Nvidia’s dominance in AI and gaming GPUs has long been undeniable. Now, the company is quietly making a play for CPU territory—starting with a custom-designed Arm chip called Vera. While the chip is initially earmarked for data centers, its arrival signals a broader strategy that could eventually reshape how processors are built, even if it won’t be powering gaming PCs anytime soon.
The Vera CPU, announced alongside a $2 billion investment in cloud infrastructure provider CoreWeave, marks Nvidia’s first standalone CPU product. Unlike its previous Arm-based Grace processor—licensed from Arm’s Neoverse V2 design—Vera is built on Nvidia’s own custom cores, codenamed Olympus. With 88 cores and a power draw of just 50 W, it delivers twice the performance of Grace, which featured 72 cores. That efficiency is striking: an 88-core chip consuming only half the power of many high-end server CPUs.
But where does Vera fit in the grand scheme? For now, it’s a data center play, competing with Intel’s Xeon and AMD’s EPYC chips, as well as cloud-native offerings like Amazon’s Graviton. CoreWeave’s $6 billion hardware commitment—partially fueled by Nvidia’s investment—underscores the chip’s immediate market focus: AI workloads and high-performance computing. Yet the bigger question lingers: Will Vera’s architecture ever trickle down to consumer devices?
Nvidia has hinted at a future for Arm in PCs with its upcoming N1X processor, but that chip is based on Arm’s Cortex-X925 and Cortex-A725 cores—not Olympus. The N2 successor, if it follows rumors, might bridge that gap, potentially adopting Nvidia’s custom design. However, even if the hardware arrives, software support—particularly for games—remains a hurdle. Arm’s PC ecosystem is still in its infancy, and Nvidia’s push will require more than just raw performance to succeed.
- Cores: 88 (custom Olympus architecture)
- Power: 50 W
- Performance: Double that of Grace (72-core)
- Target market: Data centers (AI/cloud workloads)
- Licensing: Custom Nvidia cores (not Arm Neoverse)
The Vera CPU’s arrival is a calculated move. By controlling its own core design, Nvidia can optimize for AI acceleration—a strength it already wields in GPUs. Yet the transition to consumer markets will demand more than technical prowess. For now, Vera is a data center tool, but its existence hints at a future where Nvidia’s vision for Arm extends far beyond the server rack.
One thing is clear: The chipmaker isn’t content with dominating just one segment. If Vera succeeds, it could redefine how CPUs are built—whether in data halls or living rooms remains to be seen.
