You’re out of free articles.
Log in
To continue reading, log in to your account.
Create a Free Account
To unlock more free articles, please create a free account.
Sign In or Create an Account.
By continuing, you agree to the Terms of Service and acknowledge our Privacy Policy
Welcome to Heatmap
Thank you for registering with Heatmap. Climate change is one of the greatest challenges of our lives, a force reshaping our economy, our politics, and our culture. We hope to be your trusted, friendly, and insightful guide to that transformation. Please enjoy your free articles. You can check your profile here .
subscribe to get Unlimited access
Offer for a Heatmap News Unlimited Access subscription; please note that your subscription will renew automatically unless you cancel prior to renewal. Cancellation takes effect at the end of your current billing period. We will let you know in advance of any price changes. Taxes may apply. Offer terms are subject to change.
Subscribe to get unlimited Access
Hey, you are out of free articles but you are only a few clicks away from full access. Subscribe below and take advantage of our introductory offer.
subscribe to get Unlimited access
Offer for a Heatmap News Unlimited Access subscription; please note that your subscription will renew automatically unless you cancel prior to renewal. Cancellation takes effect at the end of your current billing period. We will let you know in advance of any price changes. Taxes may apply. Offer terms are subject to change.
Create Your Account
Please Enter Your Password
Forgot your password?
Please enter the email address you use for your account so we can send you a link to reset your password:
Why the new “reasoning” models might gobble up more electricity — at least in the short term
What happens when artificial intelligence takes some time to think?
The newest set of models from OpenAI, o1-mini and o1-preview, exhibit more “reasoning” than existing large language models and associated interfaces, which spit out answers to prompts almost instantaneously.
Instead, the new model will sometimes “think” for as long as a minute or two. “Through training, they learn to refine their thinking process, try different strategies, and recognize their mistakes,” OpenAI announced in a blog post last week. The company said these models perform better than their existing ones on some tasks, especially related to math and science. “This is a significant advancement and represents a new level of AI capability,” the company said.
But is it also a significant advancement in energy usage?
In the short run at least, almost certainly, as spending more time “thinking” and generating more text will require more computing power. As Erik Johannes Husom, a researcher at SINTEF Digital, a Norwegian research organization, told me, “It looks like we’re going to get another acceleration of generative AI’s carbon footprint.”
Discussion of energy use and large language models has been dominated by the gargantuan requirements for “training,” essentially running a massive set of equations through a corpus of text from the internet. This requires hardware on the scale of tens of thousands of graphical processing units and an estimated 50 gigawatt-hours of electricity to run.
Training GPT-4 cost “more than” $100 million OpenAI chief executive Sam Altman has said; the next generation models will likely cost around $1 billion, according to Anthropic chief executive Dario Amodei, a figure that might balloon to $100 billion for further generation models, according to Oracle founder Larry Ellison.
While a huge portion of these costs are hardware, the energy consumption is considerable as well. (Meta reported that when training its Llama 3 models, power would sometimes fluctuate by “tens of megawatts,” enough to power thousands of homes). It’s no wonder that OpenAI’s chief executive Sam Altman has put hundreds of millions of dollars into a fusion company.
But the models are not simply trained, they're used out in the world, generating outputs (think of what ChatGPT spits back at you). This process tends to be comparable to other common activities like streaming Netflix or using a lightbulb. This can be done with different hardware and the process is more distributed and less energy intensive.
As large language models are being developed, most computational power — and therefore most electricity — is used on training, Charlie Snell, a PhD student at University of California at Berkeley who studies artificial intelligence, told me. “For a long time training was the dominant term in computing because people weren’t using models much.” But as these models become more popular, that balance could shift.
“There will be a tipping point depending on the user load, when the total energy consumed by the inference requests is larger than the training,” said Jovan Stojkovic, a graduate student at the University of Illinois who has written about optimizing inference in large language models.
And these new reasoning models could bring that tipping point forward because of how computationally intensive they are.
“The more output a model produces, the more computations it has performed. So, long chain-of-thoughts leads to more energy consumption,” Husom of SINTEF Digital told me.
OpenAI staffers have been downright enthusiastic about the possibilities of having more time to think, seeing it as another breakthrough in artificial intelligence that could lead to subsequent breakthroughs on a range of scientific and mathematical problems. “o1 thinks for seconds, but we aim for future versions to think for hours, days, even weeks. Inference costs will be higher, but what cost would you pay for a new cancer drug? For breakthrough batteries? For a proof of the Riemann Hypothesis? AI can be more than chatbots,” OpenAI researcher Noam Brown tweeted.
But those “hours, days, even weeks” will mean more computation and “there is no doubt that the increased performance requires a lot of computation,” Husom said, along with more carbon emissions.
But Snell told me that might not be the end of the story. It’s possible that over the long term, the overall computing demands for constructing and operating large language models will remain fixed or possibly even decline.
While “the default is that as capabilities increase, demand will increase and there will be more inference,” Snell told me, “maybe we can squeeze reasoning capability into a small model ... Maybe we spend more on inference but it’s a much smaller model.”
OpenAI hints at this possibility, describing their o1-mini as “a smaller model optimized for STEM reasoning,” in contrast to other, larger models that “are pre-trained on vast datasets” and “have broad world knowledge,” which can make them “expensive and slow for real-world applications.” OpenAI is suggesting that a model can know less but think more and deliver comparable or better results to larger models — which might mean more efficient and less energy hungry large language models.
In short, thinking might use less brain power than remembering, even if you think for a very long time.
Log in
To continue reading, log in to your account.
Create a Free Account
To unlock more free articles, please create a free account.
And more of this week’s top renewable energy fights across the country.
1. Otsego County, Michigan – The Mitten State is proving just how hard it can be to build a solar project in wooded areas. Especially once Fox News gets involved.
2. Atlantic County, New Jersey – Opponents of offshore wind in Atlantic City are trying to undo an ordinance allowing construction of transmission cables that would connect the Atlantic Shores offshore wind project to the grid.
3. Benton County, Washington – Sorry Scout Clean Energy, but the Yakima Nation is coming for Horse Heaven.
Here’s what else we’re watching right now…
In Connecticut, officials have withdrawn from Vineyard Wind 2 — leading to the project being indefinitely shelved.
In Indiana, Invenergy just got a rejection from Marshall County for special use of agricultural lands.
In Kansas, residents in Dickinson County are filing legal action against county commissioners who approved Enel’s Hope Ridge wind project.
In Kentucky, a solar project was actually approved for once – this time for the East Kentucky Power Cooperative.
In North Carolina, Davidson County is getting a solar moratorium.
In Pennsylvania, the town of Unity rejected a solar project. Elsewhere in the state, the developer of the Newton 1 solar project is appealing their denial.
In South Carolina, a state appeals court has upheld the rejection of a 2,300 acre solar project proposed by Coastal Pine Solar.
In Washington State, Yakima County looks like it’ll keep its solar moratorium in place.
And more of this week’s top policy news around renewables.
1. Trump’s Big Promise – Our nation’s incoming president is now saying he’ll ban all wind projects on Day 1, an expansion of his previous promise to stop only offshore wind.
2. The Big Nuclear Lawsuit – Texas and Utah are suing to kill the Nuclear Regulatory Commission’s authority to license small modular reactors.
3. Biden’s parting words – The Biden administration has finished its long-awaited guidance for the IRA’s tech-neutral electricity credit (which barely changed) and hydrogen production credit.
A conversation with J. Timmons Roberts, executive director of Brown University’s Climate Social Science Network
This week’s interview is with Brown University professor J. Timmons Roberts. Those of you familiar with the fight over offshore wind may not know Roberts by name, but you’re definitely familiar with his work: He and his students have spearheaded some of the most impactful research conducted on anti-offshore wind opposition networks. This work is a must-read for anyone who wants to best understand how the anti-renewables movement functions and why it may be difficult to stop it from winning out.
So with Trump 2.0 on the verge of banning offshore wind outright, I decided to ask Roberts what he thinks developers should be paying attention to at this moment. The following interview has been lightly edited for clarity.
Is the anti-renewables movement a political force the country needs to reckon with?
Absolutely. In my opinion it’s been unfortunate for the environmental groups, the wind development, the government officials, climate scientists – they’ve been unwilling to engage directly with those groups. They want to keep a very positive message talking about the great things that come with wind and solar. And they’ve really left the field open as a result.
I think that as these claims sit there unrefuted and naive people – I don’t mean naive in a negative sense but people who don’t know much about this issue – are only hearing the negative spin about renewables. It’s a big problem.
When you say renewables developers aren’t interacting here – are you telling me the wind industry is just letting these people run roughshod?
I’ve seen no direct refutation in those anti-wind Facebook groups, and there’s very few environmentalists or others. People are quite afraid to go in there.
But even just generally. This vast network you’ve tracked – have you seen a similar kind of counter mobilization on the part of those who want to build these wind farms offshore?
There’s some mobilization. There’s something called the New England for Offshore Wind coalition. There’s some university programs. There’s some other oceanographic groups, things like that.
My observation is that they’re mostly staff organizations and they’re very cautious. They’re trying to work as a coalition. And they’re going as slow as their most cautious member.
As someone who has researched these networks, what are you watching for in the coming year? Under the first year of Trump 2.0?
Yeah I mean, channeling my optimistic and Midwestern dad, my thought is that there may be an overstepping by the Trump administration and by some of these activists. The lack of viable alternative pathways forward and almost anti-climate approaches these groups are now a part of can backfire for them. Folks may say, why would I want to be supportive of your group if you’re basically undermining everything I believe in?
What do you think developers should know about the research you have done into these networks?
I think it's important for deciding bodies and the public, the media and so on, to know who they’re hearing when they hear voices at a public hearing or in a congressional field hearing. Who are the people representing? Whose voice are they advancing?
It’s important for these actors that want to advance action on climate change and renewables to know what strategies and the tactics are being used and also know about the connections.
One of the things you pointed out in your research is that, yes, there are dark money groups involved in this movement and there are outside figures involved, but a lot of this sometimes is just one person posts something to the internet and then another person posts something to the internet.
Does that make things harder when it comes to addressing the anti-renewables movement?
Absolutely. Social media’s really been devastating for developing science and informed, rational public policymaking. It’s so easy to create a conspiracy and false information and very slanted, partial information to shoot holes at something as big as getting us off of fossil fuels.
Our position has developed as we understand that indeed these are not just astro-turf groups created by some far away corporation but there are legitimate concerns – like fishing, where most of it is based on certainty – and then there are these sensationalized claims that drive fears. That fear is real. And it’s unfortunate.
Anything else you’d really like to tell our readers?
I didn’t really choose this topic. I feel like it really got me. It was me and four students sitting in my conference room down the hall and I said, have you heard about this group that just started here in Rhode Island that’s making these claims we should investigate? And students were super excited about it and have really been the leaders.