I think AI will eventually decide humans are too stupid to live. AI doesn't have to eat (other than electrons) and it doesn't have to earn meney and it doesn't engage in activity that is purposely designed to destroy valuable things and it doesn't engage in activity that is in direct conflict with its own notion of what's best. It will end up being the god they all have been wishing for. Thank you for your attention to this matter.
I'm no expert but it sounds like the power suck is for calculations AND refrigeration. Then, the water is for further cooling.
I broke down and asked AI.
AI data centers use large amounts of water mainly for cooling. The servers running AI models generate enormous heat, especially the specialized GPUs used for training and inference. If that heat is not removed continuously, the hardware can fail or throttle performance.
Here’s where the water goes:
Cooling towers and evaporation
Most large data centers use industrial cooling systems similar to those in hospitals or power plants. Water circulates to absorb heat, then some of it evaporates in cooling towers to carry the heat away.
Evaporation is the key reason consumption is high:
Water that evaporates is permanently lost from the local supply.
A hyperscale AI campus may use millions of gallons per day during hot weather.
Why AI increases demand
Traditional cloud computing already required cooling, but AI workloads are unusually power-dense:
AI training clusters pack thousands of GPUs into tight spaces.
GPUs can consume 5–10x more power per rack than older servers.
More electricity consumed = more waste heat = more cooling required.
For example:
A normal enterprise server rack might use 5–10 kilowatts.
Advanced AI racks can exceed 100 kilowatts.
Water is often the most efficient cooling method
Air cooling alone becomes difficult at very high densities. Water-based cooling is attractive because:
Water transfers heat far better than air.
It lowers electricity use for cooling.
It can improve server performance and lifespan.
Many newer AI facilities use:
Direct-to-chip liquid cooling
Rear-door heat exchangers
Immersion cooling in specialized fluids
These methods can reduce total cooling energy, but they still often depend on water somewhere in the system.
The “hidden” water use: electricity generation
There’s also indirect water consumption:
Data centers use enormous amounts of electricity.
Power plants (especially nuclear, coal, and some natural gas plants) also use water for cooling.
So AI water impact includes both:
Water used at the data center itself
Water used to generate the electricity feeding it
Why this has become controversial
Large AI campuses are increasingly being built in dry regions where electricity is cheap and land is available, including parts of:
Arizona
Texas
Nevada
Critics worry about:
Competition with local drinking water supplies
Agricultural water shortages
Strain during droughts
Companies argue that:
Modern systems recycle much of their cooling water
Some facilities use reclaimed wastewater instead of potable water
AI can help optimize energy grids, weather forecasting, and infrastructure efficiency
Typical scale
Exact numbers vary widely by design and climate, but estimates often cited are:
A single AI query can indirectly consume a small bottle’s worth of water when accounting for cooling and electricity generation.
Large AI training runs may require millions of gallons over their duration.
The hottest climates generally require the most cooling water because evaporation rates rise sharply with temperature.
Great, so AI is going to add to global warming too.
Yes — AI data centers will contribute to global warming unless their electricity and cooling systems become much cleaner. The main issue is not the computers themselves, but the energy required to run and cool them.
There are several major climate impacts:
1. Electricity consumption
AI systems require enormous computing power:
Training large models can consume as much electricity as a small town over weeks or months.
Running AI continuously (“inference”) across millions of users may ultimately use even more energy than training.
If that electricity comes from fossil fuels like coal or natural gas, the data center produces significant carbon emissions.
2. Cooling systems
Cooling itself uses energy:
Pumps
Chillers
Cooling towers
Fans
In hot climates, cooling demand rises substantially.
3. Construction footprint
Building AI infrastructure also has a carbon cost:
Concrete and steel production
Semiconductor manufacturing
Backup generators
Transmission lines and substations
Advanced AI chips are especially energy- and resource-intensive to manufacture.
How large could the impact become?
Analysts expect AI electricity demand to grow very rapidly this decade. Utilities and grid operators in places like:
Virginia
Texas
Arizona
have already reported major increases in projected power demand due partly to AI and cloud computing growth.
Some forecasts suggest global data centers could consume a few percent of all world electricity within several years if current trends continue.
But there’s another side
AI could also help reduce emissions in other sectors by improving:
Electrical grid management
Building efficiency
Industrial processes
Logistics and shipping
Weather forecasting
Scientific research
Battery and materials discovery
So the net climate effect depends on whether:
AI energy demand grows faster than clean energy supply, or
AI-driven efficiencies reduce more emissions elsewhere than the data centers create.
Why companies are racing for nuclear and renewables
Many large tech companies are now investing heavily in:
Solar
Wind
Battery storage
Small modular nuclear reactors
Long-term nuclear power agreements
because they know future AI growth may otherwise overwhelm clean-energy goals.
One important nuance
Not all AI computing is equal:
Simple AI tasks can be relatively efficient.
Massive video generation, large-scale training, and always-on AI assistants consume far more energy.
Efficiency improvements in chips and cooling are happening quickly, but demand is currently growing even faster.
Eamus Catuli~AC 000000000101010202020303010304 020405....Ahhhh, forget it, it's gonna be a while.
... Discontent surrounding the rapid growth in the construction of data centers - fueled by AI demand - has been growing in the United States, with many grassroots movements successfully delaying and cancelling over $130 billion worth of buildouts.
Many of the local groups opposing data centers have been bipartisan and largely working class, with other notable efforts coming from both Democrat and Republican representatives....
What happened to the Chinese psyop?
Earlier in June, several high-ranking Republican officials backed by fossil fuel and cryptocurrency advocacy groups sent an open letter calling on FBI Director Kash Patel to investigate evidence that data center opposition was being fueled by "foreign influence campaigns targeting artificial intelligence development in the U.S.".
The evidence given in the letter pointed the finger at China, claiming that Chinese propaganda was being funneled into the "environmental left" to assist in "hijacking MAGA" to pass legislation opposing the construction of data centers.
... The wave of Republican opposition to data centers is likely to cause a rift in President Trump's support base, with many supporting the President himself and his planned buildout of fossil fuel energy production, but opposed to the rapid expansion of data centers and the growing negative sentiment around AI.
Opps.
If you want to achieve greatness stop asking for permission.
1312. ETTD. 86 47.
... Every university in America is now grading two populations, students and students plus ChatGPT, on one curve. The honest kids in Serrano's class watched a 96 average get set by machines, then sat a real final against it. The cheaters lost nothing. That's the incentive structure now, and it grades itself.
Ftr, in this DonOLD year 2.2 Idk why we are celebrating 250, either.
... The rate of growth of this “shadow grid” of custom power plants, some of which are big enough to fuel entire cities, is so enormous that the only global entity installing more gigawatts of gas plants than Texas is China, according to environmental group Global Energy Monitor.
On a national scale, scientists are still racing to wrap their heads around the environmental footprint of our new AI obsession. Cornell researchers found that at the current rate of AI growth, the burgeoning industry could represent 24 to 44 million metric tons of carbon dioxide emissions by 2030, the equivalent of adding five to ten million cars to US roadways.
Meanwhile, ongoing projects like the one in Abilene, Texas, the starting point of president Donald Trump’s flagship $500 billion Stargate project, are obtaining environmental permits that are usually reserved for small businesses like gas stations or dry cleaners, as Wired reports. The project’s facilities feature a whopping 62 diesel backup generators, making it orders of magnitude bigger than smaller ventures that use just one or two....
If you want to achieve greatness stop asking for permission.
1312. ETTD. 86 47.
Ya know, I guess I'm missing the boat on the AI revolution.
My dealings with it in the service industry haven't been stellar, sometimes exceedingly frustrating.
Guess it's good for looking up recipes, but can anyone explain how it's really going to improve the quality of anyone's life?
I think I've had at least as good results dealing with AI than I've had speaking with Habib in India who also has a script to follow. I don't have all the answers to all the ways AI may affect our lives, but your question reminds me of myself when the early personal computers came out. I wanted one, and bought one, but remember saying numerous times that "once you have one, you can think of lots of stuff to do with it, but I don't see that it's really necessary to have one." That was back when it probably took longer to pay bills on Quicken than it would have just to write them out, but Quicken was sooooo cool.
Might require:
Not sure the tribe or Weaverville would put up with that, and electric cooling would multiply the power demands.
There are thirsty children in China.
