Cornelis takes $205M to make GPU networks less passive — and Nvidia less dominant

Cornelis, a company building networking technology to help AI chips communicate more effectively, announced Monday that it has raised $205 million in a funding round led by IAG Capital Partners.

Alongside the funding, the company announced a product called Active Compute Fabric, a network technology that targets a stubborn inefficiency in AI computing: much GPU time is wasted waiting for data to arrive. Cornelis hopes to address that with a networking fabric that lets chips process and send information at the same time.

The company, which spun off from Intel in 2020, is competing with Nvidia by offering an open architecture, meaning customers can use a variety of GPU and accelerator hardware with Cornelis' networking fabric. Nvidia chips can technically run on other networking fabrics, but they are optimized to run on Nvidia's own software, making it much easier — and more enticing — for customers to use Nvidia's full GPU stack. Cornelis is part of a new wave of AI infrastructure companies emerging to break apart Nvidia's market dominance, piece by piece (or chip by chip, one could say).

Cornelis is not starting from a blank slate. It has already started shipping its product and is working on a new generation of it, expected to be released later this year.

The round is the company's most visible bet yet that the interconnect layer — the plumbing that moves data between accelerators — is a place where an open alternative can win customers who do not want to be locked into a single vendor's stack.