County with 37 Data Centers Asks Schools to 'Conserve Electricity'
404media.co
404media.co
Lawrence Berkeley National Lab recently did an analysis on electricity prices in the US [1] and found that most of the rate increase in Virginia was attributable to the VCEA, and that load growth had a mitigating effect on price increases.
If you look at the overall report (not just Virginia), the places where electricty costs are rising the fastest are generally not the same places where lots of new datacenters are being built. It's easy to blame datacenters, but there are many factors at play here.
[1] https://emp.lbl.gov/publications/factors-influencing-recent-...
It's so strange to me that the argument previously was "we don't have enough energy generation for EVs and heat pumps to electrify and decarbon" but data centers are thought of as must run load that everyone has to suffer in some way to enable (through increased rates or risk of blackouts), when they have very little positive impact for everyone except a small minority investing in them.
> It's easy to blame datacenters, but there are a lot of factors at play here.
It is because they are the problem. We need as much clean energy as quickly as possible to mitigate climate change, we do not need data centers, broadly speaking.
(if you replaced all of the farmland/ag land, the size of the state of Oregon, harvested for ethanol with solar, you would have more electrical generation than all current US electrical generation combined as of this comment; this is simply a question of will, proven by China's solar PV deployment rates [installing ~90-100GW of solar PV per month])
That seems slightly weird, but that sounds like there's some large fixed costs that they can spread over the entire subscriber base, so the extra data centers are picking up some of those fixed costs.
TLDR Humans need electricity to live, data center loads are a luxury that can wait for power to be provided, when available.
This is because the argument that we didn't have the grid capacity for EVs, heat pumps, and residential solar was never sincere. You could tell because the followup was never "and we should invest more in the grid" but rather the reactionary "and that's why we can't use EVs."
The same people would be opposed to data centers, if not for the fact that the AI buildout is making them all rich.
A factory produces tangible goods, and tends to get less hate because of that reality.
In other words, large “base loads” like data centers could both reduce the average power bill AND contribute to capacity shortages and load shedding.
They often have an individual contract with the utility and participate in load regulation: when you need liquefy a few tons of steel, those heaters have a lot of thermal inertia. If A/C loads are high they'll turn the power down, if wind output is high, they'll turn it up, and so on.
Do data centers participate in the same sort of dynamic pricing and power adjustment? I understand that they're spinning up and powering down instances on demand, and that those demands are somewhat outside of their control, but are they able (and willing, and desirous of reducing their electric bills) to dynamically adjust compute in response to utility rates?
There is not a good picture in aggregate though so it creates all kinds of narratives.
A few of them in the future pipeline swear they will have nuclear but I don't see that coming to fruition soon.
This is texas, they can't even form regulations that don't different between counties, or figure out zoning laws in Houston.
They are mad that they aren't getting special treatment. They want to be treated better than the aluminum smelting plant.
Many regions/markets have relatively stable retail rate structures, i.e., residential and most commercial customers don’t pay day-to-day and hour-to-hour electricity market prices. Their per-kWh and per-kW rates are adjusted on much longer time scales, like once a year.
So, big base load customers sign contracts in one year, resulting in better utilization of grid assets, so prices stay lower in the next year(s). But if those customers are not flexible (or are not on a rate structure that encourages them to be flexible), they could contribute to short-term capacity shortages this year. And that will have no impact on many rate payers’ bills, ant least not before rates get adjusted later.
Henrico County currently has 37 data-centres with ~2 gigawatts capacity (expected to reach 3 gigawatts).
Apparently, 1 MW can power approximately 834 homes annually. So 2 gigawatts would be closer to powering > 1.6 million homes.
Surely, that kind of concentrated demand is going to affect electricity distribution costs for everyone, which is what we are seeing now.
"It's a poorly designed grid". PJM might serve 65M+ users, but 2GW isn't easily able to be shipped across their grid. Dominion is proposing a 180mi 525kV HVDC line (ironically enough, from southern VA to northern VA) that can handle up to 3GW. But that's a $5 billion project. That number goes up. A PJM project proposal to ship power from NJ to VA is in the order of $12B.
So yeah, the grid can handle it, but now you're foisting twenty billion dollars of costs on to your users for the benefits of 37 customers. That's a ... pretty raw ... deal for the rest of them.
(And yes, PJM already has a grid that can supply some of those needs, and the reality isn't proposing that they run a new HVDC line just for these projects, but the capacity has to come from somewhere).
Can it handle it today? What about next year with the approved interconnects? What about in 2030? Does "handle it" include summer load or just average, what if temps are 5 degrees hotter than power models ran? What if PJM can't solve it's high priority overloads this year?
People want to talk about the grid as a binary system vs what it really is.
If serving those 37 customers forces choices that increase risk or cost for real customers (citizens, critical infrastructure like hospitals), it's a bad deal.
It seems like power utilities are optimized to solve problems, and serving these 37 customers can be framed as a solvable problem more easily than "Keeping the grid affordable and reliable" can.
Same goes for proof, which I'm sure would exclude power flow models if they were available for said debate.
Edit: We can agree the article is not comprehensive, and doesn't have data PJM and utilities have not published.
SemiAnalysis: https://newsletter.semianalysis.com/p/are-ai-datacenters-inc...
LBNL: https://emp.lbl.gov/publications/factors-influencing-recent-...
If you disagree, point to the analysis or data. Suggesting I’m “pretending” doesn’t address the argument. The article cherry-picks one county while largely ignoring the broader PJM pricing mechanics affecting millions of customers. Do better. Between pricing mechanics, input cost inflation and yes increase in demand has impact but not like the article likes to weave it.
You can't 8760 where and when overloads happen elsewhere, unplanned downtime can't be identified in (or out of) the ISO, you can't forcast demand, you can't plan pre-purchasing of power before it's on demand. You can't speculate on if reconductoring could save money, or remove overloads.
You want to look at rate increases as the core driver. So much so you dump it on replies to others comments. Too many variables are opaque for me to have a discussion about what is the 'largest' factor when each item on the above list could represent 10 million to 8 billion+ in cost.
But I can tell you I won't be convinced that new large loads, and who approved them are not a contributing factor because no one has "proven" it in real time. That's the kind of logical falacy that results in dismissal of real issues.
404 media has never been interested in being good journalism, but rather driving signups to to their website by people who don't know what proper journalism looks like
- electricity costs are rising
- there are many data centers
- claims data centers raise costs
- never actually shows how these new data centers are driving up electricity costs
Funny. They imply causation but never show it. If they had evidence that data centers caused this 25% increase, or a large percentage of it, dont you think they would show it?
Actually, you could write the exact same article but sub out data centers with the clean energy rollout.
Social media is always trying to make things sound way worse than they are and journalists are always in a rush to latch onto those trending keywords, and artificially spin headlines vaguely connected to the topic. So if you only skim headlines it's easy to think something is a bigger problem than it is.
Then, datacenters (demand increased, to put it mildly).
The explanation that one chooses for this is a function of one's other politics but the fact is, one has become a political rallying cry (opposition to demand increase) and the other (blame for supply decrease) has not, and this is the discussion we find ourselves in.
I've suspected that the energy regulations and the ruling party's close connection with Dominion Energy (the Governor recently attempted to fire the chair of Virginia Tech's board and replace him with the CEO of Dominion) have had an impact on power use more than data centers themselves.
Nb. Still many do not. Also, closed-loop systems use more electricity to operate, so it's not a cost-free solution.
One of the biggest problems with investing in a renewable grid is curtailment and managing demand or production spikes. In California, they already have to turn off solar plants during peak hours to help manage overproduction.
Even if you have massive quantities of energy storage, you don't have enough "inertia" in the grid to keep power levels consistent as DC power sources flick on and off.
Data centers can act as massive variable loads that can ramp up or down that turn excess energy into something vaguely profitable. And they can also help to create demand for more supply of intermittent green energy sources.
For one thing, energy demand hasn’t been increasing at the consumer level per capita. So you either have a growing population or growing commercial usage (like with data centers) or both driving energy demand.
Is there some literature that points to an idea that it’s faster to build a gas or coal power plant compared to a renewable one? My understanding is that solar is cheaper than gas, right? Even solar plus grid scale batteries are in a similar ballpark. And let’s not forget that other infrastructure like transmission lines and substations do not have any concept of renewable versus not.
The last thing I’ll point out is that we literally have to switch to renewables. There’s no long term versus short term, really, when we think about it. If we don’t make the switch we’ll actually put all of this planet in a lot bigger trouble than asking people to set their air conditioner thermostats a couple degrees higher.
I'm not an expert, but my understanding is that solar has a much larger up-front cost than, say, a natural gas plant -- but because there are no ongoing fuel costs, it is expected to be cheaper per kWh over the lifetime of the plant. So this is a great investment in the long term, but there are significant costs that must be absorbed in the short term -- and that's what is impacting prices right now.
Motion controlled lights are always timed badly, incredibly annoying to have them switch off when you are sitting still working or taking a duece.
How about the janitor shuts off the lights after everyone goes home?
I say that to make a point about magnitude of annoyances. Sure it’s annoying. So are potholes.
It does seem like a slightly smarter device could be built. The times it happens to me, and I wave my hands in the air to turn lights back on, it’s not that annoying. When I’m there for 3 more hours and I have to do it every 5-10 minutes. That’s more annoying. So the simple thing to do is to program the thing to incrementally increase its timer length. 5/10/20/30 minutes might be less annoying. Also if motion is detected a short time after the no motion timer triggers, that’s probably a sign there was someone present the entire time. Can adjust the logic with that in mind too. The current devices are fine it’s the logic that was lazy.
Things like motion sensor lights, faucets that turn off so quickly that you have to hold your hands in weird position to wash them, 1.6 gallon flush toilets are all completely avoidable things that are forced on us by a misguided sense of doo gooding.
These so called innovations are WORSE than what they replaced, less reliable, and often are so badly designed they accomplish the exact opposite of what they are intended to do.
A increasing retry time motion sensor light switch is a good idea. But then again, wherever these things are installed, they go with the cheapest thing that to meet code and get whatever energy certification.
Zero Fs are given by the people who don't have to use the room. The worst thing will be what is used. That is my experience.
Oh you want to change the timersetting? Guess what that's on a dip switch behind the face plate, and you need to find and shut off the breaker and pull the thing from the wall. Put in a ticket to facilities, we'll get to it never
Yes let's make you flush 5 times to save 0 water over a conventional while we run the sprinklers to keep our immaculate lawn green during a 105 degree summer.
Want to take a nice 5 gpm shower? No, you need to let almond farmers drain the river before it gets to the ocean so everybody can have almond milk. Stay in there twice as long buddy, don't know what to tell you.
The easier thing to do with motion timers would be to only enable them not during work hours. All of the other logic you want to program in them is something easily done today, but definitely not as long as they've been around.
The country roads that met up to the "big" 2 lane FM where I grew up were all posted 20mph. I remember one evening when my mom got pulled over because the deputy said she was swerving so much that she had to have been drunk. My mom looked at him and politely told him "you swerved around the same potholes I did". My mom never drank.
Also, you really seem stuck on trying to tell me what is annoying or not to the point you are now annoying me
Funny how your mom got pulled over by a police officer serving around potholes that he never noticed because public officials never used those roads. That silly city slicker stuff I suppose. But my point was about control mechanisms and you probably also fail to understand that in a rural setting you were the control mechanism. You are responsible for reporting the potholes. Even us city slickers have that ability. If that information is not acted on or there’s no budget for repairs; well that’s a whole different issue and you should consider that potholes was used as a weak analogy in this discussion about lighting. You keep wanting to argue purely about potholes, which is actually quite annoying.
$3.86 million claim from a pothole. I'm normally pretty skeptical of damage claims, but that is a pretty nasty pothole.
A light switch that doesn’t respond immediately might be the most frustrating technology life.
My machine had a hard reboot that first night... I lost my unsaved work and at that point I made it a point of religiously saving my work each evening when I went home... because each night my machine rebooted.
One day it was rather quiet. Might have been day before a long weekend, but it was a slow day in the building - very few people were walking about. I was working... and then my machine lost power. I stood up to figure out what was going on and my machine got power back. Ok... followed power cords to the wall. It was plugged into a gray outlet (rather than white outlet). The gray outlets were hooked up to the motion sensor that was for the hall lighting.
it's 'Jevonsian insanity'.
Part of solving that may be in what the article touches on - how to get the generation built before the DC shows up rather than as a promise after.
https://www.wric.com/news/virginia-news/dominion-energy-rate...
You are describing the normal process of purchasing things and how prices work. There is no nefarious passing on.
We should redistribute money from the wealthy to households, but I'm not sure if rate price controls beyond what already exists is the correct way to do it. The real issue with these data centers is tax jurisdiction shopping.
We're talking about capital and interest expenses for the benefit of a single customer but paid for by all other customers too. And also about Billions of $$ of commercial risk shifted to the shoulders of the public.
ChatGPT is the #1 most popular app on the app store, with nearly a billion weekly active users.
Perhaps we should charge AI subscription "tax" to pay for more renewable energy.
This is already significantly greater than the number of paid subscribers to search engines. AFAIK that's just Kagi, and they only have ~100k subscriptions.
Personally I see no contradiction in using a product despite wanting policies that prevent it: it just implies that you understand the tragedy of the commons.
February: https://www.roanokerambler.com/water-authority-releases-goog...
April: https://cardinalnews.org/2026/04/14/former-roanoke-rambler-o...
This tells me they know what they're doing is unpopular and are willing to squash any opposition.
these are still physical assets that can be destroyed in a riot
I appreciate 404 media's mission but isn't there enough stupid shit existing naturally in the world for them to illuminate that we don't need to do this?
Like 37 data centres in a small rural county?
So it might be a small county but it sure ain't rural!
The whole thing seems pretty overblown: County where energy prices are up 25% sends a memo asking employees to conserve electricity doesn't seem worth writing about. If prices are up 25%, I bet the datacenter guys are also working on efficiency. The county isn't asking datacenter peeps to conserve energy, because the county isn't paying their electricity bill.
From the headline, I thought this was going to be schools that signed up to participate in demand response in return for reduced electric rates are being asked to reduce their demand. Growing up in socal, most of the schools were on demand response programs, and sometimes we'd have reduced lighting as a result. I wouldn't expect a lot of datacenters to participate in demand response programs, so the angle would be 'the schools have to turn off their lights, but the datacenters don't do anything' ... ignoring the cost savings the schools signed up for; some datacenters could participate though --- large operators can move traffic and shutdown, idle or limit power for most of the servers, or can switch to local generation; but facilities for small hosting / colocation probably don't have enough insight into their customer loads to move traffic and might not want to run their generators.
So the best way to keep money coming in is to read the vibes of social media, and print stories that fuel those fires. Basically manufacture stories using well established marketing and propaganda techniques to maximize click rates.
Ice cream man selling ice pops in the park becomes "Man seen using treats to lure young children to his van in the park".
Honest headline, criminally misleading takeaway.
You could basically write the same article but swap data centers for clean energy upgrades.
- rationing request anecdote
- clean energy law dictating cleaner power
- unexamined claim clean energy is increasing prices nationally
- state there is a 25% price increased
Its
- “A” and “B”
- claim “A” causes “B”
- restate “B”
Did I show “A causes B”?
Really makes you think: should you really support efforts like this when there are better and more honest groups out there? (Cough EFF *cough)
This whole just throw it on the grid and residential consumers pick up the increased infra costs is insane
These companies are investing heavily in new energy development?
https://techcrunch.com/2025/03/30/data-centers-love-solar-he...
> Meta kicked off the year with a 200-megawatt solar deal with multinational electric utility Engie. The purchase went toward a solar farm near one of the company’s existing data centers in Texas. At the time of the deal, Meta already had over 12 gigawatts of generating capacity in its renewable portfolio.
> Stargate AI partnership between OpenAI, Oracle, and SoftBank Group was reported by Bloomberg to be powered, at least in part, by solar. SB Energy, which is part of SoftBank’s portfolio, is expected to develop solar installations backed by grid-scale batteries.
> Meta closed out the month with another massive solar deal, this one with Spanish renewable developer Zelestra. The contract was for 595 megawatts of capacity.
> Meta continued its streak in February, investing in a 505-megawatt solar project with Cypress Creek Renewables, which is developing the massive installation in Coleman County, Texas — about 150 miles northwest of Austin.
> Microsoft entered the fray in February, too. The company has long been a buyer of renewable energy to power its operations, and added another 389 megawatts of solar in a deal with EDP Renewables North America.
> Amazon also made a big purchase, backing a hybrid project on the Iberian Peninsula that includes wind, solar, and pumped-hydroelectric storage. The deal included 476 megawatts total, of which 212 megawatts are solar.
> Outside of the U.S., data center operators have also been investing in solar. In India, CtrlS built its own 125-megawatt facility in two phases, the first half of which was finished in June 2024 with the second completed in early February. In South America, Telecom Argentina agreed to buy electricity from a 130-megawatt solar farm developed by MSU Green Energy.
> Microsoft added another three solar developments in March, again focusing on the Midwest. The projects span Illinois, Michigan, and Missouri, and they’re being developed by AES. Together, they will provide Microsoft with 475 megawatts of capacity, adding to its considerable 34-gigawatt portfolio.
> Cisco got in the game with a 100-megawatt deal with X-Elio, a solar developer owned by Brookfield, an asset manager that has bet big on renewables. The power purchase agreements see Cisco buying capacity from two different Texas solar projects.
> Meta added another 200 megawatts of solar to its portfolio in March in a deal with RWE. The solar farm will be built just southeast of Austin.
> In Italy, data center operator Data4 signed a 10-year deal with utility Edison Energia to buy power from a 148-megawatt solar farm northwest of Rome.
That was all in 2025^
There's also been big investment in nuclear and natural gas
All that generation capacity is great, but you need to get the electrons to the datacenter. And in most jurisdictions the transmission part of the grid is a shared cost. Mostly...there is technically some obligation to pay for upgrades but it doesn't really work.
e.g.
>Over 95% of the projects identified passed all of their transmission connection costs onto local people’s electricity bills
It's a real privatize profits/benefit, socialize losses/costs type setup.
[0] https://www.ucs.org/about/news/billions-dollars-unreported-d...
Just like all developments since collective ideas went out of fashion...
Compared to my local utility (and most of Canada), the demand charge is low, but the per kWh charge is high. NSPower's general tariff is $9.089/kW + 15.738c/kWh for the first 200/mth, and 12.674c/kWh after that. Equivalent rates in Quebec and Manitoba are about half that. https://www.nspower.ca/your-business/save-money-energy/busin...
Current CAD/USD is 0.7, so subtract 30% for NSPower's rates in USD.
Some of the data centers now run disconnected on gas turbines 24/7, which is better for electricity prices but they can be big nuisance for people living nearby.
Without some sort of mitigation, the costs keep rising and it'll drive families away from these cities and/or counties to avoid the cost hikes. This is akin to what we're seeing in a lot of major cities with rent, people are living further and further away from where they work, paying taxes in other forms (time, public transport costs, gas costs for their car inc. wear & tear, etc).
Hardly seems fair or right.
This is literally what they believe in. This is what they have been doing for nearly 50 years. It's the entire purpose of neoliberalism: corporations are more important than citizens.
It is a near certainty that power rates would be cheaper for these cities if they removed themselves from the PG&E pool. Right now they look to be on the hook to help pay for all of the (long deferred) power line under-grounding that the recent state wildfires have proven is necessary. Much of that under-grounding is to get power into remote locations, and does nothing for most people (other than the implicit reduction of wildfires, which is a complicated subject).
But there is a second side to that coin: without the big cities full of people (which are relatively cheap to service), all of the needed under grounding costs are going to fall to rural California areas, and they simply don't have the population or finances to pay for that.
Personally I am in favor of some mixture. I would make the utilities all completely non-profit, with no investors to demand returns (the current system has perverse incentives). I would also start looking at some drastic limitations on where the public pays for power lines. Yes that would make some rural locations financial impossible to draw power to, but that would probably be a part of a real-plolitik re-evaluation of where people can afford to live. This is probably going to line up pretty closely with pushing people out of fire-prone places that should also be pretty much un-insurable anyways.
That said, this is only part of the picture. The other big part is that California has very aggressive tort law and CA juries basically view PG&E as a giant piggy bank.
The list they give is overwhelmingly dominated by one item:
“Turn off your lights when leaving your workspace, including when you leave for the day. Turn off your computers/laptops at the end of each workday. If your workspace has windows, adjust the blinds to manage heat from sunlight. Unplug any appliances, chargers, or other electrical items when they are not in use. Please limit use of (or refrain altogether from using) space heaters. A typical space heater alone can cost the county from $150 to $300 per year in electricity costs.”
$20 of AI tokens over a month? Probably somewhere between, on average, 40-320 W, depending on how you weight the cost of training and which recent-ish model you're using.
Tokenmaxxers? They're the heavy users. $2k/month (or whatever) gets you a lot of electricity through those GPUs.
Those poor battlefields. How rarely it is that anyone thinks of their needs.
obviously the average goes up in this circumstance...
Pepperidge farm remembers
but sure would be nice if it would cause an exponential acceleration of fusion development in the meanwhile
however that still has a law of theromodynamics problem of pumping heat into atmosphere
maybe exponential advancement of solar but they've already figured out that cannot improve more than another several percent, and manufacturing is already near peak efficiency
The AI bubble can’t pop soon enough.
1) https://www.404media.co/the-tokenpocalypse-is-here-companies...