This article feels like the perfect distillation of a uniquely American problem.

Some weird tech startup proposing a novel solution based on a product that isn't even in it's production phase yet. Lots of pretty 3d renders and a wall of (what appears to be AI written) corpo-speak proposing some crazy technology that will revolutionize x.

It looks cool -- don't get me wrong -- but how is this going to get power online faster than just installing solar and batteries?

▲ Xylakant 3 hours ago | parent | next [-]

Did I miss something or does the article not even say how much gas they need as an input to generate the 42MW? I see they deride conventional turbines for needing cooling, but the reason they do is to increase the temp differential between hot and cold end of the turbine because some clever fellow named Carnot figured out that the amount of energy you can extract depends on this. Instead it seems that they just full-tilt run a supersonic turbine and blow the hot exhaust with all its energy into the air. So what’s the efficiency of this?

▲

azalemeth 2 hours ago | parent | next [-]

Indeed. My understanding of modern powerstation gas turbines is that they all basically run _at_ the Carnot efficiency eta = (1-T_cold / T_hot) and that rather than chasing marginal gains in how close to that theoretical limit you actually are the biggest differentiators are on maintenance intervals and reliability, which collectively have quite a large effect on eta...

▲

bob1029 2 hours ago | parent | prev [-]

> how much gas they need as an input to generate the 42MW

If you don't have a pipeline, the lower bound is something like 10 LNG tanker trucks per day for each turbine at 42MW. Natural gas is incredibly efficient to transport in liquid form so you could theoretically get away with this for a little while.

	▲	Xylakant an hour ago \| parent [-]
		The question I’m asking is slightly tangent to how to feed the required gas. It’s “How many MW of gas do I need to feed in to get one MW of electricity.” And they’re pointedly avoiding any statement about this.

▲ flohofwoe 3 hours ago | parent | prev | next [-]

> Some weird tech startup proposing a novel solution based on a product that isn't even in it's production phase yet

It's not even a novel solution, jet engines as stationary emergency 'power stations' goes back to at least the 1950s (e.g. search for TURBOLEKT here: https://de.wikipedia.org/wiki/VEB_Entwicklungsbau_Pirna).

▲ rgmerk 7 hours ago | parent | prev | next [-]

To be fair, you end up needing insane amounts of batteries if you want to run 24/7/365 just on solar, particularly if you insist on building your data centres in places with dark winters.

▲

mattmaroon 6 hours ago | parent | next [-]

Wind is better than solar in many places and somewhat reduces the need for batteries

	▲	butvacuum 4 hours ago \| parent [-]
		From ERCOT's stats- wind is complimentary. But, I can't find any hard data on intraday/hourly power usage for AI it seems reasonable to assume that night time use will be lower though. And so it doesn't have to be looked up: Wind seems to peak at dawn/dusk when solar is not delivery much power, solar peaks in line with air-conditioning load, and there's a miniscule amount of grid scale battery to hold up the grid during a short gap between solar and wind. The batteries are recharged with solar. At least that was the pattern this summer- I need to check now that it's winter.

▲

pjc50 an hour ago | parent | prev [-]

You can't run a turbine 24/7 either, they require maintenance windows.

	▲	nandomrumber an hour ago \| parent [-]
		Yes, you can run a turbine 24/7, just not 365 days a year. For gas turbines, n+1 is probably good enough for up to n=10, then n+2 and so on. If one breaks down or is undergoing maintenance you have a spare. Solar can’t work like this. Even if you build 2n solar capacity, you still have a not insignificant fraction of each day with no power. Meanwhile a gas turbine can be running continuously for week to months between service intervals. Just add batteries? Ok, but that’s no longer solar, and comes with not insignificant additional costs and maintenance etc.

▲ conradev 2 hours ago | parent | prev | next [-]

  It is thought up to 35 turbines were present on the site of xAI’s existing data center, generating 422MW of power.

That is a few square miles of solar panels, which I don't think is quicker to install than the 35 turbines.

https://www.datacenterdynamics.com/en/news/xai-removes-some-...

▲ rob74 3 hours ago | parent | prev | next [-]

Pah! Solar and batteries?! Have you been living under a rock for the last 12 months? Any startup that dares to suggest solutions based on solar and batteries (not to mention windmills) is sure to attract the ire of the Trump administration, so they'd better keep quiet and hope he doesn't notice them!

Actually, renewables seem to be such a no-no that the Boom blog even avoids mentioning them in the sentence "Meanwhile China is adding power capacity at a wartime pace — coal, gas, nuclear, everything" - even though China added overwhelmingly more renewable capacity last year than anything else: according to https://climateenergyfinance.org/wp-content/uploads/2025/03/..., solar increased by 43% from Feb. 2024 to Feb. 2025, wind increased by 17.6%, hydro by 3.5%, while thermal and nuclear increased by 3.9% and 6.9% respectively.

▲ ruined 4 hours ago | parent | prev | next [-]

by the way, china achieved the trendline in that comparison graph by installing solar and batteries

▲ shrubble 9 hours ago | parent | prev | next [-]

You get 42MW inside the footprint of what looks like 2 truck trailers, that you can park in the parking lot next to the electrical transformers. Virtually no permitting or installation required.

▲

_carbyau_ 9 hours ago | parent | next [-]

Yes...ish, I largely agree that the footprint is smaller per MW and quite a boon.

But 42MW energy doesn't come from nowhere, fuel needs to be considered. And there everyone has their own constraints.

The AI companies will likely care about $ and little else.

Engineers will point out that 42MW fuel takes up space and supply on an ongoing basis.

Other people will be worried about the externalities of burning 42MW of something vs solar panels and batteries etc.

You can't please all of the people.

	▲	ehnto 5 hours ago \| parent [-]
		Decent for large scale backup perhaps? Or remote plants (almost always mining in the middle of nowhere). Remote plants have fuel logistics already. Another fit might be somewhere like singapore which is very space poor but very trade connected. But they're currently building a ocean power cable to Australia where they will tap a massive solar farm or existing grid. It probably fits some use cases better than any alternatives, but for powering cities and suburbia I think renewables still make heaps of sense when space is available somewhere that can join the grid.

▲

javascriptfan69 9 hours ago | parent | prev | next [-]

I think a 42MW turbine might run into some permitting issues regarding safe noise levels.

▲

renewiltord 5 hours ago | parent | next [-]

Yes yes, we will surround entire turbine in piezoelectric substrate and extract energy from vibrations. It is solved problem. Then we use energy to distill fuel from CO2 in air, making it carbon neutral. Resulting fuel we will put in turbine. Zero loss generator. Can build it in cave with scraps.

	▲	sevenoftwelve 3 hours ago \| parent [-]
		Look, you can't write stuff like that any more. It took me three minutes to figure out you where joking.

▲

gorgoiler 7 hours ago | parent | prev [-]

Possibly, but I suspect mobile turbines (aircraft) are unquietened (noisy) by design because they don’t really need to be quiet at 35000ft.

Presumably a static turbine is minimizing noisy thrust in exchange for torque while also exhausting through an expansion chamber surrounded by deflective earthworks or some other shielding. (Although the one in the article is indeed all outside in the open.)

▲

adgjlsfhk1 5 hours ago | parent | next [-]

actually they've down much quieter in the past 40 years. e.g. the 787 dreamliner has wavy bits on the exit of the nozzle that reduce efficiency by 1% in exchange for quieter operation because making the engine quieter reduces the amount and weight of noise insulation in the cabin

	▲	gorgoiler 10 minutes ago \| parent [-]
		Oh of course, I didn’t mean to say that they weren’t as quiet as they need to be, only that there ought to be some obvious noise reduction opportunities once the requirement to be airborne is gone, and once the machine is being used for torque instead of thrust.

▲

mattmaroon 6 hours ago | parent | prev [-]

No, they’ve been intentionally designing them to be quieter for decades because they are in hearing distance for quite a lot of miles during takeoff and landing. I suspect you can better insulate one on land though since you’re less constrained on size and weight.

▲

lostlogin 2 hours ago | parent | prev | next [-]

> Virtually no permitting or installation required.

I hope that isn’t correct.

Noise, emissions, fuel storage, heat. There are issues that would have me annoyed if that thing appeared next door.

▲

lonelyasacloud 2 hours ago | parent | prev | next [-]

How does the fuel get to it?

Building roads and running tankers is expensive. Ditto pipelines unless very close to suitable sources.

Especially when the moment these go online at any scale the price of natural gas starts getting jacked even further.

▲

robocat an hour ago | parent | prev | next [-]

A site already needed backup generators.

Presumably there's some benefit there too.

▲

sroussey 2 hours ago | parent | prev [-]

Not getting permits, and no permits required are two different things.

Unless you got cash, then it’s the same.

▲ sho_hn 7 hours ago | parent | prev [-]

> This article feels like the perfect distillation of a uniquely American problem.

I think at this point LinkedIn culture is fairly globalized. Though America may be to blame for getting it there, largely via Deloitte & co originally. It's originally the language of managerialism.