CScale Lands $145 Million Series C for Gigawatt-Scale AI Interconnect

By Amara Osei

CScale has raised $145 million in Series C funding to build optical interconnect technology for gigawatt-scale AI infrastructure, according to Pulse 2.0. The round ranks among the larger hardware bets placed this cycle on the plumbing beneath AI datacenters — not the chips themselves, but the links between them.

The number matters more than the press release. Gigawatt-scale AI clusters — single installations drawing a gigawatt or more of power — don't exist in most of the world yet. CScale is raising capital today for a build-out whose customers are still scaling toward that threshold. The $145 million signals investor confidence that the interconnect layer becomes a bottleneck, and a market, before the gigawatt era fully arrives.

What does optical interconnect actually solve?

Training and inference at AI's frontier scale happens across thousands of accelerators working as one machine. Copper links carry signals between those chips, but copper hits hard limits on distance, bandwidth and power draw as clusters grow. Optical interconnect replaces electrical links with light, moving data farther and faster while cutting the energy spent on movement rather than compute.

That trade-off drives the investment case. At gigawatt scale, every percentage point of power diverted to moving data is power not sold to computation. Interconnect vendors who fix that arithmetic sell into the fastest-growing capital-expenditure line in technology — datacenter build-outs for large AI model training.

Who else is in the race?

Pulse 2.0's report frames CScale's raise within a crowded optical interconnect field. Major AI hardware players have pushed optics deeper into their rack architectures, and a wave of funded startups targets the same bottleneck from different angles: co-packaged optics, optical circuit switches and new fiber formats.

CScale's $145 million Series C puts it in the serious-money tier of that field. Series C capital typically funds scaling — production capacity, engineering headcount and customer deployment — rather than early research. The raise implies CScale believes its technology is ready to move from lab to datacenter floor.

Why does the round size matter?

Optical hardware is capital-hungry in a way software isn't. Fabrication, precision assembly and qualification with hyperscale buyers all burn cash before revenue arrives at scale. A $145 million Series C gives CScale runway to survive that gauntlet without a forced exit or a down round if deployment cycles run long, as hardware cycles often do.

The round also reads as a vote on AI infrastructure spending itself. Investors writing nine-figure checks for interconnect startups are underwriting a future in which AI clusters keep growing past today's largest installations. If that growth stalls — through power constraints, chip supply or cooling costs — the demand case for gigawatt-class interconnect weakens with it.

What comes next?

The reporting from Pulse 2.0 confirms the raise and its purpose: building optical interconnect for gigawatt-scale AI. The open questions are deployment specifics — which datacenter operators CScale works with, what production volumes it targets and when its hardware ships at scale. Those details will decide whether the $145 million buys CScale a seat at the table when the first true gigawatt clusters come online, or a long wait for the market to catch up to the thesis.