Cornelis Networks raises $205m to advance AI networking technology
The funding supports Cornelis’s effort to keep GPUs and other accelerators fed with data as AI clusters scale.
Reducing time GPUs spend waiting for data
Cornelis Networks builds open networking fabrics for AI and high-performance computing clusters. Its Active Compute Fabric is designed to let GPUs and accelerators exchange data while they work, addressing a bottleneck that can leave expensive processors idle while they wait for inputs.
The $205 million financing will advance the company’s AI networking technology and product development. Cornelis is already shipping networking products and is developing the new fabric for operators and chip vendors building larger AI systems, where communication between processors can determine how much useful work a cluster delivers.
Use of funds
Advance its AI networking technology and product development