Cornelis Networks raises $205M for open AI networking fabric against Nvidia
The AMW Read
Incremental funding update for an Intel-spinout networking vendor; $205M is below mega-round thresholds but segment-relevant as an open-fabric challenge to Nvidia interconnect lock-in.
Cornelis Networks raises $205M for open AI networking fabric against Nvidia
Cornelis Networks, an AI infrastructure company that spun out of Intel in 2020, announced a $205 million funding round led by IAG Capital Partners. With the raise, it unveiled Active Compute Fabric, networking technology aimed at a familiar cluster inefficiency: GPUs often sit idle waiting for data to arrive. Cornelis’s pitch is a fabric that lets chips process and send information at the same time. The company says it has already begun shipping and is working on a new generation expected later this year.
The money is aimed at the interconnect layer where AI data-center performance often bottlenecks, not at another model lab. Nvidia chips can run on third-party networks, but they are optimized for Nvidia’s own software and stack, which makes the full Nvidia path easier for buyers to adopt. Cornelis is competing with an open architecture that works across a variety of GPUs and accelerators—part of a broader wave of infrastructure vendors trying to unbundle Nvidia’s dominance piece by piece, here specifically on networking rather than on silicon alone.
For builders and investors, the concrete test is multi-vendor cluster wins: whether Active Compute Fabric shows up in deployments that mix accelerators and still reduce wasted GPU wait time. If open fabrics become a procurement option beside Nvidia’s networking stack, interconnect becomes another contested surface in AI infrastructure—alongside chips, power, and cloud capacity—rather than a default Nvidia attachment.