Remix.run Logo
epistasis a day ago

Well what I'm saying that is that on the Texas grid, solar and storage and wind are the cheapest energy, and being deployed in massive amounts because only on the Texas can an investor make money by providing the cheapest energy. (For most utilities, they take a fixed rate of profit and are incentivized to use the most expensive possible energy if they can get away with it.)

So Texas is not a laggard when it comes to clean energy, they are actually driving clean energy forward the most, because clean energy is the cheapest and most profitable energy. And that's despite Texas having natural gas that's insanely cheap right from Henry Hub.

What this tells me is that like most hyperscalers, Microsoft is not price sensitive on the electricity side, because energy costs are tiny compared to the massive capital costs of the GPUs. But why would they go this direction? What political influence would make Microsoft choose more expensive electricity, when in the past they've been fairly good at driving clean energy forward in their data center power choices, and they'd pay a premium on energy costs to go with clean energy?

tech_ken a day ago | parent | next [-]

I can think of a few angles that might have pushed them towards gas, mainly (a) they wanted on-demand generation cap, (b) they didn't want to get into the batteries game at the volume they'd require, or (c) they didn't want to deal with securing the space needed to produce 2.6GW of solar. Also yeah they're definitely not price-sensitive, any of the hyperscalars is more than happy to pay extra to get exactly what they want.

edit: for example that EIA list of new solar projects you linked indicates that the largest battery installations going up in '26 are all ~500MW, and that there are only four of them (of that size). I think the energy intensity of a multi-GW datacenter is the main reason that they're not going for solar here.

chasd00 a day ago | parent | prev | next [-]

> But why would they go this direction?

sibling comment has it, they want to do power generation on site and not connected to the grid and all the PITA that come with that. Further, they can pitch power independence to the locals which removes a big argument from the anti-datacenter crowd. Finally, the power gen i saw at Stargate in Abilene TX which was maybe 10 units (if that's what they're called) took up maybe 30 acres of land so they're not very big compared to the rest of the campus.

phil21 a day ago | parent | prev | next [-]

> What this tells me is that like most hyperscalers, Microsoft is not price sensitive on the electricity side,

What this should likely tell you, is that you are missing information and have an incomplete picture of the situation.

Or it could be MSFT just likes to spend extra money for no reason because they are simply stupid. I'm gonna go with the former though.

I'd be interested in all these behind the meter setups for large 24x7 loads that are being built using solar+battery though. I haven't heard of one personally, but I must be lacking information on the subject since you seem so certain these are common?

Symbiote a day ago | parent [-]

Maybe this counts, although it isn't built yet:

https://www.datacenter-forum.com/edora/eurowind-energy-and-e...

There are many datacenters in Denmark with decade-plus power purchase agreements from a specific wind farm, but with so much news recently I can't find if there's any operational with their own supply.

phil21 11 hours ago | parent [-]

Certainly interesting! Much closer to what I feel is actual green power w/o the greenwashing such as the PPA's other facilities tend to use while actually just buying from the grid as a whole to deal with the "hard" parts of reliable power.

My frustration is that reliable power is the expensive part. Anyone can stand up some nameplate capacity in renewables, net out their numbers and pretend they are 100% green power in a press release. All while drawing from the "free" battery the grid supplies them at night or seasonally and letting someone else deal with the dirty part of it all. It's just not very interesting to me, as it's mostly marketing.

This one does seem to be grid-tied, but is much closer to the vision of actually being powered down to that last 5% via on-site batteries and wind. Power is like many things - the last 5% of reliability (or 1%, whatever) is the hard and expensive solve.

I remain skeptical batteries are going to end up being a solution for major datacenter buildouts in most locations. But I've been wrong before, and would love to be wrong about this one perhaps the most!

HDThoreaun a day ago | parent | prev [-]

Solar makes sense for utility generation because demand goes down at night. For datacenter usage demand is effectively constant, so theyd need a fuckload of batteries which is where all the cost goes. It doesnt make sense to power 2.6 GW overnight fully on batteries. Much simpler, a.k.a faster, to just buy a plot next to an area with excess gas and build the whole thing there.