NVIDIA CEO Jensen Huang Affirms Co-Existence of x86 and ARM in Rack-Scale Solutions; Intel Foundry Adoption Status Uncertain

During a recent webcast, NVIDIA’s CEO Jensen Huang outlined the details of a significant new partnership with Intel, alongside Intel’s CEO Lip-Bu Tan. This landmark deal aims to enhance collaboration in both the datacenter and consumer CPU sectors, signaling a pivotal shift in how both companies approach their markets.

NVIDIA CEO Discusses ARM Roadmap and Potential Role of TSMC

In what has been described as a ‘blockbuster’ agreement, NVIDIA has invested $5 billion in Intel’s common stock, marking a celebration of their new partnership. Following the announcement, both CEOs held a special call to address questions from reporters covering the implications of this collaboration, particularly the strategic reasons behind NVIDIA’s venture into x86 technology and whether they might source chips from Intel Foundry Services (IFS).

Intel CEO Lip-Bu Tan — Image Credits: Intel’s CEO Lip-Bu Tan

Huang revealed that NVIDIA’s current rack-scale products are capable of achieving up to NVL72 configurations with ARM-based CPUs. However, using x86 datacenter CPUs like Intel’s Xeon limits configurations to NVL8 due to the reliance on a PCIe-based interface. This partnership intends to overcome that limitation by integrating NVLink with Intel’s datacenter CPUs, enabling NVIDIA to provide both ARM and x86 options to its clientele. NVIDIA estimates this untapped market could be worth around $30 billion, which they plan to explore through this collaboration.

And so this architecture, the NVLink 72 rack scale architecture is only available for the Vera CPU that we build, the ARM CPU that we build. And for the x86 ecosystem, it’s really unavailable except with server CPUs over PCI Express.

And that has limitations in how far you could scale these scale-up systems. And so the first opportunity is that we can now with Intel x86 CPU, integrate it directly into NVLink ecosystem and create these rack-scale AI supercomputers.

Another crucial aspect of this partnership is NVIDIA’s role as the primary consumer of Intel’s datacenter CPUs while also being involved in the distribution of Intel’s RTX GPU chiplets, which will be integrated into PC system-on-chips (SoCs).This mutual commitment to collaboration signifies that both firms are investing deeply in this alliance. When questioned about the potential use of Intel Foundry for upcoming semiconductor production, Huang acknowledged ongoing cooperation with IFS but emphasized the undeniable significance of TSMC, suggesting that, for the time being, collaborations with IFS might be limited.

Intel's 18A Process Node — Intel’s 18A wafer | Image Credits: Intel

I think Lip-Bu and I would both say that TSMC is a world-class foundry. And in fact, we’re both very successful customers of TSMCs. And I can’t — you just can’t overstate the magic that is TSMC. But today, our conversation today, our partnership today is completely focused.

There is potential for Intel Foundry to play a role through advanced packaging technology, particularly Jensen’s mention of ‘Foveros’ packaging. This innovation could facilitate the integration of RTX GPU chiplets with Intel’s CPU components for future PC SoCs. While the specific frameworks of the NVIDIA-Intel partnership are still being defined, it is evident that both companies favor utilizing TSMC for their immediate production needs, especially concerning the promising developments expected from Intel’s upcoming 18A and 14A nodes.

Source & Images