Hands-on / the power & water lab
Busy computers get hot. Cooling gear carries the heat away. Try a pretend building and see how the choices connect.
Make it yours / public campaign details
Your name and message appear above the lab. People can try the same learning tool, whichever position you choose.
Your details travel in the share link. Use public campaign information. This builder does not submit a campaign record to the poll or inquiry inbox.
Servers do work and save files. They use electricity and make heat.
A middle balance in this invented example. Real systems can change with the weather.
Invented teaching settings, held steady for one year. What did this show? ↓
IT: 87.6m kWh Support: 26.3m kWh
Power is how fast energy is used. MW means megawatts. 1 MW is 1,000 kilowatts. Energy adds up over time. A kWh is one kilowatt used for one hour.
PUE compares whole-building power with IT power. At 1.30, 1 unit for computers needs 0.3 extra units for cooling and other support. WUE counts liters consumed at the site for each IT kWh.
IT energy = average IT MW × 1,000 × 8,760 hours. Whole-building energy = IT energy × PUE. Water consumed = IT energy × on-site WUE. U.S. gallons = liters ÷ 3.785411784.
The model excludes off-site water used to generate electricity, peak demand, bills, noise, pollution, and local supply. Support power includes more than HVAC. A reused liquid loop does not mean the whole plant uses no water.
Let’s tell the story / in plain words
At your settings, this pretend building uses 113.9 million kWh of energy in one year. It consumes 43.8 million liters of water at the site in one year. That is about 120,000 liters a day, on average.
Some energy does the computer work. The rest helps run cooling and other building gear. A kWh counts energy. A liter measures water.
These are made-up teaching settings. The math holds them steady for a whole year. A real building changes with its work, weather, and cooling plan.
Each story below runs the computers at your chosen 10 MW. Only the cooling example changes. This helps you compare the choices fairly.
These three settings were invented to teach a tradeoff. They do not rank all dry, hybrid, or evaporative systems. Real results depend on the design and the place. Water used to make electricity somewhere else is outside this model.
The big lesson: A cooling choice can change energy and water needs. Adding neighbors reminds us to ask how a real place and its people may be affected.
Try another choice ↑Community listening / adults 18+
You tried the lab. Now tell us what matters to you. No name or email is needed.
This is an open, voluntary learning poll about a hypothetical nearby data center. It is not a permit vote or a representative survey. A ZIP code shows a broad area, not your exact location.
Explore the community questions ↗A building behind your screen
A center might be one room or a whole campus. Not every center runs AI. HVAC means heating, ventilation, and air conditioning. In these buildings, cooling is a big part of the job.
Put the scale in context
Estimated U.S. data-center electricity use in 2023.
Estimated share of U.S. electricity in 2023.
Estimated direct U.S. data-center water consumption in 2023.
Historical estimates from the 2024 DOE/LBNL report. TWh means a billion kilowatt-hours. Direct consumption is different from water taken from a source and excludes water used to produce electricity elsewhere.
GOTDATA.INFO / EVERYDAY LIFE. EVERYDAY DATA.
Your next tap, trip, photo, or game can start a data story. Explore the pictures, share a question, and add your own Got Data? image.
