1 hour ago
17
For the better part of a decade, the semiconductor performance race had a clear winner. Nvidia’s GPUs dominated AI training so thoroughly that the company’s market cap ballooned past $3 trillion, and the word “chip” became almost synonymous with “graphics processor.” But a funny thing happened on the way to GPU hegemony: CPUs started mattering again.
The culprit is a shift in how AI actually gets used. Training a model is one thing, a massively parallel task that GPUs were born to handle. But running that model, the inference side, involves sequential decision-making that leans heavily on traditional processors. And as agentic AI systems gain traction, handling multi-step reasoning and real-world tasks, the humble CPU has found itself back in the spotlight.
The bottleneck nobody expected
Nvidia itself is making the case for why CPUs matter. Dion Harris from Nvidia stated in March 2026 that “CPUs are becoming the bottleneck” in AI workflows. That’s a remarkable admission from a company that built its empire on the idea that GPUs were the only chips that mattered for AI.
But Nvidia isn’t just diagnosing the problem. It’s selling the cure. The company has rolled out its Grace and Vera CPU lines, with the Vera chip specifically designed for inference and agentic AI workloads. Nvidia expects $20 billion in CPU revenue in 2026 from these offerings.
The Vera CPU doesn’t try to replace GPUs. Instead, it’s designed to complement them, handling the sequential logic and orchestration tasks that trip up parallel processors. In English: GPUs are great at doing millions of simple calculations at once, but when an AI agent needs to think through steps one at a time, that’s CPU territory.
Intel and AMD aren’t sitting still
The incumbents have noticed. Intel holds about 60% of the data center CPU market, with AMD at roughly 24% and Nvidia at just 6% as of early 2026.
AMD finds itself in a more interesting position. The company has been steadily gaining data center share against Intel with its EPYC server processors, and it also competes with Nvidia in the AI GPU market through its Instinct accelerators. Now it faces Nvidia coming at it from the CPU side too.
Events like Nvidia’s GTC 2026 and Computex 2026 have become showcases for these new CPU capabilities, with each company unveiling enhancements targeted specifically at AI applications.
What this means for investors
Nvidia’s $20 billion CPU revenue projection for 2026 tells investors something important. The company sees its future not just as a GPU maker but as a full-stack AI infrastructure provider. That ambition puts it in direct competition with Intel and AMD on their home turf, while those companies simultaneously try to challenge Nvidia on GPUs.
The transition toward inference workloads is the key variable to watch. Training large models requires enormous GPU clusters, but inference happens everywhere, on every query, every API call, every time an AI agent takes an action. The volume of inference compute is growing far faster than training compute, and that’s precisely the workload where CPUs play a critical role.
The risk, of course, is that Nvidia’s integrated approach, selling matched CPU-GPU systems, proves so compelling that customers stop buying processors from Intel and AMD altogether. Nvidia has a track record of entering adjacent markets and winning. Its networking business followed a similar pattern after the Mellanox acquisition.
Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our Editorial Policy.
English (US) ·