Fresh Capital Influx for Cornelis as it Challenges Current AI Hardware Standards

Cornelis, a specialist in high-performance networking for artificial intelligence, recently secured a significant financial boost by raising $205 million. This latest investment round was spearheaded by IAG Capital Partners and signals a growing push to provide alternatives to the current market leader, Nvidia. Originally a division of Intel, Cornelis became an independent entity in 2020 and has since focused on optimizing how artificial intelligence processors interact with one another during complex computational tasks.

Along with the new funding, the company introduced its Active Compute Fabric technology. This innovation is designed to solve a persistent bottleneck in data centers: the tendency for powerful graphics processing units (GPUs) to remain idle while waiting for information to travel across a network. By enabling hardware to transmit and process data simultaneously, the new fabric aims to maximize the efficiency of expensive AI hardware. Unlike the proprietary ecosystem maintained by Nvidia, which often functions best when its own chips and networking software are used together, Cornelis promotes an open architecture. This strategy allows businesses to mix and match different accelerators and GPUs while still utilizing the Cornelis networking layer.

The startup is positioning itself as a key player in a new movement of infrastructure providers seeking to erode Nvidia’s near-monopoly on the AI sector. By offering a platform that does not lock customers into a single vendor's hardware stack, Cornelis offers greater flexibility for organizations building large-scale AI systems. The company has already begun distributing its hardware to clients and is currently developing a follow-up generation of its technology, which is slated for release before the end of the current year. As the demand for generative AI and massive machine learning models continues to grow, the industry is increasingly looking for ways to streamline data movement and reduce the total cost of ownership for high-performance computing clusters.

---

Read original at TechCrunch.