Cornelis Raises $205M and Launches Its Open AI Network Fabric

Cornelis has begun delivering its Active Compute Fabric and is preparing a new version expected later this year. The company, spun out from Intel in 2020, has raised $205 million from IAG Capital Partners to support its roadmap. Its approach is based on an open architecture aimed at reducing GPU wait times by enabling parallel processing and data transmission.

Ongoing Deliveries and Next Generation Announced for This Year

Cornelis has started shipping its Active Compute Fabric product. At the same time, the company is developing a new generation of this technology, with a launch planned for later this year.

Technological Openness Against Nvidia's Optimized Stack

Cornelis positions itself against Nvidia by offering an open architecture that allows customers to use different GPUs and accelerators with its networking fabric. Nvidia chips, while usable on other fabrics, are designed to work optimally with Nvidia's software, simplifying the use of the entire Nvidia GPU stack for customers. Cornelis is part of a new generation of AI infrastructure companies seeking to reduce Nvidia's dominance in the market.

$205 Million Raised and Details on Active Compute Fabric

Cornelis announced on Monday a funding round of $205 million, led by IAG Capital Partners. The company introduced Active Compute Fabric, a networking technology designed to decrease GPU wait times when receiving data. This networking fabric allows chips to process and transmit information simultaneously. Cornelis develops networking solutions for artificial intelligence chips and was founded in 2020 from Intel.