Texas data center boom could erase decades of progress on air quality | Hacker News Reader

Texas data center boom could erase decades of progress on air quality

texastribune.org

55 pointshn_acker30 comments

The best thing Texas could do (for both data center proliferation and air quality) would be to have ERCOT (it's in-state-only grid operator, which avoids most federal regulations) join up with the rest of the continental United States' electric infrastructure [0]

This would allow for renewables (mainly WIND in Texas) to be sold out-of-state, and would also mitigate peaking events (e.g. winter 2021 disaster) – this would reduce the need for in-state peakers (instead: buy power from another US state).

But as the saying goes: "everything is bigger in Texas" – including the self-induced isolation. Doing this would require Texas to submit to federal energy regulations... which I'm doubtful will ever happen voluntarily.

src: retired Texas datacenter electrician (IBEW)

[0] <https://en.wikipedia.org/wiki/North_American_power_transmiss...>

[Texas' isolated grid operator:] <https://www.ercot.com/>

You can do that now with dc lines. There was a 5gw 3 way intercpnnect almoat over the finish line maybe a decade ago. Being regulated by the feds is not necessary or even desireable for this.

Look up the "Midnight Connection" (when a local grid operator in Texas connected to Oklahoma just to test the legality).

At present, Texas is allowed (and does) minimal energy transfers, using mostly DC-DC ties (but also one AC interconnect in Rusk Co, Texas, for emergencies); but the limit is a small single-digit percentage of usage (<3%, IIRC).

>There was a 5gw 3 way intercpnnect almoat over the finish line maybe a decade ago.

Trés Amigos (the project you refer to) was disbanded in 2016. Its major backer "backed out."

5GW is nothing for ERCOT. Last year, the peak power used was just under 90GW. This year, all of the existing currently filed datacenter projects add another 400GW to that. And that’s just for the state of Texas.

There’s no way that anyone on this planet can add over 400GW of production in a single year. Not even if you put the entire combined capacity of the planet together, could you add another 400GW to that in a single year.

This is going to take decades to add 400GW to the ERCOT or US grid, regardless of how they are interconnected. Over 90% of these datacenter plans will just never happen. But they will do untold amounts of destruction to the planet, all the while as they are trying to push through the ones that do finally get built.

That’s why the AI bubble will collapse.

Interconnection requests are the wrong way to look at it. In every power pool interconnecti(n requests are always a large multiple of whatever actually eventually connects for many reasons but a big one is testing multiple locati)ns for a large load. Also 5gw in a 90gw peak load system is massive.

This random 3pm on a Texas Tuesday, Texas' demand is just over 60GW;

>>[Even] 5gw in a [60gw] peak load system is massive.

What you originally wrote is also true, but for normal operating conditions:

This would save Texians so much money by not having to use natural gas peakers (but the generators' owners, themselves, are making bank... so they ARE incentived to use nat.gas), and each facility makes BANK when the Texas grid is peaking at 80GW+ (e.g. 2021 winter storms, the costliest natural disaster in Texas' history, would have been entirely avoidable == buy electricity from the rest of America).

After the 2021 storm, the top-rate for a MWH was reduced -66% (but is still orders of magnitude more expensive than normal ERCOT operating compensations; the Texas power grid operators are incentivized to tax the grid maximally).

The trendy desire to blame data centers for everything is nauseating. Data centers aren't poluting the air. Power plants are.

Since Texas has been leading the nation on renewables because it has a fairly free power market amd renewables are the cheapest power available, I'd expect this surge in demand to be eventually replaced by more renewables. In the meantime, extra capacity comes from old plants that can be spun up quickly.

Data centers often come with their very own natural gas turbines right outside to provide power the grid can't supply. This is what happened at xAI's data center in Memphis.

Correct. They will operate until the grid can supply reliable power. Give it a few years. There is a massive backlog to grid connectivity and generation construction.

That massive backlog already exists for what is being built now, not things in the pipeline. Your looking at 10+ years at getting that tied in.

I don't think you have any clue on the size of the projects being suggested. We are talking 3 times the size of the current texas grid which is twice as large as the California grid. It is a mind shattering amount of power that would be required if it could all be built.

>Data centers aren't poluting the air. Power plants are.

Currently for Datacenter's to move forward in Texas they must have some local power production so you see some of the newer upcoming ones featuring 20+ portable gas turbines on site. Others feature mixed solar/portable gas turbine.

Those will all go to renewables over time, just as all of texas has been, because it's cheaper.

This is temporary because of the surge in power. Renewables will be built to handle the surge if it becomes a new baseline.

That doesn't really help the people who suffer serious health issues from the surge of pollution in the neighborhood while waiting for the nearby data center to eventually maybe switch to renewable energy.

When?

The equipment for grid connections is already 5 years behind in Texas. Are we talking about 20 or 30 years from now? Might as well be a lifetime.

Grid stuff is 5yr out everywhere, including my "we don't even have datacenter projects to speak of because out power is just that expensive" region. That's just how the industry moves.

Previous poster: This is temporary because of the surge in power

You: Grid stuff is 5yr out everywhere

The article: Grid connections being years out is going to cause a huge increase in pollution because datacenters are setting up local gas operatings at the gigawatt scale.

I think a lot of people are missing the entire point. The next 10-20 years is going to have a huge spike in pollution (and CO2 release) that is going to affect humans and the grid at large.

Nope. There’s this thing called the “Oil and Gas Lobby”, which basically owns the entire state. Renewables will continue to be blamed for everything that goes wrong, and those projects will continue to get killed. This will persist for decades and decades yet to come.

The only way that changes is when those guys finally start to get themselves planted in the ground.

Natural gas is better than coal. Natural gas exhausts can be sequestered. But since we can't even break the coal habit let alone embrace renewables and clear the path for nuclear as needed, good luck! Stupid meatbags.

Love to know who this actually triggers and what's inaccurate about the above? Coal is super good and clean? Natural gas emissions can't be sequestered despite numerous examples of exactly that? How dare I even suggest nuclear? Or is it just downright America-hating unpatriotic and letting the evil drag queens win? Genuinely curious.

It's election season, and data centers are being thrown around ads continuously to scare the olds into voting.

It's one bridge away from selling them Old Glory robot insurance.

"Data centers are everywhere and they consume old people's medicine for fuel."

The trendy desire to blame obesity for preventable death is nauseating. Obesity doesn't kill people. Heart attacks and cancers do.

FFS.

Your analogy is that data centers are like obesity because they require power. Are Aluminum smelters also like obesity?

A better analogy would be to blame exercise for food shortages because it makes people hungry.

Let's not ignore the years before LLM-oriented data centers actually if ever switch to renewable energy, never mind whether the switch will "reverse" the long-term effects of high pollution during the first years. Refining GP's and your analogies:

Imagine if an overweight person could only lose significant weight if they temporarily underwent obesity for a few years. While many overweight people exist, we should actively prevent more people from becoming overweight, and we should change the socioeconomic and political circumstances that contribute to excessive weight gain.

Imagine if a moderate exercise session causes a person to feel severe hunger pains for a few hours that do not immediately abate upon eating. Such an unfair biological condition would incentivize the person to avoid moderate or vigorous exercise. If the person attempts to power through the pain, the pain may trick them into eating more than they need to at the expense of their wallets, their health, and collective food supply. We should encourage such individuals to set aside longer times for light exercise if their finances and work allow them to, teach people to reduce their extra calorie margins without sacrificing necessary nutrients (especially because taking in less extra most-things is easier than removing extra most-things), and research palliative and curative medical treatments for this condition.

> Data centers aren't poluting the air. Power plants are.

"Guns don't kill people, people do."

Neither argument holds water.

Reply on news.ycombinator.com