Nvidia Claims Its New Vera ARM CPU Is 80% Faster Than Leading x86 CPUs

Nvidia’s Next Big Bet: Redefining the CPU for the AI Era

For years, Nvidia has dominated the artificial intelligence boom through its powerful GPUs, which have become the backbone of modern AI training and inference. Now, the company is making an even bolder move: challenging the long-standing dominance of x86 processors in the data center.

At GTC Taipei 2026, Nvidia unveiled the Vera CPU, a custom ARM-based processor designed specifically for the rapidly growing world of agentic AI. According to Nvidia, Vera delivers an average 1.8x performance improvement—or roughly 80% faster task completion—compared with leading x86 CPUs, positioning it as one of the company’s most ambitious infrastructure products to date.

The announcement signals a major shift in how data centers may be built in the coming years, as AI companies increasingly seek hardware optimized not just for training large models but also for running autonomous AI agents capable of reasoning, coding, using tools, and executing complex workflows.

Why Nvidia Built Vera

The AI industry is entering a new phase.

Early AI systems primarily focused on answering questions or generating content. Newer systems, often described as agentic AI, are designed to perform multi-step tasks independently. These systems can analyze information, write and execute code, interact with software tools, and evaluate outcomes before deciding what to do next.

Such workloads place enormous demands on CPUs, which coordinate data movement, execute application logic, run programming environments, and manage orchestration tasks around GPUs.

Nvidia CEO Jensen Huang believes this trend fundamentally changes computing requirements.

“AI agents will be the largest users of computing,” said Jensen Huang, founder and CEO of NVIDIA. “Vera is the first CPU designed for that future — built to run agentic AI at hyperscale with extraordinary performance, efficiency and programmability.”

Rather than adapting a traditional server processor for AI workloads, Nvidia developed Vera specifically for this emerging market.

Inside the Vera Architecture

Vera is a massive processor built around Nvidia’s new Olympus CPU cores, which use the ARM instruction set architecture.

Key specifications include:

• 88 Olympus ARM cores
• 176 threads per socket through Spatial Multithreading
• Support for up to 1.5TB of LPDDR5X memory
• Up to 1.2TB/s memory bandwidth
• 1.8TB/s NVLink-C2C connectivity when paired with Nvidia GPUs

The processor is designed to accelerate workloads such as:

• Agentic AI execution
• Reinforcement learning
• Data analytics
• Database processing
• Code compilation
• Python and Java workloads
• AI orchestration systems

According to Nvidia, Vera’s architecture enables it to process instructions more efficiently, anticipate workload behavior, and move data rapidly across large-scale AI environments. This helps reduce bottlenecks that often occur when CPUs feed data to AI accelerators.

Beyond CPUs: Vera as Part of the Rubin Platform

Vera is not arriving alone.

The processor serves as the CPU component of Nvidia’s new Vera Rubin platform, which combines Vera CPUs with next-generation Rubin GPUs to create highly integrated AI infrastructure.

One flagship configuration is the NVIDIA Vera Rubin NVL72, which includes:

• 36 Vera CPUs
• 72 Rubin GPUs
• NVLink-C2C interconnects delivering up to 1.8TB/s communication bandwidth between CPUs and GPUs

This architecture is designed to eliminate traditional CPU-GPU bottlenecks by allowing processors and accelerators to communicate at unprecedented speeds.

The result is a platform optimized for massive AI factories, where thousands of GPUs work together to train models and run AI services at scale.

Massive Scale for AI Factories

Nvidia is also introducing standalone Vera-based infrastructure.

The company has designed a dedicated Vera CPU Rack containing:

• 256 Vera CPUs
• 22,528 CPU cores
• 45,056 threads

Such systems are intended for enterprise AI deployments, cloud computing environments, data processing clusters, and reinforcement learning platforms.

The scale reflects Nvidia’s belief that future AI infrastructure will require enormous amounts of CPU compute alongside GPUs.

Industry executives increasingly agree. As agentic AI workloads become more common, CPUs are playing a larger role in coordinating tasks, managing workflows, and handling supporting computations around AI accelerators.

Taking Aim at Intel and AMD

Although Nvidia did not publicly identify which processors it used in its comparisons, the performance claims clearly position Vera against server CPUs from Intel and AMD, the two dominant players in the x86 market.

Nvidia states that Vera delivers:

• Up to 1.8x faster task completion than leading x86 processors
• Significant gains in energy efficiency
• Higher throughput for AI-centric workloads

The company also cited benchmarking results from Phoronix, which reportedly showed Vera leading across workloads including:

• Code compilation
• Python execution
• Java applications
• Database processing
• Agentic AI tasks

If these performance advantages hold up in independent testing, Vera could become one of the most serious challenges yet to Intel and AMD in the data center CPU market.

Major Customers Are Already On Board

Perhaps the most significant aspect of the announcement is the list of organizations planning to deploy Vera.

Among the early adopters are some of the world’s largest AI companies:

• Anthropic (Claude)
• OpenAI (ChatGPT)
• SpaceXAI (Grok)

Major hyperscalers are also exploring or deploying the platform, including:

• ByteDance
• CoreWeave
• Oracle Cloud Infrastructure
• Lambda
• Nebius
• Nscale

Anthropic’s Head of Compute, James Bradbury, emphasized the importance of expanding compute resources for advanced AI systems:

“Scaling compute is an important accelerant for the growth of models. We’re excited to see Vera emerge as a promising part of the ecosystem when solving for agentic workloads.”

Oracle also highlighted Vera’s role in supporting next-generation AI environments and high-throughput reasoning workloads.

The New York Stock Exchange Sees Potential

AI companies are not the only organizations interested in Vera.

The New York Stock Exchange (NYSE) is working with Redpanda and HPE to explore infrastructure built around Nvidia’s new CPU technology.

NYSE President Lynn Martin noted that the exchange processes more than 1.1 trillion messages every day, making latency and throughput critical priorities.

“The NYSE processes more than 1.1 trillion messages per day, and in collaboration with Redpanda and HPE, using NVIDIA Vera CPUs, we will be scaling our capacity while further optimizing latency to power a high-performance, resilient and AI-ready market infrastructure.”

The interest from financial markets illustrates that Vera’s potential extends beyond AI labs into industries that depend on extremely high-performance computing.

A Growing Ecosystem of Hardware Partners

Nvidia is not building Vera systems alone.

Several of the world’s largest server manufacturers plan to offer Vera-based products, including:

• Dell Technologies
• HPE
• Lenovo
• Supermicro

Additional system builders include:

• ASUS
• Foxconn
• Gigabyte
• Pegatron
• Quanta Cloud Technology
• Wistron
• Wiwynn
• Compal

This broad ecosystem support could accelerate adoption by making Vera available through established enterprise hardware channels.

What Vera Means for the Future of Computing

The launch of Vera represents more than a new processor.

It reflects a broader transformation in data center design. For decades, general-purpose x86 CPUs formed the foundation of enterprise computing. Nvidia is now arguing that AI-driven workloads require a fundamentally different approach.

By combining custom ARM CPUs, advanced GPUs, high-bandwidth memory, and ultra-fast interconnects into a tightly integrated platform, Nvidia is attempting to create an end-to-end AI infrastructure stack that competitors may struggle to match.

Industry analysts estimate that the addressable CPU market for AI infrastructure could reach hundreds of billions of dollars over time, making the CPU market one of Nvidia’s most important future growth opportunities.

Conclusion

Nvidia’s Vera CPU marks a major escalation in the battle for AI infrastructure leadership. With claims of 80% faster performance than leading x86 processors, an ARM-based architecture optimized for agentic AI, and backing from companies such as OpenAI, Anthropic, Oracle, and the New York Stock Exchange, Vera is positioned as far more than a supporting component in Nvidia’s ecosystem.

Whether the processor ultimately lives up to Nvidia’s ambitious performance claims will depend on real-world deployments and independent benchmarks. Yet one thing is already clear: the AI boom is no longer just a GPU story. The CPU is becoming a strategic battleground again, and Nvidia intends to be one of its leading contenders.