Nvidia‘s entry into the Central Processing Unit (CPU) market with its Vera chip is fundamentally altering the established competitive landscape that historically pitted Intel and AMD against each other. Rather than attempting to defeat established processors in direct comparisons, Nvidia is repositioning itself by selling complete, integrated AI computing systems.
The Shift in Computing Strategy
Historically, the data center market operated with a clear division of labor: Intel Corporation and Advanced Micro Devices Inc. supplied the CPUs, while Nvidia provided Graphics Processing Units (GPUs), which were initially designed for graphics but proved highly effective for the massive parallel processing required by artificial intelligence (AI).
This traditional model is now changing. Nvidia has expanded its offerings to include networking components, entire server rack solutions, software, and cloud services. The introduction of the Vera server CPU moves the company directly into the domain that provided decades of dominance to Intel and enabled AMD’s successful EPYC product line.
However, the core insight from this market shift is that Nvidia is not focused on winning isolated processor benchmarks. Instead, the company is aiming to convince customers that the simplest and most effective way to deploy AI is by purchasing an entire system designed and sold by Nvidia, which already incorporates the Vera CPU.
This strategic pivot has gained momentum since the unveiling of Vera. Major players like Amazon Web Services (AWS) and Nvidia have announced plans to deploy Vera CPU-based infrastructure within AWS. Furthermore, Nvidia has secured interest from numerous AI laboratories, cloud providers, and server manufacturers who plan to support the processor.
Nvidia’s History with Arm Architecture
Nvidia’s interest in server CPUs predates the Vera chip. The company has long utilized technology from Arm Holdings plc in processors used for mobile, robotics, embedded, and automotive applications. This interest peaked in September 2020, when Nvidia agreed to acquire Arm from SoftBank for $40 billion, giving it ownership of a core architectural foundation used widely across the semiconductor industry.
Although regulatory hurdles caused the acquisition to terminate in February 2022, resulting in Nvidia forfeiting $1.25 billion and recording an acquisition-related charge of approximately $1.35 billion, the company retained a 20-year license for Arm architecture. Subsequently, Nvidia invested approximately $100 million in Arm’s 2023 public offering, maintaining a continuing strategic and engineering commitment through proprietary Arm-based development.
The Vera CPU, which contains 88 custom Olympus cores based on the Arm architecture, represents a significant evolution from Nvidia’s earlier data-center CPU, Grace. Vera was specifically engineered to manage the diverse workloads associated with AI, including agent orchestration, data analytics pipelines, data processing, sandboxed code execution, and CPU-intensive tasks surrounding accelerated computing.
Market Validation and Adoption
Vera’s adoption is being validated by a range of key industry partners. These include Anthropic, OpenAI, SpaceXAI, ByteDance, CoreWeave, and Oracle Cloud Infrastructure, which are listed among the initial companies planning to adopt the technology. Additionally, major OEMs such as Dell Technologies, Hewlett Packard Enterprise, Lenovo, and Supermicro, alongside several Asian original-design manufacturers, are developing standalone systems featuring the Vera CPU.
Crucially, the most significant validation comes from AWS. Following an announcement in August, AWS plans to deploy two million additional Nvidia GPUs between 2027 and 2028 and simultaneously integrate Vera CPU-based infrastructure into its services. AWS’s willingness to support Vera, given that it already develops its own Graviton processors, demonstrates that custom and merchant CPUs can coexist effectively by addressing different computational requirements.
The Broader AI Infrastructure Opportunity
Nvidia’s market entry coincides with a period of massive growth in data center spending, fueled by the AI boom. The Information Network estimates that global information technology spending will climb from about $6.3 trillion in 2026 to $8.6 trillion by 2030, representing an annual growth rate of roughly 8.0%. Data-center system spending is projected to rise from approximately $790 billion to over $2.0 trillion.
The primary driver of this growth is AI. The market research suggests that AI data-center system revenue will escalate from $560 billion in 2026 to $1.68 trillion in 2030, achieving a Compound Annual Growth Rate (CAGR) of 31.6%. While AI accelerator revenue is projected to grow from $350 billion in 2026 to $970 billion in 2030, the dedicated AI CPU revenue stream is growing even faster, moving from $38 billion in 2026 to $155 billion in 2030, reflecting a CAGR of 42.1%.
This rapid growth in CPU revenue does not imply that CPUs will surpass GPUs in total dollar value, but rather that CPUs are becoming one of the fastest-expanding and most critical components within the overall AI system architecture. By offering Vera, Nvidia is expanding the revenue it can capture from each AI installation across its existing product lines—including GPUs, networking, interconnects, systems, and software—rather than simply replacing GPU revenue with CPU revenue.
