The University of Rhode Island’s AI lab estimates that GPT-5 averages just over 18 Wh per query, so putting all of ChatGPT’s reported 2.5 billion requests a day through the model could see energy usage as high as 45 GWh.
A daily energy use of 45 GWh is enormous. A typical modern nuclear power plant produces between 1 and 1.6 GW of electricity per reactor per hour, so data centers running OpenAI’s GPT-5 at 18 Wh per query could require the power equivalent of two to three nuclear power reactors, an amount that could be enough to power a small country.
I have an extreme dislike for OpenAI, Altman, and people like him, but the reasoning behind this article is just stuff some guy has pulled from his backside. There’s no facts here, it’s just “I believe XYX” with nothing to back it up.
We don’t need to make up nonsense about the LLM bubble. There’s plenty of valid enough criticisms as is.
By circulating a dumb figure like this, all you’re doing is granting OpenAI the power to come out and say “actually, it only uses X amount of power. We’re so great!”, where X is a figure that on its own would seem bad, but compared to this inflated figure sounds great. Don’t hand these shitty companies a marketing win.
It takes less energy to dry a full load of clothes
40 watt-hours? That’s the energy usage of a very small laptop.
This bubble needs to pop, the sooner the better.
I think AI power usage has an upside. No amount of hype can pay the light bill.
AI is either going to be the most valuable tech in history, or it’s going to be a giant pile of ash that used to be VC capital.
It will not go away at this point. Too many daily users already, who uses it for study, work, chatting, looking things up.
If not OpenAI, it will be another service.
Those same things were said about hundreds of other technologies that no longer exist in any meaningful sense. Current usage of a technology, which in this specific case I would argue is largely frivolous anyway, is not an accurate indicator of future usage.
Can you give some examples of those technologies? I’d be interested in how many weren’t replaced with something more efficient or convenient.
That capital was ash earlier this year. The latest $40 Billion-with-a-B financing round is just a temporary holdover until they can raise more fuel. And they already burned through Microsoft, who apparently got what they wanted and are all “see ya”.
How the hell are they going to sustain the expense to power that? Setting aside the environmental catastrophe that this kind of “AI” entails, they’re just not very profitable.
Not just”not profitable”, they don’t make any money at all. Loss only.
Look at all the layoffs they’ve been able to implement with the mere threat that AI has taken their jobs. It’s very profitable, just not in a sustainable way. But sustainability isn’t the goal. Feudal state mindset in the populace is.
I don’t care how rough the estimate is, LLMs are using insane amounts of power, and the message I’m getting here is that the newest incarnation uses even more.
BTW a lot of it seems to be just inefficient coding as Deepseek has shown.
BTW a lot of it seems to be just inefficient coding as Deepseek has shown.
Kind of? Inefficient coding is definitely a part of it. But a large part is also just the iterative nature of how these algorithms operate. We might be able to improve that via code optimization a little bit. But without radically changing how these engines operates it won’t make a big difference.
The scope of the data being used and trained on is probably a bigger issue. Which is why there’s been a push by some to move from LLMs to SLMs. We don’t need the model to be cluttered with information on geology, ancient history, cooking, software development, sports trivia, etc if it’s only going to be used for looking up stuff on music and musicians.
But either way, there’s a big ‘diminishing returns’ factor to this right now that isn’t being appreciated. Typical human nature: give me that tiny boost in performance regardless of the cost, because I don’t have to deal with. It’s the same short-sighted shit that got us into this looming environmental crisis.
Coordinated SLM governors that can redirect queries to the appropriate SLM seems like a good solution.
Powered by GNU Hurd
And it sucks even worse.
Bit of a clickbait. We can’t really say it without more info.
But it’s important to point out that the lab’s test methodology is far from ideal.
The team measured GPT-5’s power consumption by combining two key factors: how long the model took to respond to a given request, and the estimated average power draw of the hardware running it.
What we do know is that the price went down. So this could be a strong indication the model is, in fact, more energy efficient. At least a stronger indicator than response time.
Isnt it just worse than 4 tho? If they didnt make it cheaper, nobody would pay…
That’s a terrible metric. By this providers that maximize hardware (and energy) use by having a queue of requests would be seen as having more energy use.
OpenAI just needs to harness lightning. Incoming weather control tech.
I don’t buy the research paper at all. Of course we have no idea what OpenAI does because they aren’t open at all, but Deepseek’s publish papers suggest it’s much more complex than 1 model per node… I think they recommended like a 576 GPU cluster, with a scheme to split experts.
That, and going by the really small active parameter count of gpt-oss, I bet the model is sparse as heck.
There’s no way the effective batch size is 8, it has to be waaay higher than that.
And perhaps even more importantly, the per-token cost of GPT-5’s API is less than GPT-4’s. That’s why OpenAI was so eager to move everyone onto it, it means more profit for them.
I don’t believe api costs are tied all that closely to the actual cost to openAI. They seem to be selling at a loss, and they may be selling at an even greater loss to make it look like they are progressing. The second openAI seems like they have plateaued, their stock evaluation will crash and it will be game over for them.
I based my argument on actual numbers that can be looked up and verified. You “believe” that they “seem” to be doing something else. Based on what?
To be fair, OpenAI’s negative profitability has been extensively reported on.
Your point stands though; there’s no evidence they’re trying to decrease revenue. On the contrary, that would be a huge red flag to any vested interests.
Their point is that those API prices might not match reality, and the prices may be artificially low to build hype and undercut competitors. We don’t know how much it costs OpenAI, however we do know that they’re not making a profit.
Or it might not. It would be a huge short term risk to do so.
As FaceDeer said, that we truly don’t know.
OpenAI are not profitable today, and don’t estimate they’ll be profitable until 2029, so it’s almost guaranteed that they’re selling their services at a loss. Of course, that’s impossible to verify - since they’re a private company, they don’t have to release financial statements.
That’s not what I’m saying. They’ve all but outright said they’re unprofitable.
But revenue is increasing. Now, if it stops increasing like they’ve “leveled out”, that is a problem.
Hence it’s a stretch to assume they would decrease costs for a more expensive model since that would basically pop their bubble well before 2029.
Sure, they might not. But he gives no basis for saying that other than what he “believes.”
People in this community, and on the Fediverse in general, seem to be strongly anti-AI and would like to believe things that make it sound bad and unprofitable. So when an article like this comes along and says exactly what you want to believe it’s easy to just nod and go “knew it!” Rather than investigating the reasons for those beliefs and risking finding out something you didn’t want to know.
that make it sound bad and unprofitable
It is unprofitable, though.
OpenAI recently hit $10 billion in ARR and are likely to hit $12.7b by the end of the year, but they’re still losing a lot of money. They don’t think they’ll make a profit until 2029, and only if they hit their target of $125 billion revenue. That’s a huge amount of growth - 10x in 4 years - so I’m interested as to if they’ll actually hit it.
Okay, make it sound worse and even more unprofitable.
Making their AI models cheaper to run (such as by requiring less electricity) is one step along that path to profitability.
How does OpenAI getting less money (with a cheaper model) mean more profit? Am I missing something?
Usually, companies will make their product say 25% cheaper to produce, then sell it to the public at a 20% discount (while loudly proclaiming to the world about that 20% price drop) and pocket that 5% increase in profits. So if OpenAI is dropping the price by x, it’s safe to assume that the efficiency gains work out to x+1.
Thanks! This makes sense, however OpenAI are not yet profitable. It’s definitely possible that they’re losing less money with the new models, though.
That “not profitable” label should be taken with a grain of salt. Startups will do all the creative accounting they can in order to maintain that label. After all, don’t have to pay taxes on negative profits.
In the end, it still means their losses are greater than their profits.
They’ve still got taxes they need to pay, too - things like payroll taxes, real estate taxes, etc.
If the model is cheaper to run then they are able to reduce the price without reducing profit, which gives them an advantage over competitors and draws in more customer activity. OpenAI is far from a monopoly.
Of course there are comments doubting the accuracy, which by itself is valid, but they are merely doing it to defend AI. IMHO, even at a fifth of the estimates, we’re talking humongous amounts of power, all for a so-so search engine, half arsed chatbots and dubious nsfw images mostly. And let’s not forget: it may be inaccurate and estimates are TOO LOW. Now wouldn’t that be fun?
but they are merely doing it to defend AI.
No they’re not, you can agree the research is garbage without defending AI. It literally assumes everything. GPT5 could be using eight times the power. It could be using half the power. It could be using a quadrillion times the power. Nobody knows, because they keep it secret.
It’s highly unlikely they reduced power usage—one of the most consistent criticisms of LLM and image generation—without advertising it.
It’s highly unlikely they would bring more attention to one of the biggest issues AI is causing even if they did make it slightly better
40Wh or 18Wh which is it?
That’s my old gaming PC running a game for 2min42sec-6minutes … Roughly.
they vibe calculated it.
But we get a huge increase in accuracy, from 30% to 30.5%! And it only took 5x the energy consumption!
