Article makes a huge deal about humming/buzzing but there’s no way that’s a real thing right? What would be the reason it would go beyond like any HVAC-ish thing on a large building? Lost me at the first sentence :(
It's was super easy to measure using a Sensirion Differential Pressure sensor. They have a few extremely sensitive models that have a 100Hz sampling rate (50Hz useable after Nyquist) - also: you can sense washing machines (from vibrating buildings) with it 2 blocks away.
The product listing says something about a manifold, it looks like a raw component with 4 pins or so.
SDP600-25Pa --- I2C ---> ESP32 --- UDP over Wifi --> Linux Host with simple Python recording script
Very rough pin-pointing is possible: plastic kitchen funnel which is attached to the sensor using a short hose.
AI datacenters have much larger cooling needs and often use generators
Ahem.
https://www.politico.com/news/2025/05/06/elon-musk-xai-memph...
And you'd need an insane amount of solar panels to actually recharge those batteries in any kind of reasonable time too, so you expose yourself to a massive risk if you had two out of power events within say 12 hours. So you'd probably build all of those batteries and solar panels but you'd still need to have emergency generators ready to go anyway.
Critical loads require generators, batteries don’t cut it. Data centers want the most reliable backup power they can with the longest runtime. Battery storage density is not high enough to back up a 500MW+ data center for any length of time without a comical amount of batteries.
When the NEC allows critical, equipment, and life safety branch at hospitals to be backed up with batteries and solar panels, battery storage will be at a point where battery backup of data centers is feasible. Right now it isn’t.
It goes way beyond normal building HVAC levels, AI has pushed many DC's from 8 kW to 17 kW a rack in just a few years.
IIRC the average medium office building pulls about 22 kilowatt-hours (kWh) per square foot of floorspace per year. So their cooling needs are tiny.
17 kW per rack for AI/ML HPC was workable >5 years ago. If you're not budgeting for double or triple that number nowadays you're not capacity planning properly.
It'd be nice to have some hard data on it though. Sometimes the hum is "god damn, that is annoying!" and other times it's someone saying "20 miles that way they built a DC and now it makes the cell phone tower effects twice as bad in this area" when reality was it doesn't show up as audible a half mile away.
For example,
I'm fairly surprised that there aren't zoning laws that prevents datacenters from being built where people live. When we built a fairly small datacenter we had to place it in an industrial area with no housing and no offices.
If that's surprising, you should look into how infamous the city of Houston's lack of zoning laws has made it.
Another fact is just the sheer power density of those problematic north virginia datacenters. I'd bet us-east-1 is not an issue (old building and lower power density), but the newer AI ones are. Just take a look at how much AI clusters eat: a single DGX H200 box with 8 GPUs is 10kW. Most facilities provide 10kW for a whole rack, not 8Us. You're looking at 60kW racks, which is a mental power density: a single aisle trivially gets over the MW threshold. You used to feed rooms with megawatts. Heck, DC5 has 24MW of power, that's only 20000 H200 (again, think hyperscaler scale).
Still about cooling, the load profile is even different. DC5 is a general purpose datacenter, where the load is not full blast. Your AI datacenter has the GPU clusters full blast all the time. That's a LOT of power.
I happen to know pretty well the Scaleway infra and visited others of their datacenters. You can stand centimeters from the noise-dampening wall surrounding the dry coolers and not hear a thing; while almost needing noise protection within the wall.
There's going to be variation based on how the data center is constructed.
The ones I've seen (which is certainly not all of them, so take it with a grain of salt) in the geographic area that this particular article is about are closed buildings that should be pretty well noise regulated individually, but that's not always the case -- some companies (like Marathon/MARA) have data centers using what are effectively open air designs that don't contain sound very well at all and can easily generate significant noise pollution around them.
And some companies (eg. xAI/grok) are a bit of a hybrid where the main part of the data center is enclosed but they park many loud methane turbines outside of the building for power and those make a lot of noise (and air pollutants).
The basic TL;DR is that it is possible to build data centers that aren't an absolute nuisance to the population around them, but there are plenty of data centers out there that don't meet that goal. And the more these companies try to scale up quickly to meet their perceived "AI" needs, the worse things are getting in terms of noise pollution, air pollution and competing for resources (like water and energy) with local residents, etc.