I agree with your link that the unsustainable use of water for almond farms causes problems, and agree that these problems are quite similar to that of water use for AI, so both should have high scrutiny. Quoting your link:
> Almonds use roughly the same amount of water as total indoor water use in California. This is unsustainable.
> There are simply too many almonds being planted in areas with unsustainable water supplies (minimal to no groundwater, uncertain or no surface water). ...
> We’re not opposed to growing almonds; we’re opposed to growing them unsustainably and taking water away from the environment and our cities. Almonds are a luxury and export crop; approximately 70% of the state’s crop is shipped abroad for the express benefit of a handful of industrial-scale growers and hedge funds.
> Water is a public trust resource, and Californians must regain control of it as they face a future certain to be beleaguered by drought, low reservoirs and over-drafted aquifers. We can't continue to subsidize agribusiness plutocrats with the water ratepayers need for basic survival.
Water for AI also uses unsustainable water supplies, directs water away from the environment and our cities, and is subsidized for big tech plutocrats paid for by water ratepayers. It also sure seems to me that most AI use falls in the "luxury" category.
tempfile 18 minutes ago [-]
This comparison sounds to me like it is downplaying the significance of the numbers. California's almond-growing is another gross environmental crime, which is regularly controversial in California.
bigDinosaur 10 minutes ago [-]
That's because the Californian almond industry is exclusively located in California, unlike the global data centre industry.
exizt88 31 minutes ago [-]
Water consumption of AI is a non-issue. Why even spend research and policymaking resources on this?
trolleski 24 minutes ago [-]
For who?
sevenzero 26 minutes ago [-]
Misconception of the year. If you can guarantee the steam it produces rains down in areas that actually re-join the water cycle easily, you might be right (even though you still have to cover the just in time demand). You can not guarantee that though.
Best regards, European person where lakes dry out atm.
perks_12 11 minutes ago [-]
They are not drying up because of AI datacenters and especially not because of European datacenters.
While man-made climate change is undoubtedly real, the yearly "our lakes dry up"-media cycle is not proof of it at all. Check if the return to normal in November will make the news.
boxed 13 minutes ago [-]
I think you mean "re-join the FRESHWATER cycle"? Because obviously it rejoins the global water cycle.
throwfaraway135 1 hours ago [-]
I agree that water is an important resource but the wording of "consumption" always rubbed me the wrong way. Is the water contaminated with chemicals? Maybe heat? Can people or wildlife consume it? Can you use rainwater?
swiftcoder 46 minutes ago [-]
> I agree that water is an important resource but the wording of "consumption" always rubbed me the wrong way
While the Earth is (mostly[1]) a closed system when it comes to water, any particular region on Earth is not a closed system with respect to clean drinking water.
Deplete groundwater reserves faster than they refill? No more water for you. Raise the ambient temperature far enough that the rainclouds go somewhere cooler? No more water for you. As long as your usage is below the replacement rate, all seems well. Imbalance the system, and you have indeed "consumed" your local water supply.
> Is the water contaminated with chemicals? Maybe heat?
Typically yes. Definitely yes.
> Can people or wildlife consume it?
With enough water purification infrastructure, sure. I'm not aware of very many datacenters installing such infrastructure.
> Can you use rainwater?
It's all rainwater, whether directly or indirectly. The question is what would have benefited from that rainwater had it not been redirected to a datacenter?
> Imbalance the system, and you have indeed "consumed" your local water supply.
Right, but this is not what is reported by "water consumption". It would be much better to have some kind of index (probably multiple) that estimates how close a particular feature of the water system is to a problem. Is the temperature too high, is there too high a concentration of some contaminant, etc etc.
113 1 hours ago [-]
You can't drink it once it's gone through a bunch of mouldy pipes, no.
garciaperales 10 minutes ago [-]
cf
Rendered at 10:48:05 GMT+0000 (Coordinated Universal Time) with Vercel.
The real issue with AI datacenters is the huge amounts of wasted energy, after decades of energy efficiency efforts.
This is about as much as the annual consumption of the California almond farms.
(An acre-foot is approximately 1200 cubic meters.)
https://www.c-win.org/cwin-water-blog/2022/7/11/california-a...
> Almonds use roughly the same amount of water as total indoor water use in California. This is unsustainable.
> There are simply too many almonds being planted in areas with unsustainable water supplies (minimal to no groundwater, uncertain or no surface water). ...
> We’re not opposed to growing almonds; we’re opposed to growing them unsustainably and taking water away from the environment and our cities. Almonds are a luxury and export crop; approximately 70% of the state’s crop is shipped abroad for the express benefit of a handful of industrial-scale growers and hedge funds.
> Water is a public trust resource, and Californians must regain control of it as they face a future certain to be beleaguered by drought, low reservoirs and over-drafted aquifers. We can't continue to subsidize agribusiness plutocrats with the water ratepayers need for basic survival.
Water for AI also uses unsustainable water supplies, directs water away from the environment and our cities, and is subsidized for big tech plutocrats paid for by water ratepayers. It also sure seems to me that most AI use falls in the "luxury" category.
Best regards, European person where lakes dry out atm.
While man-made climate change is undoubtedly real, the yearly "our lakes dry up"-media cycle is not proof of it at all. Check if the return to normal in November will make the news.
While the Earth is (mostly[1]) a closed system when it comes to water, any particular region on Earth is not a closed system with respect to clean drinking water.
Deplete groundwater reserves faster than they refill? No more water for you. Raise the ambient temperature far enough that the rainclouds go somewhere cooler? No more water for you. As long as your usage is below the replacement rate, all seems well. Imbalance the system, and you have indeed "consumed" your local water supply.
> Is the water contaminated with chemicals? Maybe heat?
Typically yes. Definitely yes.
> Can people or wildlife consume it?
With enough water purification infrastructure, sure. I'm not aware of very many datacenters installing such infrastructure.
> Can you use rainwater?
It's all rainwater, whether directly or indirectly. The question is what would have benefited from that rainwater had it not been redirected to a datacenter?
[1]: https://en.wikipedia.org/wiki/Atmospheric_escape
Right, but this is not what is reported by "water consumption". It would be much better to have some kind of index (probably multiple) that estimates how close a particular feature of the water system is to a problem. Is the temperature too high, is there too high a concentration of some contaminant, etc etc.