Scale-up interconnect startup CScale launches with $188M in funding

CScale Inc., a startup that develops optical interconnects for data centers, launched today with $188 million in funding.

The company raised about three-quarters of the capital through a recently closed Series C round. It was co-led by Atreides Management, Valor Equity Partners and Premji Invest. The funds were joined by several other backers including Nvidia Corp. and Intel Capital.

CScale was founded in 2023 by prominent electrical engineer Sanjai Kohli. He previously founded SiRF Technologies Inc., an early supplier of GPS chips for vehicle navigation systems. Kohli is CScale’s chief technology officer, while the company’s CEO is long-time data center executive Martin Lund. Lund earlier led Cisco Systems Inc.’s chip engineering group and Broadcom Inc.’s switch business.

CScale’s interconnect technology is designed to link together the accelerators that make up an artificial intelligence cluster. According to the company, its hardware is optimized to power large scale-up cluster configurations. Such setups can provide better performance than current designs.

An AI data center is split into infrastructure units called scale-up domains. Today, a scale-up domain is usually a single server rack. The chips in the rack effectively function as one large computer. Adding more processing capacity requires companies to deploy additional scale-up domains.

Data moves more slowly between different scale-up domains than between accelerators in the same scale-up domain. In March, Nvidia revealed that it’s working to remove that bottleneck. The chipmaker is developing hardware that will make it possible to replace multiple compact scale-up domains with one large scale-up domain. Such a setup will enable data to move faster between accelerators, which should boost AI workloads’ performance.

CScale’s interconnect is designed to ease the development of large scale-up clusters. The company says that its technology can link together thousands of accelerators across dozens of racks. CScale will move data between chips via optical links, which are significantly faster than traditional copper wiring.

A malfunction in one of the optical links that powers a scale-up cluster can disrupt many other components. CScale says its interconnect is capable of “containing” such failures, which increases cluster uptime. The company didn’t specify how it achieves that reliability boost.

According to a job posting, CScale plans to ship its technology as part of a system-on-chip that will feature an onboard central processing unit. The CPU might be needed to run network automation software. CScale describes its product as an “integrated light engine.”

“As scale-up systems move toward gigawatt-class deployments, interconnect bandwidth means nothing without system reliability,” said Atreides Management Managing Partner and Chief Investment Officer Gavin Baker. “Every optical failure is a compute failure. At this scale, failures are not hypothetical — they are inevitable. CScale is designing reliable scale-up optics for next-generation AI factories.”

CScale will use its new funding to enhance its technology and accelerate commercialization initiatives.

Photo: Unsplash