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:
It took the market about a week to catch up to the fact that the Chinese artificial intelligence firm DeepSeek had released an open-source AI model that rivaled those from prominent U.S. companies such as OpenAI and Anthropic — and that, most importantly, it had managed to do so much more cheaply and efficiently than its domestic competitors. The news cratered not only tech stocks such as Nvidia, but energy stocks, as well, leading to assumptions that investors thought more-energy efficient AI would reduce energy demand in the sector overall.
But will it really? While some in climate world assumed the same and celebrated the seemingly good news, many venture capitalists, AI proponents, and analysts quickly arrived at essentially the opposite conclusion — that cheaper AI will only lead to greater demand for AI. The resulting unfettered proliferation of the technology across a wide array of industries could thus negate the energy efficiency gains, ultimately leading to a substantial net increase in data center power demand overall.
“With cost destruction comes proliferation,” Susan Su, a climate investor at the venture capital firm Toba Capital, told me. “Plus the fact that it’s open source, I think, is a really, really big deal. It puts the power to expand and to deploy and to proliferate into billions of hands.”
If you’ve seen lots of chitchat about Jevons paradox of late, that’s basically what this line of thinking boils down to. After Microsoft’s CEO Satya Nadella responded to DeepSeek mania by posting the Wikipedia page for this 19th century economic theory on X, many (myself included) got a quick crash course on its origins. The idea is that as technical efficiencies of the Victorian era made burning coal cheaper, demand for — and thus consumption of — coal actually increased.
While this is a distinct possibility in the AI space, it’s by no means a guarantee. “This is very much, I think, an open question,“ energy expert Nat Bullard told me, with regards to whether DeepSeek-type models will spur a reduction or increase in energy demand. “I sort of lean in both directions at once.” Formerly the chief content officer at BloombergNEF and current co-founder of the AI startup Halcyon, a search and information platform for energy professionals, Bullard is personally excited for the greater efficiencies and optionality that new AI models can bring to his business.
But he warns that just because DeepSeek was cheap to train — the company claims it cost about $5.5 million, while domestic models cost hundreds of millions or even billions — doesn’t mean that it’s cheap or energy-efficient to operate. “Training more efficiently does not necessarily mean that you can run it that much more efficiently,” Bullard told me. When a large language model answers a question or provides any type of output, it’s said to be making an “inference.” And as Bullard explains, “That may mean, as we move into an era of more and more inference and not just training, then the [energy] impacts could be rather muted.”
DeepSeek-R1, the name for the model that caused the investor freakout, is also a newer type of LLM that uses more energy in general. Up until literally a few days ago, when OpenAI released o3-mini for free, most casual users were probably interacting with so-called “pretrained” AI models. Fed on gobs of internet text, these LLMs spit out answers based primarily on prediction and pattern recognition. DeepSeek released a model like this, called V3, in September. But last year, more advanced “reasoning” models, which can “think,” in some sense, started blowing up. These models — which include o3-mini, the latest version of Anthropic’s Claude, and the now infamous DeepSeek-R1 — have the ability to try out different strategies to arrive at the correct answer, recognize their mistakes, and improve their outputs, allowing for significant advancements in areas such as math and coding.
But all that artificial reasoning eats up a lot of energy. As Sasha Luccioni, the AI and climate lead at Hugging Face, which makes an open-source platform for AI projects, wrote on LinkedIn, “To set things clear about DeepSeek + sustainability: (it seems that) training is much shorter/cheaper/more efficient than traditional LLMs, *but* inference is longer/more expensive/less efficient because of the chain of thought aspect.” Chain of thought refers to the reasoning process these newer models undertake. Luccioni wrote that she’s currently working to evaluate the energy efficiency of both the DeepSeek V3 and R1 models.
Another factor that could influence energy demand is how fast domestic companies respond to the DeepSeek breakthrough with their own new and improved models. Amy Francetic, co-founder at Buoyant Ventures, doesn’t think we’ll have to wait long. “One effect of DeepSeek is that it will highly motivate all of the large LLMs in the U.S. to go faster,” she told me. And because a lot of the big players are fundamentally constrained by energy availability, she’s crossing her fingers that this means they’ll work smarter, not harder. “Hopefully it causes them to find these similar efficiencies rather than just, you know, pouring more gasoline into a less fuel-efficient vehicle.”
In her recent Substack post, Su described three possible futures when it comes to AI’s role in the clean energy transition. The ideal is that AI demand scales slowly enough that nuclear and renewables scale with it. The least hopeful is that immediate, exponential growth in AI demand leads to a similar expansion of fossil fuels, locking in new dirty infrastructure for decades. “I think that's already been happening,” Su told me. And then there’s the techno-optimist scenario, linked to figures like Sam Altman, which Su doesn’t put much stock in — that AI “drives the energy revolution” by helping to create new energy technologies and efficiencies that more than offset the attendant increase in energy demand.
Which scenario predominates could also depend upon whether greater efficiencies, combined with the adoption of AI by smaller, more shallow-pocketed companies, leads to a change in the scale of data centers. “There’s going to be a lot more people using AI. So maybe that means we don’t need these huge, gigawatt data centers. Maybe we need a lot more smaller, megawatt-size data centers,” Laura Katzman, a principal at Buoyant Ventures, told me. Katzman has conducted research for the firm on data center decarbonization.
Smaller data centers with a subsequently smaller energy footprint could pair well with renewable-powered microgrids, which are less practical and economically feasible for hyperscalers. That could be a big win for solar and wind plus battery storage, Katzman explained, but a boondoggle for companies such as Microsoft, which has famously committed to re-opening Pennsylvania’s Three Mile Island nuclear plant to power its data centers. “Because of DeepSeek, the expected price of compute probably doesn’t justify now turning back on some of these nuclear plants, or these other high-cost energy sources,” Katzman told me.
Lastly, it remains to be seen what nascent applications cheaper models will open up. “If somebody, say, in the Philippines or Vietnam has an interest in applying this to their own decarbonization challenge, what would they come up with?” Bullard pondered. “I don’t yet know what people would do with greater capability and lower costs and a different set of problems to solve for. And that’s really exciting to me.”
But even if the AI pessimists are right, and these newer models don’t make AI ubiquitously useful for applications from new drug discovery to easier regulatory filing, Su told me that in a certain sense, it doesn't matter much. “If there was a possibility that somebody had this type of power, and you could have it too, would you sit on the couch? Or would you arms race them? I think that is going to drive energy demand, irrespective of end utility.”
As Su told me, “I do not think there’s actually a saturation point for this.”
Log in
To continue reading, log in to your account.
Create a Free Account
To unlock more free articles, please create a free account.
On Energy Transfer’s legal win, battery storage, and the Cybertruck
Current conditions: Red flag warnings are in place for much of Florida • Spain is bracing for extreme rainfall from Storm Martinho, the fourth named storm in less than two weeks • Today marks the vernal equinox, or the first day of spring.
A jury has ordered Greenpeace to pay more than $660 million in damages to one of the country’s largest fossil fuel infrastructure companies after finding the environmental group liable for defamation, conspiracy, and physical damages at the Dakota Access Pipeline. Greenpeace participated in large protests, some violent and disruptive, at the pipeline in 2016, though it has maintained that its involvement was insignificant and came at the request of the local Standing Rock Sioux Tribe. The project eventually went ahead and is operational today, but Texas-based Energy Transfer sued the environmental organization, accusing it of inciting the uprising and encouraging violence. “We should all be concerned about the future of the First Amendment, and lawsuits like this aimed at destroying our rights to peaceful protest and free speech,” said Deepa Padmanabha, senior legal counsel for Greenpeace USA. The group said it plans to appeal.
The Department of Energy yesterday approved a permit for the Calcasieu Pass 2 liquified natural gas terminal in Louisiana, allowing the facility to export to countries without a free trade agreement. The project hasn’t yet been constructed and is still waiting for final approvals from the independent Federal Energy Regulatory Commission, but the DOE’s green light means it faces one less hurdle.
CP2 was awaiting DOE’s go-ahead when the Biden administration announced its now notorious pause on approvals for new LNG export facilities. The project’s opponents argue it’s a “carbon bomb.” Analysis from the National Resources Defense Council suggested the greenhouse gases from the project would be equivalent to putting more than 1.85 million additional gas-fueled automobiles on the road, while the Sierra Club found it would amount to about 190 million tons of carbon dioxide equivalent annually.
President Trump met with 15 to 20 major oil and gas executives from the American Petroleum Institute at the White House yesterday. This was the president’s first meeting with fossil fuel bosses since his second term began in January. Interior Secretary Doug Burgum and Energy Secretary Chris Wright were also in the room. Everyone is staying pretty quiet about what exactly was said, but according to Burgum and Wright, the conversation focused heavily on permitting reform and bolstering the grid. Reuters reported that “executives had been expected to express concerns over Trump’s tariffs and stress the industry view that higher oil prices are needed to help meet Trump’s promise to grow domestic production.” Burgum, however, stressed that oil prices didn’t come up in the chat. “Price is set by supply and demand,” he said. “There was nothing we could say in that room that could change that one iota, and so it wasn’t really a topic of discussion.” The price of U.S. crude has dropped 13% since Trump returned to office, according to CNBC, on a combination of recession fears triggered by Trump’s tariffs and rising oil output from OPEC countries.
The U.S. installed 1,250 megawatts of residential battery storage last year, the highest amount ever and nearly 60% more than in 2023, according to a new report from the American Clean Power Association and Wood Mackenzie. Overall, battery storage installations across all sectors hit a new record in 2024 at 12.3 gigawatts of new capacity. Storage is expected to continue to grow next year, but uncertainties around tariffs and tax incentives could slow things down.
China is delaying approval for construction of BYD’s Mexico plant because authorities worry the electric carmaker’s technology could leak into the United States, according to the Financial Times. “The commerce ministry’s biggest concern is Mexico’s proximity to the U.S.,” sources told the FT. As Heatmap’s Robinson Meyer writes, BYD continues to set the global standard for EV innovation, and “American and European carmakers are still struggling to catch up.” This week the company unveiled its new “Super e-Platform,” a new standard electronic base for its vehicles that it says will allow incredibly fast charging — enabling its vehicles to add as much as 249 miles of range in just five minutes.
Tesla has recalled 46,096 Cybertrucks over an exterior trim panel that can fall off and become a road hazard. This is the eighth recall for the truck since it went on sale at the end of 2023.
This fusion startup is ahead of schedule.
Thea Energy, one of the newer entrants into the red-hot fusion energy space, raised $20 million last year as investors took a bet on the physics behind the company’s novel approach to creating magnetic fields. Today, in a paper being submitted for peer review, Thea announced that its theoretical science actually works in the real world. The company’s CEO, Brian Berzin, told me that Thea achieved this milestone “quicker and for less capital than we thought,” something that’s rare in an industry long-mocked for perpetually being 30 years away.
Thea is building a stellarator fusion reactor, which typically looks like a twisted version of the more common donut-shaped tokamak. But as Berzin explained to me, Thea’s stellarator is designed to be simpler to manufacture than the industry standard. “We don’t like high tech stuff,” Berzin told me — a statement that sounds equally anathema to industry norms as the idea of a fusion project running ahead of schedule. “We like stuff that can be stamped and forged and have simple manufacturing processes.”
The company thinks it can achieve simplicity via its artificial intelligence software, which controls the reactor’s magnetic field keeping the unruly plasma at the heart of the fusion reaction confined and stabilized. Unlike typical stellarators, which rely on the ultra-precise manufacturing and installment of dozens of huge, twisted magnets, Thea’s design uses exactly 450 smaller, simpler planar magnets, arranged in the more familiar donut-shaped configuration. These magnets are still able to generate a helical magnetic field — thought to keep the plasma better stabilized than a tokamak — because each magnet is individually controlled via the company’s software, just like “the array of pixels in your computer screen,” Berzin told me.
“We’re able to utilize the control system that we built and very specifically modulate and control each magnet slightly differently,” Berzin explained, allowing Thea to “make those really complicated, really precise magnetic fields that you need for a stellarator, but with simple hardware.”
This should make manufacturing a whole lot easier and cheaper, Berzin told me. If one of Thea’s magnets is mounted somewhat imperfectly, or wear and tear of the power plant slightly shifts its location or degrades its performance over time, Thea’s AI system can automatically compensate. “It then can just tune that magnet slightly differently — it turns that magnet down, it turns the one next to it up, and the magnetic field stays perfect,” Berzin explained. As he told me, a system that relies on hardware precision is generally much more expensive than a system that depends on well-designed software. The idea is that Thea’s magnets can thus be mass manufactured in a way that’s conducive to “a business versus a science project.”
In 2023, Thea published a technical report proving out the physics behind its so-called “planar coil stellarator,” which allowed the company to raise its $20 million Series A last year, led by the climate tech firm Prelude Ventures. To validate the hardware behind its initial concept, Thea built a 3x3 array of magnets, representative of one section of its overall “donut” shaped reactor. This array was then integrated with Thea’s software and brought online towards the end of last year.
The results that Thea announced today were obtained during testing last month, and prove that the company can create and precisely control the complex magnetic field shapes necessary for fusion power. These results will allow the company to raise a Series B in the “next couple of years,” Berzin said. During this time, Thea will be working to scale up manufacturing such that it can progress from making one or two magnets per week to making multiple per day at its New Jersey-based facility.
The company’s engineers are also planning to stress test their AI software, such that it can adapt to a range of issues that could arise after decades of fusion power plant operation. “So we’re going to start breaking hardware in this device over the next month or two,” Berzin told me. “We’re purposely going to mismount a magnet by a centimeter, put it back in and not tell the control system what we did. And then we’re going to purposely short out some of the magnetic coils.” If the system can create a strong, stable magnetic field anyway, this will serve as further proof of concept for Thea’s software-oriented approach to a simplified reactor design.
The company is still years away from producing actual fusion power though. Like many others in the space, Thea hopes to bring fusion electrons to the grid sometime in the 2030s. Maybe this simple hardware, advanced software approach is what will finally do the trick.
The Chinese carmaker says it can charge EVs in 5 minutes. Can America ever catch up?
The Chinese automaker BYD might have cracked one of the toughest problems in electric cars.
On Tuesday, BYD unveiled its new “Super e-Platform,” a new standard electronic base for its vehicles that it says will allow incredibly fast charging — enabling its vehicles to add as much as 249 miles of range in just five minutes. That’s made possible because of a 1,000-volt architecture and what BYD describes as matching charging capability, which could theoretically add nearly one mile of range every second.
It’s still not entirely clear whether the technology actually works, although BYD has a good track record on that front. But it suggests that the highest-end EVs worldwide could soon add range as fast as gasoline-powered cars can now, eliminating one of the biggest obstacles to EV adoption.
The new charging platform won’t work everywhere. BYD says that it will also build 4,000 chargers across China that will be able to take advantage of these maximum speeds. If this pans out, then BYD will be able to charge its newest vehicles twice as fast as Tesla’s next generation of superchargers can.
“This is a good thing,” Jeremy Wallace, a Chinese studies professor at Johns Hopkins University, told me. “Yes, it’s a Chinese company. And there are geopolitical implications to that. But the better the technology gets, the easier it is to decarbonize.”
“As someone who has waited in line for chargers in Pennsylvania and New Jersey, I look forward to the day when charging doesn’t take that long,” he added.
The announcement also suggests that the Chinese EV sector remains as dynamic as ever and continues to set the global standard for EV innovation — and that American and European carmakers are still struggling to catch up. The Trump administration is doing little to help the industry catch up: It has proposed repealing the Inflation Reduction Act’s tax credits for EV buyers, which provide demand-side support for the fledgling industry, and the Environmental Protection Agency is working to roll back tailpipe-pollution rules that have furnished early profits to EV makers, including Tesla. Against that background, what — if anything — can U.S. companies do to catch up?
The situation isn’t totally hopeless, but it’s not great.
BYD’s mega-charging capability is made possible by two underlying innovations. First, BYD’s new platform — the wiring, battery, and motors that make up the electronic guts of the car — will be capable of channeling up to 1,000 volts. That is only a small step-change above the best platforms available elsewhere— the forthcoming Gravity SUV from the American carmaker Lucid is built on a 926-volt platform, while the Cybertruck’s platform is 800 volts — but BYD will be able to leverage its technological firepower with mass manufacturing capacity unrivaled by any other brand.
Second, BYD’s forthcoming chargers will be capable of using the platform’s full voltage. These chargers may need to be built close to power grid infrastructure because of the amount of electricity that they will demand.
But sitting underneath these innovations is a sprawling technological ecosystem that keeps all Chinese electronics companies ahead — and that guarantees Chinese advantages well into the future.
“China’s decisive advantage over the U.S. when it comes to innovation is that it has an entrenched workforce that is able to continuously iterate on technological advances,” Dan Wang, a researcher of China’s technology industry and a fellow at the Paul Tsai China Center at Yale Law School, told me.
The country is able to innovate so relentlessly because of its abundance of process knowledge, Wang said. This community of engineering practice may have been seeded by Apple’s iPhone-manufacturing effort in the aughts and Tesla’s carmaking prowess in the 2010s, but it has now taken on a life of its own.
“Shenzhen is the center of the world’s hardware manufacturing industry because it has workers rubbing shoulders with academics rubbing shoulders with investors rubbing shoulders with engineers,” Wang told me. “And you have a more hustle-type culture because it’s so much harder to maintain technological moats and technological differentiation, because people are so competitive in these sorts of spaces.”
In a way, Shenzhen is the modern-day version of the hardware and software ecosystem that used to exist in northern California — Silicon Valley. But while the California technology industry now largely focuses on software, China has taken over the hardware side.
That allows the country to debut new technological innovations much faster than any other country can, he added. “The comparison I hear is that if you have a new charging platform or a new battery chemistry, Volkswagen and BMW will say, We’ll hustle to put this into our systems, and we’ll put it in five years from now. Tesla might say, we’ll hustle and get it in a year from now.”
“China can say, we’ll put it in three months from now,” he said.“You have a much more focused concentration of talent in China, which collapses coordination time.”
That culture has allowed the same companies and engineers to rapidly advance in manufacturing skill and complexity. It has helped CATL, which originally made batteries for smartphones, to become one of the world’s top EV battery makers. And it has helped BYD — which is close to unseating Tesla as the world’s No. 1 seller of electric vehicles — move from making lackluster gasoline cars to some of the world’s best and cheapest EVs.
It will be a while until America can duplicate that manufacturing capability, partly because of the number of headwinds it faces, Wang said.