Nvidia Vera CPU Challenges Intel and AMD in Data Center

【New AI Report】Jensen Huang’s computing empire starts a new change with the Vera CPU. This move may completely change how AI data centers are built and how much they cost.

2 billion US dollars, $87.20 per share. CoreWeave’s stock jumped 12% in one day, and its market value passed 52 billion US dollars.

This is a strong hit to the tech industry.

Jensen Huang’s investment in CoreWeave is not just about buying stock. It is a “ticket” for computing factories in the next five years: a plan to build more than 5 gigawatts of AI factories by 2030, and up to 6 billion US dollars in Nvidia hardware purchases.

The most important news is that the Vera CPU is now available as an independent product for the first time.

This means Nvidia no longer only sells GPUs. It wants to control the “heart” of data centers and directly compete with Intel and AMD in their main market.

undress ai removerThe Computing Ambition Behind the 2 Billion Dollar Investment

On January 26, 2026, Nvidia CEO Jensen Huang announced that Nvidia would buy 2 billion US dollars of CoreWeave’s Class A common stock at $87.20 per share. Right after the news, CoreWeave’s stock rose 12%, and its market value went over 52 billion US dollars.

Behind this deal is Nvidia’s strong plan to enter the CPU market.

This investment will help CoreWeave reach its big goal: building more than 5 gigawatts (1 gigawatt = 1 billion watts) of AI factories by 2030. CoreWeave will buy up to 6 billion US dollars of Nvidia hardware, including the new Vera CPU chips.

As a cloud service provider that works closely with Nvidia, CoreWeave will be the first undress ai promo code customer to use Nvidia’s Vera CPU as a separate infrastructure choice.

This means Nvidia is now selling CPUs as standalone products, not just as part of a combined system. It directly challenges Intel and AMD’s leading position in the data center processor market.

Under the partnership, CoreWeave will get first access to Nvidia’s next-generation computing tools, ai furry porn including the Rubin platform, Vera CPU, and BlueField storage system.

Jensen Huang said about the investment: “This investment shows our confidence in CoreWeave’s growth, management team, and business model.” But he also said the main goal is to bring together both companies’ engineering teams to speed up the launch of new computing infrastructure.

At the time of this deal, Nvidia faced growing competition from large tech companies that build their own AI processors. For example, Google’s TPU is now used by companies like Anthropic.

OpenAI is working with chip designer Broadcom to create its own AI accelerator. At the same time, it has a deal with AMD, Nvidia’s main rival, to buy GPUs.

With this move, Jensen Huang strengthens his partnership with CoreWeave. In the past three years, CoreWeave has become one of the most active cloud service providers focused on meeting big tech’s demand for Nvidia chips. Before this new stock purchase, Nvidia already owned 3.3 billion US dollars worth of CoreWeave stock. After this purchase, Nvidia’s ownership in CoreWeave goes above 11%.

The Big Plan Behind Vera CPU’s Independent Launch

Before now, Nvidia’s CPUs were only part of a larger system, sold together with GPUs. The independent launch of the Vera CPU means Nvidia is changing from a “component seller” to a “platform builder.”

Nvidia said about the investment: as AI agent tools become more popular, server CPUs have become a key factor limiting overall system performance. More GPU power alone is no longer enough. A matching high-performance CPU platform is needed.

AI agents are different from older AI models. They run non-stop, keep track of data, and make complex decisions. This requires much higher memory speed and better coordination between processors.

The Vera CPU is built for these exact tasks. Its main goal is to handle the most difficult AI and computing jobs.

In technical specs, the Vera CPU has 22.7 billion transistors. It marks a huge step forward in Nvidia’s processor design.

The chip uses a next-generation custom Arm design. It has 88 custom Olympus ARM cores and 176 threads. It also includes Nvidia’s new “Spatial Multi-Threading” technology.

Unlike traditional hyper-threading, spatial multi-threading splits core resources physically. This lets the Vera CPU run 176 tasks at the same time with stable performance.

For memory, Vera has 1.5 TB of system memory — three times more than the older Grace CPU. It uses SOCAMM LPDDR5X technology to reach 1.2 TB/s memory bandwidth. With NVLink-C2C connection tech, its coherent memory speed hits 1.8 TB/s, three times faster than Grace.

Vera’s cache is also greatly upgraded. Each core has 2MB of L2 cache (twice Grace’s) and 162MB of shared L3 cache (42% more). This helps Vera share data with paired GPUs much faster, boosting overall system efficiency.

The Vera CPU uses only 50W of power. For an 88-core CPU, this is very low. It will help data centers use less energy and stay cooler.

For cloud providers like CoreWeave that focus on AI workloads, buying Vera CPUs separately means more flexibility. They can mix and match computing resources based on customer needs, instead of buying a full rack system. By offering the Vera CPU alone, Nvidia gives customers a way to avoid traditional x86 system limits.

Nvidia’s next-generation Vera CPU will work with the next-gen Rubin GPU. The Rubin GPU is expected to use HBM4 memory with 22 TB/s bandwidth — 2.75 times faster than the Blackwell GPU. For home and office use, Nvidia is also making ARM-based N1/N1X CPUs for future AI PCs.

CoreWeave’s stock price has become a sign of how excited investors are about AI. Since its IPO last year, the stock has more than doubled. Still, it is down about half from its highest point last June. Jensen Huang’s 2 billion dollar bet today is not only a vote of confidence in CoreWeave but also a strong belief in the future of AI computing power.