CScale, which is developing optical interconnect for accelerators, emerges from stealth with a $145M Series C led by Atreides, Valor Equity, and Premji Invest

Optical interconnects designed for continuity will become a standard component in future AI data centers.

CScale, a startup developing optical interconnect technology for large-scale AI systems, has emerged from stealth and secured $145 million in Series C funding. The round was co-led by Atreides Management and Premji Invest, with participation from Valor Equity Partners and Premji Invest. This funding brings CScale's total capital raised to $188 million, including earlier investments from Sutter Hill Ventures, Maverick Capital, Nvidia, and Intel Capital. The company aims to build high-bandwidth, low-latency optical interconnects designed to allow thousands of AI accelerators to function as a single, massive computing unit. CScale's technology focuses on ensuring continuity of operation even when individual optical links fail, a critical requirement for future gigawatt-scale AI data centers expected to house massive numbers of accelerators across numerous racks. The company was founded in 2023 by Sanjai Kohli and is led by CEO Martin Lund, who has previous executive experience at Broadcom, Microsoft, and Cisco.

The substantial funding and strategic investor interest underscore a growing industry consensus that AI infrastructure scaling is fundamentally a systems and interconnect problem, not just a compute one. Companies like CScale are betting that the sheer scale of future AI deployments necessitates novel optical solutions to manage communication between thousands of accelerators reliably. This suggests a significant market opportunity for specialized interconnect hardware that prioritizes uptime and fault tolerance over raw speed alone.

This development impacts how hyperscalers and AI hardware providers will approach building exascale AI systems. It signals a shift toward architecting for inevitability of component failures, demanding more robust and self-healing interconnect fabrics. CScale's focus on architectural judgment for system-level value, rather than just component performance, will likely drive a new generation of interconnect solutions that are integral to AI factory economics and reliability.