Huang credits Vera with unlocking Nvidia’s $200B agentic AI market

Vera unlocks – Nvidia CEO Jensen Huang used the company’s earnings call to frame its March-launched Vera CPU as the start of a new $200 billion market, arguing that agentic AI and robotics will push billions of CPU-based “agent” systems into everyday computing.

By the time Jensen Huang got to the most audacious part of Nvidia’s pitch, he wasn’t talking about incremental growth. He was talking about a whole new spending category.

On Wednesday’s earnings call. after Nvidia posted another record-breaking quarter with $81.6 billion in revenue and forecast $91 billion for the next. Huang pointed to Nvidia’s new CPU product. Vera. introduced in March. He said Vera could open a “brand new $200 billion TAM for Nvidia. ” a market Nvidia “has never addressed before. ” and he tied that opportunity directly to agentic AI and robotic physical AI.

Huang’s message landed on a sensitive fault line for investors: CPUs.

Nvidia has long been synonymous with GPUs, while CPU markets were historically dominated by Intel and AMD. Nvidia has made CPUs before, but Huang’s company is not usually viewed as a CPU contender. That’s why the CPU question keeps resurfacing—especially as more players decide they want a piece of the AI chip stack.

Last month, Amazon Web Services highlighted a large contract it signed with Meta for “millions” of Amazon’s homegrown AI CPUs. Amazon CEO Andy Jassy has been clear that he thinks AWS can build AI chips—both GPUs and CPUs—at least as well, and possibly better, than Nvidia.

Against that backdrop, Huang argued that Vera changes the matchup. He said Vera is sold on its own but bundled with Nvidia’s Rubin GPU, and that it’s designed for agents. Huang described Vera as “the world’s first CPU, purpose-built for agentic AI,” and called it a “major new growth driver.”

He said the logic is different for agent workloads. In his framing, the “thinking” part of an AI model uses GPUs. Agents, though, mostly run on CPUs. Those agents will then use their own form of CPU-driven computing—his expectation is that they’ll function like the next generation of “PCs. ” built around the tools agents rely on.

Huang specifically said Vera is built to process tokens as fast as possible. He contrasted that with classic cloud architecture CPU designs, built around “cores”—the ability to run multiple instances of apps quickly rather than pushing token processing speed.

But the biggest credibility test wasn’t the pitch itself. It was the question of why Nvidia would be the go-to source when major cloud providers and startups are also pursuing AI chip development.

Huang offered a sales number meant to bridge that gap. He said Nvidia has already sold $20 billion worth of standalone Vera CPUs this year, “and we’re only at the beginning.”

Then he expanded the market picture again, returning to sheer scale. Huang said the world has “a billion users. ” and that his sense is the world is going to have “billions of agents. ” even if that growth arrives gradually. He added that those billions of agents will use tools—tools that will resemble PCs. “just like us humans using PCs today.”.

“We’re going to need a lot more CPUs,” Huang said.

And in his concluding sweep, he positioned Nvidia at the center of that shift: “The world is rebuilding computing for agentic AI and robotic physical AI. Nvidia sits at the center of these transitions.”

Nvidia Jensen Huang Vera CPU Rubin GPU agentic AI AI chips earnings call TAM AWS Meta Andy Jassy cybersecurity