Atoco Demonstrates Water Harvesting With Simulated Data-Center Waste Heat
The startup says it is discussing pilots with major tech companies. Its demonstration produces up to 300 liters daily, far short of large facilities’ cooling needs.
Loading page…
The startup says it is discussing pilots with major tech companies. Its demonstration produces up to 300 liters daily, far short of large facilities’ cooling needs.
Listen to this story
Atoco is exploring whether data-center heat can power atmospheric water harvesting, but its Irvine demonstration used a simulated heat source rather than a live server facility. The prototype’s output and economics remain far from proving hyperscale usefulness: Atoco reports up to 300 liters a day from heat equivalent to a 20-kilowatt data center, while targeting a 1,000-liter-a-day commercial unit and orders by the end of 2026. The proposal could turn cooling waste into a water resource, but scale, cost, and customer demand are still unverified.
Atoco says it has deployed five grid-connected prototype trials with partners in the United States and Gulf Cooperation Council countries.
CEO Samer Taha says the system can release water using heat as low as 100°F and operate at relative humidity in the teens.
Atoco estimates production costs at $5 per metric ton, above the $2-or-less cost cited for newer desalination plants.
Atoco has demonstrated a machine that pulls water from air using heat meant to mimic a data center’s waste output. In an October 7 WIRED report, CEO Samer Taha disclosed talks with major technology companies about pilots. The Irvine demonstration offers a working prototype—not yet a proven way to supply large computing facilities.
Data centers were not part of Atoco’s thinking four or five years ago, Taha told WIRED. Their rapid buildout changed that. Server facilities must dispose of heat, and many cooling systems also consume water. Atoco’s proposal is to use one of those burdens to help address the other.
During a September demonstration at its Irvine facility, Atoco piped heat at around 150 degrees Fahrenheit into a 20-foot-tall machine. No data center was nearby. The heat source simulated one, and WIRED observed water flowing from the unit’s spigot.
The machine uses metal-organic frameworks, or MOFs: porous materials that act like molecular sponges. Air passes through material tailored to attract water molecules, which stick inside its pores. Heat then releases the captured water so the material can repeat the cycle.
Atoco says its system works when relative humidity is in the teens, and Taha says heat as low as 100 degrees Fahrenheit can release the water. The harvesting operation itself does not require electricity, although a solar panel powers other systems on the demonstration unit.
We actually want it; we need it. We give you back water, the same product that is causing the issue.
Samer Taha, Atoco CEO, speaking to WIRED about data-center waste heat
The demonstration unit can produce up to 300 liters daily, using heat equivalent to that from a 20-kilowatt data center. Hyperscale facilities are orders of magnitude larger, and some data-center cooling systems consume millions of gallons per day. Atoco envisions installing multiple harvesting units together.
The company says it has deployed five grid-connected prototype trials with partners in the United States and Gulf Cooperation Council countries. It is developing a commercial offering targeting up to 1,000 liters daily and hopes to begin taking orders by the end of 2026.
Those plans do not yet come with named data-center customers. Taha declined to identify the technology companies in pilot discussions or disclose what a unit would cost.
Taha puts Atoco’s levelized production cost at $5 per metric ton of water. Newer desalination plants can produce water for $2 per ton or less. He says reaching that lower cost could take Atoco three to five years.
Seth Cohen, UC Irvine’s physical sciences dean, cautions that even well-designed MOFs need energy to release water without an outside heat source. Atoco’s technology remains untested at scale, with commercial manufacturing and meaningful water volumes still to establish.
Loading discussion...
Join the conversation
Explain when reuse would justify paying more.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.