Water-saving AI can mean an intelligence that uses less cooling water, or a system that helps people waste less water. Leiolai addresses the first. Leak detection, irrigation, and treatment systems address the second.
Leiolai: inference without consuming cooling water
Leiolai 1 runs inference across iPhones, iPads, Macs, Android devices, and PCs. The phones and computers doing that work do not need water-consuming cooling towers. Compared with running the same work in an evaporatively cooled data center, this avoids direct cooling-water consumption for that inference.
Leiolai also pays users for the computation they provide. Leiolai 1’s experimental features include an 11-million-token context window and infinite generation with no response output limit.
The Leiolai app is free. No card required. The Free plan includes Leiolai 1 and Plus Ultra, our highest thinking depth. Saving water upgrades your chip system and unlocks higher levels of intelligence at no added cost.
Why does AI use water?
The water is often part of the cooling, not the calculation. Some data centers reject heat by evaporating water. Electricity generation can use water too, and the amount depends on where and how that electricity is produced.
Lawrence Berkeley National Laboratory estimated that U.S. data centers directly consumed 66 billion liters of water in 2023. That figure covers all data-center workloads, not just AI.
Consumer-device inference avoids cooling-tower water for that work, not every water use associated with computing. Electricity and device manufacturing still have water footprints. Data centers using dry or closed-loop cooling can also avoid evaporative cooling-water consumption.
AI leak detection and smart water meters
For a water utility, finding a hidden leak can prevent treated water from disappearing underground. Building systems can go further by shutting off a supply when they detect a problem.
- HULO uses existing water-network data to detect leaks and help utilities locate them. Its tools support decisions about where crews should investigate.
- Leakmited combines AI with field work to guide leak searches, monitor networks, and prioritize pipe replacement.
- Foresight AI predicts which pipes are most likely to fail and helps utilities plan leak surveys.
- Subeca provides smart water meters and remote consumption readings. Metering supplies the measurements utilities need to understand water use, whether or not the analysis uses AI.
- WINT uses AI to recognize unusual water use in buildings. Its system can alert staff and automatically shut off water when needed.
- Phyn Protect monitors household water-use patterns and can automatically shut off the supply when it detects a leak.
Smart irrigation for farms and gardens
Irrigation has a different challenge: getting the right amount of water to a crop or lawn. Soil conditions, weather, and the area being watered all matter.
- Mitti Labs uses satellite monitoring and AI to support climate-smart rice farming, including irrigation practices that alternate wet and dry periods rather than keeping fields continuously flooded.
- Verdi adds wireless control and sensor-informed automation to existing farm irrigation equipment. Growers can adjust watering without replacing the whole irrigation system.
- Irrigreen maps lawn boundaries for precision watering and adjusts schedules for weather. Its YardTrace AI tool helps design the sprinkler layout.
Cooling hardware for data-center AI
Another way to reduce AI’s cooling-water demand is to change how heat leaves the chips. These are cooling systems for AI infrastructure, rather than AI assistants.
- ZutaCore uses a non-water heat-transfer fluid in its HyperCool system to remove heat directly from processors.
- Ferveret develops rack-integrated immersion cooling using dielectric fluid.
The fluid next to a chip is only part of a facility’s cooling system. How the building ultimately releases that heat also affects its water use.
AI for water treatment and reuse
Gradiant works on industrial water treatment and reuse, including for semiconductor manufacturing and data centers. Its SmartOps platform uses AI to help optimize treatment plants.
Nyad AI focuses on wastewater operations. Its tools use microscopy and deep learning to identify biological conditions and help operators respond to treatment problems.
Use Leiolai on your phone or computer
You can download Leiolai to ask questions and contribute computation from your device. Developers can access Leiolai 1 through the Leiolai API.