Yesterday, Nvidia announced a $5 billion investment in Intel, alongside a joint development agreement that will allow both companies to break into markets they previously couldn’t reach.
In the PC segment, Nvidia dominates discrete GPUs for workstations and gaming, but laptops have always been out of reach. With roughly 150 million units sold annually, this lucrative market has long been locked down by Intel and AMD, thanks to their integrated GPUs within system-on-chips (SoCs). That changes now: Nvidia’s RTX GPU tiles/chiplets will be fused into Intel’s SoCs, enabling laptops to ship with the combined firepower of Intel CPUs and Nvidia GPUs.
In the data center, Intel will adapt its Xeon CPUs to Nvidia’s needs, integrating them seamlessly into Nvidia’s AI infrastructure. The key move here is replacing the industry-standard PCI Express with NVLink, Nvidia’s proprietary interconnect that distributes workloads across multiple GPUs and CPUs. By purchasing these custom-designed CPUs directly, Nvidia will become one of Intel’s largest customers helping Intel regain some share in the data center market (where it has been losing ground to AMD and ARM) and simultaneously strengthening NVLink’s position as the interconnect of choice in high-performance computing.
This partnership comes on the heels of Intel securing $8.9 billion in U.S. government support, and Nvidia’s financial backing could not be more timely. Since 2021, Intel has poured enormous sums into catching up with TSMC in advanced chip manufacturing, but the effort has strained its balance sheet: debt has ballooned, cash reserves have dwindled, and returns have yet to materialize. Even if there’s no mention of a foundry collaboration in the announcement, Nvidia naming Intel as a strategic CPU partner could open the door to a manufacturing relationship between the two companies and signal that Nvidia believes Intel may finally be close to turning its manufacturing around.
The long-delayed 18A node, once hailed as Intel’s comeback bet, failed to attract meaningful customers due to poor yields. But the next iteration, 14A, could mark a turning point—restoring Intel’s competitiveness and securing its place among the very few companies still capable of producing leading-edge semiconductors.
