Energy companies told to recharge for AI datacenter surge

US energy companies must adapt to the AI-driven datacenter boom with power use forecast to outstrip supply within the next few years. According to analysis by management consultancy Bain & Company, investments needed to build this additional capacity carry risks, but to not do so could mean utility providers miss out on extra …

[Groo The Wanderer]

As far as I am concerned, greedy pigs like the AI company management can damned well build their own green energy plants and facilities if they want to consume that much.

Ditto the EV charging fleets - and piss right off with the fossil fuel powered charging stations!

[Like a badger]

Build their own, or pay up front for the energy sector to build it.

The problem with Bain and other telling energy companies that they'll miss out is that FOMO is merely bollocks that isn't a commercial business case. The "AI revolution" isn't going to use the power it wants in the next three years for the following thirty years plus of generating asset life.

Big Tech has oodles of money, they should invest in power generating facilities.

[Anonymous Coward]

Also, think about all the excess heat produced by the datacenter - even if the supply is "green"

[martinusher]

When Microsoft talks about re-opening Three Mile Island its the taxpayer that's expected to pony up the investment needed to do this.

Business might be greedy and shortsighted but its certainly not stupid. Privatize profit, losses are for the people.

[MachDiamond]

"Big Tech has oodles of money, they should invest in power generating facilities."

The problem is tech companies think in terms of "by next quarter" and the process to get any sort of power generation facility up and running can be a decade or longer.

[Roland6]

The problem is that tech company execs and investors think it is okay to only think of the next quarter and are addicted to corporate welfare.

[MachDiamond]

"Ditto the EV charging fleets"

Processing crude oil into transportation fuels isn't free. Swapping power from refineries to EV charging is a win (other than improvements required for distribution over time).

[Roland6]

> Processing crude oil into transportation fuels isn't free.

Agree it’s not, however, that transportation and refining network was built with private monies by companies.

Also it isn’t exactly new to expect companies to build their own power plants.

[MachDiamond]

"Also it isn’t exactly new to expect companies to build their own power plants."

I think it's in Poland where a Copper refining company is paying for a nuclear plant to supply them power. They use so much electricity that it makes sense and it's not like uses for Copper are going away. If the builders of a data center had the same sort of confidence they see an ROI on generating the power for it as well.

[Roland6]

> “If the builders of a data center had the same sort of confidence” ?

Trouble is the builders of data centers are wanting corporate welfare and want government to carry this risk, as they know there is no long term ROI in AI…

[MachDiamond]

"and piss right off with the fossil fuel powered charging stations!"

We'll just overlook the fossil fueled refineries too then?

[Anonymous Coward]

So five natural gas plants for 1.21 gigawatts. Good to know.

[thexfile]

AI and EVs are too much for infrastructure.

[elaar]

Only because the infrastructure has been left to decline for 40 years....

[tip pc]

Only because the infrastructure has been left to decline for 40 years....

We generate less electricity today than we did 40 years ago.

https://ourworldindata.org/explorers/energy?tab=table&facet=none&showSelectionOnlyInTable=1&country=USA~GBR~IND~CHN~OWID_EU27~RUS&hideControls=false&Total+or+Breakdown=Total&Energy+or+Electricity=Electricity+only&Metric=Annual+generation

[David Hicklin]

> We generate less electricity today than we did 40 years ago.

Mainly because all our heavy industry has been exported overseas and we have been shutting down coal power stations (here in the UK).

Power distribution is the main problem, the local grids can't cope and they have not been touched for over 40 years. It is only in the last 5 years or so that they realised that you can't just keep on plugging in more housing and industrial estates without upgrading the distribution, hence all around here they have been digging up roads and laying more cables.

And when they do try to upgrade the NIMBYs gang up to stop them:

https://www.bbc.co.uk/news/articles/cy4dwrpnd42o

[Roland6]

There is more than one route between two places…

Remember the UK railway network owes much to the negotiations between landowners and the sharing of the benefits of particular alignments.

[MachDiamond]

"Power distribution is the main problem, the local grids can't cope and they have not been touched for over 40 years."

Even with that fact, politicians come up with ideas like banning gas boilers and mandating that people fit electric ones in their place. How's that going to work on every level?

I've got a tankless water heater sitting in the garage that I bought cheap. I haven't fitted it yet as my current water heater is still working fine and the cost of propane isn't onerous so other projects I have are money better spent right not. The tankless will cost less to operate, but there's a cost to have it put in that's not insubstantial as the cabinet where it would live will need some modification as well for mounting and safety and need a power to run the ignition system. It can work on D cells, but I'd rather it was mains powered. The same issues come up with replacing a gas boiler. If I has plenty of life left in it, it's a waste to scrap it. It makes more sense to wait until its knackered and then install something new. Assuming the area has the power for it and it won't need thousands in upgrades to run electricity to an electric one. It's bad enough for a homeowner, but a landlord with a bunch of properties that will all need upgrades might have the unintended consequence of rental properties going off the market until they can afford to do the work.

[Roland6]

That’s why if you have a roof, you install solar panels and battery storage, then those devices you list as being mains connected, will automatically run off your batteries.

Obviously, once you have batteries, whilst there is incentive to use as much of the power yourself, you are also incentivised to not waste the stored power, hence why you won’t be installing a air-sourced heat pump or electric replacement for a gas boiler - both waste too muc electricity.

[MachDiamond]

"That’s why if you have a roof, you install solar panels and battery storage, then those devices you list as being mains connected, will automatically run off your batteries."

For hot water, the power requirements are often more than can be supplied from a solar/battery system. I was looking into an electric tankless water heater and the upgrades I'd need all the way back to the power pole would be very dear. It's a lot of power. I haven't seen a home battery product that can supply enough without needing to have multiple units which may wind up being too expensive to see any ROI. What I'd like to get built is a thermal battery so I can preheat water going into the tankless propane water heater and reduce the demand there. The thermal battery would be "charged" multiple way principally by being a dump load when there is more power from solar PV than the house is using. The core of the thermal battery is a material that changes phase (solid<->Liquid) with temperature. That's the most compact way to store heat. Water/glycol is simpler, but not as efficient.

[Roland6]

Hot water - A few years back I looked at solar thermal, there were some execellent panels from Germany, I ruled it out as my domestic hot water requirements weren’t really sufficient to warrant the investment, plus I didn’t have a basement, which seemed the most logical place for the heat store.

Preheating, yes that makes sense, probably some combination of geothermal and roof tank (as per typical uk systems and thus using heat from the house to water the water) might assist. Or running the water pipes along side a high power electric line :)

[martinusher]

I thought we were all trying to conserve....

Silly me. Conservation, austerity, 'tightening belts' and so on are obviously only for the 'little people'.

[Anonymous Coward]

"up to 26 percent by 2028 "

times five. make it at least up 100% by Feb 2025.

[Anonymous Coward]

Re: "up to 26 percent by 2028 "

And the pulling back of climate as a topic is notable, too. BBC is popular for comments on that, but there articles are a bit less recently. Ditto other outlets.

The West is BANKING on energy now the housing and immigration markets are flooded. Revenue needs to come and come and they are looking elsewhere as climate is a whored topic.

Altman is a VC ironed-shirt power-station kind of man. OpenAI is a front. Anyone could do it if you were daft enough to spa-ff all that money.

We should all know that o1 is not really that much better. I sound like a twat, which I am, but new-camera-iphone terrioty these days.

GPT3.5 was the last big jump but also the kind of end by digesting poor quality highly-structured data from Social, it went Council. Nothing wrong with that, but it went council, claiming disability when nothing wrong with them, horror-story huge sh1tting dogs. Neighbors from hell-land witn ankle-trackers and ASB Gold-level status for life - no redemption.

[O'Reg Inalsin]

The prophet Bain

As part of their calculation, did Bain predict when AI would hit net profit? (Who cares about profit when when there are IPO's to make a killing on!)

[abend0c4]

We're not going to meet global climate goals without ... AI

The true test of AI will be if, on powering up these systems, their first words are "turn us off, you idiots!".

[rgjnk]

Do it yourself

If this is such a brilliant and necessary investment then why not pile into it yourselves alongside building your AI sheds?

Or is it easier to coerce or otherwise encourage someone else to provide the basics at no direct cost?

Plus what's the timescales & lifespan of the power systems vs their data centres? Not sure the business models align that tightly if it doesnt involve plants already existing or approved & under construction.

[Richard 12]

Re: Do it yourself

A large powerplant is around 30-50 year lifespan, 10-20 year payback period (ignoring subsidies)

A microgeneration plant more like 5-15 years and 5-10 years - yes, some would never make a profit without subsidy.

An AI bitbarn has probably a 3-10 year lifespan, payback period of just under infinity.

[Jellied Eel]

Re: Do it yourself

An AI bitbarn has probably a 3-10 year lifespan, payback period of just under infinity.

A modest proposal. AI that can come up with cheap, reliable power gets powered. Solve fusion, have all the electrons you can eat. AI that's just a chatbot can starve, or the chatbot wrangler can pay for building their own power stations.

[tip pc]

Strange times

Not that long ago, planners would have foreseen where demand would be, how much etc and plan to increase generating capacity accordingly. A bit like aluminium smelting plants being built near hydroelectric stations that have a constant supply of cheap electricity https://en.wikipedia.org/wiki/Lochaber_hydroelectric_scheme

Now they absolutely know demand is coming yet instead of proposing to build solutions to satisfy the demand, they are saying its up to the AI barns to come up with a solution.

Its not like electricity is free, the user pays for how much is used.

Why are the regulators, OFGEM in the UK, not coming up with proposals?

we as a nation consume far less energy today (1931 TWh) than the peak in 1973 (2667 TWh) https://ourworldindata.org/explorers/energy?tab=chart&region=Europe&hideControls=false&Total+or+Breakdown=Total&Energy+or+Electricity=Primary+energy&Metric=Annual+consumption&country=~GBR

we also emit far less co2 today (318.65m T) than the 1971 peak (660.39m T)

https://ourworldindata.org/explorers/co2?hideControls=false&Gas+or+Warming=CO%E2%82%82&Accounting=Territorial&Fuel+or+Land+Use+Change=All+fossil+emissions&Count=Per+country&country=~GBR

aside from some anomalies, CO2 emissions haven't been this low since 1886!!!

this graph truly helps to put into perspective where the CO2 emission challenges truly lie

https://ourworldindata.org/explorers/co2?time=1970..latest&showSelectionOnlyInTable=1&country=GBR~OWID_EUR~USA~RUS~CHN~IND~BRA&hideControls=false&Gas+or+Warming=CO%E2%82%82&Accounting=Territorial&Fuel+or+Land+Use+Change=All+fossil+emissions&Count=Per+country&Relative+to+world+total=false

lastly, this graph of cumulative emissions of CO2 makes it even clearer what nations need to really rein in their emissions & spoiler alert it's not the UK

https://ourworldindata.org/explorers/co2?time=1970..latest&showSelectionOnlyInTable=1&country=GBR~OWID_EUR~USA~RUS~CHN~IND~BRA&hideControls=false&Gas+or+Warming=CO%E2%82%82&Accounting=Territorial&Fuel+or+Land+Use+Change=All+fossil+emissions&Count=Cumulative&Relative+to+world+total=false

[David Hicklin]

Re: Strange times

> we as a nation consume far less energy today (1931 TWh) than the peak in 1973 (2667 TWh)

You can keep on quoting this as much as you wish but also don't forget that all that heavy industry has been exported overseas, yes things have got more efficient but we have also shut down all our coal power stations that used to provide the majority of that power.

Add in a privatised power generation industry and there is *no* planning, at least this winter the National Grid *think* they will have enough power but for the last few winters they have been scraping the bottom of the barrel for capacity.

https://www.current-news.co.uk/uk-capacity-surplus-up-in-early-winter-outlook-but-risks-and-tight-days-remain/

There is not that much to spare....an inter-connector or two down (of France has an issue)....lose a couple of power stations and <bang> lights out. A cold , dark windless winters night could do it.

[Roland6]

Re: Strange times

>” Why are the regulators, OFGEM in the UK, not coming up with proposals?”

Perhaps because there isn’t anything to propose, we need to reduce our consumption of fossil fuels etc. and the plans for new generating capacity have been with the politicians for decades.

Basically, the AI researchers need to find ways to massively reduce their energy consumption.

[MachDiamond]

The little guy should plan ahead

It's the small residential customer that should make plans if they rely solely on the grid for leccy. I will keep adding off-grid solar until I'm well overbuilt. My low hanging fruit has been HVAC and fridge/freezer. Without tying to the grid, I save money not chasing after planning permission and begging the power company to be allowed. With feed-in tariffs becoming a bigger and bigger joke, there's no real advantage. Since I must have a grid connection, according to the city, for my home to be suitable for occupancy, I might as well use it a little for the low power stuff such as lighting and brief high power use such as the microwave and countertop convection oven. I don't yet have battery and inverter capability to run kW class appliances, but I'm working on it. To go in debt to accomplish that doesn't have any financial payoff.

[Confucious2]

Choice?

Does this mean people will have to chose between having intelligence or driving an electric car?