NVIDIA Vera CPU Benchmarks Outperform Intel, AMD x86 Rivals

NVIDIA’s Vera CPU, its first fully in-house Arm-based processor for agentic AI, has emerged as a serious competitor to Intel and AMD in early performance benchmarks. Phoronix’s latest tests, published May 26, reveal Vera delivering a 1.5x overall performance advantage over a 128-core x86 processor, solidifying NVIDIA’s ambitions to dominate AI-focused data center infrastructure.

The Vera CPU packs 88 custom Olympus cores, supporting the Armv9.2 instruction set, and achieves up to 1.2 TB/s of memory bandwidth—double that of traditional CPUs. It also brings efficiency gains, consuming less than 30 watts of memory power compared to over 100 watts for DDR5-equipped alternatives. This combination of raw power and efficiency positions Vera as a purpose-built solution for AI factories managing massive workloads like code compilation, runtime orchestration, and database queries.

Breaking Down the Benchmarks

Phoronix’s tests highlighted Vera’s edge in real-world developer tasks. For example, it compiled a default Linux kernel in just 20 seconds, the fastest result in its class. On a per-core basis, Vera was twice as fast as a 128-core x86 CPU in the same task. Additionally, in memory bandwidth testing, Vera sustained 90% of its peak capacity during STREAM TRIAD tests—more efficient than any CPU Phoronix has tested to date.

Michael Larabel, founder of Phoronix, called the results a “heavy-hitting punch” against Intel Xeon and AMD EPYC processors. He noted that Vera’s performance represents “the most formidable competition” to x86 CPUs from any Arm-based architecture.

Market Implications for NVIDIA

Vera’s success comes at a critical time for NVIDIA as it looks to solidify its positioning in the AI hardware market. The company has designed Vera not as a general-purpose CPU, but as a vertically integrated processor optimized for AI workloads and tightly coupled with its next-generation Rubin GPUs. This approach directly targets hyperscale AI users like OpenAI, Anthropic, and Oracle Cloud Infrastructure, all of which have reportedly started deploying Vera-based systems.

For NVIDIA investors, Vera’s potential to disrupt the $50 billion server CPU market represents a significant growth opportunity. Shares of NVIDIA (NVDA) closed at $214.86 on May 26, reflecting a slight 0.26% dip on the day, though the stock has posted strong gains year-to-date due to its dominant position in AI hardware.

What’s Next?

NVIDIA plans to roll out the Vera CPU through partners in the second half of 2026. The processor will be available in single- and dual-socket configurations, with options for both air- and liquid-cooled deployments. Early adopter feedback indicates strong interest among leading AI companies and cloud service providers.

As the AI arms race heats up, Vera’s performance has raised the stakes for Intel and AMD. Both companies face increasing pressure to innovate as NVIDIA further entrenches itself as the go-to provider for next-generation AI infrastructure. Investors and industry stakeholders will be watching closely for additional adoption announcements and competitive responses in the months ahead.

Image source: Shutterstock