Nvidia chief Jensen Huang took the stage and revealed the company’s new Vera CPU, a chip built specifically to run AI agents. The announcement pushes Nvidia deeper into processor territory that has long belonged to rivals, signaling a shift in how the company sees its role in the AI stack.
A CPU Built for AI Agents
The Vera CPU isn’t a general-purpose chip. Huang described it as designed from the ground up for the workloads that AI agents generate — reasoning, planning, tool use, and multi-step decision-making. That’s different from the GPUs Nvidia is famous for, which handle parallel math for training and inference.
By specializing the CPU, Nvidia is betting that agentic AI will demand a different kind of compute, one where the CPU isn’t just a supporting actor but a lead performer. The company didn’t release detailed specs or a launch timeline, but the chip’s existence alone tells competitors that Nvidia plans to own more of the hardware pipeline.
Redefining AI Infrastructure
Right now, most AI deployments rely on a split: GPUs accelerate the heavy math, and standard CPUs orchestrate the work. The Vera CPU could blur that line. If Nvidia can offer a CPU that handles agent logic efficiently, data centers might start buying whole Nvidia systems instead of mixing components from different vendors.
That’s a bet on infrastructure. Nvidia already sells complete server racks with its GPUs, networking gear, and software. Adding its own CPU means fewer parts come from outside suppliers. For hyperscale customers like cloud providers, a one-stop shop could simplify procurement and cut power overhead.
Market Share and Revenue Prospects
The move targets a chunk of the server CPU market that’s currently dominated by Intel and AMD. Even a modest slice of that business would add billions to Nvidia’s top line, which is already ballooning from AI GPU sales. Huang didn’t offer revenue projections, but analysts will watch for early customer adoption as a signal.
Nvidia’s expansion into CPUs also creates a new moat. If AI agents become the next wave of software, having both the GPU and the CPU on the same architecture could give Nvidia a performance edge that competitors can’t easily replicate. The company isn’t just selling chips anymore — it’s selling a platform that covers the whole AI lifecycle.
The Vera CPU is still in the early stages, and Nvidia hasn’t said when it will ship or how much it will cost. What’s clear is that Huang believes the AI agent market is large enough to warrant a custom chip. The rest of the industry is now on notice: Nvidia is coming for the processor socket, too.
