from the something-you'd-see-in-the-ACME-catalog dept.
Cast-off turbines generate up to 48 MW of electricity apiece:
Faced with multi-year delays to secure grid power, US data center operators are deploying aeroderivative gas turbines — effectively retired commercial aircraft engines bolted into trailers — to keep AI infrastructure online.
According to IEEE Spectrum, facilities in Texas are already spinning up units based on General Electric's CF6-80C2 and LM6000, the same turbine cores once found on 767s and Airbus A310s. Vendors like ProEnergy and Mitsubishi Power have turned these into modular, fast-start generators capable of delivering 48 megawatts apiece, enough to support a large AI cluster while utility-scale infrastructure lags.
Fast, loud, and anything but elegant, these "bridging power" units come from vendors like ProEnergy, which offers trailerized turbines built around ex-aviation cores that can spin up in minutes to meet energy demand. Meanwhile, Mitsubishi Power's FT8 MOBILEPAC, which derives from Pratt & Whitney jet engines, delivers a similar output in a self-contained footprint designed for fast deployment.
While this might not be the cheapest, and certainly not the cleanest, way to power racks, it's a viable stopgap for companies racing to hit AI milestones while local substations and modular nuclear power deployments remain years away.
[...] In one of the more visible examples, OpenAI's parent group is deploying nearly 30 LM2500XPRESS units at a facility near Abilene, Texas, as part of its multi-billion-dollar Stargate project. Each unit spins up to 34 megawatts, fast enough to cold-start servers in under ten minutes.
Also see: Data Centers Look to Old Airplane Engines for Power
« Breaking Down Rare Earth Element Magnets for Recycling | Meet Mico, Microsoft's AI Version of Clippy »
Related Stories
A blog post covers why datacenters in space are a terrible, horrible, no good idea. Thermal management is just the beginning of the long list of challenges which make space an inferior environment for data centers.
In the interests of clarity, I am a former NASA engineer/scientist with a PhD in space electronics. I also worked at Google for 10 years, in various parts of the company including YouTube and the bit of Cloud responsible for deploying AI capacity, so I'm quite well placed to have an opinion here.
The short version: this is an absolutely terrible idea, and really makes zero sense whatsoever. There are multiple reasons for this, but they all amount to saying that the kind of electronics needed to make a datacenter work, particularly a datacenter deploying AI capacity in the form of GPUs and TPUs, is exactly the opposite of what works in space. If you've not worked specifically in this area before, I'll caution against making gut assumptions, because the reality of making space hardware actually function in space is not necessarily intuitively obvious.
Previously:
(2025) The Data Center Resistance Has Arrived
(2025) Microsoft: the Company Doesn't Have Enough Electricity to Install All the AI GPUs in its Inventory
(2025) China Submerges a Data Center in the Ocean to Conserve Water, is That Even a Good Idea?
(2025) Data Centers Turn to Commercial Aircraft Jet Engines Bolted Onto Trailers as AI Power Crunch Bites
(2025) The Real (Economic) AI Apocalypse is Nigh
(2025) Real Datacenter Emissions Are A Dirty Secret
... and more.
(Score: 4, Informative) by jelizondo on Saturday October 25, @05:45PM (4 children)
There is already opposition [npr.org] to data centers in several states due to higher electricity costs, excessive use of water and noise.
Imagine a freakin' jet engine running 24/7 in your neighborhood!
(Score: 5, Informative) by crm114 on Saturday October 25, @06:07PM (1 child)
(Score: 2) by krishnoid on Saturday October 25, @06:16PM
Which is a favorite hangout place for hot rodder billionaires who are intermittently revving their private jet engines 24/7.
There used to be people arguing that electric vehicles would overwhelm the grid. Probably still are.
(Score: 4, Informative) by RS3 on Saturday October 25, @10:13PM
From a web search:
"Typical sound levels Near the unit:
A typical gas turbine package may be guaranteed to meet an average of 85 dBA at 1 meter. [slightly louder than a vacuum cleaner]
A small micro gas turbine can produce 67-109 dBC.
[Adding: dBA (or dB(A)) is sound pressure weighted on a scale that roughly matches average human hearing perception. dBC is no weighting- pure sound power.]
The maximum sound pressure level can be as high as 92.5 dB(A) near the turbine inlet plenum.
At a distance (with silencing):
With an enclosure and silencers, the sound level at the property line can be significantly reduced, for example, to 40-45 dBA in some cases, especially in quiet conditions.
Noise levels can drop considerably with distance, with levels around 32-40 dB measured at neighboring properties in one case. "
Very reasonable, but certainly should not be located near residences (but maybe underground?).
(Score: 3, Informative) by mcgrew on Saturday October 25, @11:41PM
Indeed. I worked on the flightline in the USAF half a century ago, and the gasoline MD-3 ground power generators for the C-141 cargo planes were relatively quiet; quieter than a Harley motorcycle, anyway. But the dash sixties that had F-15 engines to power the giant C-5As were LOUD.
What did you expect when you voted for a convicted felon, peace and rainbows?
(Score: 5, Informative) by VLM on Saturday October 25, @06:17PM (27 children)
I got to visit a Ma Bell central office less than a decade after divestiture at a new at that time RBOC (not a scared-straight program, if it was it clearly didn't work LOL it was a rare EE class field trip as the prof had serious "connections") and we visited the room with their turbine generator. Very small source of a very large amount of power. What I recall is rows of hearing protection equipment and roughly infinite signs about noise levels and some kind of "If the yellow light is on you must evac within X seconds or wear hearing pro" system.
The CO manager was very proud that the model of turbine could burn any liquid or gas fuel it didn't care about. In the long run you don't want soot on the blades or something but it was flexifuel before flexifuel was "cool". We asked him to start it up but it would have been too many OSHA violations so he just kind of laughed.
The device was small, like the size of a small desk for a couple MW. The room and equipment in it was pretty large so it might be a net loss compared to a compact outdoor diesel. Air (and exhaust) handling is not cheap or small nor light. Its also loud.
Something to keep in mind is industrial chillers are not as quiet as a luxury house refrigerator, either. COs (and data centers) can be pretty darn loud almost up to the point of needing hearing pro anyway even without a turbine generator.
Even water cooled mainframes are loud because everything ELSE around them is air cooled.
(Score: 5, Funny) by JoeMerchant on Saturday October 25, @07:59PM (23 children)
When fantasizing about ways to power an EV, a used market APU from a commercial jet came to mind. Put that sucker on a trailer and fire it up whenever you need more range. Bonus: you can bleed vacuum for more cold air than you'll ever know what to do with. And you'll never need a horn.
🌻🌻🌻 [google.com]
(Score: 2, Interesting) by anubi on Saturday October 25, @09:08PM (9 children)
I will never need a horn?
I guess my ears will be completely shot.
If it wasn't for the noise, seems jet engines would make an ideal power source for data centers and large complexes needing a lot of thermal control.
A problem I see with jet engines is their speed, Wouldn't the frequency of a conventional six pole three phase generator be way over conventional frequency? I take it the engines will have to drive alternators to get DC output. I could see alternator rotors made using another alternator on the same shaft that the main armature winding uses instead of the slip rings used in automotive alternators so as to completely eliminate brush contacts.
On paper, interesting...ibe heard of selsyn motors doing something like this to eliminate slip rings and brushes to the rotor.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 5, Interesting) by RS3 on Saturday October 25, @09:50PM (4 children)
They get geared down.
Most people think of "jet engine" as in what's on a jet plane, which we know are loud. But these generators would use a "gas turbine", like what's used to power many helicopters, battle tanks, and even trains (locomotives). They're nowhere nearly as loud as a thrust producing jet engine. In a way they seem somewhat quieter than piston engine driven helicopters, but I'm no expert, and the rotor blade noise is the overwhelming sound. Point is you don't have the roar of the thrust, which of course includes the bypass fan in planes. Also you can use muffling on gas turbine exhaust, which you obviously can't on a jet plane without clobbering thrust.
Planes typically use 400 Hz, but also AFAIK use generators because they're considered more reliable (no diodes to fail). Again, not expert, but did a bit of search. And 400 Hz would be better for overall system efficiency, but I'm not sure if anyone is doing that. I think I remember a spec for some large computer, maybe from long ago, that used 400 Hz power... Yep, Cray-1 used 400 Hz, and I'm sure other mainframes did too. Not sure how they got that 400 Hz, maybe a motor-alternator set (?).
(Score: 4, Interesting) by anubi on Saturday October 25, @11:34PM (3 children)
My problem with gears is that they are a mechanical wear item. Planetary gears help a lot, but still there is that metal-on-metal sliding that goes on during meshing.
I concur with your 400 Hz suggestion. I'd go 400Hz three phase as a lot of transformer manufacturers know all about it. Dust off those old aircraft power distribution technologies for a new use. They already are tooled for the smaller transformers, as well as 400 Hz is clean power as far as EMI/Harmonics go...and the magnetic components are much smaller, lighter, and more efficient, now that we have new alloys of core materials with low hysteresis loss.
At the point of load, I'd probably consider a three phase delta to six phase "star" secondary for ease of filtering to DC. Could even go to 12 phase using creative transformer winding to minimize ripple and maybe improve power factor. I haven't looked into that, but don't put it past me to feed this concept into my simulator and see what I get. If it looks promising, build one and tweak it IRL. It looks linear. It oughta scale.
And probably consider mag-amp designs in the voltage regulators ( some PC power supplies use mag-amp techniques in the output filtering chokes - they take on some of the characteristics of the "swinging choke" used to compensate for copper loss in some of those old vacuum tube power supply filters. Weight was not a prime concern, but simplicity and reliability were.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 3, Interesting) by RS3 on Sunday October 26, @12:56AM (2 children)
Cool and agree on all points. I know about those interesting inductors in PC power supply secondaries, in older ones anyway. I never thought of them as a mag-amp but it makes sense. Interesting permeability curve. You might know they used a similar thing in CRT TVs. It's been a bit too long but IIRC they had something to do with smoothing the flyback output that was used to drive the horizontal deflection yoke coil. Or maybe some other thing between vertical and horizontal- I know it was deflection compensation of some type. Oh yeah, just remembered inductors with a permanent magnet core in some TVs.
I didn't know about the "swinging choke" compensating for copper loss. I have to look into that much more. I've dabbled in that stuff most of my life, and of course vacuum tubes continue in a niche market for very high-end audiophiles and of course guitar amps. And there's no substitute for tubes (sorry EU- "valves" (thermionic emission valves if you will?)) in the output, with big fat iron transformers. There are pretty good simulators but it's just not the same.
You may know of "saturable reactor" (from mag-amp of course) transformers used for regulation. I have a few, including two in an old Sola mainframe power supply. They have a special winding that connects to a capacitor and AFAIK it sort of makes a soft clipping. "Ferro-resonant" was the term I just remembered. (that's from a long time ago, memory still works!)
Back to gears, I don't know, I've seen automatic transmissions go more than 300,000 miles. More highway (less shifting) of course but the gears themselves seem to last well. Good (and older) differentials in cars and trucks can last pretty much forever if aligned properly and oil changed occasionally. I think the only thing that'll hurt gears like those is loose bolts and/or worn bearings that allow gear misalignment.
A quick web search says gas turbines can be on the order of 3000 - 3600 RPM for power generation, therefore perfect for alternators. Some are as high as 20-25 K RPM, which could be pretty easily geared down if they're more powerful (thinking big).
(Score: 4, Insightful) by Unixnut on Sunday October 26, @11:23PM
I was always told that well oiled gears should hardly wear at all (after all the oil is there to prevent metal-metal contact), which is one of the reasons that an engine oil pump usually outlasts everything else. Usually any failures are due to a failure of filtering (allowing foreign object damage), neglecting oil changes/quality, or a loss of oil resulting in excessive wear.
Gears with jet turbines have been used since the beginning, with turboprop and turboshaft engines for example (some used epicyclic gearing, others standard straight cut ones), and gear wear was not a major issue AFAIK. Only thing that I've heard noted is the noise from the gears spinning at high speed can be quite loud (probably why pretty much everyone moved from straight cut to epicyclic gearing).
Also, for JoeMerchant: The "Jet turbine attached to an EV" has been thought of already. I remember EV hybrids being touted with a microturbine instead of a piston engine. I think it didn't catch on because turbine efficiency improves with size, and as such the smaller the turbine the less efficient it is. At the microturbine size, a conventional piston engine seems to be a better proposal than a turboshaft engine.
The one such concept car that stuck in my mind was from Jaguar, and it looks like it has a wiki page [wikipedia.org] about it.
(Score: 3, Interesting) by anubi on Monday October 27, @12:33AM
"I didn't know about the "swinging choke" compensating for copper loss."
As current goes up, voltage loss across the inductor increases as current increases, but the inductor core begins to saturate and it's reactance goes down.
The TV coil is probably the "pincushion" coil, which uses both vertical and horizontal correction to compensate for projection onto a relatively flat viewing plane. This became obvious with the newer and thinner wide-deflection angle CRTs. Didn't show up much on long oscilloscope tubes.
Those coils wound on a magnet were because they wanted the coil to work as if the tube DC plate current didn't exist. They could vary plate current with biasing, so the magnet and tube bias was chosen to null out the asymmetry of the hysteresis curve. Again trying to compensate for projection onto a "flat" viewing screen.
Hope that helps. That was my understanding of it as told to me by an old service tech I asked the same question to. I was completely puzzled by the concept of a "polarized" coil, even though I never had one go bad.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 2) by JoeMerchant on Saturday October 25, @11:55PM
APUs include a 440hz generator.
🌻🌻🌻 [google.com]
(Score: 5, Informative) by VLM on Sunday October 26, @02:18PM (2 children)
The biggest VFD I'm aware of is 100 MW from ABB although thats a special one time off thing that made the news at NASA.
Technologically we're getting VERY close to installing VFDs at power plants at the supply side. Like those Honda generators that run at random speeds to generate DC and the onboard VFD/UPS/Inverter electronics turn it into perfect 120 single phase sine waves.
Things are going to get weird in the electrical power world when we can replace giant 60 Hz AC power transformers with a little box full of silicon. We've already replaced all the small ones on the demand side (the little wall warts we used to have before switching supplies).
Coincidentally steam turbines big enough for an entire coal plant are usually direct coupled, there's a lot of design motivation there. No gears at the powerco, usually.
One big problem with turbines is look at the compressor map the surge line (where you essentially stall out the blades and it converts from a pump to an egg beater) is right next to the operating line which is usually right next to the peak efficiency line which is usually next to the max turbine inlet temp line. You can't really "hot rod" a turbine because they're usually maxed out from every direction at precisely one operating point. They don't throttle well at all. You can engineer them to have poorer performance in exchange for better throttling performance.
Compressor surge is pretty entertaining for non-technical viewers. It looks very much like a hollywood explosion with fire and burning liquid fuel shooting out, sometimes out both ends simultaneously. They're built to do that but not repeatedly. You can blow turbine vanes off if it sets up an oscillation and they wiggle back and forth every revolution. Anyway the point is you can make a "stable" steam engine or internal combustion engine using minimal mechanical regulators if any, but jets "pretty much need engine computers" you can make a hideously complicated mechanical/hydraulic one but generally turbines need computers and they tend to only work really well at one operating point.
Interesting to think about optimizing a backup generator turbine. Don't care about weight, don't care about every tiny percent of efficiency, it aint getting maintained never until after it fails during an outage, all kinds of interesting engineering tradeoffs.
(Score: 1) by anubi on Sunday October 26, @11:16PM (1 child)
That is quite interesting.
I got the opportunity to study a Chevron gas-turbine oil tanker up close, many years ago, and really blew it, as I was so ignorant, and confused over what I was supposed to be doing. I was with a senior engineer diagnosing erratic engine shutdowns. My lack of experience was quite evident. I guess I was like a three year old trying to help, while everything I did just complicated everything.
I had no idea how serious this problem really was, or how to go about diagnosis.
One thing I did get is that the "gearing" between turbine and propeller was the ratio of the number of poles in the generator and poles on the motor. The "clutch" was whether the motor had sufficient excitation to run synchronously or asynchronous. The amount of mechanical coupling. It was most efficient if synchronous, and when running synchronously, it's power factor could be tweaked by excitation current.
We took advantage that the tanker mostly cruised at constant velocity, so everything was optimized for that. I believe "Reverse" was just swapping two of the three phases of motor power.
Such elegance. Back in the early 80's.
I was overwhelmed. I mean I was just getting used to 7400 TTL and thought 4000 CMOS was black art. Most of what I was familiar with was those Markus books of analog circuit designs. Seems I did know about every analog chip Burr-Brown, National Semiconductor, or Analog Devices made, and a few others. The Intel 8080 had just come out. $400. I was assembling a kit of IMSAI 8080.
I knew nothing about turbines, thermodynamics, and the operation of physical-inertial systems.
I just remembered all the frustration I experienced being so damm ignorant, as I read your reply, and actually comprehended it. It made me aware of the gravity of the matter Chevron was facing, and the fear and frustration the ship's captain was experiencing. While I felt so useless and ignorant.
It was humbling for me to read your post, sure brought back a lot of embarrassing memories for me and the ignorance and arrogance of my younger days, and why I value understanding so much ( aka "tinkering" with things until getting it down pat - what makes it tick? ). It sure was no fun being ignorant and letting everyone else that believed in me down
Thanks for posting that. It tripped off not only a reflection of turbine operation, but also a self examination and gratefulness to others for trying to help me when I was too ignorant to comprehend.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 2, Interesting) by anubi on Monday October 27, @12:49AM
Oh, incidentally, discussions like this is exactly why I treasure membership in S/N so much. Since my layoff/retirement from the industry, I have found it nearly impossible to find interesting discussions, and the major social sites discuss nothing I find worthwhile to participate in. I am almost 75, so it's a little late in the game for me to show up at personnel.
There isn't really anything else I have much interest in.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 2) by mcgrew on Saturday October 25, @11:46PM (12 children)
But that removes one of the many advantages of EVs. These days, there are two reasons for a noisy vehicle: 1. It's a motorcycle; noise is a safety feature. 2. you can't afford to get your car's exhaust system repaired.
What did you expect when you voted for a convicted felon, peace and rainbows?
(Score: 2) by JoeMerchant on Saturday October 25, @11:53PM (11 children)
This APU on a trailer is planned for an electric Miata, pretty much a four wheel motorcycle when it comes to visibility to SUVs.
🌻🌻🌻 [google.com]
(Score: 2) by RS3 on Sunday October 26, @01:02AM (10 children)
Years ago I had a college friend who ended up working for Pratt & Whitney. He seriously wanted me to put a gas turbine in my '68 Firebird. Incredible power to weight. Remember there was an experimental turbine car way back- Chrysler 1963 or so? I keep thinking about doing that...
So would the APU be onboard the Miata? Or just to charge it?
For visibility you could do like boats- have a mast with flashing lights. :)
My main car isn't much bigger than Miata. I rely on my attentiveness and the car's turbo.
(Score: 2, Interesting) by Anonymous Coward on Sunday October 26, @03:26AM (4 children)
> Remember there was an experimental turbine car way back- Chrysler
Also a variety of race cars, including the turbine that nearly won the Indy 500 -https://en.wikipedia.org/wiki/STP-Paxton_Turbocar and this Lotus nearly won the next year, https://en.wikipedia.org/wiki/Lotus_56 [wikipedia.org]
From memory, they suffered from slow response to changes in throttle. OK at Indy where the cars lap at (or near) full throttle all the time. No good for road courses, or city driving where engine speed needs to change all the time.
This road racer, https://en.wikipedia.org/wiki/Howmet_TX [wikipedia.org] used a waste gate to keep the turbine speed up for quicker throttle response...but on a couple of occasions it jammed, effectively leaving the throttle stuck wide open (and causing a crash). I saw these run at Watkins Glen and much later when one of them was restored and appeared at the Goodwood Festival of Speed, very impressive.
The proposed trailer charger is a good application--runs full throttle until charging is done, then shuts off.
(Score: 4, Informative) by RS3 on Sunday October 26, @03:59AM (3 children)
All very interesting. I need to learn more about gas turbines. I would have thought they could spool up pretty fast. Maybe not fast enough for racing. My current main car has a turbo. It certainly has lag, and it's never the same- I wonder what the ECM is doing. But for sure you can pre-charge it by the good old hold the brake and give it some gas before letting off the brake if for some reason I need a quick start. If the engine is above 2-3 k rpm, throttle and turbo response is usually pretty good. It would be okay for racing providing all other cars had similar engines / turbos / ECM etc.
I'm just starting to think this through. You may (probably) know cars have more and more transmission gears. And of course there are now CVT. Reliability seems to be problematic, but that aside, the idea is the engine doesn't have to vary RPM as much and can be tuned for maximum efficiency in that narrow range. Point being- maybe a gas turbine would work much better with modern transmissions.
And/or do like locomotives where the gas turbine drives a generator which powers the traction motors. With today's motors and controllers it might be a very viable system.
BTW, I'm not trying to usurp BEVs- I think and hope they're the future, but combustion engines are likely to be around for a while and maybe a gas turbine would emit less CO2?
Speaking of turbines, do y'all know that Titanic had 2 4-cylinder steam engines, but also had a 3rd, turbine, powered by the piston engines' exhaust steam? And that the turbine put out slightly more horsepower than each of the piston engines - powered by the low pressure exhaust steam.
(Score: 1) by anubi on Sunday October 26, @11:30PM (2 children)
Did that turbine power the ship's generator?
I know a movie is not reality, but I was taken by those images of the ship sinking, all engine compartments obviously flooded, yet the lights were still on as it went down.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 2) by RS3 on Monday October 27, @12:23AM (1 child)
No, and largely because you need to vary its speed, which wouldn't do well with electricity generation.
Search AI says there were 4 steam-driven main 400 KW generators (100VDC), and 2 auxiliary generators, also steam-driven. So as long as they had steam they had electricity.
The aux generators powered emergency lights, which were on separate circuits.
The workers did all they could to keep the steam boilers running, not only to keep lights on, but to keep the steam-driven bilge pumps running.
(Score: 2, Interesting) by anubi on Monday October 27, @02:56AM
I had a feeling, given the era, they would be DC, given speed control could be done by varying field current via rheostat. Again speculation. I haven't looked it up.
I was thinking the hotel load on the generators would be reasonably constant, and they could control both generator field and steam supply. ( Armchair quarterbacking again...). I oughta do some research on how the Titanic worked. The Chevron supertanker I got to "help" fix ( "help" in quotes - the senior engineer knew what he was doing - I didn't . Basically, I watched ) is a marvel of design elegance.
But I learned a lot and had a better feel for practical industrial designs for stuff that works in the field, which is an extension for that which works as a mathematical model on paper ( but omits a helluva lotta stuff to simplify elementary classroom discussions ). The devil's in the details. Ignore them with great peril, as I later learned.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 3, Interesting) by JoeMerchant on Sunday October 26, @02:28PM (1 child)
The APU plan was to put it on a trailer, so you've got the APU and its fuel back there providing battery charging when needed, and if you ever want to just run around town for short trips you can ditch the trailer.
Power definitely helps, I turbocharged the 1.6 liter Miata in 1997, then did a 3.0V6 swap in 2021... it enables a different kind of defensive driving.
🌻🌻🌻 [google.com]
(Score: 3, Informative) by JoeMerchant on Monday October 27, @01:56AM
The impractical side, unless you enjoy those noise cancelling earmuffs with bluetooth audio inside:
🌻🌻🌻 [google.com]
(Score: 4, Informative) by VLM on Sunday October 26, @02:45PM (2 children)
They goofed around for a quarter century in total IIRC and gave up in '79 it won't be happening for emissions reasons.
Turbine efficiency and power out go up fast with exhaust temp and, unfortunately, heat plus oxygen plus nitrogen equals massive NOx emissions and there seems to be no way to make a gas turbine car that would pass anything near emissions.
Its like comparing lemons and limes, they're both citrus, but... AFAIK the euros seem to think 0.4 g/KWh or better is "good" whereas the Americans measure PPM of the exhaust that boils down to something like "grams of NOx per KWh" I would guess.
GE used to advertise a gas turbine that broke the 60% efficient level, twice as good as burning gasoline in an internal combustion engine, mostly by running hot as hell, and inevitably heat+oxygen+nitrogen equals nitrogen oxides which give the EPA fits.
For 200 F-22 the emissions don't really matter, its like having one extra thunderstorm per year or something like that, but for 200 million cars it would become quite an issue.
(Score: 2) by RS3 on Sunday October 26, @03:24PM
Oh, I didn't do that research. I wonder how bad it is, like could it be catalyzed? Maybe like Diesels are with DEF injection, maybe into the combusters or just after + exhaust cat? Or some other similar chemical process + cat converter?
(Score: 2) by RS3 on Sunday October 26, @03:32PM
Okay, just did some research and it looks like DEF injection is done, post combustion, then cat converter. At least that's what an AI thinks, I haven't done further research...
(Score: 2) by driverless on Sunday October 26, @05:31AM (2 children)
Similar story here, 1980s visit to a data centre that had a reinforced-concrete underground room with two fighter-aircraft turbines in it, possibly from A-4s but it was too long ago to remember exactly. They had all sorts of safety protocols in place to make sure they couldn't be fired up with anyone present.
(Score: 2) by VLM on Sunday October 26, @01:59PM (1 child)
Maybe. The easy way would be something like a Embraer Tucano (or supertucano although those are too new). Its turbo prop so unscrew the airplane propeller and screw on the generator alternator and the conversion is done.
Or an old Huey helicopter engine is probably "cheap"
(Score: 2) by driverless on Monday October 27, @08:26AM
Yeah, you're right, looked up the details and the dates are wrong for them to be A-4 engines. Whatever they were there were two of them and they were ex-military, and were used because it was what was available at that moment.
(Score: 5, Insightful) by sjames on Saturday October 25, @07:43PM (5 children)
Right now, AI is practically free (or actually free) to the user in order to promote it and get people hooked on it. But given how much power it consumes, that can't last. Soon Wall Street will get itchy and so will come "Th Great Monitization" where 90% of the users will say "It's cool, but not THAT cool". With adoption falling well short of the pundit's breathless estimates, the price for those who are determined to use it will soar just to break even. And so we'll get the next big tech bust.
(Score: 0) by Anonymous Coward on Saturday October 25, @08:04PM
And then we'll see the next big tech bailouts. Our Wall Street sociopaths have nothing to fear
And we can enjoy the bright lights and soothing rumble of distant turbines
(Score: 4, Interesting) by Anonymous Coward on Saturday October 25, @08:05PM (3 children)
> Right now, AI is practically free (or actually free) to the user in order to promote it and get people hooked on it. But given how much power it consumes, that can't last. Soon Wall Street will get itchy and so will come "Th Great Monitization" where 90% of the users will say "It's cool, but not THAT cool".
Right now (according to a panel discussion I just sat in on) the crash (literally) is happening to the driverless taxi business. Waymo and the others have been developing long enough that now it's crunch time. Prices will go up, cost saving corners are being cut.
What came out is that the public accident stats that we read all leave out the rear end crashes where the AI car phantom-brakes and causes a crash (or a chain reaction crash in heavy traffic). These are about half of all the crashes that involve self driving cars. Also, don't be fooled by accident statistics that compare the AI taxis with Uber and Lyft--the ride sharing services have accident rates several times higher than the general population (the drivers are trying to keep up with the reporting required through their phone--heavily distracted). The only thing that keeps this out of the news is low speeds in city traffic, limiting the damage and injury.
The other piece that is kept hidden is the remote human operators--many, many of them, required to get the AI cars out of situations that the software can't handle. Even when the operation centers are local, the latency is such that the cars can not be driven anywhere near highway speeds, and some of the remote operators are in Asia (cheaper) where latency is a second or more.
(Score: 3, Funny) by anubi on Saturday October 25, @11:57PM (2 children)
On Overcoming Neophobia:
Feed the cat free kibble.
Get it used to eating it, and liking it.
( Free. Anonymous. No registration.)
Now, start putting the kibble in an open box.
( Get it accustomed to trivial terms and conditions )
Replace the box with a trap.
( Change the terms and conditions )
( It's now your cat ).
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 1, Funny) by Anonymous Coward on Sunday October 26, @12:59AM (1 child)
Depends on what you are trying to trap!
When a mama skunk and three babies showed up in our semi-rural yard, I just put out an open box trap, no bait. One baby wandered in, I guess out of curiosity, and was trapped. The next day I transported it (under a plastic tarp!) to a remote wooded area, then reset the trap. Only took four nights to get all three. Mama never came back.
Is there a moral to this story?
(Score: 1) by anubi on Sunday October 26, @01:40AM
In my case, they are trapping humans.
They call it "customer lock-in".
Moral:
Mama skunk is smart. She only had to lose one litter of kittens to learn about the trap.
Most humans, well, not so smart. They don't recognize the trap yet and get snared into signing things and pressing buttons they shouldn't, then the debt collectors start showing up.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 5, Insightful) by Undefined on Saturday October 25, @08:02PM (4 children)
In the meantime, the human brain only consumes about 5 watts.
Perhaps LLMs aren't the way to go about this particular pursuit.
I use a dedicated preprocessor to elaborate abbreviations.
Hover to reveal elaborations.
(Score: 1, Touché) by Anonymous Coward on Saturday October 25, @09:16PM
As well as it runs on farmed fuel!
(Score: 1, Insightful) by Anonymous Coward on Saturday October 25, @09:59PM (1 child)
Not sure where you got that- quick web search turns up many references saying more like 20 watts. Which still isn't a lot of course.
So you're suggesting we would do much better to somehow connect some kind of biological brain? I like it. Star Trek episode "Spock's Brain" [wikipedia.org].
(Score: 2) by Undefined on Tuesday October 28, @01:04PM
Estimates vary from 5 watts to 20 watts; hard to say what's right, but we're still in the ballpark no matter what. Even if it was 100 watts, it's still ridiculously less than the corporate LLMs are chewing up. Though a shout out to local LLMs as they consume much, much less per instance.
Also, the brain does a lot of things an artificial brain does not have to that aren't involved in thought per se. So some of that energy doesn't apply to what it takes to think. Examples include regulating heart rate, breathing, heart rate, blood pressure, balance, touch, pain, temperature and resource control (such as glucose and water.)
So whatever energy is required there doesn't really count towards what a putative AI would need to do — and some things we can do with tech, like vision and sound generation are much more efficient already than how the brain does it, so at least some of that can be offloaded to subprocess hardware and software.
I use a dedicated preprocessor to elaborate abbreviations.
Hover to reveal elaborations.
(Score: 1) by anubi on Monday October 27, @03:29AM
But the AI has the knowledge ( memory ) of nearly everything found on the internet and then some.
It can also process human prompts, parsing our native languages, quite well. Especially given all the nuances of our language, where the meanings of words are often a function of context.
It can also reply to us in the format we understand best...our native language, again with the proper nuances.
The only thing I see lacking seems to be cognition of the situation ( very debatable ), love, empathy.
I think of it as a super-knowledgeable but "agentic" entity. "Agentic" in the sense of Stanley Milgram's "Obedience to Authority". In the agentic state, humans are obedient to the giver of orders, executing orders with no regard for morals or ethics if so instructed. Typically seen in military hierarchies, where they must follow regulations.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 4, Interesting) by chucky on Saturday October 25, @08:08PM (3 children)
So, there’s a rotating turbine, attached to a generator I suppose. What happens with all the heat these engines produce? And airplane engines produce thrust, where is this going? I didn’t read very carefully, but didn’t see anything about co-generation. How does it really look?
(Score: 3, Informative) by RS3 on Saturday October 25, @10:04PM
As I mention above, these would be gas turbines, not thrust producers.
Yes, they'll make heat, which will obviously heat the atmosphere. They are typically 60+% efficient so they'll make less waste heat than many other power generation methods.
(Score: 4, Interesting) by mcgrew on Saturday October 25, @11:55PM
These generators aren't some new high tech stuff, they have been powering parked aircraft being serviced for almost as long as there have been aircraft; at least since WWII. Both commercial and military jets use them. Google is a useful tool. [ctipft.com] This looks nothing like the ones we had in 1971.
What did you expect when you voted for a convicted felon, peace and rainbows?
(Score: 2) by VLM on Sunday October 26, @02:02PM
Strange idea I have not checked the math on because I'm still half asleep:
The last thing a hot data center needs is cogeneration heat. However: Ammonia vapor refrigeration. Doesn't require high temps would seem to be a good use for turbine waste heat at a data center.
The capital costs might be too icky to consider (might be cheaper to install a larger alternator and run the same cooling system off the electricity)
(Score: 5, Insightful) by Gaaark on Saturday October 25, @08:34PM (1 child)
I say "Go ahead and approve this.... so long as it is right next to the CEO's Primary Residence where they HAVE to live and sleep so long as the jet engines are powered up."
Other than that, feck off. We have done without AI for how many generations now? Just like cell-phones, AI is NOT a necessity. But hey, lets create more noise pollution and push more heat into the environment just to make some ass-hat richer.
Yeah.
....
fuck.
--- Please remind me if I haven't been civil to you: I'm channeling MDC. I have always been here. ---Gaaark 2.0 --
(Score: 1, Insightful) by Anonymous Coward on Saturday October 25, @09:45PM
You're silly. Remember "drill baby drill" and how imperative it was to rapidly approve all those fracking permits for "energy independence and national security"? Well, except for those that would be in the general region of the CEO of the energy companies? [yahoo.com]
(Score: 5, Insightful) by SomeGuy on Saturday October 25, @09:28PM (2 children)
So what are people even DOING with magical "AI" that is so precious that requires burning up entire solar systems just to fuel it?
They used it to write a bullshit filled letter?
Creating moar fake porn?
Creating "original" pictures and video that reek of copyright violations?
Really? What? And is it worth it? I don't think so.
(Score: 1) by anubi on Sunday October 26, @02:08AM (1 child)
We haven't seen the end of this yet.
Apparently, they are already running out of training data.
Apparently, information retrieval isn't all that resource intensive. Training is.
I think this does have the potential to be a great public resource. The Oracle. Filled with everything known.
Hopefully remaining publicly available, like the public library. The caveat is we become dependent on it yet none of us can maintain it: A meme of a lot of science fiction stories.
That's just my take on it. I may think it's great for organizing, storing, and generating reports, I still question it's "wisdom", as that aspect is still not evident to me, but, like a compiler, it's a useful tool.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 1, Touché) by Anonymous Coward on Sunday October 26, @03:36AM
> I think this does have the potential to be a great public resource. The Oracle. Filled with everything known.
Best to check this out with a question to The Internet Oracle--
https://internetoracle.org/gateway.cgi [internetoracle.org]
(Score: 3, Informative) by Sourcery42 on Monday October 27, @05:29PM
The reality is the LM6000 is so commonly used in combined heat and power applications that Baker Hughes has a fleet of the damn things so you can swap to a rental while your LM6000 is out for overhaul with minimal downtime https://www.bakerhughes.com/gas-turbines/aeroderivative-technology/lm6000pf [bakerhughes.com] This is a pretty natural extension of a capability that already existed. They aren't bolting jet engines to flatbeds to generate power here, the designed intent of these machines is combustion turbines for power generation.
Wind, solar, or nuclear it ain't, but a nice little CHP burning natural gas and co-located with the data center could be quite efficient - nice base load, cuts out transmission losses. Reading between the lines I doubt these turbines are getting installed with the heat recovery steam generator, steam turbine, combustion promoter, and selective catalytic reduction to make them high efficiency and low emissions, but those details are not present.
I guess the point is more, "look how much energy they're using!" I hope society is getting a good return on investment tying up all that capital and using those megawatts, but I am not convinced.