Concerns about large data centres in Scotland near homes

[Sheldonsheher]

Why is there not more outcry about data centres in Scotland. As far as I can see it’s a resource grab by tech giant companies to cream off Scotland’s natural resources. They won’t provide much benefit and will decimate local communities. For example hermiston gait right next to a student campus and homes. They will drain Scotland’s resources. Make a mockery of net zero and hose pipe bans. Yes to small data centres away from homes no to mega data centres next to people’s homes.

AI is yet to show any real profit.

As soon as investment in it collapses there will be no funds to build hyperscale data centres.

It's just not possible to realise ROI on AI, and as soon as this becomes inescapable there will be a collapse.

It would be a huge mistake to invest public money time or resources in something you know is going to fail.

We may well be in an AI asset bubble.

The idea that we therefore "know AI is going to fail" is absurdly overconfident.

AI providers see massive revenue growth. The reason they are not profitable is because they are investing massive amounts in expanding capacity.

www.exponentialview.co/p/is-ai-a-bubble

As for "no way that it can generate ROI":

For many tasks, AI is already, demonstrably, much quicker and cheaper than a comparable human for a similar task. That's the absolute base case for ROI, assuming no improvements in AI technology.

In fact, model capabilities have continued to grow rapidly, to the point where they are now, to take just one example, happily solving mathematical conjectures that the world's top mathematicians puzzled over for decades or centuries.

Let's just take that one, very narrow, skill. If you don't see how mathematical ability can be economically valuable then you've not been paying attention for the past four decades, where finance and then tech have been built on the back of that, into two of the largest and most lucrative sectors for humans to work in.

The more interesting question is what happens to our economy and our society when that lucrative work, and its returns, go not to labour but to capital.

If we want a functioning society, I'd rather our policymakers worry about the future of work, the future of meaning, and the future of our tax base.

The issue is that deciding we won't have data centres does not solve this for us. It simply puts us out of the room when these things are being shaped.

And you really think AI centres will pay appropriate tax? Look at all the Tech companies avoiding tax right now, those based in Ireland for example.
And if these data centres are so important to our society you can bet the private companies who own them will demand low taxes or else. Just as you say Trump would do to us.

My O/H will soon lose their job to AI. It uses him and his colleagues to train on. When I see what AI produces it reduces choice, flexibility and nuance to the client. It's actually worse than that but I'm assuming it will improve before launch. The problem is that when the client uses it in the future they most likely won't be aware this has happened which will suit companies well.

It's like the chap said about AI models being used - the models, the hairdressers, make up artists, hire of the venue, technicians etc etc all lose. Except the few who control the AI.

They could use the big park across the road from the parliament...

After all, if it's okay to build one next to Auchtertool, then it must be okay to build one at Holyrood.

And you are not posting any links to the dangers and how they destroy people's health.

I get it re energy use and that it's potentially a bubble. But you are claiming health risks with no data to back it up.

The reason some tech companies are hard to tax is that they don't have many physical assets.

Data centres are a physical asset. In a world where computing power is abundant then switching the work to other data centres elsewhere is easy, making data centres hard to tax again. Right now, and for the forseeable future, computing power is extremely constrained.

It's not that the data centres are important to our society. It is that AI is likely to become extremely important to our society.

Scenario 1: they've got something we really need (AI). We have got something they need (physical data infrastructure, plus a serious AI talent pool).

Scenario 2: they've got something we really need (AI). We don't have anything they need.

In which scenario do we come off worst, against these firms you think act only in their self-interest? (that's not the bit I'm contesting).

AI may well destroy many many jobs, and with it the tax base.

Scenario 3: the jobs are gone (either because our firms replaced them with AI, or because they got out-competed by other firms that did) and with it the tax revenue. The economic growth has shifted to AI firms, a decent chunk of which are based in the UK where we get to benefit from, and redistribute, that growth.

Scenario 4: the jobs are gone (either because our firms replaced them with AI, or because they got out-competed by other firms that did) and with it the tax revenue. The economic growth has shifted to AI firms, who owe tax only to Taipei, Texas and California.

Which works out better for everyone who will need public services in the decades ahead?

I'm not making light of the issues above. I think they are really serious and important, and the future may be frightening. I think worrying that data centres turning the water brown is a trivial distraction from the issues that matter.

But none of that addresses the basic economic problem.

AI is creating fewer jobs. There are therefore fewer people able to buy tokens. Companies can still buy tokens, but their revenue will drop if there is no job creation, and they won't be able to afford the more expensive AI option, no matter how many times better it is.
So how will it make money?

All the money it has is speculative. Last year only 0.2% of US stock market growth could not be traced back to AI. The market is hugely overheated: the stock price to dividend ratio is now 44, so it costs 44 times the income of the share. Last 2 times it was this high were before the dotcom and 1929 crashes. After 1929, shares traded at 15 times income.

So last time this happened, shares became worth less than 40% of what was paid for them relative to income. If all this value is wiped off AI, it will not be able to continue to build data centres, nor raise capital to do so as it is now. So any money invested in them will be wasted.

It isn't a case of speculating that AI will fail, it is in the process of doing so and only the overheated market is prolonging the bubble.

Firstly, I'll hedge by saying I'm highly unconfident in my own, or anyone's, ability to foresee how this turns out in the long term (say, twenty or fifty years).

On creating fewer jobs:

Earlier automating technologies didn't grind to a halt, or stop economic growth, because they put everyone out of a job and thus noone could afford them.

If you think AI will, you need a story for why it may be radically different this time.

I think AI might be radically different, though I doubt it will be so in a way that makes it all peter out.

On AI valuations: I think AI is possibly in a local bubble (as in, we may be going wrong in predicting which firms or assets will hold long term value, and/or there will be a sharp correction along the way).

My personal best guess is that long-term, AI is not a bubble at all, and may well be massively undervalued.

People are very bad at predicting exponentials (think of Covid), but AI is becoming more capable exponentially. It is also being adopted more, although overall, AI usage remains low compared to where it could be (think of, say, how many people use the internet now, compared to in the early years).

Unclear what a 40% crash would do. Certainly not pretty, though a lot of AI funding is via private credit (which implies its own problems).

But if the technology is here to stay, data centres will be needed.

I think ultimately they will migrate to wherever can give them a good balance of cheap energy + stable politics. I'd like that to be the UK.

That means lots more nuclear (and other) power. Luckily, investing massively in energy won't be wasted regardless, because we are already so badly lacking.

If we can build the network effects then we may get to keep more of the industry than we 'deserve' on those grounds. Currently we do have quite a lot of AI (probably the biggest frontier AI talent pool outside of California), so it's a matter of keeping it and growing it.

There is no contradiction at all between doing that and regulating/governing carefully. Indeed, having a decent AI sector is the only way we can effectively govern it.

I've only recently become aware of the impact of these data centres and it's insane. People must find out the implications and do what they can to stop them. It seems that Ai trumps the needs of humans and the environment.

I'm afraid it's the wrong AI thing to worry about (imo).

If you want nightmare scenarios look at cyber and bioterrorism risks, potential labour impacts, geopolitics and loss-of-control.

Sadly, none of these can be addressed by the UK building fewer data centres. And a few of the bad risks are made actively much worse by us not having a strong domestic AI sector under our own control and a place at the grown-ups table when the rules are being set (we won't be at the top table, but we do find ourselves, perhaps surprisingly, at the next table down, and giving that up would be a very risky thing to do).

It’s possible to worry about a range of actual and potential consequences of AI, including living with the infrastructure. It simply isn’t correct to describe the impact of living near very large data centres as ‘minor’.

There is an editorial about the impact of noise on health in the Guardian today. The fact people live near motorways and under flight paths now is not proof that adding more noise into to mix is ok. Far from it.

Asking if it’s worth it is a legitimate question - although of course we never get a chance to ask it.

@GeneralPeter I think its interesting that you say that worrying if data centres turn water brown is trivial.
We're in the middle of a heatwave, wildfires raging across the UK and Europe. We're being warned about climate change and lack of water.
We're also paying extra for energy and green levy taxes. High taxes on new diesel cars. Costly electric cars . Told to reduce energy consumption possible blackouts blah blah blah
Carrots and sticks to the public all the time in a cost of living crisis.
Then the next thing we're told we have to have huge energy and water consuming data centres all over our food producing prime farmland .
Meanwhile a very rich man selling EVs gets even richer with AI along with his buddies.

There's a log of contradiction coming from the " Elites" in the telling us what we do and don't need and apparently we're supposed to just acquiesce and not worry about a little brown water.

This thread is about data centres though. If you want to talk about other aspects of AI, start a new thread.

EDIT: having read your other updates to this thread you sound like you might just be here to stoke division.

My point on data centres is that: i) they are important economically and strategically, and ii) the risks are low in comparison to pretty much any other industrial use of land, and iii) in relation to worries about AI, it's barking up the wrong tree, iv) and if you want to protect the UK from some of those very serious other risks, you need data centres.

On the final point, if I start a thread about risks of the MMR jab, is you mentioning the risk of measles off-topic?

On your update: genuinely, I'd like to understand why you think that.

I'm not trying to stir any kind of division, other than in the normal sense that if someone has a view on an MN board and we disagree with it, the board is the place to say that.

If I have an ulterior motive, I guess it's to raise awareness of some of the wider societal risks that this technology presents. That doesn't seem to me like stirring division. I do, genuinely, think that those countries that eschew an AI industry (e.g. by blocking data centres) will find themselves exposed to almost all the important risks of AI that they would have had, but with none of the leverage they need to manage them. I like this country, and I think that would be a shame.

Obviously I might be barking up the wrong tree, like people who sounded the alarm about climate change in the early days might have been. You can take or leave what you like from my views. But 'no discord on the discussion boards please' is a weird stance to me.

Scotland has a huge amount of wilderness, I'm really not concerned. So many places to build data centres without encroaching on built up areas. The Non-Domestic Rates income, employment etc will boost the economy. If Scotland want independence being a world leader in AI infrastructure will be a good thing (disclaimer - I'm not for independence).

I would rather we all focused our energy on cleaning up our seas and beaches, vastly reducing our use of plastic especially in the fast fashion sector which is endless amounts of cheap clothes made with acrylic sending microplastics into the water system with every laundry load.

They are not proposing to build them away from built up areas though due to grid connections etc.

I find the arguments about the one at Eurocentral really hypocritical, most of people complaining stay in the newer estates in Chapelhall and Carnbroe, they don't want to have to look at it thats all and they are jumping on the bandwagon. 20 years ago there was nothing but fields where these estates are, so now they know how all the villagers feel about their houses, now they can't get a nursery place or an appointment with the doctor, but god forbid these poor folk need to look at and share there energy supply with a small data centre, they won't be able to put all their roofline lights on all night or have their hot tubs running all year round. They weren't complaining when they were planning on putting a petrol station and drive thru there though. They wouldn't have to drive to far to fill their massive 4×4.
No one even batted an eyelid at the one thats been operational there for a couple of years, no ones complained about dirty water or a humming sound. But now its fashionable to protest against them here we are!

You sound very bitter that will serve them right for wanting to live in a house! Never mind kids because your parents live in a new build you shouldn’t not mind a large massive ( note not small) humming data centre blighting your childhood and turning the water brown. The people who lived here before you think you deserve it:

You still have not posted anything about this humming or brown water.

I am not saying you are wrong here, just that it is usual to post a data link when claims such as this are made.

Putting aside all the AI bubble stuff, and I agree, it is a potential bubble, some of these points you make do not make sense.

Quote :" Any tax goes to UK government. They have already had £420 billion of oil revenue and have spent every penny and more, so that the UK is in vast amounts of debt. It is exceptionally foolish to think they wouldn't do the same with any windfall from data centres."

Go to near any political thread here and it will be loaded with posts saying government must increase revenue without taxing workers. But here you seem to be saying Government waste tax money so what is the point of them collecting it ? And also, of course government spend tax income. That is why they collect tax. And there would be a bit of an outcry if they collected tax and did not spend it. Here the tax would come from the companies buying electricity.

Quote :" The likely result is that tech companies will get massive government grants, start building data centres everywhere, pollute groundwater and disrupt electrical supply."

Yes. Grants are usually connected to investments. Look at any party manifesto to see their plans to attract investment. How much the grants will be, no idea.

However, quickly looking up the projects, what I am seeing is that these companies are interested in Scotland because it has excess green energy at times. And after so much investment in green energy, the way to get costs back and get costs down etc, is to sell the electricity, and these people want to buy it for their data centers.

If I change "data centers" to "factories", does that make it sound different ? That is what they are after all. And it appears these companies want to set up in Scotland because it has the recourses that they need.

Or it all might be vapourware. That is possible. But I have not seen that mentioned here.

@Sheldonsheher

As you have not posted about where the brown water comes from, I had a look and found this :

4 Ways To Eliminate Data Center Water Pollution (datacenterknowledge.com)

Maybe not a great site for reference use, but it has enough detail. Here is how it affects water :

"Data centers can inadvertently pollute water through chemical runoff from evaporative cooling systems, including biocides, corrosion inhibitors, and heavy metals that accumulate at scale when facilities discharge up to 5 million gallons daily."

Potential pollutants are :

"Biocide
The most common pollutant in data center water systems is biocide – chemicals designed to prevent the growth of bacteria, mold, and algae. Evaporative cooling systems create warm, moist environments that are ideal for the organic buildup. To combat this, data center operators often use biocides such as chlorine dioxide or bromine to maintain system cleanliness and efficiency."

"Corrosion Inhibitors
Another category involves chemicals that prevent corrosion within cooling systems. These inhibitors, such as various types of phosphates, are crucial for maintaining the integrity of the system's components and ensuring its long-term functionality."

"Heavy Metals
Heavy metals, such as copper, zinc and occasionally lead, represent a third pollutant category. Unlike biocides and corrosion inhibitors, heavy metals are not intentionally added to cooling systems. Instead, they originate from the materials used to construct the systems themselves. Over time, metal components can degrade and release trace amounts of these substances into the water."

So it is basically cooling towers. And they add to water use as well, because cooling towers evaporate water.

But as a PP said, Scotland does not, or has not, used cooling towers in thermal power stations. They use river water, or sea water. And a data center is basically the same as a thermal power plant in that it has lots of heat to dispose of. Water goes in cold, come out warm, or hot. No chemicals because it feeds through.

So these data centers, if they do use evaporative cooling, will be no different to existing cooling systems already in use. And they are regulated by law. Laws, by the way, that the USA is rather lax on.

Are there any complaints or campaigns against existing cooling towers ?

https://aprs.scot/stop-the-hyperscale-ai-data-centres/

Have you considered selling tinfoil hats to those who live near data centres OP?

Thank you for the petition link, I have signed it!

Can't believe the number of people on this thread who think its a good idea to take an unquantifiable risk near your home so that a tech company can make money and avoid tax. Utter madness.

It's why I'm glad a moratorium is being voted on. Once the Scottish Parliament reconvenes, a ban will give time for evidence on safety to accumulate, and for the AI bubble to burst. They will stop building these sites once investment stops.