The MIT spinout that looked at the plume drifting off a cooling tower and saw clean water - then built the physics to grab it back.
Walk up to a power plant on a cold morning and the first thing you notice is the column of white rising off the cooling towers. Most people read it as exhaust. Infinite Cooling reads it as inventory. That plume is nearly distilled water - water the plant already paid to pump, treat, and evaporate - and for a century the industry has watched it float away. A small team in Malden, Massachusetts spends its days making sure it doesn't.
Today Infinite Cooling is roughly seventeen people sitting at an unlikely intersection: heavy industry, atmospheric physics, and machine learning. It sells hardware that bolts onto cooling towers already standing, and software that tells operators exactly how much water and energy those towers are bleeding. Its customers are the least sentimental buyers on earth - power plants, data centers, refineries, manufacturers - the kind who don't sign purchase orders for vibes. They sign because the math works.
Cooling towers are the quiet glutton of industrial water use. In most power plants they are the single largest consumer of water on site, and the way they cool is by evaporation - which is a polite way of saying they boil off enormous volumes of fresh water on purpose. The water that leaves as visible plume is some of the cleanest on the property. It is also, traditionally, gone forever.
This was tolerable when water was cheap and nobody was counting. That era has quietly ended. Drought, tightening discharge rules, and the arrival of water-hungry data centers turned an ignored line item into a constraint. The industry's answer for decades was a shrug and a bigger water bill. Which, as business strategies go, has the virtue of being simple.
By one estimate, the water lost as plume across the world's cooling towers adds up to roughly 400 billion gallons a year. That is not a rounding error. That is a resource the size of a small inland sea, evaporating in plain sight, every year, while everyone agreed there was nothing to be done about it.
CAPTION: Somewhere, an accountant is weeping into a spreadsheet shaped like a cloud.
In 2016, Maher Damak was a PhD student in mechanical engineering at MIT, running experiments on how to control sprays and fog using electric fields, alongside his advisor, professor Kripa Varanasi. The work was about getting droplets to go where you want - useful for things like pesticide spraying. Then Damak and co-founder Karim Khalil noticed something. The same trick that steered a fog could, in principle, collect one. Charge the droplets, give them a surface to land on, and a cloud becomes a collection problem.
A cooling tower plume is just a very large, very industrial fog. The leap from lab bench to power plant smokestack is the entire company.
They incorporated the idea, entered MIT's startup gauntlet, and in 2018 won the MIT $100K Entrepreneurship Competition - the kind of credential that opens doors that a clever PDF cannot. Varanasi stayed on as chairman, lending the academic gravity; Damak took the CEO seat; Khalil ran the technology. Three people, one observation about charged droplets, and a conviction that the fix had been hovering overhead the whole time.
The flagship is WaterPanel. It is a retrofit - a structure of charged collection surfaces installed at the top of an existing cooling tower. As plume rises, the system charges the droplets and pulls them onto the panels, where they coalesce and drain as recovered water. In ideal conditions it can capture up to about 80 percent of the escaping vapor. At nuclear scale, with rougher conditions, the realistic recovery runs closer to 1 to 15 percent of evaporated water - still a meaningful number when the denominator is a 3.6GW plant. The recovered water comes out remarkably pure, which means it can go straight back into the loop and cut treatment costs at the same time.
WaterPanel doesn't ask a plant to build a new tower or change how it operates. It bolts onto what's already there. For an industry that treats downtime as a four-letter word, "we'll add it to your existing equipment" is far more persuasive than "we'll reinvent your facility."
CAPTION: The most radical thing about the technology may be that it requires you to demolish nothing.
The second idea is TowerPulse, an AI-powered analytics and monitoring platform. Sensors plus machine learning give operators a live read on how their cooling system is performing - where it's wasting energy, where it's wasting water, and what's about to break. Infinite Cooling pitches gains on the order of higher output, lower energy use, and lower water use. The hardware captures water; the software makes sure you knew you were losing it in the first place. There is also a plume abatement retrofit for facilities that simply need the visible cloud gone to satisfy regulators - low power, no giant heated-air system required.
It is one thing to win a campus pitch competition. It is another to get the world's largest nuclear power producer to put your equipment on a working reactor. In 2024, EDF began testing WaterPanel at its 3.6GW Bugey Nuclear Power Plant in France, with a trial running from August 2024 into March 2025. The goal: measure exactly how much water comes back, how pure it is, and whether the system disrupts anything. Nuclear operators do not run experiments for fun. They run them when the downside of doing nothing has gotten expensive.
Behind the EDF headline sits a less glamorous list: utilities, Fortune 500 facilities, national laboratories, and a stack of federal grants from the DOE and NSF that funded the early science. The Series A, $12.25 million led by Material Impact in 2021, was the moment the capital markets agreed the fog experiment had a balance sheet. Total funding sits around $16.5 million - lean for hardware this ambitious.
Strip away the patents and the pitch decks and Infinite Cooling is built on a single, stubborn idea: industrial cooling should be a closed loop. Water should go in, do its job, and come back - not vanish into the sky while a plant trucks in more. The company frames its purpose as mitigating water scarcity by eliminating losses in the most water-intensive industries. That is a tidy sentence for an enormous problem, and the founders seem comfortable that the gap between the two is exactly the size of the opportunity.
Here is the uncomfortable part. The demand for cooling is not flattening - it is accelerating. The AI build-out runs on data centers, data centers run hot, and hot infrastructure runs on water. Every new server hall is another set of cooling towers, another plume, another column of clean water headed for the sky. The problem Infinite Cooling formed to solve is not shrinking with time. It is compounding.
Which makes the company's bet less a clever piece of physics and more a wager on arithmetic. As water gets scarcer and cooling gets hungrier, the value of the water already on site - the water in the plume - only goes up. Infinite Cooling didn't invent the scarcity. It just refused to accept that the cleanest water on the property was un-catchable.
So go back to that cold morning and the white column rising off the towers. The old story ends there: water leaves, nobody counts, the bill arrives. Infinite Cooling is writing a different ending - one where the plume thins, the water drains back into the loop, and the cloud that used to mean loss starts to mean nothing at all. The steam, finally, has somewhere to be.