A new partnership aims to turn one of tech’s biggest headaches into a source of clean water. Data center waste heat will drive a new kind of desalination, and the idea starts with a simple swap. Endeavour as well as Trevi Systems announced the partnership on October 22, 2026. The partners will pair a ThermalWorks closed loop waterless cooling platform with forward osmosis technology. Trevi Systems brings the membrane science, and Endeavour brings the data center heat.
Forward osmosis pulls water through a membrane using a draw solution. Heat at 60 to 80 degrees Celsius regenerates the draw solution for reuse. Data center waste heat, as a result, becomes the engine of the whole process, not a burden. The system, for example, can turn brackish groundwater into clean water for daily use. Each megawatt of computing capacity could yield up to 200,000 gallons a day from brackish groundwater.
Oilfield produced water offers a second feedstock, and the yield drops to about 50,000 gallons a day per megawatt. A 100 megawatt campus could produce up to 20 million gallons a day under these projections. The process turns leftover brine into solids, and that design means zero discharge to local waterways. Data center waste heat will power this cycle, and engineers will tune each stage for steady output.

A data center that makes water flips a familiar story on its head. Critics often call data centers thirsty, and this design answers that charge with output, not just savings. The first deployment will run at Endeavour’s own data centers in the United States. Teams will learn from real operations, meanwhile, and they will refine the system step by step. Data center waste heat could, in addition, help towns near large campuses secure a steadier water supply.
Honesty matters here, and readers should note one key limit of the announcement. These figures are company projections, however, not measured output from a commercial AI data center yet. No commercial AI data center has published measured results from this exact pairing to date. The partners, for instance, still need to prove scale, cost and reliability in daily operation. That proof will decide whether towns and developers embrace the idea or wait for stronger evidence.
Data center waste heat, for example, can make up to 31 liters of clean water for every kilowatt hour of computing. Conventional desalination plants spend heavy energy on brine concentrators and crystallizers, and Trevi says its approach can cut that energy demand by up to 85 percent. Trevi also claims up to 98 percent water recovery, since the process turns leftover brine into solids instead of liquid waste.

The scale of computing demand makes the idea matter beyond one campus. Data centers used 415 terawatt hours of electricity in 2024, according to the International Energy Agency, and that figure could rise to about 945 terawatt hours by 2030. Captured heat from that future fleet could, in theory, produce more than half of the 130 million cubic meters of clean water every desalination plant on Earth makes each day. A conventional evaporatively cooled facility, for instance, uses about 1.8 liters of water per kilowatt hour, so a waterless cooling loop already changes the starting point.
Endeavour will prove the system first at its own data centers in the United States, and broader work with hyperscalers and other operators will follow. Founder and chief executive Jakob Carnemark says Trevi’s process runs on heat that is normally thrown away, and that shift changes the arithmetic for water scarce regions. Data center waste heat could become the largest new source of fresh water on Earth, if early deployments prove the economics at scale.
The promise still excites water experts, and the logic of reuse feels hard to ignore. Data center waste heat must clear the same bar as any new water source: safety, cost and trust. The partners have set a clear opening step, and AI centres show how fast computing demand keeps growing. Readers can read the full report through the original story linked at the end of this article. Clean water from wasted heat would help many people, and that hopeful goal deserves careful and honest testing.
