Login to Continue Learning
NVIDIA’s CEO, Jensen Huang, provided details about the company’s collaboration with Intel during a webcast. He also discussed potential partnerships with Intel Foundry Services (IFS).
NVIDIA CEO Reiterates No Changes to ARM Roadmap; TSMC Likely Primary Foundry
Yesterday, NVIDIA and Intel announced a significant partnership, focusing on datacenter and consumer CPU segments. NVIDIA invested $5 billion in Intel’s common stock to celebrate the deal. During their special call, both CEOs answered questions from multiple media outlets.
Key points included:
– Why NVIDIA chose an x86 partnership.
– Possibility of using Intel Foundry for chip needs.
Jensen Huang revealed that current ARM-based CPUs can support NVL72 configurations with NVLink, while x86 datacenter CPUs like Intel’s Xeon are limited to NVL8 due to PCIe interfaces. The deal aims to integrate NVLink into Intel’s datacenter CPUs, offering both ARM and x86 options.
Quoting Jensen:
– “The NVLink 72 rack scale architecture is only available for the Vera CPU we build. For the x86 ecosystem, it’s unavailable except with server CPUs over PCIe.”
– “This limits scalability in these systems. Intel x86 CPUs can now integrate directly into the NVLink ecosystem, creating rack-scale AI supercomputers.”
NVIDIA will be the primary consumer of Intel’s datacenter CPUs and seller of RTX GPU chiplets for a PC SoC. The partnership is mutually beneficial, with both firms committed to deep cooperation.
Jensen was asked about using Intel Foundry instead of TSMC. He stated that while they have been working with IFS, TSMC remains crucial.
One potential involvement for Intel Foundry could be advanced packaging technology like Foveros, which might combine RTX GPU and CPU chiplets in a PC SoC. For now, both companies seem comfortable using TSMC.
