- CScale exits stealth with a $145M Series C led by Atreides and Valor.
- NVIDIA and Intel Capital join as its first strategic investors.
- It is building optical links meant to keep AI clusters running when lasers fail.
In a gigawatt-scale AI data centre, one failed laser barely registers. Multiply the links across hundreds of thousands of accelerators, the chips that run AI models, and failures that are rare on any single link become a constant condition of the fleet.
CScale, a Palo Alto startup that exited stealth on September 30, has raised $145 million in Series C funding. It is building optical connections that carry data between chips as light, designed to contain those failures without interrupting computation.
Atreides Management led the round alongside Valor Equity Partners, and Premji Invest joined as a co-lead. Sutter Hill Ventures, which has backed CScale since its inception, continued its support alongside existing investor Maverick Silicon. NVIDIA and Intel Capital joined as its first strategic investors. The round takes total funding to $188 million and will fund development and commercialisation of its optical interconnect platform.
Why one dead laser matters at scale
CScale works on what the industry calls scale-up: the high-bandwidth, ultra-low-latency links that let accelerators work together like a single, much larger computer. In the gigawatt-class data centres expected by the end of the decade, one scale-up domain will span thousands of tightly coupled accelerators across dozens of racks.
The startup, founded in 2023, is building an integrated light engine for reliable optical connectivity at that scale, with an architecture designed to contain failures rather than propagate them to the workload.
CEO Martin Lund puts the philosophy bluntly. “As AI scale-up domains extend across dozens of racks, optical interconnect becomes essential. ,” he said. “At that scale, reliable, predictable communication is fundamental to system economics. Easier part replacement improves serviceability, not continuity. We’re designing the interconnect for continuity. Lasers will fail. Compute shouldn’t.”
Lead investor Gavin Baker, managing partner and chief investment officer at Atreides Management, calls it a systems problem. “AI infrastructure is no longer just a compute problem. It is a systems problem,” he said. “As scale-up systems move toward gigawatt-class deployments, interconnect bandwidth means nothing without system reliability. 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, unlocking the full performance of optics without trade-offs.”
Who’s behind the company
Martin Lund, CEO, brings around three decades of experience in silicon and networking. He previously built Broadcom’s switching business into a billion-dollar franchise and held senior executive positions at Microsoft and Cadence. Most recently, he led Cisco’s Common Hardware Group, overseeing silicon, hardware systems and optics, including Silicon One.
Sanjai Kohli, CTO and founder, co-founded SiRF, which helped bring GPS technology to mass-market devices and earned him the 2010 European Inventor Award. He later founded Inovi, which was acquired by Facebook in 2014.
A crowded AI connectivity market
CScale enters a rapidly growing optical interconnect market as AI infrastructure companies raise increasingly large rounds to address the data-movement bottleneck.
Celestial AI’s $250 million funding highlights the company’s Photonic Fabric platform, which was designed to overcome the limitations of traditional copper-based connectivity. Celestial AI later agreed to be acquired by Marvell for $2.35 billion. Lightmatter’s $400 million funding valued it at $4.4 billion and took its total funding to $850 million.
More recently, TFN covered Eliyan’s $145 million Series C, as the startup expanded into electro-optical interconnects for next-generation AI systems.
CScale is approaching the same infrastructure bottleneck with a different emphasis: keeping optical connectivity operational as AI clusters become larger, more distributed and increasingly difficult to service.
“AI infrastructure is no longer just a compute problem. It is a systems problem,” said Gavin Baker, Managing Partner and Chief Investment Officer at Atreides Management. “As scale-up systems move toward gigawatt-class deployments, interconnect bandwidth means nothing without system reliability. Every optical failure is a compute failure. At this scale, failures are not hypothetical—they are inevitable.”
“At Premji Invest, we partner with teams that think big and have the discipline to build enduring companies, and CScale is exactly that. We are proud to co-lead this financing and support CScale as they bring transformative solutions to market,” added Sandesh Patnam, Managing Partner at Premji Invest.
