Nvidia's next growth story is no longer just AI training. Jensen Huang is pushing the company into a broader market for agentic and physical AI, and Vera is now central to that pitch.
Nvidia's latest earnings call did more than confirm another blockbuster quarter. It also gave investors a clearer read on where Jensen Huang thinks the company's next layer of growth lives, and it is not in the same narrow frame that powered the first wave of AI demand.
Nvidia said Vera opens a brand new $200 billion total addressable market for the company, a market it had not previously addressed. That opportunity is tied to agentic AI and robotic physical AI, which is a notable shift because it suggests Nvidia wants to sell more than accelerators for model training and inference. It wants to define the stack that sits underneath the next generation of AI systems.
The framing matters because Nvidia has already become the default supplier for the AI buildout that has dominated the last two years. Reuters reported that the company posted first-quarter revenue of $81.62 billion, beating analyst estimates, and guided to $91 billion for the next quarter. That is still an enormous business, but Huang's language says the company is looking for the next expansion point before the current one slows.
The new market opportunity is tied to Vera, Nvidia's CPU product, which the company positioned as a core part of its next chapter. Nvidia has described Vera as the world's first CPU purpose-built for agentic AI, and it is expected to be sold both as part of Vera Rubin systems and as a standalone CPU offering.
That is strategically important because it moves Nvidia closer to the parts of computing that have traditionally belonged to Intel, AMD, and the major cloud platforms. Huang's argument is simple enough: if AI agents need to do work continuously, and if those systems increasingly run outside the GPU-heavy training phase, then CPUs built for that workload become part of the same commercial opportunity. In other words, Nvidia is trying to own the full loop, not just the training step.
Nvidia also said it has visibility into nearly $20 billion in total CPU revenue this year. That is not the same as completed Vera sales, and it should be treated as management guidance rather than independently verified demand. Still, it reinforces the point Nvidia was making on the call. This is not being framed as an experiment. It is being presented as the start of a new revenue pool.
The market is also broadening because Huang is no longer talking only about software-like AI demand. He is tying Nvidia to physical AI, the same phrase that has increasingly come up in industry discussions around robotics, autonomous systems, digital twins, and industrial automation. Nvidia has been building this position through products such as Cosmos world models, robotics tools, and industrial partnerships, which gives the latest earnings commentary a practical shape, not just a promotional one.
From AI chips to platforms
The bigger story is that Nvidia is trying to be seen as a platform company with a much wider capture area. TechCrunch reported in March that Huang said Nvidia's AI chip opportunity could reach at least $1 trillion through 2027, a figure that already stretched well beyond the company's legacy image as a GPU vendor. The $200 billion Vera market is smaller than that, but it points in a different direction. It is less about the brute-force scale of model training and more about the everyday computing layer that AI agents and physical machines may require.
That distinction matters for startups and investors. If Nvidia's thesis is shifting toward agents, robotics, and the infrastructure that supports them, then capital and product development are likely to follow. Startups building simulation tools, robotics software, industrial AI applications, and the tooling around agent deployment may find themselves pulled into Nvidia's orbit, whether they are designing for factories, autonomous systems, or sovereign AI infrastructure.
It also means Nvidia's influence is becoming more structural. When the company makes a strong bet, suppliers, cloud providers, and startups often move with it because the ecosystem has already learned to follow the hardware roadmap. That is what gives Huang's comments weight. He is not describing a future market from the sidelines. He is telling the market where the next distribution channel is likely to form.
For now, the immediate takeaway is that Nvidia's earnings story is no longer just about beating estimates. The company is using each quarter to redraw the boundaries of its own addressable market, and Huang's latest call suggests the next boundary will be built around agentic systems and the physical world they are expected to run in. If he is right, the next big Nvidia opportunity will not simply be another faster chip. It will be the infrastructure layer that makes machines act a little more like workers and a little less like tools.
Also read: Nvidia's $43 Billion bet on startups tightens its grip on the AI stack • Nvidia's startup bets are turning its AI dominance into a strategic moat • Lawsuit and a $2.8bn bet: xAI's Memphis buildout exposes AI's physical limits
