower total cost of ownership, or faster time-to-deployment for specific AI use cases, it may gain leverage with large customers even without fully dethroning x86 in every segment.

Nvidia also has a broader track record in custom and platform-oriented chip designs, particularly around data center networking and storage acceleration. The inclusion of BlueField-4 STX in the same platform description underscores that point, since network and storage bottlenecks can throttle AI training and inference even when GPU compute is abundant. That “system” framing is a key reason the CPU move is being interpreted as strategically meaningful rather than a standalone product launch.

Still, important details remain unconfirmed in the materials reviewed here. The coverage does not provide direct excerpts from Nvidia’s own investor relations filings or a detailed specification sheet for Vera CPU performance metrics, customer design wins, production volumes, or timing beyond the platform update references. For investors and buyers, the next test will be tangible adoption indicates, such as named deployments, benchmark results in relevant workloads, and whether the CPU revenue trajectory becomes a company-stated target in official guidance rather than an estimate discussed in third-party analysis.