NVIDIA Corporation announced today its intention to invest $5 billion in Intel Corporation as part of a strategic partnership aimed at jointly developing next-generation chips.
The collaboration spans both data centers and consumer markets. In the server segment, Intel will design customized central processing units (CPUs) optimized to work efficiently with NVIDIA graphics processing units (GPUs). For the consumer sector, Intel plans to launch compact personal computer processors equipped with NVIDIA chips.
Following the announcement, shares of both companies rose. Intel's stock surged more than 20%, while NVIDIA gained 3.8%.
"This deal benefits NVIDIA more than Intel," said Dave Vellante, co-founder of SiliconANGLE Media and chief analyst at theCUBE Research. "Ironically, Intel's stock gained significantly more than NVIDIA's today, highlighting a fundamental misunderstanding of the market. This gives NVIDIA a direct entry into Intel's PC ecosystem and helps accelerate enterprise AI adoption through a hybrid CPU/GPU approach."
Vellante added, "A $5 billion investment is substantial, but it doesn't address Intel's most pressing challenge—resolving its foundry issues. Frankly, Intel's design business is solid but weighed down by financial issues in its foundry division. I would have preferred to see the initial $5 billion investment directed toward a wafer-scale volume commitment for the foundry; that's the missing piece here."
Currently, data center operators use PCIe interconnects to link NVIDIA GPUs with Intel CPUs in servers. As part of the collaboration, the chipmakers will offer customers an alternative to PCIe by using NVLink. Originally developed by NVIDIA, NVLink is an interconnect technology that enables higher throughput between chips. The fifth-generation version now offers more than ten times the bandwidth of PCIe.
NVIDIA plans to integrate the custom chips developed with Intel into its "AI infrastructure platform." The GPU manufacturer sells a range of artificial intelligence systems under the DGX series brand. Its most advanced DGX system, the GB300, combines 72 GPUs with 36 of its internally developed Grace CPUs.
NVIDIA CEO Jensen Huang stated during today's media briefing that the chipmaker will become a major customer for Intel. The company will resell the CPUs it purchases to end users. Huang co-hosted the briefing with Intel CEO Pat Gelsinger and emphasized that NVIDIA will continue developing custom CPUs based on the Arm architecture.
Huang also revealed that NVIDIA plans to release its upcoming Vera chip, which succeeds the Grace CPU used in the GB300 system, along with the next-generation custom CPU. He confirmed recent reports that NVIDIA is developing a CPU named N1, which will power the company's DGX Spark system—a compact version of its DGX data center hardware designed for developers.
"This is a strategic move for NVIDIA, which now fills a gap in its chip portfolio—the CPU," said Holger Mueller, analyst at Constellation Research. "If NVIDIA can integrate its chips with Intel's server offerings, it will gain additional traction in on-premises data centers. Previous partnerships served as pathways (such as with HPE), but chip-level partnerships don't last forever. CxOs should be aware of lock-in risks, but for now, those managing Intel server farms are likely pleased."
In the consumer market, Intel plans to release compact PC systems integrated with NVIDIA's GPU chips. These chips will be derived from the company's RTX series of discrete consumer graphics cards. The latest RTX models include AI cores based on the Blackwell architecture, which powers NVIDIA's data center GPUs.
NVIDIA plans to use the $5 billion investment to purchase Intel common stock at a price of $23.05 per share, representing a discount of more than 5% compared to Intel's closing price on Wednesday. The transaction is expected to give NVIDIA ownership of approximately 4% of Intel, making it one of the chipmaker's largest shareholders.
