NVIDIA To Become The Worlds Leading CPU Supplier With Vera Hitting $20 Billion Revenue This Year

NVIDIA, known for being the world's leading GPU supplier, will also become the lead CPU supplier with its Vera platforms this year.

Vera CPU Demand Is So Insane That NVIDIA Will Become Both The World's Leading GPU and CPU Supplier This Year

NVIDIA recently announced that its Vera CPUs were in full production, hand-delivering the first CPU racks to major AI firms such as OpenAI, SpaceX, Anthropic, and Oracle. Vera is a fundamental component of the Extreme Co-Design ecosystem powering the Rubin platform, but with Vera, NVIDIA is entering into a new market for the first time: Standalone CPUs.

The ARM-based CPU with 88 custom Olympus cores is all set to deliver some big upgrades with 50% better performance, twice the performance per watt (efficiency), and four times the density per rack vs traditional x86 CPUs. It's a CPU that is purpose-built for Agentic AI and Inference domains.

• What it is — NVIDIA’s first custom CPU, designed for agentic AI
• What it handles — Orchestration, tool-calling, RL workloads, data analytics, agent sandboxing, long-context state management
• Who it’s for — AI labs, cloud providers, and enterprises running agentic AI at scale
• Core specs — 88 custom Olympus cores, 1.2 TB/s memory bandwidth, 50% faster per-core under full load

As such, NVIDIA says that Vera CPUs open up a brand new $200B TAM, and they are estimating a total $20B in CPU revenue alone this year, with the majority driven by Vera. With this demand, NVIDIA is on the path to becoming the world's leading CPU supplier, surpassing AMD and Intel, both of which are also seeing huge CPU demand driven by Agentic AI workloads.

Built on custom ARM cores and co-designed end-to-end with Rubin GPUs and NVLink, Vera will deliver up to 1.5x faster performance per core, 2x performance per watt, and 4x density per rack compared to x86-based alternatives.

Vera CPU opens a brand new $200 billion town for NVIDIA, a market we have never addressed before, and every major hyperscale and system maker is partnering with us to get it deployed. We have visibility to nearly $20 billion in total CPU revenue this year, setting us up to become the world-leading CPU supplier.

Colette Kress - NVIDIA CFO

The more important consideration to make is that the $20B number isn't for all of Vera CPUs, it's just for the Standalone CPU.

As we know, Vera has many different applications; Jensen highlights four of them. The Vera CPU will be the host CPU for Rubin racks, featuring two Vera chips connected to four GPUs. NVIDIA also has entry-level NVL4 racks, which feature Intel Xeon CPUs. NVIDIA will be shipping millions of Rubin GPUs, which are now in full production, and first shipments are slated for Q3 2026.

Then we have Vera with CX9 for storage and Vera with CX9 for security. Lastly, there's the standalone CPU, and once again, that's what is accounted for in the $20B figure. This far exceeds AMD's EPYC & Intel Xeon CPU figures for this year.

But there are also some major constraints in the way for Vera. One of those is going to be getting enough supply, as Jensen states that it will be constrained throughout the entire life of Vera Rubin, and the other is going to be memory. Vera relies heavily on LPDDR5X, which is heavily in demand due to the AI supercycle. And while NVIDIA is investing heavily to overcome these constraints, the demand keeps on swelling, and both Vera and Vera Rubin require lots of memory.

The 20 billion is for standalone CPU. And remember, we have Vera is used in three ways as a standalone CPU, four ways. Let me just start with the one that you already know. The first way is Vera Rubin. And we'll sell millions of Rubins, and every two of them is connected to a Vera. And of course, we price those too. And they're properly priced. And so that's number one use case.

The second use case is Vera standalone CPU. The third is Vera with CX9 and the software stack for storage. And then Vera with CX9 with a software stack for security and compute isolation, and confidential computing. And so each one of those use cases is built on Vera. And my sense is that we'll be supply-constrained throughout the entire life of Vera Rubin.

And Vera was designed to be an agentic CPU. The CPUs of the past were designed to have many cores so that it could be easily rentable. People rented cores. Well, agents don't rent cores. They just want the work to be done fast. The economics of the past was dollars per core.

Jensen Huang - NVIDIA CEO

Talking about the Groq 3 LPX, which is an Inference boosting LPU with large amounts of SRAM, Jensen called it a niche product that is designed to offer low-latency and high-token rate.

The use case of LPX is fairly infant at this point, and it is only usable by firms that have a requirement for different types of tokens. Currently, these "Premium" tokens are way less than 20% of the total tokens generated, and NVIDIA is working towards the eventuality when these tokens will be 20% of the total output, but the LPX solution will remain a niche product for NVIDIA for quite some time.

The LPX is designed for low latency and high token rate. But its throughput is low. Its model size capacity is low. And its context processing, its ability to absorb a lot of context, for example, for software coding, for agentic workloads, its ability to absorb a great deal of context is lower. And so the challenge is simply, and I've explained before that the use case for LPX is not broad.

It's, you know, intended for somebody who has a fairly large portfolio of different types of token services. And for the high token rate, maybe these services are quite premium. And the number of customers is not significant, but the token rate is very high. And so that remains exactly consistent with what I've said before. And I still expect that.

And so I expect that LPX and other SRAM-based decode-focused, high token rate-generated focused accelerators will always be a niche product for some time to come.

Jensen Huang - NVIDIA CEO

NVIDIA is set to dominate the AI hardware landscape like never before. With its groundbreaking Vera CPU now in full production, the company is poised to become the world’s leading supplier of both GPUs and CPUs this year, unlocking a massive new $200 billion market while generating nearly $20 billion in standalone CPU revenue alone.

Purpose-built for agentic AI with unmatched performance, efficiency, and density, Vera isn’t just competing with traditional x86 CPUs — it’s redefining the future of AI infrastructure. Demand is so strong that NVIDIA expects to be supply-constrained, cementing its position as the undisputed leader powering the next era of Agentic AI.

Follow Wccftech on Google to get more of our news coverage in your feeds.