Omen AI raises $31M Series A to monitor liquid cooling fluid in AI data centers

Omen AI raises $31M Series A to monitor liquid cooling fluid in AI data centers

Omen AI builds real-time fluid monitoring systems for liquid-cooled AI data centers. The company raised a $31M Series A led by Nava Ventures, with strategic investors including Sheryl Sandberg and C-suite executives from CoreWeave and other data center operators.

“We do blood monitoring for mission critical machines. We work with almost every Neo Cloud that's out there in some form of pilot and deployment. Nine months ago, folks were just starting to get these systems online and now you're starting to see folks who are running dozens, maybe hundreds of megawatts worth of liquid cooled compute, and they're starting to see failures.”

The problem

Liquid cooling systems in AI clusters typically run a water-glycol mix, either ethylene glycol or propylene glycol, circulated in closed loops around GPU racks. The fluid degrades over time as pumps and seals wear, leaching copper, chromium, and other contaminants into the loop. Heat accelerates the breakdown, and once the glycol-to-water ratio degrades far enough, biological growth can take hold, corroding chips and clogging pipes.

The current standard for catching this is sending fluid samples to an external lab and waiting two to three weeks for results. By then, the only real option is a full flush, which takes four to five hours per rack. On a loop running 1,000 GPUs, that's 5,000 GPU hours of downtime per rack — costing what Laberge estimates at millions of dollars per incident.

What Omen AI does

Omen AI's sensor plugs directly into the rack manifold via standard OCP quick-disconnect fittings, drawing power and network connectivity from the cabinet itself. Installation takes a few minutes. The device monitors fluid chemistry daily, giving operators enough signal to make smaller, targeted interventions — bio-shocking the loop, adjusting the glycol-to-water ratio — before a full flush becomes necessary.

Go-to-market

Omen AI is currently working with nearly every major neo-cloud operator in the US in some form of pilot or deployment. Laberge says the sales motion has shifted noticeably in the past nine months: early on, customers weren't sure what their problems were; now, with hundreds of megawatts of liquid-cooled compute online, failures are accumulating and urgency is real. Some of its partners also operate internationally, and Omen AI is evaluating a handful of projects in Europe.

Company

The business has grown from essentially one person at seed — raised roughly eight months ago — to 20 employees today, with plans to reach 40 by the end of 2026. Capital from the Series A is going toward headcount and manufacturing scale.