For decades, the structure of the data center market was clearly segmented: Intel and Advanced Micro Devices (AMD) supplied the central processing units (CPUs) that powered servers, while Nvidia focused on graphics processing units (GPUs), components originally designed for visual tasks but found to be ideal for the massive parallel calculations required by artificial intelligence (AI).
However, this traditional division of labor is undergoing a dramatic transformation. Nvidia has expanded its reach far beyond GPUs, venturing into networking, complete server rack solutions, software, and cloud services. With the introduction of its Vera server CPU, Nvidia is now directly competing in the market segment that long provided dominance for Intel and enabled AMD’s successful EPYC lineup.
While the immediate assumption is that Nvidia is attempting to capture CPU market share from its rivals, the more profound conclusion is that the company is redefining *how* that market share will be acquired. The success of Vera does not depend on outperforming every Intel Xeon or AMD EPYC processor in a direct benchmark comparison. Instead, Nvidia aims to convince customers that the most efficient and straightforward method for deploying AI workloads is through a complete system designed by Nvidia, which inherently includes the Vera CPU.
Nvidia’s Strategic Entry into the CPU Market
The introduction of Vera has solidified this strategic shift, moving the discussion beyond mere product roadmaps and into tangible purchasing plans. For instance, Amazon Web Services (AWS) and Nvidia have announced plans to deploy Vera CPU-based infrastructure to AWS. Furthermore, Nvidia has secured commitments from various entities, including AI laboratories, cloud service providers, and server manufacturers, indicating support for the processor.
Nvidia’s foray into server CPUs is not unprecedented. The company has utilized technology from Arm Holdings plc (NASDAQ: ARM) for years in processors used in embedded, mobile, and automotive applications. In September 2020, Nvidia entered into an agreement to acquire Arm from SoftBank for $40 billion. The acquisition was opposed by regulators due to the dependence of many Nvidia competitors on Arm technology, and the deal was eventually terminated in February 2022. During the termination, Nvidia forfeited $1.25 billion, resulting in an acquisition-related charge of approximately $1.35 billion.
Despite the failed takeover, Nvidia retained a 20-year license for the Arm architecture. The company continued its strategy by investing in Arm’s 2023 public offering and developing proprietary Arm-based processors. While Nvidia’s first major data center CPU was named Grace, Vera represents a more significant second step. Vera incorporates 88 custom Olympus cores built on the Arm architecture and is specifically engineered for AI-related tasks, including agent orchestration, data processing, analytics pipelines, sandboxed code execution, and other CPU-intensive aspects surrounding accelerated computing.
Industry Validation and Market Growth Projections
The initial list of companies expected to adopt Vera includes Anthropic, OpenAI, SpaceXAI, ByteDance, CoreWeave, and Oracle Cloud Infrastructure. Additionally, manufacturers such as Dell Technologies, Hewlett Packard Enterprise, Lenovo, and Supermicro, alongside several Asian original-design manufacturers, are developing standalone systems utilizing the Vera CPU. These commitments demonstrate that Vera is not restricted to proprietary Nvidia racks.
A key validation point is AWS. In August, AWS and Nvidia announced plans for AWS to deploy two million additional Nvidia GPUs between 2027 and 2028 and to incorporate Vera CPU-based infrastructure into its services. Given that AWS already develops its own Graviton processors, its willingness to support Vera suggests that custom CPUs and commercially available merchant processors can coexist effectively when they address different computing requirements.
The Expansive AI Data Center Market
Nvidia is entering the CPU market precisely as artificial intelligence is fundamentally reshaping data center spending. Projections estimate that global information technology spending will increase from about $6.3 trillion in 2026 to $8.6 trillion in 2030, representing an annual growth rate of approximately 8.0%. Data center systems spending is anticipated to grow even faster, rising from roughly $790 billion to over $2.0 trillion.
AI is the primary driver of this growth. Forecasts indicate that AI data center systems revenue will increase from $560 billion in 2026 to $1.68 trillion in 2030, with a Compound Annual Growth Rate (CAGR) of 31.6%. While accelerators remain the largest component, the dedicated AI CPU segment is projected to grow even more rapidly. AI CPU revenue is expected to rise from $38 billion in 2026 to $155 billion in 2030, representing a CAGR of 42.1%, significantly outpacing the 29.0% CAGR projected for accelerators. This rapid growth confirms that CPUs are taking on increasingly vital roles in inference management, data preparation, orchestration, and agentic computing, making them a critical, rapidly expanding part of the AI system.
In summary, Nvidia’s strategy is not merely a peripheral effort to diversify from GPUs. The company is positioning itself within one of the fastest-growing sectors of the AI infrastructure market, thereby expanding the total revenue it can capture from every single AI installation by integrating its own CPU, networking, software, and system components.
