NVIDIA and Intel Announce $5B Collaboration #
The semiconductor industry was shaken when NVIDIA and Intel announced a $5 billion equity deal, with NVIDIA purchasing Intel stock to cement their partnership. The agreement spans data center processors, consumer PC SoCs, and future AI infrastructure. Following the announcement, Jensen Huang (NVIDIA CEO) and Pat Gelsinger (Intel CEO) hosted a webcast to outline the collaboration’s vision.
Expanding NVLink to x86 CPUs #
At the heart of the partnership is a plan to integrate NVLink directly into Intel Xeon processors.
- Today, NVIDIA’s NVL72 architecture scales efficiently with ARM-based CPUs.
- By contrast, PCIe-based x86 CPUs are limited to NVL8, restricting scalability in rack-level AI supercomputers.
With Intel as a partner, NVIDIA aims to bring Xeon into the NVLink ecosystem, opening a $30B+ market opportunity. This move would give data centers more flexibility, supporting both ARM and x86 architectures in large-scale AI deployments.
Two-Way Collaboration #
The deal is not one-sided:
- Intel CPUs → will be adopted in NVIDIA-driven AI systems.
- NVIDIA GPUs → RTX will be supplied to Intel for integration into future PC SoCs.
This shows a deep product-level integration, not just a loose market partnership.
Foundry Question: Intel vs. TSMC #
A key question has been whether NVIDIA will shift more of its manufacturing to Intel Foundry Services (IFS).
Huang’s remarks provide clarity:
- NVIDIA already works with IFS in packaging, particularly Intel’s Foveros 3D packaging for future PC chips.
- However, he emphasized:
“TSMC remains our world-class foundry partner.”
This confirms that TSMC retains the lead in wafer fabrication, while Intel’s role is currently focused on advanced packaging rather than primary chip production.
Intel’s Process Roadmap #
Intel is betting big on its 18A node, which offers:
- 36% lower power at the same frequency compared to Intel 3
- 25% higher frequency at the same power
- 30% higher transistor density
Future 14A technology could push these gains further, giving Intel a chance to attract clients like NVIDIA for specific products or heterogeneous packaging projects.
Strategic Takeaways #
This collaboration sends a clear signal:
- NVIDIA → Needs x86 CPUs with NVLink support to grow its AI supercomputing footprint.
- Intel → Wants to leverage NVIDIA’s ecosystem to regain relevance in high-performance computing.
Though rivals in GPUs, the two companies are finding synergies in data centers. This “co-opetition” could shape the next era of AI infrastructure and semiconductor strategy.