Comment by BearOso
1 hour ago
The disparity is that older data centers were for storing and serving data. They have a lot of bandwidth and storage, but aren't consuming much energy. Plus, they are there to provide quicker access to data, to provide a service to actual people.
New AI data centers are obviously compute-oriented instead. That means they consume a lot more power, like hundreds of megawatts. For perspective, on average a U.S. household uses only 1.2KW. Because of this, cooling needs are significantly increased, and while water usage is often overblown, noise volume is a major problem. At 100m, it often exceeds the 90dB threshold for hearing loss.
Then there's the fact that millions of dollars of GPUs are inside these things, so security means lots of bright lights at all times of day.
These are just problems with the physical presence of a data center, not to mention the electric bill hikes, land value loss, environmental effects, etc. And there's no obvious positives to one going up. No new jobs. No tax money because it's often elided with incentives from bribing local officials.
With all the negatives and seemingly no benefits, ask yourself who wants these data centers, and why? If the only way to improve AI is using more and more compute like this, what is the end-game?
[delayed]
> At 100m, it often exceeds the 90dB threshold for hearing loss
Are you serious? Is there any data for this? This would make the whole area unlivable
do you have data on no new jobs/tax revenue?
I’m under the impression that most data centers have very few staff on site, almost zero.
Sure, some temporary jobs during construction, but I bet it’s nothing like a retail store or factory in operation…
>For perspective, on average a U.S. household uses only 1.2KW.
This is always kind of funny to me. How's an average US house supposed to provide context for industrial compute?
How much energy does a car factory use? What about a chip fab? A hospital? A battery manufacturer?
What about an apartment of the same physical dimensions as the proposed "data" center?
Any of those measurements would provide meaningful context that would allow the audience to decide what they'd like built in that space if energy consumption is a concern of theirs.
Comparing a multi-million dollar server farm almost deliberately designed to turn electrons into dollars to a US house usually designed to require as little electricity as possible provides...well I can't decide if it's zero or next to zero context. Like, sure, that power powers say thousands of homes, but my own home consumes more energy than is really required to keep me alive. If the concern is efficiency, I'd be just as inclined to abolish new freestanding homes in favor of efficient apartment complexes as I would be to disfavor a data center over a neighborhood. But those up in arms about data center power consumption are not also rallying for fewer lawns, so the argument can't be purely about power consumption.
>With all the negatives and seemingly no benefits, ask yourself who wants these data centers, and why?
What do you mean no benefits? Do you think local governments are signing these contracts and listening to their constituents yelling at them for the fun of it? The company building it profits from compute contracts. The device mfgs profit from purchasing agreements. The city with the land profits via both legitimate and alleged payments from the corp building the center. To suggest that there are seemingly no benefits just because you aren't receiving the benefits is naive. No one's doing all this for free.
>If the only way to improve AI is using more and more compute like this, what is the end-game?
You could have and probably could still say the same about the internet in general. We've been exponentially building out storage, delivery, and compute for decades. If the only way to improve the web is more and more servers, what is the end game?
You could say the same about just about any growth industry in history. If the only way to reduce the price of bananas is more banana farms, what is the end game?
And the answer is the same in every case. There is not infinite demand for AI and the accompanying infra just as their is not an infinite demand for bananas. If their is not an infinite demand, there is no call for infinite supply increase.
Demand for AI will peak, plateau, and crash, over time just as the tulips, beanie babies, pokemon cards, bitcoin, and everything else does. If you're not in to tulips or pokemon, the craze seems stupid. Because the craze seems stupid, you do not feed it. So long as there are people not feeding it (in order to buy bananas or beanie babies instead of AI subscriptions), its growth and presence will be modulated just as any other thing available for purchase.