CScale exits stealth with $145M for gigawatt-scale AI interconnect
What's the deal? CScaleDealroom has a profile for this one. Try Dealroom → exited stealth with $145 million in Series C funding to develop and commercialize optical interconnect for AI scale-up — the high-bandwidth, low-latency connections that let accelerators work together like one much larger computer. The round brings the Palo Alto company's total funding to $188 million.
Who's backing it? Atreides ManagementDealroom has a profile for this one. Try Dealroom →, Valor Equity Partners, and Premji Invest co-led the round. Sutter Hill Ventures, a backer since inception, and existing investor Maverick SiliconDealroom has a profile for this one. Try Dealroom → also joined, while NVIDIA and Intel Capital came in as CScale's first strategic investors.
What's the endgame? In the gigawatt-class data centers expected by the decade's end, a single scale-up domain will span thousands of accelerators across dozens of racks. CScale is building its interconnect to contain optical failures without interrupting compute.
What could go wrong? Across hundreds of thousands of accelerators, failures rare on any single link become a constant fleet-level condition, degrading parts of the data center unless contained. "Lasers will fail. Compute shouldn't," said chief executive officer Martin Lund.
Who's building it? Lund previously built Broadcom's switching business into a billion-dollar franchise and led Cisco's Common Hardware Group. Founder and chief technology officer Sanjai Kohli co-founded SiRF, which helped bring GPS to the mass market, and later founded InoviDealroom has a profile for this one. Try Dealroom →, acquired by FacebookDealroom has a profile for this one. Try Dealroom → in 2014.
The signal: "AI infrastructure is no longer just a compute problem. It is a systems problem," said Gavin Baker, managing partner at Atreides Management. As deployments push toward gigawatt scale, investors are betting that interconnect reliability — not raw bandwidth — decides whether these AI factories perform.
Read more: businesswire.com
Image credit: CScale
