NVIDIA Targets 1,000 Vera Rubin AI Racks Per Day as Grace CPU Momentum Accelerates
NVIDIA is preparing for a massive expansion of its AI infrastructure business, with the company reportedly aiming to build as many as 1,000 Vera Rubin racks per day. The target comes as demand for NVIDIA’s Grace CPUs continues to grow across AI servers, hyperscale data centers, and next-generation cloud computing deployments.
For years, NVIDIA was best known as a graphics processor company. Today, that image has changed dramatically. The company is no longer focused only on GPUs; it is positioning itself as a full-stack AI systems provider, offering CPUs, GPUs, networking, software, and complete rack-scale platforms designed for the world’s largest AI workloads.
That shift is becoming increasingly clear with NVIDIA’s CPU roadmap. Grace marked the company’s first major push into high-performance Arm-based CPUs for data centers, and it appears to have found strong traction in the AI market. Vera, the CPU at the heart of the Vera Rubin platform, is expected to take that strategy to a much larger scale.
According to NVIDIA’s Vice President of HPC and Hyperscale Systems, Ian Buck, the company has already shipped hundreds of thousands of Grace standalone servers. NVIDIA has also previously stated that it has shipped more than 2.5 million Grace chips, showing how quickly its CPU business has grown alongside the AI boom.
Now, NVIDIA appears to be preparing for even stronger demand with Vera Rubin. The company is working with more than 300 global partners to ensure the platform can be manufactured, assembled, and deployed at the speed required by cloud providers and AI companies. These partnerships are crucial as demand for AI computing continues to move from traditional training clusters toward larger, more complex agentic AI systems.
NVIDIA’s Senior Vice President of Hardware Engineering, Andrew Bell, has reportedly said that around a dozen manufacturing partners will be capable of producing up to 1,000 Vera racks per day. If achieved, that production rate would represent one of the most aggressive AI infrastructure ramps in the industry.
Based on rough estimates, such a scale could represent an enormous revenue opportunity, potentially reaching hundreds of billions of dollars in quarterly hardware value. However, that figure would be shared across NVIDIA and its manufacturing, supply chain, and system partners. Even so, the scale highlights how important rack-level AI systems have become to NVIDIA’s long-term growth strategy.
The Vera Rubin platform is designed to serve the next wave of AI data centers, where performance, efficiency, and system-level integration matter more than ever. Instead of selling individual chips alone, NVIDIA is increasingly focused on complete AI factories: integrated systems that combine CPUs, GPUs, memory, networking, cooling, and software into tightly optimized platforms.
This approach gives NVIDIA a major advantage in the AI infrastructure market. Customers building large AI clusters want more than raw silicon; they want complete, scalable systems that can be deployed quickly and operated efficiently. Vera Rubin is intended to meet that demand by delivering a high-performance platform built specifically for modern AI workloads.
NVIDIA began ramping volume production of the Vera Rubin platform recently, and the company is expected to become one of the world’s leading CPU suppliers as its AI server business expands. NVIDIA is reportedly aiming for around $20 billion in annual CPU revenue, a remarkable milestone for a company that was once viewed almost entirely as a GPU maker.
Competition, however, is intensifying. AMD is also preparing its next-generation EPYC Venice processors, which are expected to target high-performance servers and AI data center deployments. As more chipmakers chase the same opportunity, the battle for dominance in AI infrastructure will become even more aggressive.
The next several months could be pivotal for the server CPU and AI hardware markets. NVIDIA is betting that Vera Rubin will build on the success of Grace and become a cornerstone of future AI data centers. If the company can truly scale production to 1,000 racks per day, it would mark a major step in NVIDIA’s transformation from a chip supplier into one of the most powerful full-stack AI infrastructure companies in the world.
