And I’m arguing that the main mitigation strategy should be opposing it at every opportunity. Don’t start the compromise by giving in. The Overton Window on data centers needs to be dragged heavily to the “these a really bad” side.
1 Like
Like micro said if you don’t build it they won’t come
My bad. I misunderstood your post. Ignore withdrawn. 
1 Like
That’s the core of our difference. If proper regulations were drafted and they were enforced on data centre developers then I wouldn’t oppose them for the reasons I said earlier. I’m not totally opposed to AI in general.
If I was ruler of the world there would be a global pause on all hyperscale campuses over a couple hundred megawatts until regulations can be drafted and the science can be done into their actual impacts. Maybe like 5 years.
Those regulations would include use of green energy, water recycling, noise and heat mitigations like immersion cooling, legal arrangements where local authorities can create special tax classes for data centres and required security bonds for reclamation.
This is what I’m getting at Generative AI Is an Engineering Disaster - The Atlantic
"“But with the massive investment behind the current bloated approach, there may not be much will to change. Ilya Sutskever, a co-founder and former chief scientist at OpenAI, said in a November interview that companies take the brute-force approach “because it gives you a very low-risk way of investing your resources.” It’s harder, he argued, to invest in research that would reengineer a product currently accruing trillion-dollar valuations. Those who suspect we are in an AI-driven bubble economy have pointed out that the profitability of these companies remains an open question, largely because of the high cost and inefficiency of the technology.”
2 Likes
AI will be harder/impossible to stop. Very few organizations had to be regulated for the other bans and the regulated groups weren’t denied massive amounts of money. Still plenty of money in unleaded gas. Data centers are already being built in poor countries that won’t be able to resist the money.
Should it be slowed down? Yeah, I think so. But
2 Likes
It seems contradictory that you’re saying that being wholly anti data center isn’t realistic because they’re going to happen anyway, yet also say this:
None of this is going to happen either.
1 Like
That’s just not true. Those things are happening in many jurisdictions.
Several U.S. jurisdictions have recently introduced data centre specific regulations. California and Minnesota are tightening oversight of water use, Ohio and Utah require large data centres to pay for grid upgrades rather than passing costs to ratepayers, Virginia has strengthened utility and planning rules to manage rapid growth, and New York temporarily paused approval of very large data centres while developing new standards.
I guess time will tell if these are sufficient.
If they are, it will be in no small part because there were a ton of people like me who were like “fuck this shit, no more of this ever” and got to effective mitigation plans as a compromise.
1 Like
I’ll take a run at this. From an energy perspective
All data centre energy use to be completely covered by new green energy. This means
- Solar and battery (or wind) on-site to match full demand. Maximum 5% pull from the grid in any given week. OR
- Incremental renewable energy contracts/PPAs. 100% match if time of use can be tracked in the jurisdiction, or 150% if not
- Incrementality needs to be proved. Something like PPA contract needs to be signed before financial close on the renewables and storage project. (I’d accept non incremental PPAs for say year 1 and 2)
- Grid studies to be complete showing full incremental cost of connection. Data centre to pay for all required grid upgrades including transmission.
- All DCs to have smart demand management to a) respond to market price signals and b) be controllable by grid operators to support grid stability and renewables expansion
The challenge being that these are the types of regulation that are inherently difficult to implement without corporate capture. They show all the traits.
- They are very complex and the detail of the regulations matters a lot.
- A small number of rich actors get the benefit from DCs, while a large number of people get the downsides, spread out in complex ways over space and time.
This is one of those weird situations where the “mature and reasonable” approach SEEMS like it should be all about sensible well thought through regulation.
But the reality, as is often the case, is that that middle of the road is actually hopelessly politically naive.
So the well informed and thought out response to DCs actually needs to embrace the burn it all down extremism, because only simple ideas will cut through and address a problem like this.
4 Likes
This.
1 Like
I wish I could elaborate but this is very close to what I’m proposing. When I’m done this I’ll share the final results when it’s public.
The issue with green energy is it’s not continuous as you know so we are proposing centres need to do a mix of BESS and contracting offsets.
Ideally those that do behind-the-fence power should be required to over generate and put extra power onto the grid.
Did you miss the second part of my post? Lol
Are you proposing that contracts should be incremental?
The issue I see is data centres come in. Buy up existing renewable offsets that only cover the day, while driving massive total energy uplift with obvious increase in system carbon all while stressing the fuck out of the grid and making renewables expansion difficult.
Then claiming “oh we are renewable” based on generation that existed before they came in.
Do data centers need to use fresh water or can it be any water and why?
They don’t need fresh water but if they use waste water from another source they need onsite treatment because impurities can degrade the cooling system.
The water issue really is becoming less and less a problem as it’s possible now to build pretty huge hyperscale campuses that use very little water. The water issue was largely a combination of older technology and the insanity of building these in hot climates like Arizona. That’s partly why cold climates like Canada are attractive.
Ok you can have them
Their waste heat could be helpful to people in cold climates until AI murders everyone.
1 Like
Yes. I already pointed about why I don’t agree with the second part so didn’t need to belabour the point.
It’s no more naive to regulate data centres than it is to regulate anything. Unless your position is all regulation is a waste of time.
There are examples of using the waste heat for things like adjacent greenhouses.