Written by Admin Alex · Fact-Checked by M.Ali · Info Verified September 2026
We review and update this article regularly as new information becomes available.
TL;DR: Nvidia says its new Arm-based Vera server chip is on pace to bring in roughly $20 billion in sales this year, with that figure expected to double again in 2027. The company is using its grip on AI hardware to push into the CPU market Intel and AMD have controlled for decades, and it thinks the whole server CPU category can nearly triple in size by 2030.
Twenty billion dollars. That is roughly what Nvidia expects to collect this year from a product category it barely competed in five years ago: server CPUs. Not GPUs. Not AI accelerators. The workhorse chips that run the basic compute tasks inside a data center. Intel and AMD have split that market between them for most of the last two decades. Nvidia wants a real piece of it now, and it is showing up with an Arm-based chip called Vera.
Why a GPU Company Selling CPUs Is a Big Deal
A data center is not just racks of AI accelerators. Every server still needs a central processor to manage tasks, move data around, and run the software stack that sits underneath the flashier AI workloads. For decades, that job went almost automatically to Intel or AMD, running the x86 architecture that has anchored servers since the 1990s. Nvidia already sells the GPU. If it can also sell the CPU sitting next to it, it captures a bigger slice of every server that ships, and it can design the two chips to work as one system in a way a mixed Intel-plus-Nvidia setup never quite manages.
That is the strategic threat, in plain terms. Nvidia is not trying to win on CPU specs alone. It is trying to make the CPU part of a bundle customers were already buying anyway.
The Pitch, in Nvidia’s Own Words
Nvidia CFO Colette Kress laid out the sales pitch on a recent earnings call. Vera, she said, “completes agentic tasks 1.8x faster on the spec benchmark and provides 5x the bandwidth per watt than any other data center CPU.” Strip away the marketing polish and the claim is simple: Vera is supposed to handle AI-heavy workloads faster while using less power per unit of data moved than anything Intel or AMD currently ships. Whether that holds up under independent, apples-to-apples testing is a separate question, one Nvidia’s rivals will be racing to answer over the next year.
The Money Nvidia Is Chasing
The $20 billion figure for this year is not a rounding error next to Nvidia’s overall business, but it is still early. Nvidia expects it to double again next year. That would put Vera on track to become one of the company’s largest product lines within a couple of years, not a side experiment.
Zoom out and the ambition gets bigger. Nvidia estimates the entire server CPU category could grow into a $220 billion annual opportunity by 2030, expanding at roughly 50% a year. That is a market Intel spent decades building brick by brick. Nvidia is telling investors, plainly, that it intends to eat a meaningful chunk of it.
There is already evidence the shift is underway, and it predates Vera. Arm-based chips, the broader category Vera belongs to, already account for about 45% of data center revenue share, by Nvidia’s own estimate. Amazon, Google, and Microsoft have each built their own Arm server chips for internal use over the past several years. Vera gives Nvidia a version of that idea it can sell to everyone else, not just the hyperscalers with the engineering budgets to build their own.
Why Wall Street Is Watching the Valuation
Here is the part non-finance readers usually skip past, and should not. Nvidia trades at a price-to-earnings ratio of around 29. AMD trades near 122. Intel sits around 88. In plain terms, investors are paying far more for every dollar of current profit at AMD and Intel than they are at Nvidia, even though Nvidia is the one now moving onto their turf.
A high P/E usually means investors expect fast future growth, or it can just mean a stock’s price has run ahead of its earnings. Either way, a lower P/E sitting next to the bigger stated ambition is the kind of gap analysts tend to flag. It is part of why some read Nvidia, of all companies, as the cheaper growth story on the table right now.
None of this makes Intel and AMD irrelevant overnight. x86 chips still run the overwhelming majority of enterprise software. Switching costs are real. Nvidia has never shipped a CPU at this kind of volume before, and manufacturing at scale is its own separate challenge. Market opportunity and successful execution are not the same thing.
Bottom Line
Nvidia is not diversifying for the sake of it. It is using the leverage it already has, as the company almost nobody in AI can avoid buying from, to walk into a market it does not own and take share through bundling. If Vera’s numbers hold up outside Nvidia’s own slides, Intel and AMD are not just facing a new competitor. They are facing one that can afford to sell CPUs at thin margins just to keep customers locked into buying its GPUs, and that is a fight neither company has had to have before.
