Every time AI comes up, someone drops the water bomb. “ChatGPT is drinking the planet dry.” I used to nod along, feeling a little guilty every time I typed a prompt. Then I saw this post from an AI professional who lined up the actual numbers, and my jaw hit the floor.
The creator’s whole point is simple: we’re panicking about the wrong thing. So let’s walk through the myths this expert takes apart, one by one.
Myth 1: AI is a water hog compared to your daily life
Here’s the reality check the original poster shared. If you’re worried about AI’s water use, you’d have to give up a lot of everyday stuff first, because each of these uses more water than every AI system on Earth combined:
- Avocados — 5x more water
- Coffee — 8x more
- Almonds — 11x more
- Beer, wine, spirits — 16x more
- Olive oil — 26x more
- Eggs — 44x more
- Clothes — 80x more
- Beef — 84x more
- Cheese — 170x more
- Bread — 412x more
All measured the same way, the author notes: global water use, per year. That last one stopped me cold. Your morning toast out-drinks the entire AI industry by 400 times.
Myth 2: AI uses a scary chunk of the world’s water
According to this contributor, every AI system on the planet, ChatGPT, Claude, Gemini, all of them, uses about 0.5 km³ of water a year. Humanity uses 1,000. Do the math and AI lands at roughly 0.05% of total use.
That’s a rounding error, not a crisis. And the expert points out this figure doesn’t even account for where the technology is heading next.
Myth 3: Data centers waste water forever
This is where it gets interesting. The mind behind the post explains how older data centers cool their chips by evaporating water. It turns to steam, floats off, and it’s gone. Around 2.6 million gallons per megawatt, every single year. Fresh water used once, then wasted.
But the newest hardware doesn’t work that way. The creator describes how NVIDIA’s latest servers run one sealed loop of liquid:
- You fill it once.
- It runs hot, hotter than a hot tub, so the cool outside air handles the rest.
- The same liquid just loops around, again and again.
The result, per the post’s author: close to zero water. Up to 100% less than the old evaporation method.
The truth worth acting on
I think this reframe is genuinely useful, because it moves the conversation from guilt to facts. The point isn’t that AI is perfect or that resource use doesn’t matter. It’s about proportion. When something makes up 0.05% of the problem and its newest generation is trending toward zero, aiming your outrage there is misfiring.
The original poster closes with a line I loved: so keep sipping your almond milk coffee, with a slice of sourdough, wearing your jeans, tweeting from your iPhone about how AI water usage is ruining everything. The everyday stuff in that sentence quietly outweighs the thing you’re mad about.
How you can use this: next time the AI water argument shows up in your feed, you’ve got the actual figures. Compare like with like, check whether the data center in question uses old evaporative cooling or a sealed loop, and follow where the numbers point instead of the headline.
The full breakdown from this savvy professional has more detail on the cooling tech and the measurements. Give the original LinkedIn post a read, it’s worth your time.