Omen AI raised a $31 million Series A led by Nava Ventures to put real-time spectrometers on the cooling loops that keep AI data centers from melting down. The round, which includes CRV, Vanderbilt University, Mann+Hummel, Starhill Holdings, Hard Launch Capital, and personal checks from executives at Bridgestone, GM, Johnson Controls, and TensorWave, brings the two-year-old startup's total funding to $40 million. The pitch is unglamorous and load-bearing: as operators push GPU racks harder, the water-heavy coolant mixes that absorb the extra heat also breed bacteria, and a single contaminated rack can cost millions.
The problem is mechanical, not algorithmic. Liquid-cooled chips run on a mixture of water and an additive that suppresses bacterial growth. To squeeze more performance out of a rack, operators raise the water content for better heat absorption — and unlock a contamination problem that clogs flow. The fix today is to flush the system, which can pull a rack offline for five or six hours.
Omen's device is a tiny optical spectrometer that watches fluid chemistry continuously, flagging bacterial growth, copper or chromium from worn pumps, and silicon from failing seals before any of it becomes a shutdown. The alternative — mailing fluid samples to a lab and waiting — is what most of the industry still does.
Key facts
- 01Omen AI raised a $31M Series A led by Nava Ventures, bringing total funding to $40M since its 2024 founding.
- 02The startup's spectrometer monitors liquid-cooling fluid in real time to flag bacterial growth before it forces a flush.
- 03A single rack flush can shut down hardware for five or six hours at a potential cost of millions of dollars.
- 04Omen is working with a dozen data center customers, including AMD-based AI cloud TensorWave.
- 05Founder Zach Laberge started his first sensor company in 2020 at age 14, raising $3M before dropping out of high school.
Founder Zach Laberge took an unusual route here. He started his first company in 2020 at age 14, raising $3 million to put sensors on construction equipment and dropping out of high school to run it. After that startup wound down, he founded Omen in 2024 to apply the same real-time fluid-monitoring idea to heavy machinery, with Caterpillar dealerships as an anchor customer.
Caterpillar also happens to be a major supplier of on-site turbines and generators for data centers, and that adjacency dragged Omen into a much larger market within months.
The pivot was customer-led. Omen quickly figured out that data center buildings are full of fluid systems — HVAC, chilled water loops, and the chip-cooling rigs themselves — and that nobody had good chemistry telemetry on any of them. A dozen data center customers are now working with Omen as it builds out the product, including TensorWave, which is constructing an AI compute cloud on AMD chips.
TensorWave's interest is operational, not speculative. AMD-based AI clouds run at the bleeding edge of thermal density, and an unplanned rack flush in the middle of a training run is the kind of incident that erases margin on a contract.
Cory Rellas, a partner at Nava Ventures who joined Omen's board, said much of the firm's diligence came through introductions to Omen's existing large customers, who validated the approach before the term sheet was signed. He called it rare to see a young founder with that kind of standing among slower-moving industrial buyers.
Omen is not alone in the category. Pyxis, an established water-monitoring firm, rolled out its own data center coolant product earlier this month, signaling that on-premises fluid analytics is shifting from a niche to a contested market. Laberge argues the timing is structural: optical hardware has gotten cheap enough to deploy at scale, and signal-processing software has gotten good enough to pull a clean reading out of a noisy industrial environment.
The risk for Omen is that hyperscalers eventually build this telemetry in-house, the way they have with power distribution and custom cooling skids. Pyxis is also further along on water-treatment relationships, and a dozen customers is a beachhead, not a moat. Omen's bet is that the optical-plus-software stack moves faster than industrial incumbents and stays ahead of internal hyperscaler tooling long enough to win the colocation and neo-cloud tier first.
The interesting market signal here is where the AI infrastructure capex is finally landing. The headline numbers — Broadcom chip orders, gigawatts of power, multi-year Nvidia contracts — get the attention, but the unit economics of an AI data center increasingly turn on uptime per rack, and uptime per rack increasingly turns on coolant chemistry. A $31 million round for a spectrometer company is what the build-out looks like at the boring, expensive layer underneath the model launches.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
