AI infrastructure company Cornelis raises $205M to chip away at Nvidia’s dominance

Companies building AI models can now more easily avoid vendor lock-in with Nvidia's hardware ecosystem.

AI infrastructure company Cornelis has raised $205 million in funding to challenge Nvidia's dominance in the AI hardware market. The company announced its product, Active Compute Fabric, a networking technology designed to reduce wasted GPU time caused by data transfer delays. Cornelis's solution aims to enable simultaneous processing and data transmission by utilizing a specialized networking fabric. This approach allows customers to integrate various GPU and accelerator hardware, offering an open architecture alternative to Nvidia's proprietary ecosystem. Cornelis, which originated as a spin-off from Intel in 2020, is now shipping its product and preparing for a new generation release later this year as part of a broader trend of startups targeting Nvidia's market share.

Cornelis's $205 million funding round signifies a growing appetite for alternative AI infrastructure solutions that can break Nvidia's de facto monopoly. By focusing on an open architecture for its networking fabric, Cornelis enables greater hardware flexibility for AI developers, potentially reducing costs and increasing efficiency. This move challenges Nvidia's strategy of tightly integrating its hardware and software, which has historically made it difficult for competitors to gain traction.

The introduction of Active Compute Fabric addresses a critical bottleneck in AI computation: data latency. If successful, this technology could unlock significant performance gains for large-scale AI training and inference, impacting the cost-effectiveness and accessibility of advanced AI development. Watch for how quickly other players adopt open standards or how Nvidia responds to maintain its market position.