This $23B number that gets thrown around is not the increase to the public. The wording in the referenced report is
> Based on actual auction clearing prices and quantities and uplift MW, inclusion of existing and forecast data center load growth resulted in a combined total increase in capacity market revenue for the 2025/2026 BRA, the 2026/2027 BRA, and the 2027/2028 BRA of $23,100,955,341.
This is the increase in revenue to PJM from adding datacenter customers, and includes both the amount that datacenters paid as well as the amount that other customers paid due to higher prices from datacenters. So Fortune calling it an increase to "the public" means that they didn't read the report they are using as their source and are probably just repeating what they thought someone else meant.
Bloomberg in the past worded it as "data centers will add at least $23 billion to customer bills" in April and "added a minimum of $23 billion to customer bills" in February. Which while technically correct (datacenters are customers) seems meant to be misleading. And now that's the number that's getting thrown around as the increase to "the public".
The part I don't get is that the journalists could just give the actual number for the quantity that they are referring to (the amount that non-datacenters paid due to higher rates due to datacenter loads): when I calculated it a few months ago I think it was something like $16 billion rather than $23 billion. I feel like the story would have the same impact if the headline number was $16B as $23B, but $16B has the benefit of not being a misrepresentation of the situation.
---
Also I would definitely recommend checking out the PJM BRA report. It's a bit dense but not too hard to follow, and my personal takeaway was that the PJM market is just very dysfunctional and they are blaming the datacenters instead. I thought SemiAnalysis had a good analysis of it: https://newsletter.semianalysis.com/p/are-ai-datacenters-inc...
Or how about this framing (using your own numbers): for every dollar a datacenter is charged for electricity, regular hard-working Americans get charged $2.
Well, I think some of those hardworking Americans would be happy to pay some of those $2 because otherwise all the SaaS services and other services, platforms, and tools running in datacenters wouldn’t be accessible to them.
Or did I read this wrong and somewhere it said only datacenters running inference or training for LLMs?
The thing people are pissed about is giving better rates to datacenters.
Compare this to most of Europe (and Texas if I understood correctly) where the detacenters buy their electricity from the same market as everyone else (in Europe the spot market or futures) meaning they effectively pay the same price as everyone else.
It’s when they do some back room deal with the local public utility to get 50% off and offload the real costs to the public when people get angry.
Is there an example of this happening? Isn't the problem that datacenters are drawing electricity at the market rate and driving the cost up, rather than paying a surcharge for the difference?
I don't think the mechanism is that different, and even if it was Europe, etc would have the same issue - data centers would still bid up the price for existing capacity, and the future capacity added to accommodate the dcs would be added at a higher marginal rate too because scarcity of new supply + base load nature of dc loads. So the problem will get worse before it gets better
Concrete examples? Quick google only shows small discounts for DCs relative to other wholesale industrial consumers, which could be partially explained away by the flat load, lack of stability concerns, no need for power factor correction hardware, etc. I think 'bidding up the market' is the dominant mechanism, here in Australia (spot market + futures) consumer prices are projected to increase 26%.
> But what if the power company needs to upgrade the substation to handle the increased needs of the data center? Or secure additional sources of electricity? In these cases, the investments are part of the electricity grid that everyone uses. These costs will likely be shared among all customers.
Okay but this is a policy choice. It doesn’t have to be that way.
Where I live, if a developer wants to build a subdivision, they pay for the water and sewer lines. They pay to connect those lines to the city infrastructure. If the city infrastructure needs to be upgraded to handle the new volumes, they pay for at least a proportionate share of that too. The ongoing maintenance becomes part of the city's budget eventually, but not the costs of the build out.
All those costs go into the price of the houses built there.
And this is also part of why building "affordable" houses rarely happens. All the infrastructure costs the same whether the houses cost $100K or $1 million.
That's smart. Do they do it with roads also? That's a big one near me - developers buying hundred acre farms on unpainted 2 lane country roads and jamming in 2000 houses. Then inevitably the road becomes unusable until the city or county gets around to addressing it.
Always wondered why the county didn't require the road work, or money for it, up front.
That's negligence on the part of the county/township. Around here, every new development is required to pay for the traffic and utilities improvements that will be required once the thing is built. (And all the engineering that goes into figuring out the impact in the first place.)
In my hometown in Idaho in the 1990s and 2000s yes, this includes access roads and improvements to the surrounding area. A car dealership and Wal Mart both paid for road improvements and traffic lights as part of development.
The developer gets to at least name the roads. I worked for a city, the developers would submit names and while the city had the final call 99% of the time they would just accept whatever the developer submitted. The only times that I saw it get denied were times where it would create confusion like a similarly named road already existed or it could be construed as something profane
The “affordable” housing thing seems like such a misdirection. I can’t help but daydream that some moneyed interest somewhere fans those flames, as it looks like a dead end that can absorb endless fervor.
You know what you do if you want an affordable car? You buy used. I think most people understand Ford is never going to build another a car that costs $10k brand new, and the last new car near that price barely sold because it was so stripped.
When everything is built huge you don’t end up with homes that are cheap to heat, cool, or maintain. That’s why building affordable houses is actually a real issue.
What’s really dumb is the average number of people living in a house has tanked over time but the median new home just keeps getting larger. In an efficient market you’d expect new homes to match what buyers want, but regulatory capture has severely distorted the market.
What makes me crazy is these big houses aren't even that nice. They don't have significantly more rooms. It's all just stretched out to maximize sq ft for minimal costs. Sometimes with so many cut corners they're unsafe.
So if you have a big family or multigenerational needs then often you have to do significant remodeling or sell anyway.
If a gornment changed policy to make houses depreciate like cars, it would not be a government for long.
And the car analogy is n9t fitting, unless we talk about cars bundled with the parking spot. But then they would not depreciate that much, and the banger with a parking spot in Manhattan would cost more than the Ferrari that could only park in shitsville.
The property a home sits on and the property a car sits on would depreciate the same.
But in reality neither has to depreciate, just stop growing at these insane rates. Below the cost of inflation, until the average worker can afford a home again.
>The “affordable” housing thing seems like such a misdirection. I can’t help but daydream that some moneyed interest somewhere fans those flames, as it looks like a dead end that can absorb endless fervor.
Do you really need a conspiracy by "moneyed interest" when the general public is perfectly happy to support similarly bad policies like rent control?
It is similar in Seattle. Large residential developments have a wastewater surcharge for several years to help upgrade infrastructure. Which normally just gets added as a line item in the utilities bill they pass off to residents.
What a smart idea! As a home owner I approve of this way to keep supply out of the market (or at least make it expensive enough to prop my price up). Can we invent any more charges?
In the US this isn't even true. The costs of grid upgrades are paid new capacity up front, which actually means there is a queue for grid connections twice the US total production, with no costs to ongoing customers.
Grid upgrade costs are frequently socialized onto existing customers, especially for larger upgrades. Your central premise that interconnect charges cover everything is false.
Yes, an alternative policy could be, the same unit price for electricity, and the additional cost is covered by increased usage (at the same unit price). The entity making the decision for the increased investment (the utilities company) will take the risk and rewards that comes with the investment.
But, if it turns out to be a reward, people will revolt that for-profit company is making excess profit from something as basic as utilities.
> "concluded that expected power demand from data centers was _a_ primary reason for $23 billion in customer price increases "
Also, it's weird that he describes PJM as "the organization that monitors the PJM market" when they describe themselves as "a regional transmission organization (RTO) that coordinates the movement of wholesale electricity" [0]. So are they monitors of the market or are they the market themselves?
I don't know... maybe I'm being picky, but the article just seems off. The whole bit about how data centers could maybe game the system by using less power during peak times also doesn't make sense - that's when they also have the highest demand. Pointing to cryptominers just makes me think he doesn't get what they do, which is basically arbitrage. Of course they stop when power costs go up, it eats up all of their profits and they can simply start back up when the costs go down.
Right, you can't point to a year where price increases didn't happen, so "a primary reason" becomes meaningless. It was going to happen for other reasons if it didn't happen for this one.
by my research, everything has scaled almost linearly with population. But- there was a huge, undeniable, dip in cost of a kw/h of power when lighting started focusing on CFLs and LEDs. Grid expansion stalled, and now people are upset since it started growing again
Doesn’t the rollout of more efficient lighting also coincide with the shale gas boom? This also happened at right around the time coal plants were starting to be retired for good as well, which short-circuited some expected generation cost increases.
You also had the de-industrialization of the US happening at the same time, which also took pressure off the transmission grid.
We basically had a few things offset population growth to allow us to use our grandparents electric grid investments for far beyond their expected lifetime. We’ve finally just caught up after a free ride over the past 4-5 decades.
I overall agree with you strongly, just surprised that more efficient lighting would have that much impact on grid demand.
I find it hilarious nobody ever mentions texas w/r/t this issue. They already required new Large Load Interconect Studies- and paying the cost of new grid infra. And, the hufe influx of study reqiests lead to new laws introducing batching.
I will note- the actual generation is left to market economics. I have much more faith in that working out equitably than regulating the grid. Even so, they've had significant consumer level grod connection fee increases- which I think reflects the end of various easy houshold effeciency gains (eg: incandecent -> CFL -> LED, P4 -> 14th gen, and HVAC. mainly lighting) and privatized profits more than anything.
Except it's fundamentally not equitable: regular people can't easily relocate, and use electricity to heat and cool their homes, cook food, and have lighting.
A data center can easily outbid them on the electricity market and drive prices through the roof for a few years, but also is free to simply turn off if it doesn't like the rates.
Exactly. Over in Virginia 37 datacenters use close to 3GW of electricity. The power utilities and overarching transmitters are looking at projects to ship some of that electricity (which requires ~700kV transmission). Two projects in the works are at $18B between them.
$18B to provide redundancy and not have to require schools and local government to limit electricity use and provide a bit more slack in the powergrid is a burden that all the users get to share. Lucky them. Yes, all users benefit, but lucky break for those datacenters, getting all that redundancy for power, without a $500M/ea bill.
That's great to hear that we have at least somewhat decent voltages. I thought America was still sort of piddling around with mostly 500kV and under power distribution.
It's not a huge power multiplier (P = VI, so linear), but in principle I do love the idea that if you are going to have these massive transmissions lines we ought use the conductor well, at good high voltages.
I'm really tired of reading comments like this"this is a policy choice" when these companies are straight up bribing politicians or exploiting their conflicts of interest to make this their policy. The public doesn't have a choice when all our leaders are bought and paid for by oligarchs and their businesses.
Comments like this are even more tiring. Everyone knows that there are moneyed interests in politics. But decisions on these issues are made at a local level and so are the most addressable by local political actions.
In my area there is quite a lot of mobilized anti-data center action that is completely ineffective. There is no legal way to remove corrupt local politicians from office that prevents the harm from being done.
And I mean this as in, local government ostensibly representing the community are also the land owners and contractors selling to developers, who can't be voted out of office until next year at the earliest and whose contracts and permits can't be revoked by a subsequent administration.
It's important to ground the increase in raw numbers.
The total revenue for electricity generation was $514b in 2024. So this was a 4-5% increase in costs. And if it is being invested in better generation and our aging infrastructure, that seems fine.
Sure, but why should regular people pay for the upgrades needed to power data centers? The tech companies can surely afford that easily given their record profits for the past decade?
supply and demand. there is more demand, prices go up, the market allocates, it's the best way, no other way comes close to making the overall price level to be as cheap as possible.
Forcing all costs on to new entrants in a market is generally a bad thing for the market and causes it to not expand fast enough. Also they may just choose to not participate in the market at all. I.e they'll bring in a whole lot of natural gas generators, put them in the parking lot and call it good, which is not a good thing long term.
Why would any reasonable governing entity (I understand we live in a world where there are not many of these) allow you to run a bunch of natural gas generators in a parking lot?
Forcing all costs into externalities is also not ideal. Right now, some state's systems, force residential consumers to pay for new infrastructure for new demand while only minimal costs are paid by the new customer.
But like, just hypothetically here, what if they don't? Isn't that the framing here?
Like, don't get me wrong, if we want to start talking about more centralized organization of the economy that operates on the calculated total benefit it provides to people in general, I am personally all for that. But something tells me that is not quite the argument you are wanting to make here, right?
It's funny that you don't realize you are doing the same thing; you are giving the absolute change like the article is doing; you are just doing the change in the rate instead of the underlying quantity. A 4% change in my rate if the original rate is 4% would be a 100% rate rate change, which of course is substantial.
I can't find anything breakdown in the article. But from what I have heard, you either live somewhere with a high concentration of new data centers and see a massive price increase of as much as 50%, or more likely you don't see any increase at all.
Or you live in Maine with no datacenters and see massive price increases anyway, because God forbid we actually had a reasonable energy policy in the state.
This just isn't true. On balance, data centers are turning out to be more like the "anchor tenant" of the power grid, financing improvements for everyone.
If the climate change denying and anti renewables Manhattan Institute is running propaganda for something that is strong evidence that the ordinary person is being conned.
Consumers pay for everything -- increased RAM, SSD, CPU, GPU prices, either directly (for DIY) or indirectly (apple/android phones). We pay dramatically increased prices for every device. We pay more because servers (non AI) cost a lot more now, which means cloud (or your own DC) bills increases for all companies and they are all going to raise prices.
Electricity will cost more because AI DCs need to be cheap, all upgrades cost will be borne by consumers.
Water will be less available and/or cost more because all forms of energy (except solar, wind, hydro) need a TON of water[1]. Which was ~48 trillion gallons in 2021.
As a consequence, nearly everything will cost a lot more. And this is the price we pay for AI. AI has hijacked the supply chains built for existing uses, taken over all the low cost ones. Forcing the rest of humanity to bear the costs of building everything new at a high cost, both high CapEx/OpEx.
These externalities must be accounted for when thinking about how 'cheap' AI is.
U.S. electric power sector water withdrawals for power plant cooling: 47.7 trillion gallons of water.
The electric power sector uses a large amount of water, mostly for cooling. Thermoelectric power plants (including natural gas, nuclear, and coal plants) boil water to create steam, which spins a turbine to generate electricity. The steam leaving the turbine must be cooled back into water to be used to generate more electricity. Plants withdraw water from nearby rivers, lakes, or oceans and pass that water through the steam leaving the turbine. That process cools and condenses the steam back into water. In 2021, 73% of the utility-scale electricity generated in the United States came from thermoelectric power plants.
Isn't this the classic overcapacity leads to lower prices that also represses investments that would increase capacity. But those lower prices also stimulate new demand that lead to higher prices...which then motivate investments that increase capacity?
Perhaps I'm just spoiled because I live in the PNW, where are best use for overcapacity was to ship power off to California. But in the past, cheap hydro attracted aluminum production that then attracted also attracted a whole airplane production industry.
I think most people are just debating whether the extra demand generated by AI is worth it, they weren't necessarily debating the same thing when it came aluminum or airplane production (albeit in the 1930s).
There are a few things going on at the same time that mean that electricity cost might actually go down.
The reasons for that are complex but have to do with how electricity pricing works. In many markets the price includes a lot of taxes, fixed cost for providers and infrastructure. Generation is only a minor cost. And on top of that the prices are set in a way that isn't really that flexible.
Infrastructure utilization is a very important here. Grid operators are very conservative with their infrastructure. They want to ensure there's enough to handle the worst case. That means there are a lot of assets that are nowhere near 100% utilized (e.g. cables and long distance transmission). It also means they are very inflexible serving new demand like data centers.
Adding batteries as energy buffers enables a lot better utilization of all these assets. That enables more revenue for the same infrastructure cost. Electricity prices can actually go down if you do that right. With renewables, there is very low marginal cost for generation. It's all infrastructure cost. Anything that improves infrastructure utilization enables more customers to have power that then share the infrastructure cost.
Data centers that are currently powered by things like on site gas turbines are not being very cost efficient. There's an obvious incentive for hyper scalers to invest in infrastructure that will lower their cost. They have access to many billions. They are spending on anything that will get them energy. They are desperate to spend. And they are completely bottle necked on grid operators that are being very conservative. Hence the expensive side hustle with gas turbines. There's a big opportunity here for massive investment in better grid infrastructure. That wouldn't necessarily be payed for by consumers. But they would still benefit from better infrastructure.
Utility providers won't build fiber connectivity to homes and make massive profits ($50+/month), but they'll build unnecessary infrastructure for AI DCs and make people pay for it. As long as we do things to serve the super rich, everything is okay.
Personally, some of my electricity bills will soon be rising because of changes to how rates are computed. The added "demand charge" will penalize customers that have variable loads, while benefiting data centers with relatively constant loads.
Also IMHO, we are building way too many data centers right now. It reminds me a lot of the Y2K dot.com crash, and all of the residual dark fiber.
I spotted this article today that claims it's all in our imagination.
Isn't this the same as saying "utility regulators delaying connecting new power to the grid hiked electricity prices on the public by $23B?"
When my apples are expensive, I don't generally grumble about all the demand from pie makers. If they demand more apples, new suppliers should come in to restore the price, right?
What is the increase in jobs/GDP for those communities that have paid more in electricity? A lot of these data centers are built in places with declining population and zero economic prospects for locals within their communities so they're a huge boon.
None? There's like 12 jobs per datacenter and the revenue goes to the Valley, mostly. The utilities get some profit but I doubt that money stays very local either.
Places I know that are getting them are generally seeing big time pay (relatively that is) for construction work and then dozens to a hundred of ongoing jobs once the work is done. That doesn't sound like much but it's huge for many of these communities.
Isn't almost all of the datacenter build out for inference, rather than training? If so, what's the issue? If the electricity demand is coming from actual use, then why are people getting mad at datacenters or AI companies, rather than the actual people driving the demand? It's like getting mad at Amazon for how much they increased fuel prices, which they surely did, given all the fuel that their trucks/planes burn. Or getting mad at some global food conglomerate for making açaí berries[1] more expensive, but there's a global craze for them and the conglomerate is just catering to that demand.
[1] or whatever other "superfood" that explodes in popularity
The electricity supply is probably a lot less elastic than the gas and oil supply at this point. Also, there’s no global price of electricity like there (more or less) is for oil. Residents suffer a localized cost when a large electricity consumer moves in, and they benefit when they can force the consumer to move to another locality. Definitely NIMBYism, but the reaction is understandable.
Bruh.... I've climbed those acai trees with locals in Brazil, and when I saw the thickness and richness of real, purely ground acai, and the dilute garbage (cough sambazon cough) they sell in the US, and mentioned how much it costs in the US.... the guy froze and thought he misheard me. A local friend turned and warned me "don't tell them that, because it'll create frustration and anger he can't do anything about."
They just run to the local store after manually grinding it in their machines, and get peanuts for a bag that would sell for >$100 in the US.
Most raw materials cost less than the final consumer-facing good made from those raw materials, because of all the costs that go into turning the raw materials into a useful product to sell to customers - why should ground acai be any different? If those people climbing acai trees in Brazil know what acai costs in the US or know that the product they have is better quality than what gets exported to the US, they can in principle use this information to create a better or cheaper acai export business than the people they're currently selling their ground acai to and undercut them.
Or getting mad at oil & gas companies? They’re only supplying the fossil drugs we demand. And using their billions in profits to lobby at every level of government and society to ensure they can continue despite the scientific evidence of harm that they’ve known since the 70s? Not at all like that.
Data centers should at the very least build their own renewable energy generation. It would set the right incentives in place and encourage investment in clean energy generation solutions. It would also present a very compelling problem to direct all that new AI compute towards solving.
The challenge is timing. Power is a relatively low share of the cost of a modern data center, but the length of time it takes for approvals and construction for solar means it would substantially slow down the construction vs. using grid+natural gas turbines which are faster to build and get approval for.
I'm 100% on board with charging data centers more for their power usage. They can afford to pay the extra costs. They would much rather pay in dollars vs. in time.
It is crazy to me that somehow customers pay to cover some of the costs for datacenters. I thought everything was usage based, but apparently that is not the case everywhere and it’s a shared burden? What kind of weird crap is that
Every tech company building out datacenters would pay in a heartbeat to add commensurate power to the grid. The cost is not an objection.
The problem is, there are insane and dumb regulatory barriers to adding power plants or interconnects. THESE ARE THE SAME PROBLEMS FACTORIES FACE WHEN RESHORING PRODUCTION, you should treat datacenters as the face of reindustrialization. Instead of complaining about using resources, we need to focus on solving our inability to provide infrastructure needed to support economic growth.
Given that datacenters seem to much more manpower efficient, that seems like a poor trade, to the point it might not be worth viewing them similarly. I'm seeing info reporting about 1 job per 5000 square feet of a datacenter after completion, and 1 job per 800-1500 square feet for a factory, depending on type.
More jobs is good, but if we're going to look at this through the lens of industry returning, it's a lousy return, even before factoring in that we probably lost a lot more factories than we're gaining datacenters.
Sounds probably about right in scale. 5-10x more employees per sqft for common industrial or manufacturing businesses.
But local negative impact on community is not remotely comparable to most industrial development. Everything from traffic, noise, pollution, etc. A few bad projects aside, these things are glorified warehouses typically sited in suburban industrial parks or the middle of nowhere.
It all really just comes down to the electricity needs they demand. Otherwise it’s about as close to as free money to a community as you can get.
The most problematic thing to me with this whole deal are local tax abatements. Those should be outright illegal though for any development.
The fact datacenters are standing up extremely expensive on-site natural gas turbines due to the immense delays for interconnect.
Pretty much no facility operator wants to also pay for and operate their own power plant. They are small and expensive to operate compared to combined cycle natural gas or other sources, and lack of access to a wider grid means even more additional expenses like a additional on-site redundancy.
If they could simply pay for grid interconnect that is ready by the time facility construction is completed they would do so, in the vast majority of cases.
The money is in getting things online ASAP. Builders are effectively throwing unlimited buckets of money at all aspects of these builds at people who can get shit done fast. Power interconnect would be no exception.
I think folks have become numb to these huge numbers being thrown around in terms of how large this investment cycle/bubble is. Those numbers mean real things - like operators writing basically blank checks to jump the queue on networking equipment, wiring harnesses, etc. Those in the industry who are not these giant hyperscalers are kind of shut out of everything from cabling, network gear, HVAC, optics, etc. For any price - since our volumes are a joke compared to these huge contracts. There is a reason you see unprecedented margins for every company selling critical parts and equipment for these builds.
Heck, standard power cables have gone up about 8x in cost on us in the past 12mo. It’s been crazy to watch and hear from suppliers how they give insane quotes to these companies which get approved same-day.
It’s all a giant race, cost is currently not a primary concern if money can be used to speed anything up.
Elon (in)famously put gas turbines on pickup trucks to power the Colossus 1/2 datacenters because connecting to the grid was incredibly slow and there was limited supply. And got sued to remove them.
Absent regulation, every operator would happily do the same thing to make the problem go away.
This $23B number that gets thrown around is not the increase to the public. The wording in the referenced report is
> Based on actual auction clearing prices and quantities and uplift MW, inclusion of existing and forecast data center load growth resulted in a combined total increase in capacity market revenue for the 2025/2026 BRA, the 2026/2027 BRA, and the 2027/2028 BRA of $23,100,955,341.
This is the increase in revenue to PJM from adding datacenter customers, and includes both the amount that datacenters paid as well as the amount that other customers paid due to higher prices from datacenters. So Fortune calling it an increase to "the public" means that they didn't read the report they are using as their source and are probably just repeating what they thought someone else meant.
Bloomberg in the past worded it as "data centers will add at least $23 billion to customer bills" in April and "added a minimum of $23 billion to customer bills" in February. Which while technically correct (datacenters are customers) seems meant to be misleading. And now that's the number that's getting thrown around as the increase to "the public".
The part I don't get is that the journalists could just give the actual number for the quantity that they are referring to (the amount that non-datacenters paid due to higher rates due to datacenter loads): when I calculated it a few months ago I think it was something like $16 billion rather than $23 billion. I feel like the story would have the same impact if the headline number was $16B as $23B, but $16B has the benefit of not being a misrepresentation of the situation.
---
Also I would definitely recommend checking out the PJM BRA report. It's a bit dense but not too hard to follow, and my personal takeaway was that the PJM market is just very dysfunctional and they are blaming the datacenters instead. I thought SemiAnalysis had a good analysis of it: https://newsletter.semianalysis.com/p/are-ai-datacenters-inc...
Or how about this framing (using your own numbers): for every dollar a datacenter is charged for electricity, regular hard-working Americans get charged $2.
First they gave all the water away for free to big ag, and now they're doing the same with the power.
When they gave water to big ag, we got to buy food.
With this, we pay extra to lose our jobs.
The ag water was used to grow hay which was shipped to China and Brazil...
And corn which was used to make energy (as ethanol) at a fraction of the land efficiency of solar.
https://www.anthropocenemagazine.org/2025/04/new-study-compa...
Nbd sun will provide power for a few billion years still
Not the political hill to die on
Surely that’s a different mode for of criticism than the one in the post?
Well, I think some of those hardworking Americans would be happy to pay some of those $2 because otherwise all the SaaS services and other services, platforms, and tools running in datacenters wouldn’t be accessible to them.
Or did I read this wrong and somewhere it said only datacenters running inference or training for LLMs?
The thing people are pissed about is giving better rates to datacenters.
Compare this to most of Europe (and Texas if I understood correctly) where the detacenters buy their electricity from the same market as everyone else (in Europe the spot market or futures) meaning they effectively pay the same price as everyone else.
It’s when they do some back room deal with the local public utility to get 50% off and offload the real costs to the public when people get angry.
Is there an example of this happening? Isn't the problem that datacenters are drawing electricity at the market rate and driving the cost up, rather than paying a surcharge for the difference?
I don't think the mechanism is that different, and even if it was Europe, etc would have the same issue - data centers would still bid up the price for existing capacity, and the future capacity added to accommodate the dcs would be added at a higher marginal rate too because scarcity of new supply + base load nature of dc loads. So the problem will get worse before it gets better
Yes datacenters inccrease the prices here but they don't get a cheaper price then everyone else. There is a massive difference there.
Concrete examples? Quick google only shows small discounts for DCs relative to other wholesale industrial consumers, which could be partially explained away by the flat load, lack of stability concerns, no need for power factor correction hardware, etc. I think 'bidding up the market' is the dominant mechanism, here in Australia (spot market + futures) consumer prices are projected to increase 26%.
Building a data center in my country is so expensive only Microsoft and Google do it.
The paperwork and environmental laws just aren't worth it.
> But what if the power company needs to upgrade the substation to handle the increased needs of the data center? Or secure additional sources of electricity? In these cases, the investments are part of the electricity grid that everyone uses. These costs will likely be shared among all customers.
Okay but this is a policy choice. It doesn’t have to be that way.
Where I live, if a developer wants to build a subdivision, they pay for the water and sewer lines. They pay to connect those lines to the city infrastructure. If the city infrastructure needs to be upgraded to handle the new volumes, they pay for at least a proportionate share of that too. The ongoing maintenance becomes part of the city's budget eventually, but not the costs of the build out.
All those costs go into the price of the houses built there.
And this is also part of why building "affordable" houses rarely happens. All the infrastructure costs the same whether the houses cost $100K or $1 million.
That's smart. Do they do it with roads also? That's a big one near me - developers buying hundred acre farms on unpainted 2 lane country roads and jamming in 2000 houses. Then inevitably the road becomes unusable until the city or county gets around to addressing it.
Always wondered why the county didn't require the road work, or money for it, up front.
That's negligence on the part of the county/township. Around here, every new development is required to pay for the traffic and utilities improvements that will be required once the thing is built. (And all the engineering that goes into figuring out the impact in the first place.)
In my hometown in Idaho in the 1990s and 2000s yes, this includes access roads and improvements to the surrounding area. A car dealership and Wal Mart both paid for road improvements and traffic lights as part of development.
The developer gets to at least name the roads. I worked for a city, the developers would submit names and while the city had the final call 99% of the time they would just accept whatever the developer submitted. The only times that I saw it get denied were times where it would create confusion like a similarly named road already existed or it could be construed as something profane
The “affordable” housing thing seems like such a misdirection. I can’t help but daydream that some moneyed interest somewhere fans those flames, as it looks like a dead end that can absorb endless fervor.
You know what you do if you want an affordable car? You buy used. I think most people understand Ford is never going to build another a car that costs $10k brand new, and the last new car near that price barely sold because it was so stripped.
Houses don’t shrink as they age.
When everything is built huge you don’t end up with homes that are cheap to heat, cool, or maintain. That’s why building affordable houses is actually a real issue.
What’s really dumb is the average number of people living in a house has tanked over time but the median new home just keeps getting larger. In an efficient market you’d expect new homes to match what buyers want, but regulatory capture has severely distorted the market.
What makes me crazy is these big houses aren't even that nice. They don't have significantly more rooms. It's all just stretched out to maximize sq ft for minimal costs. Sometimes with so many cut corners they're unsafe.
So if you have a big family or multigenerational needs then often you have to do significant remodeling or sell anyway.
If a gornment changed policy to make houses depreciate like cars, it would not be a government for long.
And the car analogy is n9t fitting, unless we talk about cars bundled with the parking spot. But then they would not depreciate that much, and the banger with a parking spot in Manhattan would cost more than the Ferrari that could only park in shitsville.
The property a home sits on and the property a car sits on would depreciate the same.
But in reality neither has to depreciate, just stop growing at these insane rates. Below the cost of inflation, until the average worker can afford a home again.
>The “affordable” housing thing seems like such a misdirection. I can’t help but daydream that some moneyed interest somewhere fans those flames, as it looks like a dead end that can absorb endless fervor.
Do you really need a conspiracy by "moneyed interest" when the general public is perfectly happy to support similarly bad policies like rent control?
It's almost like "affordable hosuing" are the vacancies that the people moving into shiny new homes leave behind.
It is similar in Seattle. Large residential developments have a wastewater surcharge for several years to help upgrade infrastructure. Which normally just gets added as a line item in the utilities bill they pass off to residents.
What a smart idea! As a home owner I approve of this way to keep supply out of the market (or at least make it expensive enough to prop my price up). Can we invent any more charges?
In the US this isn't even true. The costs of grid upgrades are paid new capacity up front, which actually means there is a queue for grid connections twice the US total production, with no costs to ongoing customers.
Your post is largely false, and quite confused.
Grid upgrade costs are frequently socialized onto existing customers, especially for larger upgrades. Your central premise that interconnect charges cover everything is false.
It varies by jurisdiction in the US.
Yes, an alternative policy could be, the same unit price for electricity, and the additional cost is covered by increased usage (at the same unit price). The entity making the decision for the increased investment (the utilities company) will take the risk and rewards that comes with the investment.
But, if it turns out to be a reward, people will revolt that for-profit company is making excess profit from something as basic as utilities.
Still waiting for canadian cell phone bills to come down after all those "upgrades" that supposedly happened.
Precisely this. Incredibly annoying headline
It doesn't even match what's in the article.
> "concluded that expected power demand from data centers was _a_ primary reason for $23 billion in customer price increases "
Also, it's weird that he describes PJM as "the organization that monitors the PJM market" when they describe themselves as "a regional transmission organization (RTO) that coordinates the movement of wholesale electricity" [0]. So are they monitors of the market or are they the market themselves?
I don't know... maybe I'm being picky, but the article just seems off. The whole bit about how data centers could maybe game the system by using less power during peak times also doesn't make sense - that's when they also have the highest demand. Pointing to cryptominers just makes me think he doesn't get what they do, which is basically arbitrage. Of course they stop when power costs go up, it eats up all of their profits and they can simply start back up when the costs go down.
[0] https://www.pjm.com/about-pjm
Right, you can't point to a year where price increases didn't happen, so "a primary reason" becomes meaningless. It was going to happen for other reasons if it didn't happen for this one.
by my research, everything has scaled almost linearly with population. But- there was a huge, undeniable, dip in cost of a kw/h of power when lighting started focusing on CFLs and LEDs. Grid expansion stalled, and now people are upset since it started growing again
Doesn’t the rollout of more efficient lighting also coincide with the shale gas boom? This also happened at right around the time coal plants were starting to be retired for good as well, which short-circuited some expected generation cost increases.
You also had the de-industrialization of the US happening at the same time, which also took pressure off the transmission grid.
We basically had a few things offset population growth to allow us to use our grandparents electric grid investments for far beyond their expected lifetime. We’ve finally just caught up after a free ride over the past 4-5 decades.
I overall agree with you strongly, just surprised that more efficient lighting would have that much impact on grid demand.
I find it hilarious nobody ever mentions texas w/r/t this issue. They already required new Large Load Interconect Studies- and paying the cost of new grid infra. And, the hufe influx of study reqiests lead to new laws introducing batching.
I will note- the actual generation is left to market economics. I have much more faith in that working out equitably than regulating the grid. Even so, they've had significant consumer level grod connection fee increases- which I think reflects the end of various easy houshold effeciency gains (eg: incandecent -> CFL -> LED, P4 -> 14th gen, and HVAC. mainly lighting) and privatized profits more than anything.
Except it's fundamentally not equitable: regular people can't easily relocate, and use electricity to heat and cool their homes, cook food, and have lighting.
A data center can easily outbid them on the electricity market and drive prices through the roof for a few years, but also is free to simply turn off if it doesn't like the rates.
Build more power!
Exactly. Over in Virginia 37 datacenters use close to 3GW of electricity. The power utilities and overarching transmitters are looking at projects to ship some of that electricity (which requires ~700kV transmission). Two projects in the works are at $18B between them.
$18B to provide redundancy and not have to require schools and local government to limit electricity use and provide a bit more slack in the powergrid is a burden that all the users get to share. Lucky them. Yes, all users benefit, but lucky break for those datacenters, getting all that redundancy for power, without a $500M/ea bill.
That's great to hear that we have at least somewhat decent voltages. I thought America was still sort of piddling around with mostly 500kV and under power distribution.
It's not a huge power multiplier (P = VI, so linear), but in principle I do love the idea that if you are going to have these massive transmissions lines we ought use the conductor well, at good high voltages.
China has been doing 1MV and 1.1MV lines for a while now, which is so excellent to see. https://www.bbc.co.uk/future/article/20241113-will-chinas-ul... https://en.wikipedia.org/wiki/Ultra-high-voltage_electricity...
I'm really tired of reading comments like this"this is a policy choice" when these companies are straight up bribing politicians or exploiting their conflicts of interest to make this their policy. The public doesn't have a choice when all our leaders are bought and paid for by oligarchs and their businesses.
Comments like this are even more tiring. Everyone knows that there are moneyed interests in politics. But decisions on these issues are made at a local level and so are the most addressable by local political actions.
In my area there is quite a lot of mobilized anti-data center action that is completely ineffective. There is no legal way to remove corrupt local politicians from office that prevents the harm from being done.
And I mean this as in, local government ostensibly representing the community are also the land owners and contractors selling to developers, who can't be voted out of office until next year at the earliest and whose contracts and permits can't be revoked by a subsequent administration.
How exactly are you supposed to stop that?
You’re not. You’re reaping the consequences of your previous decisions as a community.
So what? You're saying local officials can't be corrupted by campaign donations? Because that's a hot take.
I’m saying local officials are the ones where your vote counts the most.
It's important to ground the increase in raw numbers.
The total revenue for electricity generation was $514b in 2024. So this was a 4-5% increase in costs. And if it is being invested in better generation and our aging infrastructure, that seems fine.
Sure, but why should regular people pay for the upgrades needed to power data centers? The tech companies can surely afford that easily given their record profits for the past decade?
supply and demand. there is more demand, prices go up, the market allocates, it's the best way, no other way comes close to making the overall price level to be as cheap as possible.
Forcing all costs on to new entrants in a market is generally a bad thing for the market and causes it to not expand fast enough. Also they may just choose to not participate in the market at all. I.e they'll bring in a whole lot of natural gas generators, put them in the parking lot and call it good, which is not a good thing long term.
Why would any reasonable governing entity (I understand we live in a world where there are not many of these) allow you to run a bunch of natural gas generators in a parking lot?
Probably because there are laws that exist
Because xAI does? I wasn't making a hypothetical.
Forcing all costs into externalities is also not ideal. Right now, some state's systems, force residential consumers to pay for new infrastructure for new demand while only minimal costs are paid by the new customer.
All or minimal costs? You’ve provided two contradicting statements here.
Because they like shopping on Amazon and watching Netflix?
Consumers are already paying Amazon and Netflix directly.
But like, just hypothetically here, what if they don't? Isn't that the framing here?
Like, don't get me wrong, if we want to start talking about more centralized organization of the economy that operates on the calculated total benefit it provides to people in general, I am personally all for that. But something tells me that is not quite the argument you are wanting to make here, right?
The fact they gave a raw number instead of a percent of a quantity that is a change hinted that this was the case. $23B is more impressive than 4%.
Imagine a mortgage rate raising by 4%. Still unimpressive ?
It's funny that you don't realize you are doing the same thing; you are giving the absolute change like the article is doing; you are just doing the change in the rate instead of the underlying quantity. A 4% change in my rate if the original rate is 4% would be a 100% rate rate change, which of course is substantial.
So the rate goes from say 4% to 4.16%? That's still unimpressive.
If you mean from 4% to 8%, then that's actually raising by 100%
Mortgage rates have gone up 8% just since February.
https://fred.stlouisfed.org/series/MORTGAGE30US
> So this was a 4-5% increase in costs
…spread over an entire country.
I can't find anything breakdown in the article. But from what I have heard, you either live somewhere with a high concentration of new data centers and see a massive price increase of as much as 50%, or more likely you don't see any increase at all.
Or you live in Maine with no datacenters and see massive price increases anyway, because God forbid we actually had a reasonable energy policy in the state.
>> if it is being invested in better generation and our aging infrastructure
If.
It's 4-5% increase when averaged out across the population. Some have personally faced much larger increases.
This just isn't true. On balance, data centers are turning out to be more like the "anchor tenant" of the power grid, financing improvements for everyone.
Overview article with links to actual studies: https://cityjournal.substack.com/p/data-centers-arent-raisin...
If the climate change denying and anti renewables Manhattan Institute is running propaganda for something that is strong evidence that the ordinary person is being conned.
Consumers pay for everything -- increased RAM, SSD, CPU, GPU prices, either directly (for DIY) or indirectly (apple/android phones). We pay dramatically increased prices for every device. We pay more because servers (non AI) cost a lot more now, which means cloud (or your own DC) bills increases for all companies and they are all going to raise prices.
Electricity will cost more because AI DCs need to be cheap, all upgrades cost will be borne by consumers.
Water will be less available and/or cost more because all forms of energy (except solar, wind, hydro) need a TON of water[1]. Which was ~48 trillion gallons in 2021.
As a consequence, nearly everything will cost a lot more. And this is the price we pay for AI. AI has hijacked the supply chains built for existing uses, taken over all the low cost ones. Forcing the rest of humanity to bear the costs of building everything new at a high cost, both high CapEx/OpEx.
These externalities must be accounted for when thinking about how 'cheap' AI is.
--- [1]From: https://www.eia.gov/todayinenergy/detail.php?id=56820
U.S. electric power sector water withdrawals for power plant cooling: 47.7 trillion gallons of water.
The electric power sector uses a large amount of water, mostly for cooling. Thermoelectric power plants (including natural gas, nuclear, and coal plants) boil water to create steam, which spins a turbine to generate electricity. The steam leaving the turbine must be cooled back into water to be used to generate more electricity. Plants withdraw water from nearby rivers, lakes, or oceans and pass that water through the steam leaving the turbine. That process cools and condenses the steam back into water. In 2021, 73% of the utility-scale electricity generated in the United States came from thermoelectric power plants.
Isn't this the classic overcapacity leads to lower prices that also represses investments that would increase capacity. But those lower prices also stimulate new demand that lead to higher prices...which then motivate investments that increase capacity?
Perhaps I'm just spoiled because I live in the PNW, where are best use for overcapacity was to ship power off to California. But in the past, cheap hydro attracted aluminum production that then attracted also attracted a whole airplane production industry.
I think most people are just debating whether the extra demand generated by AI is worth it, they weren't necessarily debating the same thing when it came aluminum or airplane production (albeit in the 1930s).
There are a few things going on at the same time that mean that electricity cost might actually go down.
The reasons for that are complex but have to do with how electricity pricing works. In many markets the price includes a lot of taxes, fixed cost for providers and infrastructure. Generation is only a minor cost. And on top of that the prices are set in a way that isn't really that flexible.
Infrastructure utilization is a very important here. Grid operators are very conservative with their infrastructure. They want to ensure there's enough to handle the worst case. That means there are a lot of assets that are nowhere near 100% utilized (e.g. cables and long distance transmission). It also means they are very inflexible serving new demand like data centers.
Adding batteries as energy buffers enables a lot better utilization of all these assets. That enables more revenue for the same infrastructure cost. Electricity prices can actually go down if you do that right. With renewables, there is very low marginal cost for generation. It's all infrastructure cost. Anything that improves infrastructure utilization enables more customers to have power that then share the infrastructure cost.
Data centers that are currently powered by things like on site gas turbines are not being very cost efficient. There's an obvious incentive for hyper scalers to invest in infrastructure that will lower their cost. They have access to many billions. They are spending on anything that will get them energy. They are desperate to spend. And they are completely bottle necked on grid operators that are being very conservative. Hence the expensive side hustle with gas turbines. There's a big opportunity here for massive investment in better grid infrastructure. That wouldn't necessarily be payed for by consumers. But they would still benefit from better infrastructure.
The key is unlocking these investments to happen.
Utility providers won't build fiber connectivity to homes and make massive profits ($50+/month), but they'll build unnecessary infrastructure for AI DCs and make people pay for it. As long as we do things to serve the super rich, everything is okay.
Personally, some of my electricity bills will soon be rising because of changes to how rates are computed. The added "demand charge" will penalize customers that have variable loads, while benefiting data centers with relatively constant loads.
Also IMHO, we are building way too many data centers right now. It reminds me a lot of the Y2K dot.com crash, and all of the residual dark fiber.
I spotted this article today that claims it's all in our imagination.
https://www.city-journal.org/article/data-centers-arent-rais...
Isn't this the same as saying "utility regulators delaying connecting new power to the grid hiked electricity prices on the public by $23B?"
When my apples are expensive, I don't generally grumble about all the demand from pie makers. If they demand more apples, new suppliers should come in to restore the price, right?
Do you need apples to turn on your lights? Was there a sudden new influx of pie makers? Is growing apples a government granted monopoly business?
If the answer to these questions was yes, yes, and yes then I think you would grumble.
Failures to allow faster generation/hookup rollout have suppressed supply increase relative to demand increase
What is the increase in jobs/GDP for those communities that have paid more in electricity? A lot of these data centers are built in places with declining population and zero economic prospects for locals within their communities so they're a huge boon.
None? There's like 12 jobs per datacenter and the revenue goes to the Valley, mostly. The utilities get some profit but I doubt that money stays very local either.
Places I know that are getting them are generally seeing big time pay (relatively that is) for construction work and then dozens to a hundred of ongoing jobs once the work is done. That doesn't sound like much but it's huge for many of these communities.
Data centers pay taxes don’t they? That money presumably goes into funding government services which would necessarily raise the local GDP
Only if the data center's task force of influencers, lobbyists, and arm-twisters are incompetent.
https://en.wikipedia.org/wiki/Tax_abatement
(Yes, in general, they'll pay some tax. Some. But they also add a wide variety of direct and indirect expenses, both fiscal and social.)
Isn't almost all of the datacenter build out for inference, rather than training? If so, what's the issue? If the electricity demand is coming from actual use, then why are people getting mad at datacenters or AI companies, rather than the actual people driving the demand? It's like getting mad at Amazon for how much they increased fuel prices, which they surely did, given all the fuel that their trucks/planes burn. Or getting mad at some global food conglomerate for making açaí berries[1] more expensive, but there's a global craze for them and the conglomerate is just catering to that demand.
[1] or whatever other "superfood" that explodes in popularity
The electricity supply is probably a lot less elastic than the gas and oil supply at this point. Also, there’s no global price of electricity like there (more or less) is for oil. Residents suffer a localized cost when a large electricity consumer moves in, and they benefit when they can force the consumer to move to another locality. Definitely NIMBYism, but the reaction is understandable.
It’s really just a question of saliency.
Bruh.... I've climbed those acai trees with locals in Brazil, and when I saw the thickness and richness of real, purely ground acai, and the dilute garbage (cough sambazon cough) they sell in the US, and mentioned how much it costs in the US.... the guy froze and thought he misheard me. A local friend turned and warned me "don't tell them that, because it'll create frustration and anger he can't do anything about."
They just run to the local store after manually grinding it in their machines, and get peanuts for a bag that would sell for >$100 in the US.
Most raw materials cost less than the final consumer-facing good made from those raw materials, because of all the costs that go into turning the raw materials into a useful product to sell to customers - why should ground acai be any different? If those people climbing acai trees in Brazil know what acai costs in the US or know that the product they have is better quality than what gets exported to the US, they can in principle use this information to create a better or cheaper acai export business than the people they're currently selling their ground acai to and undercut them.
Or getting mad at oil & gas companies? They’re only supplying the fossil drugs we demand. And using their billions in profits to lobby at every level of government and society to ensure they can continue despite the scientific evidence of harm that they’ve known since the 70s? Not at all like that.
https://www.investigate-europe.eu/posts/big-tech-data-centre...
https://www.politico.eu/article/europe-choose-ai-climate-goa...
Data centers should at the very least build their own renewable energy generation. It would set the right incentives in place and encourage investment in clean energy generation solutions. It would also present a very compelling problem to direct all that new AI compute towards solving.
Why not all factories? And hospitals? Schools? Offices?
The challenge is timing. Power is a relatively low share of the cost of a modern data center, but the length of time it takes for approvals and construction for solar means it would substantially slow down the construction vs. using grid+natural gas turbines which are faster to build and get approval for.
I'm 100% on board with charging data centers more for their power usage. They can afford to pay the extra costs. They would much rather pay in dollars vs. in time.
Maybe what they're doing in Oregon with POWER Act hikes on data centers is the way:
Oregon approves PGE’s 29.7% rate hike for data centers under landmark law
https://www.opb.org/article/2026/07/07/oregon-data-center-ge...
It is crazy to me that somehow customers pay to cover some of the costs for datacenters. I thought everything was usage based, but apparently that is not the case everywhere and it’s a shared burden? What kind of weird crap is that
Untrue headline.
PJM’s market monitor estimates data-center demand added $23.1 billion to regional wholesale capacity costs across three delivery years through 2028.
https://archive.ph/Ypq0o
according to my wife we paid about 1/2 of that :)
ah, welcome fellow PG&E user
Every tech company building out datacenters would pay in a heartbeat to add commensurate power to the grid. The cost is not an objection.
The problem is, there are insane and dumb regulatory barriers to adding power plants or interconnects. THESE ARE THE SAME PROBLEMS FACTORIES FACE WHEN RESHORING PRODUCTION, you should treat datacenters as the face of reindustrialization. Instead of complaining about using resources, we need to focus on solving our inability to provide infrastructure needed to support economic growth.
Given that datacenters seem to much more manpower efficient, that seems like a poor trade, to the point it might not be worth viewing them similarly. I'm seeing info reporting about 1 job per 5000 square feet of a datacenter after completion, and 1 job per 800-1500 square feet for a factory, depending on type.
More jobs is good, but if we're going to look at this through the lens of industry returning, it's a lousy return, even before factoring in that we probably lost a lot more factories than we're gaining datacenters.
Sounds probably about right in scale. 5-10x more employees per sqft for common industrial or manufacturing businesses.
But local negative impact on community is not remotely comparable to most industrial development. Everything from traffic, noise, pollution, etc. A few bad projects aside, these things are glorified warehouses typically sited in suburban industrial parks or the middle of nowhere.
It all really just comes down to the electricity needs they demand. Otherwise it’s about as close to as free money to a community as you can get.
The most problematic thing to me with this whole deal are local tax abatements. Those should be outright illegal though for any development.
> Every tech company building out datacenters would pay in a heartbeat to add commensurate power to the grid. The cost is not an objection.
Yeah. I'm going to need a source on this claim.
The fact datacenters are standing up extremely expensive on-site natural gas turbines due to the immense delays for interconnect.
Pretty much no facility operator wants to also pay for and operate their own power plant. They are small and expensive to operate compared to combined cycle natural gas or other sources, and lack of access to a wider grid means even more additional expenses like a additional on-site redundancy.
If they could simply pay for grid interconnect that is ready by the time facility construction is completed they would do so, in the vast majority of cases.
The money is in getting things online ASAP. Builders are effectively throwing unlimited buckets of money at all aspects of these builds at people who can get shit done fast. Power interconnect would be no exception.
I think folks have become numb to these huge numbers being thrown around in terms of how large this investment cycle/bubble is. Those numbers mean real things - like operators writing basically blank checks to jump the queue on networking equipment, wiring harnesses, etc. Those in the industry who are not these giant hyperscalers are kind of shut out of everything from cabling, network gear, HVAC, optics, etc. For any price - since our volumes are a joke compared to these huge contracts. There is a reason you see unprecedented margins for every company selling critical parts and equipment for these builds.
Heck, standard power cables have gone up about 8x in cost on us in the past 12mo. It’s been crazy to watch and hear from suppliers how they give insane quotes to these companies which get approved same-day.
It’s all a giant race, cost is currently not a primary concern if money can be used to speed anything up.
Elon (in)famously put gas turbines on pickup trucks to power the Colossus 1/2 datacenters because connecting to the grid was incredibly slow and there was limited supply. And got sued to remove them.
Absent regulation, every operator would happily do the same thing to make the problem go away.
"small price to pay for Massive AI Gains"
FYI to others: the above is not a quote from the article.