Hacker News Reader

Comment by konschubert | Hacker News Reader

Abilene, Texas is also a great place to build a solar power plant with some batteries and reduce the gas bills for these datacenters.

EDIT: I am not suggesting that they don’t build gas turbines or go off grid. I’m saying they can save fuel by using solar when it’s there.

View on HN

Everyone is talking about batteries but honestly you don’t need batteries.

- Data centers don’t sleep

- Data center load (for AI) could be shifted to follow the sun

- The energy requirements mean you aren’t likely to overbuild your solar farm

At night just stop running your GPUs and / or pull from the grid

Using GPUs as paperweights at night doesn't seem to be an efficient usage and it comes with its own costs. "Pull from the grid" is what they try not to do.

Seems crazy to build an expensive factory and then cheap out on the power source so it can only run when it is sunny out

<50% utilization of billions of dollars of capex. Brilliant idea.

There isn't a battery in the world big enough to provide enough power for data enter if this size.

Some batteries in this case is a bit like saying some water about the Pacific ocean.

A big AI data center uses about 1 GWh of power each night.

A large battery storage site is about 500MWh.

So this is totally doable and it’s also going to be economical as soon as the US has built enough LNG export capacity.

That's a big data enter from last decade. Today big data enters use 4GWh power in winter. So only 8 of them.

4GWh of power in what time frame?

Either way, in winter you’ll need the gas turbines, I didn’t claim otherwise.

so, half an hour's worth?

A medium AI data center uses 1GWh.. per hour.. 10-14GWh overnight (the shoulders aren't strong solar producers). Google 1GW (India), Meta 2GW (Louisiana), Meta 1GW (Texas), Stargate 1GW (UAE), Tsukuba 1GW (Japan), Gangwon Hyperscaler 1GW (South Korea), Teesside AI 6GW (UK).

https://constructionreviewonline.com/top-5-largest-data-cent...

Hi,

I think that's a fair point, I underestimated the power needs.

For a 1GW data center, the economical case is probably to build about 2 GWp of solar, plus a 5GWh battery or so.

That is not unrealistic:

https://masdar.ae/en/news/newsroom/masdar-and-consortium-par...

https://www.pv-magazine.com/2026/04/28/catl-secures-worlds-l...

On a sunny day, that will let you to power the data center fully from solar for most of the day, and partially through the night.

Obviously you need ~1GW of gas turbine capacity as well, for the non-so-sunny-days, so the economic case for solar here is dependent on the fuel price.

The Pacific Ocean is a bit of an exageration here.

There isn't a single battery this big today, but if batteries continue following the exponential growth curve they've been on then there probably will be in the next decade or so (if not sooner).

There's a bit of a difference between 'some batteries' and at least five times the size of the largest battery ever build.

There is, and "some batteries" is an under count, however:

- The ratio between "some water" and "the pacific ocean" is a lot higher than 5

- On an exponential growth curve, a factor of 5 isn't all that much.

This probably isn't feasible for a data center being built today (although they could build solar to at least reduce their fossil fuel power generation needs during the day). But it probably will be for data centers being built in 5-10 years time.

Basically it's important to differentiate between "we can't quite do this yet" and "we'll likely never be able to do this". And powering a data center with rewnewables+batteries is definitely in the former category.

You will always use gas for cloudy days or maybe in winter. I didn’t claim solar should be the sole source of power.

Some batteries? Battery-backed solar has the highest cost as a baseload power source due to high intermittency of solar power.

Do you have a source for this? I constantly point out that several cloudy days in a row often happens but am rebutted with graphs of the declining costs of solar modules and batteries

https://www.sciencedirect.com/science/article/pii/S036054422... Figure 3

When used as the sole source of electricity solar clocks at 565 EUR/MWh. Nuclear, for comparison, is at 141 EUR/MWh.

Nobody is suggesting to use solar as the sole source of electricity. That would obviously be insane and then leads to silly numbers like the ones you gave.

A MWh of solar is about 30 Euro/MWh if you levelize the capital costs.

The article discusses why LCOE is not a good estimate of the costs. Yes, combining different sources lowers the total cost (by damping intermittency). For Denmark it's offshore wind and solar in a 7 to 1 ratio (plus natural gas or biomethane power plants).

> Nobody is suggesting to use solar as the sole source of electricity.

You weren't clear on what you propose.

https://has-electricity-decoupled-yet.strommarktberatung.de

Germany has a lot of solar. It's suppressing prices below the electricity price you would get with pure gas power generation.

The electricity price is not decoupled from the gas price because there's not enough energy storage, wind turbines, interconnections (that is the energy grid can't function without natural gas). The cost of building (and operating) all this goes into SLCOE.

I was very clear, even before the edit. I wrote "reduce the gas bills". Not "cut the gas line".

Konschubert, it's widely reported that German energy prices are some of the highest in the world.

Exempli gratia: https://www.cleanenergywire.org/news/germanys-household-powe...

And the proposal to fix this is more accounting games to transfer costs to different constituents.

Well, I found the proposal clear and your response confusing.

There are reasons it might not work, not least of all political. But a link to a paper picking a bone with LCoE is talking past GP.

I wanted to remind that "just add solar" comes with the costs that are rarely mentioned.

Batteries are good to cover evening peaks or maybe a while night. Not multiple days of low solar, that’s not economical.

good point, I guess the market says we have to poison the earth

No, it means that if you expect that sticking solar panels everywhere will solve all the problems for cheap (because solar panels are cheap), then you are up for a disappointment. I am of the opinion that handling out rose-tinted glasses will not go well.

The rest of the world seems to be doing exactly that, though: Sticking solar panels everywhere.

Reply on news.ycombinator.com