Compute that gives back water.
Cascade runs AI training and inference in data centers built to turn GPU waste heat into clean water. We donate the water, and credit your company for every liter your compute produced.
AI COMPUTE
MED DESALINATION
COMMUNITY WATER
THE PROBLEM
AI runs hot. Water is running short.
The heat problem
Every watt of AI compute ends up as heat. Modern liquid-cooled GPU racks reject that heat at 60–70°C, warm enough to do useful work, yet almost all of it is dumped through chillers and cooling towers, often using more water and energy in the process.
Water is getting scarcer
Fresh water demand is rising in the same places compute is growing, and AI’s water footprint is under growing scrutiny from communities, regulators and customers.
HOW IT WORKS
We built the data center around the water.
1. Compute
Your training and inference jobs run on liquid-cooled GPUs in Cascade facilities.
2. Distill
The GPUs’ warm coolant (60–70°C) powers multi-effect distillation, a proven desalination method that runs best at exactly those temperatures.
3. Donate
Clean water goes to local community and public use. Every liter is metered and traced back to the compute that produced it.
GPU compute with a water footprint that runs in reverse.
Cascade runs AI training and inference in its own data centers, turns the heat into clean water, and credits you for every liter.
AI training and inference
GPU capacity for training runs and production inference, billed like any cloud GPU provider.
Verified water attribution
Every customer gets a report of the liters of clean water their compute produced and donated, ready for sustainability reporting and water-positive goals.
Community water
Water is donated locally, never bottled or sold. We’re working with New York State programs and partners on how that impact is verified and credited.
CONTACT
Training a model? Make it count twice.
We’re onboarding early compute customers, water and community partners, and supporters.