🚀 Omen AI: Down to the Molecule

The water that keeps AI's GPUs cool can quietly grow bacteria, shed copper, and chew through seals — and by the time an operator notices, a rack is dark for six hours and the bill runs into the millions. This week's startup, Omen AI, reads that coolant molecule by molecule, in real time. Let's check them out!

As AI racks blow past 130 kilowatts, air cooling has hit a physical wall, and data centers are now pumping liquid coolant directly over the chips to keep them alive. That water has become mission-critical plumbing — and almost nobody is watching what's happening inside it. The industry standard is to mail a coolant sample to a lab every few months and wait days for the results, which means a bacterial bloom or a spike in wear metals can fester for weeks before anyone notices. A single contaminated loop can force a five-to-six-hour shutdown that costs millions.

Enter Omen AI. The company drops a miniaturized spectrometer straight into the coolant loop and reads more than 21 elemental and biological signatures in real time — the copper and chromium that signal pump wear, the silicon from degrading seals, the bacteria before it blooms. Founder and CEO Zach Laberge is 20 years old and a high-school dropout who sold his first sensor company, Frenter, before he turned 19; CTO Matthew Maduzia runs the hardware. In June 2026 they raised a $31M Series A led by Nava Ventures, and the company says its sensors already sit in data centers managing 10–14 gigawatts of capacity. As Laberge puts it: "You're not risking huge amounts of downtime because you have no insight into what's happening chemically."

• 🌐 Website: omen.ai

• 📅 Founded: 2024 (San Francisco)

• 👥 Team: Zach Laberge (Founder & CEO), Matthew Maduzia (CTO)

• 💰 Stage: Series A — $31M led by Nava Ventures (CRV, Vanderbilt, Mann+Hummel; angels incl. Sheryl Sandberg) · ~$41.5M total, after a $10.5M seed

• 📈 Traction: Company reports ~a dozen data-center customers; sensors across operators managing ~$200B in assets; early customer TensorWave

• 🔮 Tech Trend: AI data-center infrastructure & liquid cooling

• 🎯 Target Market: AI data-center operators (plus industrial machine fleets)

• 🏢 Location: San Francisco (~20-person team)

• 🌊 The timing is physical, not hype: Liquid cooling isn't a trend a startup has to manufacture — it's being forced by watts. NVIDIA's GB200 racks pull ~130 kW and next-gen designs push toward 240 kW, far past the ~35 kW ceiling where air cooling gives up. That's why the data-center liquid-cooling market is projected to grow from $4.07B in 2026 to $27.65B by 2033 — and every one of those loops needs someone watching the water.

• 🎒 A founder who's shipped hardware since 14: Laberge started a construction-equipment sensor company at 14 and raised ~$3M before dropping out of high school. Omen is that same instinct pointed at a bigger machine — he built fluid sensors for cranes and forklifts before turning them on the racks powering AI. The backers betting on him (CRV, filtration giant Mann+Hummel, and angels like Sheryl Sandberg) seem to be wagering on the operator as much as the idea.

• 🛡️ A real wedge into an unglamorous problem: Swapping "mail a sample to a lab" for a continuous molecular readout is the kind of boring-but-essential infrastructure that compounds. The challenge will be execution: this is hardware, so scaling means manufacturing, long data-center sales cycles, and real rivals — water-monitoring incumbent Pyxis Lab shipped a competing coolant monitor the very same month Omen raised. But with ~$41.5M in the bank and customers already live, Omen has a head start on a market that's only getting hotter — literally.

• 📱 See the sensor and book a demo

• 💼 Join the team — they're hiring hardware and go-to-market roles in SF

• 📰 Read the TechCrunch story on the raise

• 🐦 Follow Zach on X

• 📧 Reach the team at hello@omen.ai