For more than a decade, Nvidia has dominated the market for graphics processing units, the specialized chips that power everything from video games to artificial intelligence training. But the company led by Jensen Huang is no longer content to reign over GPUs alone. A new deal with Meta Platforms to supply central processing units for the social media giant’s data centers marks Nvidia’s most aggressive incursion yet into territory long held by Intel and Advanced Micro Devices.
The arrangement, first reported by Business Insider, would see Nvidia supply Arm-based CPUs to Meta for use in its sprawling network of data centers. The move represents a direct challenge to Intel and AMD, which have for decades split the lucrative server CPU market between them. For Nvidia, it is the latest step in a broader strategy to become a full-stack computing provider — one that supplies not just accelerators but the general-purpose processors that form the backbone of modern cloud infrastructure.
A Chip Giant Expands Its Ambitions Beyond the GPU
Nvidia’s push into CPUs is not entirely new. The company unveiled its Grace CPU in 2022, an Arm-based server processor designed to work in tandem with its GPU accelerators. Grace was initially positioned as a complement to Nvidia’s AI chips, meant to reduce bottlenecks in data transfer between the CPU and GPU. But the Meta deal suggests Nvidia’s CPU ambitions extend well beyond a supporting role. Supplying standalone CPUs — or at least CPUs that serve as the primary processors in data center racks — puts Nvidia in direct competition with Intel’s Xeon and AMD’s EPYC product lines.
The significance of Meta as a customer cannot be overstated. The company, which operates one of the largest data center footprints in the world, has been on an extraordinary spending spree to build out AI infrastructure. Meta CEO Mark Zuckerberg has committed tens of billions of dollars in capital expenditure for 2025 and beyond, much of it directed toward AI training and inference hardware. Winning a CPU contract with a hyperscaler of Meta’s scale sends a powerful signal to the rest of the industry that Nvidia’s processors are ready for prime time.
Intel and AMD Face a New Competitor at the Worst Possible Time
For Intel, the timing could hardly be worse. The company is in the midst of a painful restructuring under CEO Lip-Bu Tan, who took over from Pat Gelsinger in late 2024. Intel has been losing market share in data center CPUs to AMD for years, and its foundry business has burned through billions in cash. The prospect of a third major competitor entering the server CPU market — and one with Nvidia’s brand cachet and customer relationships — threatens to further erode Intel’s position in what was once its most profitable business segment.
AMD, meanwhile, has been on a roll. The company’s EPYC processors have steadily gained share against Intel, and its MI300 series of AI accelerators have positioned it as the most credible alternative to Nvidia in the GPU market. But AMD’s server CPU business remains a critical profit center, and the entry of Nvidia into that market introduces a new variable. If major cloud providers and hyperscalers begin splitting their CPU orders between AMD and Nvidia — rather than AMD and Intel — it could reshape the competitive dynamics of the entire semiconductor industry.
The Arm Architecture Gains Ground in the Data Center
Nvidia’s CPU strategy is built on Arm architecture, the same instruction set that powers virtually every smartphone on the planet. Arm-based chips have been making steady inroads into the data center, driven by their superior performance-per-watt characteristics compared to the x86 architecture used by Intel and AMD. Amazon Web Services was an early mover with its Graviton processors, custom Arm-based chips designed for its own cloud infrastructure. Microsoft has followed with its own Arm-based server chips, and Google has invested heavily in custom silicon as well.
What distinguishes Nvidia’s approach is that the company is not merely building chips for internal use, as Amazon and Google have done. Nvidia is positioning itself as an external supplier of Arm-based CPUs, selling to hyperscalers and potentially to enterprise customers as well. This makes Nvidia a direct merchant silicon competitor to Intel and AMD in a way that Amazon’s Graviton, for example, is not. The Meta deal validates this merchant model and could open the door to similar arrangements with other large-scale data center operators.
The Full-Stack Play: From Chips to Software to Networking
Nvidia’s move into CPUs is part of a broader strategy to supply every major component of a modern AI data center. The company already dominates in GPU accelerators, and its acquisition of Mellanox Technologies in 2020 gave it a leading position in high-speed networking equipment — the InfiniBand and Ethernet switches that connect thousands of GPUs in large training clusters. Adding CPUs to the mix means Nvidia can offer customers an integrated package: GPU, CPU, networking, and the software stack to tie it all together.
This integrated approach has clear appeal for customers building AI infrastructure at scale. When the CPU, GPU, and network interface are all designed by the same company, the potential for optimization is significant. Nvidia’s CUDA software platform already gives it a formidable advantage in GPU computing; extending that software advantage to encompass the CPU could make it even harder for competitors to dislodge. Jensen Huang has spoken publicly about Nvidia’s vision of the “AI factory” — a data center purpose-built for AI workloads — and owning the CPU is a critical piece of that vision.
Wall Street Weighs the Implications
Investors have taken notice. Nvidia’s stock, which has surged more than tenfold since the beginning of 2023 on the strength of AI demand, could see further upside if the CPU business scales successfully. Analysts at major banks have begun modeling the potential revenue contribution of Nvidia’s Grace CPU line, though estimates vary widely depending on assumptions about market share and pricing. According to Business Insider, the Meta deal is viewed within Nvidia as a validation of the company’s ability to compete outside its traditional GPU stronghold.
For Intel and AMD shareholders, the calculus is more concerning. Intel’s stock has already been battered by years of execution missteps and market share losses. The addition of Nvidia as a CPU competitor could put further pressure on Intel’s data center revenue, which the company is counting on to fund its expensive foundry turnaround. AMD’s stock, while far more resilient, could face headwinds if investors begin to price in the risk of CPU market share losses to Nvidia alongside the existing GPU competition.
What This Means for the Broader Semiconductor Industry
The implications extend beyond the three companies directly involved. Arm Holdings, which licenses the architecture underpinning Nvidia’s Grace CPU, stands to benefit from any expansion of Arm-based chips in the data center. Every Nvidia CPU sold generates royalty revenue for Arm, reinforcing the company’s position as a critical player in the semiconductor supply chain. Arm’s stock has reflected this optimism, trading at elevated multiples as investors bet on the continued proliferation of Arm architecture beyond mobile devices.
Taiwan Semiconductor Manufacturing Company, which fabricates Nvidia’s chips, also stands to gain. TSMC is already Nvidia’s primary manufacturing partner for GPUs, and producing CPUs on advanced process nodes would add another revenue stream. The concentration of so much of the world’s most advanced chip production at TSMC — which also manufactures AMD’s processors and Apple’s mobile chips — continues to raise questions about supply chain resilience, particularly given geopolitical tensions surrounding Taiwan.
The Battle Lines Are Drawn
Nvidia’s push into CPUs with the Meta deal is not a tentative experiment. It is a calculated move by a company that has demonstrated, time and again, its ability to enter new markets and dominate them. From gaming GPUs to AI training accelerators to data center networking, Nvidia has built a track record of expansion that few technology companies can match. The server CPU market, worth tens of billions of dollars annually, represents perhaps the largest remaining opportunity for the company to grow.
Whether Nvidia can replicate its GPU success in CPUs remains an open question. Intel and AMD are formidable competitors with decades of experience in processor design, and both companies are investing heavily in next-generation products. But the Meta deal suggests that at least one of the world’s largest technology companies believes Nvidia’s CPUs are ready for deployment at scale. If others follow, the server CPU market — long a two-horse race between Intel and AMD — may be entering a new and far more competitive era.