Sharespulse
sharespulseTelegram

Gelsinger Backs $100 Million Bet Against Nvidia's AI Networking Lock-In

AI Chips·October 6, 2026

Gelsinger Backs $100 Million Bet Against Nvidia's AI Networking Lock-In

Pat Gelsinger, the former Intel chief executive, has thrown his weight behind a $100 million push to loosen Nvidia's hold on the networking layer of AI data centers. The move targets a part of Nvidia's business that gets far less attention than its GPUs but may matter just as much to its long-term dominance.

The source item is light on detail, so the specifics of the funding structure and the technology involved should be treated with caution. The core idea, though, is clear: build credible alternatives to the proprietary interconnects that tie AI clusters together, so customers are not forced to buy Nvidia end to end.

That networking layer is often described as Nvidia's moat. Its InfiniBand and NVLink technologies are designed to work tightly with its own chips, which makes the whole system faster and also harder to leave. A buyer who commits to Nvidia GPUs tends to commit to Nvidia networking as well.

This is where AMD enters the picture. The chipmaker has been the most visible rival in AI accelerators, but it has struggled to match the full-stack package Nvidia offers. Open networking standards would let AMD GPUs sit inside large clusters without a networking handicap. Any technology that makes mixed or non-Nvidia systems easier to build helps AMD's pitch to cloud providers and large AI labs.

Gelsinger's involvement adds credibility and visibility. He spent decades in the chip industry and has long argued for more open, competitive infrastructure. His backing signals that investors see the networking layer as a real opening, not just a side project.

For markets, the story is a reminder of how much of Nvidia's valuation rests on its ecosystem, not only raw chip performance. Even a well-funded challenger faces a long road. Nvidia's software, tooling and customer relationships are deeply entrenched, and $100 million is small next to the sums being spent on AI infrastructure across the industry.

Still, the direction matters for investors. If open networking matures, it could chip away at the switching costs that protect Nvidia's margins and widen the field for AMD and others. For now, it is a bet worth watching, not a threat that changes the competitive picture overnight.

Reporting based on an external source.