The clean cement industry might be facing the end of the road, before it ever really got rolling.
On Friday, the US Department of Energy announced that it was canceling $3.7 billion in funding for 24 projects related to energy and industry. That included nearly $1.3 billion for cement-related projects.
Cement is a massive climate problem, accounting for roughly 7% of global greenhouse-gas emissions. What’s more, it’s a difficult industry to clean up, with huge traditional players and expensive equipment and infrastructure to replace. This funding was supposed to help address those difficulties, by supporting projects on the cusp of commercialization. Now companies will need to fill in the gap left by these cancellations, and it’s a big one.
First up on the list for cuts is Sublime Systems, a company you’re probably familiar with if you’ve been reading this newsletter for a while. I did a deep dive last year, and the company was on our list of Climate Tech Companies to Watch in both 2023 and 2024.
The startup’s approach is to make cement using electricity. The conventional process requires high temperatures typically achieved by burning fossil fuels, so avoiding that could prevent a lot of emissions.
In 2024, Sublime received an $87 million grant from the DOE to construct a commercial demonstration plant in Holyoke, Massachusetts. That grant would have covered roughly half the construction costs for the facility, which is scheduled to open in 2026 and produce up to 30,000 metric tons of cement each year.
“We were certainly surprised and disappointed about the development,” says Joe Hicken, Sublime’s senior VP of business development and policy. Customers are excited by the company’s technology, Hicken adds, pointing to Sublime’s recently announced deal with Microsoft, which plans to buy up to 622,500 metric tons of cement from the company.
Another big name, Brimstone, also saw its funding affected. That award totaled $189 million for a commercial demonstration plant, which was expected to produce over 100,000 metric tons of cement annually.
In a statement, a Brimstone representative said the company believes the cancellation was a “misunderstanding.” The statement pointed out that the planned facility would make not only cement but also alumina, supporting US-based aluminum production. (Aluminum is classified as a critical mineral by the US Geological Survey, meaning it’s considered crucial to the US economy and national security.)
An award to Heidelberg Materials for up to $500 million for a planned Indiana facility was also axed. The idea there was to integrate carbon capture and storage to clean up emissions from the plant, which would have made it the first cement plant in the US to demonstrate that technology. In a written statement, a representative said the decision can be appealed, and the company is considering that option.
And National Cement’s funding for the Lebec Net-Zero Project, another $500 million award, was canceled. That facility planned to make carbon-neutral cement through a combination of strategies: reducing the polluting ingredients needed, using alternative fuels like biomass, and capturing the plant’s remaining emissions.
“We want to emphasize that this project will expand domestic manufacturing capacity for a critical industrial sector, while also integrating new technologies to keep American cement competitive,” said a company spokesperson in a written statement.
There’s a sentiment here that’s echoed in all the responses I received: While these awards were designed to cut emissions, these companies argue that they can fit into the new administration’s priorities. They’re emphasizing phrases like “critical minerals,” “American jobs,” and “domestic supply chains.”
“We’ve heard loud and clear from the Trump administration the desire to displace foreign imports of things that can be made here in America,” Sublime’s Hicken says. “At the end of the day, what we deliver is what the policymakers in DC are looking for.”
But this administration is showing that it’s not supporting climate efforts—often even those that also advance its stated goals of energy abundance and American competitiveness.
On Monday, my colleague James Temple published a new story about cuts to climate research, including tens of millions of dollars in grants from the National Science Foundation. Researchers at Harvard were particularly hard hit.
Even as there’s interest in advancing the position of the US on the world’s stage, these cuts are making it hard for researchers and companies alike to do the crucial work of understanding our climate and developing and deploying new technologies.
This article is from The Spark, MIT Technology Review’s weekly climate newsletter. To receive it in your inbox every Wednesday, sign up here.
The Trump administration has terminated National Science Foundation grants for more than 100 research projects related to climate change amid a widening campaign to slash federal funding for scientists and institutions studying the rising risks of a warming world.
The move will cut off what’s likely to amount to tens of millions of dollars for studies that were previously approved and, in most cases, already in the works.
Affected projects include efforts to develop cleaner fuels, measure methane emissions, improve understanding of how heat waves and sea-level rise disproportionately harm marginalized groups, and help communities transition to sustainable energy, according to an MIT Technology Review review of a GrantWatch database—a volunteer-led effort to track federal cuts to research—and a list of terminated grants from the National Science Foundation (NSF) itself.
The NSF is one of the largest sources of US funding for university research, so the cancellations will deliver a big blow to climate science and clean-energy development.
They come on top of the White House’s broader efforts to cut research funding and revenue for universities and significantly raise their taxes. The administration has also strived to slash staff and budgets at federal research agencies, halt efforts to assess the physical and financial risks of climate change, and shut down labs that have monitored and analyzed the levels of greenhouse gases in the air for decades.
“I don’t think it takes a lot of imagination to understand where this is going,” says Daniel Schrag, co-director of the science, technology, and public policy program at Harvard University, which has seen greater funding cuts than any other university amid an escalating legal conflict with the administration. “I believe the Trump administration intends to zero out funding for climate science altogether.”
The NSF says it’s terminating grants that aren’t aligned with the agency’s program goals, “including but not limited to those on diversity, equity, and inclusion (DEI), environmental justice, and misinformation/disinformation.”
Trump administration officials have argued that DEI considerations have contaminated US science, favoring certain groups over others and undermining the public’s trust in researchers.
“Political biases have displaced the vital search for truth,” Michael Kratsios, head of the White House Office of Science and Technology Policy, said to a group of NSF administrations and others last month, according to reporting in Science.
Science v. politics
But research projects that got caught in the administration’s anti-DEI filter aren’t the only casualties of the cuts. The NSF has also canceled funding for work that has little obvious connections to DEI ambitions, such as research on catalysts.
Many believe the administration’s broader motivation is to undermine the power of the university system and prevent research findings that cut against its politics.
“It certainly seems like a deliberate attempt to undo any science that contradicts the administration,” says Alexa Fredston, an assistant professor of ocean sciences at the University of California, Santa Cruz.
On May 28, a group of states including California, New York, and Illinois sued the NSF, arguing that the cuts illegally violated diversity goals and funding priorities clearly established by Congress, which controls federal spending.
A group of universities also filed a lawsuit against the NSF over its earlier decision to reduce the indirect cost rate for research, which reimburses universities for overhead expenses associated with work carried out on campuses. The plaintiffs included the California Institute of Technology, Carnegie Mellon University, and the Massachusetts Institute of Technology, which has also lost a number of research grants.
(MIT Technology Review is owned by, but editorially independent from, MIT.)
The NSF declined to comment.
‘Theft from the American people’
GrantWatch is an effort among researchers at rOpenSci, Harvard, and other organizations to track terminations of grants issued by the National Institutes of Health (NIH) and NSF. It draws on voluntary submissions from scientists involved as well as public government information.
A search of its database for the terms “climate change,” “clean energy,” “climate adaptation,” “environmental justice,” and “climate justice” showed that the NSF has canceled funds for 118 projects, which were supposed to receive more than $100 million in total. Searching for the word “climate” produces more than 300 research projects that were set to receive more than $230 million. (That word often indicates climate-change-related research, but in some abstracts it refers to the cultural climate.)
Some share of those funds has already been issued to research groups. The NSF section of the database doesn’t include that “outlaid” figure, but it’s generally about half the amount of the original grants, according to Noam Ross, a computational researcher and executive director of rOpenSci, a nonprofit initiative that promotes open and reproducible science.
A search for “climate change” among the NIH projects produces another 22 studies that were terminated and were still owed nearly $50 million in grants. Many of those projects explored the mental or physical health effects of climate change and extreme weather events.
The NSF more recently released its own list of terminated projects, which mostly mirrored GrantWatch’s findings and confirms the specific terminations mentioned in this story.
“These grant terminations are theft from the American people,” Ross said in an email response. “By illegally ending this research the Trump administration is wasting taxpayer dollars, gutting US leadership in science, and telling the world that the US government breaks its promises.”
Harvard, the country’s oldest university, has been particularly hard hit.
In April, the university sued the Trump administration over cuts to its research funding and efforts to exert control over its admissions and governance policies. The White House, in turn, has moved to eliminate all federal funds for the university, including hundreds of NSF and NIH grants.
Daniel Nocera, a professor at Harvard who has done pioneering work on so-called artificial photosynthesis, a pathway for producing clean fuels from sunlight, said in an email that all of his grants were terminated.
“I have no research funds,” he added.
Another terminated grant involved a collaboration between Harvard and the NSF National Center for Atmospheric Research (NCAR), designed to update the atmospheric chemistry component of the Community Earth System Model, an open-source climate model widely used by scientists around the world.
The research was expected to “contribute to a better understanding of atmospheric chemistry in the climate system and to improve air quality predictions within the context of climate change,” according to the NSF abstract.
“We completed most of the work and were able to bring it to a stopping point,” Daniel Jacob, a professor at Harvard listed as the principal investigator on the project, said in an email. “But it will affect the ability to study chemistry-climate interactions. And it is clearly not right to pull funding from an existing project.”
Plenty of the affected research projects do, in one way or another, grapple with issues of diversity, equity, and inclusion. But that’s because there is ample evidence that disadvantaged communities experience higher rates of illness from energy-sector pollution, will be harder hit by the escalating effects of extreme weather and are underrepresented in scientific fields.
One of the largest terminations cut off about $4 million dollars of remaining funds for the CLIMATE Justice Initiative, a fellowship program at the University of California, Irvine designed to recruit, train and mentor a more diverse array of researchers in Earth sciences.
The NSF decision occurred halfway into the 5-year program, halting funds for a number of fellows who were in the midst of environmental justice research efforts with community partners in Southern California. Kathleen Johnson, a professor at UC Irvine and director of the initiative, says the university is striving to find ways to fund as many participants as possible for the remainder of their fellowships.
“We need people from all parts of society who are trained in geoscience and climate science to address all these global challenges that we are facing,” she says. “The people who will be best positioned to do this work … are the people who understand the community’s needs and are able to therefore work to implement equitable solutions.”
“Diverse teams have been shown to do better science,” Johnson adds.
Numerous researchers whose grants were terminated didn’t respond to inquiries from MIT Technology Review or declined to comment, amid growing concerns that the Trump administration will punish scientists or institutions that criticize their policies.
Coming cuts
The termination of existing NSF and NIH grants is just the start of the administration’s plans to cut federal funding for climate and clean-energy research.
The White House’s budget proposal for the coming fiscal year seeks to eliminate tens of billions of dollars in funding across federal agencies, specifically calling out “Green New Scam funds” at the Department of Energy; “low-priority climate monitoring satellites” at NASA; “climate-dominated research, data, and grant programs” at the National Oceanic and Atmospheric Administration; and “climate; clean energy; woke social, behavioral, and economic sciences” at the NSF.
The administration released a more detailed NSF budget proposal on May 30th, which called for a 60% reduction in research spending and nearly zeroed out the clean energy technology program. It also proposed cutting funds by 97% for the US Global Change Research Program, which produces regular assessments of climate risks; 80% for the Ocean Observatories Initiative, a global network of ocean sensors that monitor shifting marine conditions; and 40% for NCAR, the atmospheric research center.
If Congress approves budget reductions anywhere near the levels the administration has put forward, scientists fear, it could eliminate the resources necessary to carry on long-running climate observation of oceans, forests, and the atmosphere.
The administration also reportedly plans to end the leases on dozens of NOAA facilities, including the Global Monitoring Laboratory in Hilo, Hawaii. The lab supports the work of the nearby Mauna Loa Observatory, which has tracked atmospheric carbon dioxide levels for decades.
Even short gaps in these time-series studies, which scientists around the world rely upon, would have an enduring impact on researchers’ ability to analyze and understand weather and climate trends.
“We won’t know where we’re going if we stop measuring what’s happening,” says Jane Long, formerly the associate director of energy and environment at Lawrence Livermore National Lab. “It’s devastating—there’s no two ways around it.”
Stunting science
Growing fears that public research funding will take an even larger hit in the coming fiscal year are forcing scientists to rethink their research plans—or to reconsider whether they want to stay in the field at all, numerous observers said.
“The amount of funding we’re talking about isn’t something a university can fill indefinitely, and it’s not something that private philanthropy can fill for very long,” says Michael Oppenheimer, a professor of geosciences and international affairs at Princeton University. “So what we’re talking about is potentially cataclysmic for climate science.”
“Basically it’s a shit show,” he adds, “and how bad a shit show it is will depend a lot on what happens in the courts and Congress over the next few months.”
One climate scientist, who declined to speak on the record out of concern that the administration might punish his institution, said the declining funding is forcing researchers to shrink their scientific ambitions down to a question of “What can I do with my laptop and existing data sets?”
“If your goal was to make the United States a second-class or third-class country when it comes to science and education, you would be doing exactly what the administration is doing,” the scientist said. “People are pretty depressed, upset, and afraid.”
Given the rising challenges, Harvard’s Schrag fears that the best young climate scientists will decide to shift their careers outside of the US, or move into high tech or other fields where they can make significantly more money.
“We might lose a generation of talent—and that’s not going to get fixed four years from now,” he says. “The irony is that Trump is attacking the institutions and foundation of US science that literally made America great.”
It’s been a little over a week since we published Power Hungry, a package that takes a hard look at the expected energy demands of AI. Last week in this newsletter, I broke down the centerpiece of that package, an analysis I did with my colleague James O’Donnell. (In case you’re still looking for an intro, you can check out this Roundtable discussion with James and our editor in chief Mat Honan, or this short segment I did on Science Friday.)
But this week, I want to talk about another story that I also wrote for that package, which focused on nuclear energy. I thought this was an important addition to the mix of stories we put together, because I’ve seen a lot of promises about nuclear power as a saving grace in the face of AI’s energy demand. My reporting on the industry over the past few years has left me a little skeptical.
As I discovered while I continued that line of reporting, building new nuclear plants isn’t so simple or so fast. And as my colleague David Rotman lays out in his story for the package, the AI boom could wind up relying on another energy source: fossil fuels. So what’s going to power AI? Let’s get into it.
When we started talking about this big project on AI and energy demand, we had a lot of conversations about what to include. And from the beginning, the climate team was really focused on examining what, exactly, was going to be providing the electricity needed to run data centers powering AI models. As we wrote in the main story:
“A data center humming away isn’t necessarily a bad thing. If all data centers were hooked up to solar panels and ran only when the sun was shining, the world would be talking a lot less about AI’s energy consumption.”
But a lot of AI data centers need to be available constantly. Those that are used to train models can arguably be more responsive to the changing availability of renewables, since that work can happen in bursts, any time. Once a model is being pinged with questions from the public, though, there needs to be computing power ready to run all the time. Google, for example, would likely not be too keen on having people be able to use its new AI Mode only during daylight hours.
Solar and wind power, then, would seem not to be a great fit for a lot of AI electricity demand, unless they’re paired with energy storage—and that increases costs. Nuclear power plants, on the other hand, tend to run constantly, outputting a steady source of power for the grid.
As you might imagine, though, it can take a long time to get a nuclear power plant up and running.
Large tech companies can help support plans to reopen shuttered plants or existing plants’ efforts to extend their operating lifetimes. There are also some existing plants that can make small upgrades to improve their output. I just saw this news story from the Tri-City Herald about plans to upgrade the Columbia Generating Station in eastern Washington—with tweaks over the next few years, it could produce an additional 162 megawatts of power, over 10% of the plant’s current capacity.
But all that isn’t going to be nearly enough to meet the demand that big tech companies are claiming will materialize in the future. (For more on the numbers here and why new tech isn’t going to come online fast enough, check out my full story.)
Instead, natural gas has become the default to meet soaring demand from data centers, as David lays out in his story. And since the lifetime of plants built today is about 30 years, those new plants could be running past 2050, the date the world needs to bring greenhouse-gas emissions to net zero to meet the goals set out in the Paris climate agreement.
One of the bits I found most interesting in David’s story is that there’s potential for a different future here: Big tech companies, with their power and influence, could actually use this moment to push for improvements. If they reduced their usage during peak hours, even for less than 1% of the year, it could greatly reduce the amount of new energy infrastructure required. Or they could, at the very least, push power plant owners and operators to install carbon capture technology, or ensure that methane doesn’t leak from the supply chain.
AI’s energy demand is a big deal, but for climate change, how we choose to meet it is potentially an even bigger one.
A California-based company called Magrathea just turned on a new electrolyzer that can make magnesium metal from seawater. The technology has the potential to produce the material, which is used in vehicles and defense applications, with net-zero greenhouse-gas emissions.
Magnesium is an incredibly light metal, and it’s used for parts in cars and planes, as well as in aluminum alloys like those in vehicles. The metal is also used in defense and industrial applications, including the production processes for steel and titanium.
Today, China dominates production of magnesium, and the most common method generates a lot of the emissions that cause climate change. If Magrathea can scale up its process, it could help provide an alternative source of the metal and clean up industries that rely on it, including automotive manufacturing.
The star of Magrathea’s process is an electrolyzer, a device that uses electricity to split a material into its constituent elements. Using an electrolyzer in magnesium production isn’t new, but Magrathea’s approach represents an update. “We really modernized it and brought it into the 21st century,” says Alex Grant, Magrathea’s cofounder and CEO.
The whole process starts with salty water. There are small amounts of magnesium in seawater, as well as in salt lakes and groundwater. (In seawater, the concentration is about 1,300 parts per million, so magnesium makes up about 0.1% of seawater by weight.) If you take that seawater or brine and clean it up, concentrate it, and dry it out, you get a solid magnesium chloride salt.
Magrathea takes that salt (which it currently buys from Cargill) and puts it into the electrolyzer. The device reaches temperatures of about 700 °C (almost 1,300 °F) and runs electricity through the molten salt to split the magnesium from the chlorine, forming magnesium metal.
Typically, running an electrolyzer in this process would require a steady source of electricity. The temperature is generally kept just high enough to maintain the salt in a molten state. Allowing it to cool down too much would allow it to solidify, messing up the process and potentially damaging the equipment. Heating it up more than necessary would just waste energy.
Magrathea’s approach builds in flexibility. Basically, the company runs its electrolyzer about 100 °C higher than is necessary to keep the molten salt a liquid. It then uses the extra heat in inventive ways, including to dry out the magnesium salt that eventually goes into the reactor. This preparation can be done intermittently, so the company can take in electricity when it’s cheaper or when more renewables are available, cutting costs and emissions. In addition, the process will make a co-product, called magnesium oxide, that can be used to trap carbon dioxide from the atmosphere, helping to cancel out the remaining carbon pollution.
The result could be a production process with net-zero emissions, according to an independent life cycle assessment completed in January. While it likely won’t reach this bar at first, the potential is there for a much more climate-friendly process than what’s used in the industry today, Grant says.
Breaking into magnesium production won’t be simple, says Simon Jowitt, director of the Nevada Bureau of Mines and of the Center for Research in Economic Geology at the University of Nevada, Reno.
China produces roughly 95% of the global supply as of 2024, according to data from the US Geological Survey. This dominant position means companies there can flood the market with cheap metal, making it difficult for others to compete. “The economics of all this is uncertain,” Jowitt says.
The US has some trade protections in place, including an anti-dumping duty, but newer players with alternative processes can still face obstacles. US Magnesium, a company based in Utah, was the only company making magnesium in the US in recent years, but it shut down production in 2022 after equipment failures and a history of environmental concerns.
Magrathea plans to start building a demonstration plant in Utah in late 2025 or early 2026, which will have a capacity of roughly 1,000 tons per year and should be running in 2027. In February the company announced that it signed an agreement with a major automaker, though it declined to share its name on the record. The automaker pre-purchased material from the demonstration plant and will incorporate it into existing products.
After the demonstration plant is running, the next step would be to build a commercial plant with a larger capacity of around 50,000 tons annually.
A new type of fuel cell that runs on sodium metal could one day help clean up sectors where it’s difficult to replace fossil fuels, like rail, regional aviation, and short-distance shipping. The device represents a departure from technologies like lithium-based batteries and is more similar conceptually to hydrogen fuel cell systems.
The sodium-air fuel cell was designed by a team led by Yet-Ming Chiang, a professor of materials science and engineering at MIT. It has a higher energy density than lithium-ion batteries and doesn’t require the super-cold temperatures or high pressures that hydrogen does, making it potentially more practical for transport. “I’m interested in sodium metal as an energy carrier of the future,” Chiang says.
The device’s design, published today in Joule, is related to the technology behind one of Chiang’s companies, Form Energy, which is building iron-air batteries for large energy storage installations like those that could help store wind and solar power on the grid. Form’s batteries rely on water, iron, and air.
One technical challenge for metal-air batteries has historically been reversibility. A battery’s chemical reactions must be easily reversed so that in one direction they generate electricity, discharging the battery, and in the other electricity goes into the cell and the reverse reactions happen, charging it up.
When a battery’s reactions produce a very stable product, it can be difficult to recharge the battery without losing capacity. To get around this problem, the team at Form had discussions about whether their batteries could be refuelable rather than rechargeable, Chiang says. The idea was that rather than reversing the reactions, they could simply run the system in one direction, add more starting material, and repeat.
Ultimately, Form chose a more traditional battery concept, but the idea stuck with Chiang, who decided to explore it with other metals and landed on the idea of a sodium-based fuel cell.
In this fuel cell format, the device takes in chemicals and runs reactions that generate electricity, after which the products get removed. Then fresh fuel is put in to run the whole thing again—no electrical charging required. (You might recognize this concept from hydrogen fuel cell vehicles, like the Toyota Mirai.)
Chiang and his colleagues set out to build a fuel cell that runs on liquid sodium, which could have a much higher energy density than existing commercial technologies, so it would be small and light enough to be used for things like regional airplanes or short-distance shipping.
Sodium metal could be used to power regional planes or short distance shipping.
GRETCHEN ERTL/MITTR
The research team built small test cells to try out the concept and ran them to show that they could use the sodium-metal-based system to generate electricity. Since sodium becomes liquid at about 98 °C (208 °F), the cells operated at moderate temperatures of between 110 °C and 130 °C (or 230 °F and 266°F), which could be practical for use on planes or ships, Chiang says.
From their work with these experimental devices, the researchers estimated that the energy density was about 1,200 watt-hours per kilogram (Wh/kg). That’s much higher than what commercial lithium-ion batteries can reach today (around 300 Wh/kg). Hydrogen fuel cells can achieve high energy density, but that requires the hydrogen to be stored at high pressures and often ultra-low temperatures.
“It’s an interesting cell concept,” says Jürgen Janek, a professor at the Institute of Physical Chemistry at the University of Giessen in Germany, who was not involved in the research. There’s been previous research on sodium-air batteries in the past, Janek says, but using this sort of chemistry in a fuel cell instead is new.
“One of the critical issues with this type of cell concept is the safety issue,” Janek says. Sodium metal reacts very strongly with water. (You may have seen videos where blocks of sodium metal get thrown into a lake, to dramatic effect). Asked about this issue, Chiang says the design of the cell ensures that water produced during reactions is continuously removed, so there’s not enough around to fuel harmful reactions. The solid electrolyte, a ceramic material, also helps prevent reactions between water and sodium, Chiang adds.
Another question is what happens to one of the cell’s products, sodium hydroxide. Commonly known as lye, it’s an industrial chemical, used in products like liquid drain-cleaning solution. One of the researchers’ suggestions is to dilute the product and release it into the atmosphere or ocean, where it would react with carbon dioxide, capturing it in a stable form and preventing it from contributing to global warming. There are groups pursuing field trials using this exact chemical for ocean-based carbon removal, though some have been met with controversy. The researchers also laid out the potential for a closed system, where the chemical could be collected and sold as a by-product.
There are economic factors working in favor of sodium-based systems, though it would take some work to build up the necessary supply chains. Today, sodium metal isn’t produced at very high volumes. However, it can be made from sodium chloride (table salt), which is incredibly cheap. And it was produced more abundantly in the past, since it was used in the process of making leaded gasoline. So there’s a precedent for a larger supply chain, and it’s possible that scaling up production of sodium metal would make it cheap enough to use in fuel cell systems, Chiang says.
Chiang has cofounded a company called Propel Aero to commercialize the research. The project received funding from ARPA-E’s Propel-1K program, which aims to develop new forms of high-power energy storage for aircraft, trains, and ships.
The next step is to continue research to improve the cells’ performance and energy density, and to start designing small-scale systems. One potential early application is drones. “We’d like to make something fly within the next year,” Chiang says.
“If people don’t find it crazy, I’ll be rather disappointed,” Chiang says. “Because if an idea doesn’t sound crazy at the beginning, it probably isn’t as revolutionary as you think. Fortunately, most people think I’m crazy on this one.”
This week, we published Power Hungry, a package all about AI and energy. At the center of this package is the most comprehensive look yet at AI’s growing power demand, if I do say so myself.
This data-heavy story is the result of over six months of reporting by me and my colleague James O’Donnell (and the work of many others on our team). Over that time, with the help of leading researchers, we quantified the energy and emissions impacts of individual queries to AI models and tallied what it all adds up to, both right now and for the years ahead.
There’s a lot of data to dig through, and I hope you’ll take the time to explore the whole story. But in the meantime, here are three of my biggest takeaways from working on this project.
1. The energy demands of AI are anything but constant.
If you’ve heard estimates of AI’s toll, it’s probably a single number associated with a query, likely to OpenAI’s ChatGPT. One popular estimate is that writing an email with ChatGPT uses 500 milliliters (or roughly a bottle) of water. But as we started reporting, I was surprised to learn just how much the details of a query can affect its energy demand. No two queries are the same—for several reasons, including their complexity and the particulars of the model being queried.
One key caveat here is that we don’t know much about “closed source” models—for these, companies hold back the details of how they work. (OpenAI’s ChatGPT and Google’s Gemini are examples.) Instead, we worked with researchers who measured the energy it takes to run open-source AI models, for which the source code is publicly available.
But using open-source models, it’s possible to directly measure the energy used to respond to a query rather than just guess. We worked with researchers who generated text, images, and video and measured the energy required for the chips the models are based on to perform the task.
Even just within the text responses, there was a pretty large range of energy needs. A complicated travel itinerary consumed nearly 10 times as much energy as a simple request for a few jokes, for example. An even bigger difference comes from the size of the model used. Larger models with more parameters used up to 70 times more energy than smaller ones for the same prompts.
As you might imagine, there’s also a big difference between text, images, or video. Videos generally took hundreds of times more energy to generate than text responses.
2. What’s powering the grid will greatly affect the climate toll of AI’s energy use.
As the resident climate reporter on this project, I was excited to take the expected energy toll and translate it into an expected emissions burden.
Powering a data center with a nuclear reactor or a whole bunch of solar panels and batteries will not affect our planet the same way as burning mountains of coal. To quantify this idea, we used a figure called carbon intensity, a measure of how dirty a unit of electricity is on a given grid.
We found that the same exact query, with the same exact energy demand, will have a very different climate impact depending on what the data center is powered by, and that depends on the location and the time of day. For example, querying a data center in West Virginia could cause nearly twice the emissions of querying one in California, according to calculations based on average data from 2024.
This point shows why it matters where tech giants are building data centers, what the grid looks like in their chosen locations, and how that might change with more demand from the new infrastructure.
3. There is still so much that we don’t know when it comes to AI and energy.
Our reporting resulted in estimates that are some of the most specific and comprehensive out there. But ultimately, we still have no idea what many of the biggest, most influential models are adding up to in terms of energy and emissions. None of the companies we reached out to were willing to provide numbers during our reporting. Not one.
Adding up our estimates can only go so far, in part because AI is increasingly everywhere. While today you might generally have to go to a dedicated site and type in questions, in the future AI could be stitched into the fabric of our interactions with technology. (See my colleague Will Douglas Heaven’s new story on Google’s I/O showcase: “By putting AI into everything, Google wants to make it invisible.”)
AI could be one of the major forces that shape our society, our work, and our power grid. Knowing more about its consequences could be crucial to planning our future.
The thousands of sprawling acres in rural northeast Louisiana had gone unwanted for nearly two decades. Louisiana authorities bought the land in Richland Parish in 2006 to promote economic development in one of the poorest regions in the state. For years, they marketed the former agricultural fields as the Franklin Farm mega site, first to auto manufacturers (no takers) and after that to other industries that might want to occupy more than a thousand acres just off the interstate.
This story is a part of MIT Technology Review’s series “Power Hungry: AI and our energy future,” on the energy demands and carbon costs of the artificial-intelligence revolution.
So it’s no wonder that state and local politicians were exuberant when Meta showed up. In December, the company announced plans to build a massive $10 billion data center for training its artificial-intelligence models at the site, with operations to begin in 2028. “A game changer,” declared Governor Jeff Landry, citing 5,000 construction jobs and 500 jobs at the data center that are expected to be created and calling it the largest private capital investment in the state’s history. From a rural backwater to the heart of the booming AI revolution!
The AI data center also promises to transform the state’s energy future. Stretching in length for more than a mile, it will be Meta’s largest in the world, and it will have an enormous appetite for electricity, requiring two gigawatts for computation alone (the electricity for cooling and other building needs will add to that). When it’s up and running, it will be the equivalent of suddenly adding a decent-size city to the region’s grid—one that never sleeps and needs a steady, uninterrupted flow of electricity.
To power the data center, Entergy aims to spend $3.2 billion to build three large natural-gas power plants with a total capacity of 2.3 gigawatts and upgrade the grid to accommodate the huge jump in anticipated demand. In its filing to the state’s power regulatory agency, Entergy acknowledged that natural-gas plants “emit significant amounts of CO2” but said the energy source was the only affordable choice given the need to quickly meet the 24-7 electricity demand from the huge data center.
Meta said it will work with Entergy to eventually bring online at least 1.5 gigawatts of new renewables, including solar, but that it had not yet decided which specific projects to fund or when those investments will be made. Meanwhile, the new natural-gas plants, which are scheduled to be up and running starting in 2028 and will have a typical lifetime of around 30 years, will further lock in the state’s commitment to the fossil fuel.
The development has sparked interest from the US Congress; last week, Sheldon Whitehouse, the ranking member of the Senate Committee on Environment and Public Works issued a letter to Meta that called out the company’s plan to power its data center with “new and unabated natural gas generation” and said its promises to offset the resulting emissions “by funding carbon capture and a solar project are vague and offer little reassurance.”
The choice of natural gas as the go-to solution to meet the growing demand for power from AI is not unique to Louisiana. The fossil fuel is already the country’s chief source of electricity generation, and large natural-gas plants are being built around the country to feed electricity to new and planned AI data centers. While some climate advocates have hoped that cleaner renewable power would soon overtake it, the booming power demand from data centers is all but wiping out any prospect that the US will wean itself off natural gas anytime soon.
The reality on the ground is that natural gas is “the default” to meet the exploding power demand from AI data centers, says David Victor, a political scientist at the University of California, San Diego, and co-director of its Deep Decarbonization Project. “The natural-gas plant is the thing that you know how to build, you know what it’s going to cost (more or less), and you know how to scale it and get it approved,” says Victor. “Even for [AI] companies that want to have low emissions profiles and who are big pushers of low or zero carbon, they won’t have a choice but to use gas.”
The preference for natural gas is particularly pronounced in the American South, where plans for multiple large gas-fired plants are in the works in states such as Virginia, North Carolina, South Carolina, and Georgia. Utilities in those states alone are planning some 20 gigawatts of new natural-gas power plants over the next 15 years, according to a recent report. And much of the new demand—particularly in Virginia, South Carolina and Georgia—is coming from data centers; in those 3 states data centers account for around 65 to 85% of projected load growth.
“It’s a long-term commitment in absolutely the wrong direction,” says Greg Buppert, a senior attorney at the Southern Environmental Law Center in Charlottesville, Virginia. If all the proposed gas plants get built in the South over the next 15 years, he says, “we’ll just have to accept that we won’t meet emissions reduction goals.”
But even as it looks more and more likely that natural gas will remain a sizable part of our energy future, questions abound over just what its continued dominance will look like.
For one thing, no one is sure exactly how much electricity AI data centers will need in the future and how large an appetite companies will have for natural gas. Demand for AI could fizzle. Or AI companies could make a concerted effort to shift to renewable energy or nuclear power. Such possibilities mean that the US could be on a path to overbuild natural-gas capacity, which would leave regions saddled with unneeded and polluting fossil-fuel dinosaurs—and residents footing soaring electricity bills to pay off today’s investments.
The good news is that such risks could likely be managed over the next few years, if—and it’s a big if—AI companies are more transparent about how flexible they can be in their seemingly insatiable energy demands.
Around 2016, natural gas overtook coal as the main fuel for electricity generation in the US. And today—despite the rapid rise of solar and wind power, and well-deserved enthusiasm for the falling price of such renewables—natural gas is still king, accounting for around 40% of electricity generated in the US. In Louisiana, which is also a big producer, that share is some 72%, according to a recent audit.
Natural gas burns much cleaner than coal, producing roughly half as much carbon dioxide. In the early days of the gas revolution, many environmental activists and progressive politicians touted it as a valuable “bridge” to renewables and other sources of clean energy. And by some calculations, natural gas has fulfilled that promise. The power sector has been one of the few success stories in lowering US emissions, thanks to its use of natural gas as a replacement for coal.
But natural gas still produces a lot of carbon dioxide when it is burned in conventionally equipped power plants. And fracking causes local air and water pollution. Perhaps most worrisome, drilling and pipelines are releasing substantial amounts of methane, the main ingredient in natural gas, both accidentally and by intentional venting. Methane is a far more potent greenhouse gas than carbon dioxide, and the emissions are a growing concern to climate scientists, albeit one that’s difficult to quantify.
Still, carbon emissions from the power sector will likely continue to drop as coal is further squeezed out and more renewables get built, according to the Rhodium Group, a research consultancy. But Rhodium also projects that if electricity demand from data centers remains high and natural-gas prices low, the fossil fuel will remain the dominant source of power generation at least through 2035 and the transition to cleaner electricity will be much delayed. Rhodium estimates that the continued reign of natural gas will lead to an additional 278 million metric tons of annual US carbon emissions by 2035 (roughly equivalent to the emissions from a large US state such as Florida), relative to a future in which the use of fossil fuel gradually winds down.
Our addiction to natural gas, however, doesn’t have to be a total climate disaster, at least over the longer term. Large AI companies could use their vast leverage to insist that utilities install carbon capture and sequestration (CCS) at power plants and use natural gas sourced with limited methane emissions.
Entergy, for one, says its new gas turbines will be able to incorporate CCS through future upgrades. And Meta says it will help to fund the installation of CCS equipment at one of Entergy’s existing natural-gas power plants in southern Louisiana to help prove out the technology.
But the transition to clean natural gas is a hope that will take decades to realize. Meanwhile, utilities across the country are facing a more imminent and practical challenge: how to meet the sudden demand for gigawatts more power in the next few years without inadvertently building far too much capacity. For many, adding more natural-gas power plants might seem like the safe bet. But what if the explosion in AI demand doesn’t show up?
Times of stress
AI companies tout the need for massive, power-hungry data centers. But estimates for just how much energy it will actually take to train and run AI models vary wildly. And the technology keeps changing, sometimes seemingly overnight. DeepSeek, the new Chinese model that debuted in January, may or may not signal a future of new energy-efficient AI, but it certainly raises the possibility that such advances are possible. Maybe we will find ways to use far more energy-efficient hardware. Or maybe the AI revolution will peter out and many of the massive data centers that companies think they’ll need will never get built. There are already signs that too many have been constructed in China and clues that it might be beginning to happen in the US.
Despite the uncertainty, power providers have the task of drawing up long-term plans for investments to accommodate projected demand. Too little capacity and their customers face blackouts; too much and those customers face outsize electricity bills to fund investments in unneeded power.
There could be a way to lessen the risk of overbuilding natural-gas power, however. Plenty of power is available on average around the country and on most regional grids. Most utilities typically use only about 53% of their available capacity on average during the year, according to a Duke study. The problem is that utilities must be prepared for the few hours when demand spikes—say, because of severe winter weather or a summer heat wave.
The soaring demand from AI data centers is prompting many power providers to plan new capacity to make sure they have plenty of what Tyler Norris, a fellow at Duke’s Nicholas School of the Environment, and his colleagues call “headroom,” to meet any spikes in demand. But after analyzing data from power systems across the country, Norris and his coauthors found that if large AI facilities cut back their electricity use during hours of peak demand, many regional power grids could accommodate those AI customers without adding new generation capacity.
Even a moderate level of flexibility would make a huge difference. The Duke researchers estimate that if data centers cut their electricity use by roughly half for just a few hours during the year, it will allow utilities to handle some additional 76 gigawatts of new demand. That means power providers could effectively absorb the 65 or so additional gigawatts that, according to some predictions, data centers will likely need by 2029.
“The prevailing assumption is that data centers are 100% inflexible,” says Norris. That is, that they need to run at full power all the time. But Norris says AI data centers, particularly ones that are training large foundation models (such as Meta’s facility in Richland Parish), can avoid running at full capacity or shift their computation loads to other data centers around the country—or even ramp up their own backup power—during times when a grid is under stress.
The increased flexibility could allow companies to get AI data centers up and running faster, without waiting for new power plants and upgrades to transmission lines—which can take years to get approved and built. It could also, Norris noted in testimony to the US Congress in early March, provide at least a short-term reprieve on the rush to build more natural-gas power, buying time for utilities to develop and plan for cleaner technologies such as advanced nuclear and enhanced geothermal. It could, he testified, prevent “a hasty overbuild of natural-gas infrastructure.”
AI companies have expressed some interest in their ability to shift around demand for power. But there are still plenty of technology questions around how to make it happen. Late last year, EPRI (the Electric Power Research Institute), a nonprofit R&D group, started a three-year collaboration with power providers, grid operators, and AI companies including Meta and Google, to figure it out. “The potential is very large,” says David Porter, the EPRI vice president who runs the project, but we must show it works “beyond just something on a piece of paper or a computer screen.”
Porter estimates that there are typically 80 to 90 hours a year when a local grid is under stress and it would help for a data center to reduce its energy use. But, he says, AI data centers still need to figure out how to throttle back at those times, and grid operators need to learn how to suddenly subtract and then add back hundreds of megawatts of electricity without disrupting their systems. “There’s still a lot of work to be done so that it’s seamless for the continuous operation of the data centers and seamless for the continuous operation of the grid,” he says.
Footing the bill
Ultimately, getting AI data centers to be more flexible in their power demands will require more than a technological fix. It will require a shift in how AI companies work with utilities and local communities, providing them with more information and insights into actual electricity needs. And it will take aggressive regulators to make sure utilities are rigorously evaluating the power requirements of data centers rather than just reflexively building more natural-gas plants.
“The most important climate policymakers in the country right now are not in Washington. They’re in state capitals, and these are public utility commissioners,” says Costa Samaras, the director of Carnegie Mellon University’s Scott Institute for Energy Innovation.
In Louisiana, those policymakers are the elected officials at the Louisiana Public Service Commission, who are expected to rule later this year on Entergy’s proposed new gas plants and grid upgrades. The LPSC commissioners will decide whether Entergy’s arguments about the huge energy requirements of Meta’s data center and need for full 24/7 power leave no alternative to natural gas.
In the application it filed last fall with LPSC, Entergy said natural-gas power was essential for it to meet demand “throughout the day and night.” Teaming up solar power with battery storage could work “in theory” but would be “prohibitively costly.” Entergy also ruled out nuclear, saying it would take too long and cost too much.
Others are not satisfied with the utility’s judgment. In February, the New Orleans–based Alliance for Affordable Energy and the Union of Concerned Scientists filed a motion with the Louisiana regulators arguing that Entergy did not do a rigorous market evaluation of its options, as required by the commission’s rules. Part of the problem, the groups said, is that Entergy relied on “unsubstantiated assertions” from Meta on its load needs and timeline.
“Entergy is saying [Meta] needs around-the-clock power,” says Paul Arbaje, an analyst for the climate and energy program at the Union of Concerned Scientists. “But we’re just being asked to take [Entergy’s] word for it. Regulators need to be asking tough questions and not just assume that these data centers need to be operated at essentially full capacity all the time.” And, he suggests, if the utility had “started to poke holes at the assumptions that are sometimes taken as a given,” it “would have found other cleaner options.”
In an email response to MIT Technology Review, Entergy said that it has discussed the operational aspects of the facility with Meta, but “as with all customers, Entergy Louisiana will not discuss sensitive matters on behalf of their customers.” In a letter filed with the state’s regulators in early April, Meta said Entergy’s understanding of its energy needs is, in fact, accurate.
The February motion also raised concern over who will end up paying for the new gas plants. Entergy says Meta has signed a 15-year supply contract for the electricity that is meant to help cover the costs of building and running the power plants but didn’t respond to requests by MIT Technology Review for further details of the deal, including what happens if Meta wants to terminate the contract early.
Meta referred MIT Technology Review’s questions about the contract to Entergy but says its policy is to cover the full cost that utilities incur to serve its data centers, including grid upgrades. It also says it is spending over $200 million to support the Richland Parish data centers with new infrastructure, including roads and water systems.
Not everyone is convinced. The Alliance for Affordable Energy, which works on behalf of Louisiana residents, says that the large investments in new gas turbines could mean future rate hikes, in a state where residents already have high electricity bills and suffer from one of country’s most unreliable grids. Of special concern is what happens after the 15 years.
“Our biggest long-term concern is that in 15 years, residential ratepayers [and] small businesses in Louisiana will be left holding the bag for three large gas generators,” says Logan Burke, the alliance’s executive director.
Indeed, consumers across the country have good reasons to fear that their electricity bills will go up as utilities look to meet the increased demand from AI data centers by building new generation capacity. In a paper posted in March, researchers at Harvard Law School argued that utilities “are now forcing the public to pay for infrastructure designed to supply a handful of exceedingly wealthy corporations.”
The Harvard authors write, “Utilities tell [public utility commissions] what they want to hear: that the deals for Big Tech isolate data center energy costs from other ratepayers’ bills and won’t increase consumers’ power prices.” But the complexity of the utilities’ payment data and lack of transparency in the accounting, they say, make verifying this claim “all but impossible.”
The boom in AI data centers is making Big Tech a player in our energy infrastructure and electricity future in a way unimaginable just a few years ago. At their best, AI companies could greatly facilitate the move to cleaner energy by acting as reliable and well-paying customers that provide funding that utilities can use to invest in a more robust and flexible electricity grid. This change can happen without burdening other electricity customers with additional risks and costs. But it will take AI companies committed to that vision. And it will take state regulators who ask tough questions and don’t get carried away by the potential investments being dangled by AI companies.
Huge new AI data centers like the one in Richland Parish could in fact be a huge economic boon by providing new jobs, but residents deserve transparency and input into the negotiations. This is, after all, public infrastructure. Meta may come and go, but Louisiana’s residents will have to live with—and possibly pay for—the changes in the decades to come.
In the high desert east of Reno, Nevada, construction crews are flattening the golden foothills of the Virginia Range, laying the foundations of a data center city.
Google, Tract, Switch, EdgeCore, Novva, Vantage, and PowerHouse are all operating, building, or expanding huge facilities within the Tahoe Reno Industrial Center, a business park bigger than the city of Detroit.
This story is a part of MIT Technology Review’s series “Power Hungry: AI and our energy future,” on the energy demands and carbon costs of the artificial-intelligence revolution.
Meanwhile, Microsoft acquired more than 225 acres of undeveloped property within the center and an even larger plot in nearby Silver Springs, Nevada. Apple is expanding its data center, located just across the Truckee River from the industrial park. OpenAI has said it’s considering building a data center in Nevada as well.
The corporate race to amass computing resources to train and run artificial intelligence models and store information in the cloud has sparked a data center boom in the desert—just far enough away from Nevada’s communities to elude wide notice and, some fear, adequate scrutiny.
Switch, a data center company based in Las Vegas, says the full build-out of its campus at the Tahoe Reno Industrial Center could exceed seven million square feet.
EMILY NAJERA
The full scale and potential environmental impacts of the developments aren’t known, because the footprint, energy needs, and water requirements are often closely guarded corporate secrets. Most of the companies didn’t respond to inquiries from MIT Technology Review, or declined to provide additional information about the projects.
But there’s “a whole lot of construction going on,” says Kris Thompson, who served as the longtime project manager for the industrial center before stepping down late last year. “The last number I heard was 13 million square feet under construction right now, which is massive.”
Indeed, it’s the equivalent of almost five Empire State Buildings laid out flat. In addition, public filings from NV Energy, the state’s near-monopoly utility, reveal that a dozen data-center projects, mostly in this area, have requested nearly six gigawatts of electricity capacity within the next decade.
That would make the greater Reno area—the biggest little city in the world—one of the largest data-center markets around the globe.
It would also require expanding the state’s power sector by about 40%, all for a single industry in an explosive growth stage that may, or may not, prove sustainable. The energy needs, in turn, suggest those projects could consume billions of gallons of water per year, according to an analysis conducted for this story.
Construction crews are busy building data centers throughout the Tahoe Reno Industrial Center.
EMILY NAJERA
The build-out of a dense cluster of energy and water-hungry data centers in a small stretch of the nation’s driest state, where climate change is driving up temperatures faster than anywhere else in the country, has begun to raise alarms among water experts, environmental groups, and residents. That includes members of the Pyramid Lake Paiute Tribe, whose namesake water body lies within their reservation and marks the end point of the Truckee River, the region’s main source of water.
Much of Nevada has suffered through severe drought conditions for years, farmers and communities are drawing down many of the state’s groundwater reservoirs faster than they can be refilled, and global warming is sucking more and more moisture out of the region’s streams, shrubs, and soils.
“Telling entities that they can come in and stick more straws in the ground for data centers is raising a lot of questions about sound management,” says Kyle Roerink, executive director of the Great Basin Water Network, a nonprofit that works to protect water resources throughout Nevada and Utah.
“We just don’t want to be in a situation where the tail is wagging the dog,” he later added, “where this demand for data centers is driving water policy.”
Luring data centers
In the late 1850s, the mountains southeast of Reno began enticing prospectors from across the country, who hoped to strike silver or gold in the famed Comstock Lode. But Storey County had few residents or economic prospects by the late 1990s, around the time when Don Roger Norman, a media-shy real estate speculator, spotted a new opportunity in the sagebrush-covered hills.
He began buying up tens of thousands of acres of land for tens of millions of dollars and lining up development approvals to lure industrial projects to what became the Tahoe Reno Industrial Center. His partners included Lance Gilman, a cowboy-hat-wearing real estate broker, who later bought the nearby Mustang Ranch brothel and won a seat as a county commissioner.
In 1999, the county passed an ordinance that preapproves companies to develop most types of commercial and industrial projects across the business park, cutting months to years off the development process. That helped cinch deals with a flock of tenants looking to build big projects fast, including Walmart, Tesla, and Redwood Materials. Now the promise of fast permits is helping to draw data centers by the gigawatt.
On a clear, cool January afternoon, Brian Armon, a commercial real estate broker who leads the industrial practices group at NAI Alliance, takes me on a tour of the projects around the region, which mostly entails driving around the business center.
Lance Gilman, a local real estate broker, helped to develop the Tahoe Reno Industrial Center and land some of its largest tenants.
GREGG SEGAL
After pulling off Interstate 80 onto USA Parkway, he points out the cranes, earthmovers, and riprap foundations, where a variety of data centers are under construction. Deeper into the industrial park, Armon pulls up near Switch’s long, low, arched-roof facility, which sits on a terrace above cement walls and security gates. The Las Vegas–based company says the first phase of its data center campus encompasses more than a million square feet, and that the full build-out will cover seven times that space.
Over the next hill, we turn around in Google’s parking lot. Cranes, tents, framing, and construction equipment extend behind the company’s existing data center, filling much of the 1,210-acre lot that the search engine giant acquired in 2017.
Last August, during an event at the University of Nevada, Reno, the company announced it would spend $400 million to expand the data center campus along with another one in Las Vegas.
Thompson says that the development company, Tahoe Reno Industrial LLC, has now sold off every parcel of developable land within the park (although several lots are available for resale following the failed gamble of one crypto tenant).
When I ask Armon what’s attracting all the data centers here, he starts with the fast approvals but cites a list of other lures as well: The inexpensive land. NV Energy’s willingness to strike deals to supply relatively low-cost electricity. Cool nighttime and winter temperatures, as far as American deserts go, which reduce the energy and water needs. The proximity to tech hubs such as Silicon Valley, which cuts latency for applications in which milliseconds matter. And the lack of natural disasters that could shut down the facilities, at least for the most part.
“We are high in seismic activity,” he says. “But everything else is good. We’re not going to have a tornado or flood or a devastating wildfire.”
Then there’s the generous tax policies.
In 2023, Novva, a Utah-based data center company, announced plans to build a 300,000-square-foot facility within the industrial business park.
Nevada doesn’t charge corporate income tax, and it has also enacted deep tax cuts specifically for data centers that set up shop in the state. That includes abatements of up to 75% on property tax for a decade or two—and nearly as much of a bargain on the sales and use taxes applied to equipment purchased for the facilities.
Data centers don’t require many permanent workers to run the operations, but the projects have created thousands of construction jobs. They’re also helping to diversify the region’s economy beyond casinos and generating tax windfalls for the state, counties, and cities, says Jeff Sutich, executive director of the Northern Nevada Development Authority. Indeed, just three data-center projects, developed by Apple, Google, and Vantage, will produce nearly half a billion dollars in tax revenue for Nevada, even with those generous abatements, according to the Nevada Governor’s Office of Economic Development.
The question is whether the benefits of data centers are worth the tradeoffs for Nevadans, given the public health costs, greenhouse-gas emissions, energy demands, and water strains.
The rain shadow
The Sierra Nevada’s granite peaks trace the eastern edge of California, forcing Pacific Ocean winds to rise and cool. That converts water vapor in the air into the rain and snow that fill the range’s tributaries, rivers, and lakes.
But the same meteorological phenomenon casts a rain shadow over much of neighboring Nevada, forming an arid expanse known as the Great Basin Desert. The state receives about 10 inches of precipitation a year, about a third of the national average.
The Truckee River draws from the melting Sierra snowpack at the edge of Lake Tahoe, cascades down the range, and snakes through the flatlands of Reno and Sparks. It forks at the Derby Dam, a Reclamation Act project a few miles from the Tahoe Reno Industrial Center, which diverts water to a farming region further east while allowing the rest to continue north toward Pyramid Lake.
Along the way, an engineered system of reservoirs, canals, and treatment plants divert, store, and release water from the river, supplying businesses, cities, towns, and native tribes across the region. But Nevada’s population and economy are expanding, creating more demands on these resources even as they become more constrained.
The Truckee River, which originates at Lake Tahoe and terminates at Pyramid Lake, is the major water source for cities, towns, and farms across northwestern Nevada.
EMILY NAJERA
Throughout much of the 2020s the state has suffered through one of the hottest and most widespread droughts on record, extending two decades of abnormally dry conditions across the American West. Some scientists fear it may constitute an emerging megadrought.
About 50% of Nevada currently faces moderate to exceptional drought conditions. In addition, more than half of the state’s hundreds of groundwater basins are already “over-appropriated,” meaning the water rights on paper exceed the levels believed to be underground.
It’s not clear if climate change will increase or decrease the state’s rainfall levels, on balance. But precipitation patterns are expected to become more erratic, whiplashing between short periods of intense rainfall and more-frequent, extended, or severe droughts.
In addition, more precipitation will fall as rain rather than snow, shortening the Sierra snow season by weeks to months over the coming decades.
“In the extreme case, at the end of the century, that’s pretty much all of winter,” says Sean McKenna, executive director of hydrologic sciences at the Desert Research Institute, a research division of the Nevada System of Higher Education.
That loss will undermine an essential function of the Sierra snowpack: reliably delivering water to farmers and cities when it’s most needed in the spring and summer, across both Nevada and California.
These shifting conditions will require the region to develop better ways to store, preserve, and recycle the water it does get, McKenna says. Northern Nevada’s cities, towns, and agencies will also need to carefully evaluate and plan for the collective impacts of continuing growth and development on the interconnected water system, particularly when it comes to water-hungry projects like data centers, he adds.
“We can’t consider each of these as a one-off, without considering that there may be tens or dozens of these in the next 15 years,” McKenna says.
Thirsty data centers
Data centers suck up water in two main ways.
As giant rooms of server racks process information and consume energy, they generate heat that must be shunted away to prevent malfunctions and damage to the equipment. The processing units optimized for training and running AI models often draw more electricity and, in turn, produce more heat.
To keep things cool, more and more data centers have turned to liquid cooling systems that don’t need as much electricity as fan cooling or air-conditioning.
These often rely on water to absorb heat and transfer it to outdoor cooling towers, where much of the moisture evaporates. Microsoft’s US data centers, for instance, could have directly evaporated nearly 185,000 gallons of “clean freshwater” in the course of training OpenAI’s GPT-3 large language model, according to a 2023 preprint study led by researchers at the University of California, Riverside. (The research has since been peer-reviewed and is awaiting publication.)
What’s less appreciated, however, is that the larger data-center drain on water generally occurs indirectly, at the power plants generating extra electricity for the turbocharged AI sector. These facilities, in turn, require more water to cool down equipment, among other purposes.
You have to add up both uses “to reflect the true water cost of data centers,” says Shaolei Ren, an associate professor of electrical and computer engineering at UC Riverside and coauthor of the study.
Ren estimates that the 12 data-center projects listed in NV Energy’s report would directly consume between 860 million gallons and 5.7 billion gallons a year, based on the requested electricity capacity. (“Consumed” here means the water is evaporated, not merely withdrawn and returned to the engineered water system.) The indirect water drain associated with electricity generation for those operations could add up to 15.5 billion gallons, based on the average consumption of the regional grid.
The exact water figures would depend on shifting climate conditions, the type of cooling systems each data center uses, and the mix of power sources that supply the facilities.
Solar power, which provides roughly a quarter of Nevada’s power, requires relatively little water to operate, for instance. But natural-gas plants, which generate about 56%, withdraw 2,803 gallons per megawatt-hour on average, according to the Energy Information Administration.
Geothermal plants, which produce about 10% of the state’s electricity by cycling water through hot rocks, generally consume less water than fossil fuel plants do but often require more water than other renewables, according to some research.
But here too, the water usage varies depending on the type of geothermal plant in question. Google has lined upseveral deals to partially power its data centers through Fervo Energy, which has helped to commercialize an emerging approach that injects water under high pressure to fracture rock and form wells deep below the surface.
The company stresses that it doesn’t evaporate water for cooling and that it relies on brackish groundwater, not fresh water, to develop and run its plants. In a recent post, Fervo noted that its facilities consume significantly less water per megawatt-hour than coal, nuclear, or natural-gas plants do.
Part of NV Energy’s proposed plan to meet growing electricity demands in Nevada includes developing several natural-gas peaking units, adding more than one gigawatt of solar power and installing another gigawatt of battery storage. It’s also forging ahead with a more than $4 billion transmission project.
But the company didn’t respond to questions concerning how it will supply all of the gigawatts of additional electricity requested by data centers, if the construction of those power plants will increase consumer rates, or how much water those facilities are expected to consume.
NV Energy operates a transmission line, substation, and power plant in or around the Tahoe Reno Industrial Center.
EMILY NAJERA
“NV Energy teams work diligently on our long-term planning to make investments in our infrastructure to serve new customers and the continued growth in the state without putting existing customers at risk,” the company said in a statement.
An added challenge is that data centers need to run around the clock. That will often compel utilities to develop new electricity-generating sources that can run nonstop as well, as natural-gas, geothermal, or nuclear plants do, says Emily Grubert, an associate professor of sustainable energy policy at the University of Notre Dame, who has studied the relative water consumption of electricity sources.
“You end up with the water-intensive resources looking more important,” she adds.
Even if NV Energy and the companies developing data centers do strive to power them through sources with relatively low water needs, “we only have so much ability to add six gigawatts to Nevada’s grid,” Grubert explains. “What you do will never be system-neutral, because it’s such a big number.”
Securing supplies
On a mid-February morning, I meet TRI’s Thompson and Don Gilman, Lance Gilman’s son, at the Storey County offices, located within the industrial center.
“I’m just a country boy who sells dirt,” Gilman, also a real estate broker, says by way of introduction.
We climb into his large SUV and drive to a reservoir in the heart of the industrial park, filled nearly to the lip.
Thompson explains that much of the water comes from an on-site treatment facility that filters waste fluids from companies in the park. In addition, tens of millions of gallons of treated effluent will also likely flow into the tank this year from the Truckee Meadows Water Authority Reclamation Facility, near the border of Reno and Sparks. That’s thanks to a 16-mile pipeline that the developers, the water authority, several tenants, and various local cities and agencies partnered to build, through a project that began in 2021.
“Our general improvement district is furnishing that water to tech companies here in the park as we speak,” Thompson says. “That helps preserve the precious groundwater, so that is an environmental feather in the cap for these data centers. They are focused on environmental excellence.”
The reservoir within the industrial business park provides water to data centers and other tenants.
EMILY NAJERA
But data centers often need drinking-quality water—not wastewater merely treated to irrigation standards—for evaporative cooling, “to avoid pipe clogs and/or bacterial growth,” the UC Riverside study notes. For instance, Google says its data centers withdrew about 7.7 billion gallons of water in 2023, and nearly 6 billion of those gallons were potable.
Tenants in the industrial park can potentially obtain access to water from the ground and the Truckee River, as well. From early on, the master developers worked hard to secure permits to water sources, since they are nearly as precious as development entitlements to companies hoping to build projects in the desert.
Initially, the development company controlled a private business, the TRI Water and Sewer Company, that provided those services to the business park’s tenants, according to public documents. The company set up wells, a water tank, distribution lines, and a sewer disposal system.
But in 2000, the board of county commissioners established a general improvement district, a legal mechanism for providing municipal services in certain parts of the state, to manage electricity and then water within the center. It, in turn, hired TRI Water and Sewer as the operating company.
As of its 2020 service plan, the general improvement district held permits for nearly 5,300 acre-feet of groundwater, “which can be pumped from well fields within the service area and used for new growth as it occurs.” The document lists another 2,000 acre-feet per year available from the on-site treatment facility, 1,000 from the Truckee River, and 4,000 more from the effluent pipeline.
Those figures haven’t budged much since, according to Shari Whalen, general manager of the TRI General Improvement District. All told, they add up to more than 4 billion gallons of water per year for all the needs of the industrial park and the tenants there, data centers and otherwise.
Whalen says that the amount and quality of water required for any given data center depends on its design, and that those matters are worked out on a case-by-case basis.
When asked if the general improvement district is confident that it has adequate water resources to supply the needs of all the data centers under development, as well as other tenants at the industrial center, she says: “They can’t just show up and build unless they have water resources designated for their projects. We wouldn’t approve a project if it didn’t have those water resources.”
Waterbattles
As the region’s water sources have grown more constrained, lining up supplies has become an increasingly high-stakes and controversial business.
More than a century ago, the US federal government filed a lawsuit against an assortment of parties pulling water from the Truckee River. The suit would eventually establish that the Pyramid Lake Paiute Tribe’s legal rights to water for irrigation superseded other claims. But the tribe has been fighting to protect those rights and increase flows from the river ever since, arguing that increasing strains on the watershed from upstream cities and businesses threaten to draw away water reserved for reservation farming, decrease lake levels, and harm native fish.
The Pyramid Lake Paiute Tribe considers the water body and its fish, including the endangered cui-ui and threatened Lahontan cutthroat trout, to be essential parts of its culture, identity, and way of life. The tribe was originally named Cui-ui Ticutta, which translates to cui-ui eaters. The lake continues to provide sustenance as well as business for the tribe and its members, a number of whom operate boat charters and fishing guide services.
“It’s completely tied into us as a people,” says Steven Wadsworth, chairman of the Pyramid Lake Paiute Tribe.
“That is what has sustained us all this time,” he adds. “It’s just who we are. It’s part of our spiritual well-being.”
Steven Wadsworth, chairman of the Pyramid Lake Paiute Tribe, fears that data centers will divert water that would otherwise reach the tribe’s namesake lake.
EMILY NAJERA
In recent decades, the tribe has sued the Nevada State Engineer, Washoe County, the federal government, and others for overallocating water rights and endangering the lake’s fish. It also protested the TRI General Improvement District’s applications to draw thousands of additional acre‑feet of groundwater from a basin near the business park. In 2019, the State Engineer’s office rejected those requests, concluding that the basin was already fully appropriated.
More recently, the tribe took issue with the plan to build the pipeline and divert effluent that would have flown into the Truckee, securing an agreement that required the Truckee Meadows Water Authority and other parties to add back several thousand acre‑feet of water to the river.
Whalen says she’s sensitive to Wadsworth’s concerns. But she says that the pipeline promises to keep a growing amount of treated wastewater out of the river, where it could otherwise contribute to rising salt levels in the lake.
“I think that the pipeline from [the Truckee Meadows Water Authority] to our system is good for water quality in the river,” she says. “I understand philosophically the concerns about data centers, but the general improvement district is dedicated to working with everyone on the river for regional water-resource planning—and the tribe is no exception.”
Water efficiency
In an email, Thompson added that he has “great respect and admiration,” for the tribe and has visited the reservation several times in an effort to help bring industrial or commercial development there.
He stressed that all of the business park’s groundwater was “validated by the State Water Engineer,” and that the rights to surface water and effluent were purchased “for fair market value.”
During the earlier interview at the industrial center, he and Gilman had both expressed confidence that tenants in the park have adequate water supplies, and that the businesses won’t draw water away from other areas.
“We’re in our own aquifer, our own water basin here,” Thompson said. “You put a straw in the ground here, you’re not going to pull water from Fernley or from Reno or from Silver Springs.”
Gilman also stressed that data-center companies have gotten more water efficient in recent years, echoing a point others made as well.
“With the newer technology, it’s not much of a worry,” says Sutich, of the Northern Nevada Development Authority. “The technology has come a long way in the last 10 years, which is really giving these guys the opportunity to be good stewards of water usage.”
An aerial view of the cooling tower fans at Google’s data center in the Tahoe Reno Industrial Center.
GOOGLE
Indeed, Google’s existing Storey County facility is air-cooled, according to the company’s latest environmental report. The data center withdrew 1.9 million gallons in 2023 but only consumed 200,000 gallons. The rest cycles back into the water system.
Google said all the data centers under construction on its campus will also “utilize air-cooling technology.” The company didn’t respond to a question about the scale of its planned expansion in the Tahoe Reno Industrial Center, and referred a question about indirect water consumption to NV Energy.
The search giant has stressed that it strives to be water efficient across all of its data centers, and decides whether to use air or liquid cooling based on local supply and projected demand, among other variables.
Four years ago, the company set a goal of replenishing more water than it consumes by 2030. Locally, it also committed to provide half a million dollars to the National Forest Foundation to improve the Truckee River watershed and reduce wildfire risks.
Microsoft clearly suggestedin earlier news reports that the Silver Springs land it purchased around the end of 2022 would be used for a data center. NAI Alliance’s market real estate report identifies that lot, as well as the parcel Microsoft purchased within the Tahoe Reno Industrial Center, as data center sites.
But the company now declines to specify what it intends to build in the region.
“While the land purchase is public knowledge, we have not disclosed specific details [of] our plans for the land or potential development timelines,” wrote Donna Whitehead, a Microsoft spokesperson, in an email.
Workers have begun grading land inside a fenced off lot within the Tahoe Reno Industrial Center.
EMILY NAJERA
Microsoft has also scaled down its global data-center ambitions, backing away from several projects in recent months amid shifting economic conditions, according to various reports.
Whatever it ultimately does or doesn’t build, the company stresses that it has made strides to reduce water consumption in its facilities. Late last year, the company announced that it’s using “chip-level cooling solutions” in data centers, which continually circulate water between the servers and chillers through a closed loop that the company claims doesn’t lose any water to evaporation. It says the design requires only a “nominal increase” in energy compared to its data centers that rely on evaporative water cooling.
Others seem to be taking a similar approach. EdgeCore also said its 900,000-square-foot data center at the Tahoe Reno Industrial Center will rely on an “air-cooled closed-loop chiller” that doesn’t require water evaporation for cooling.
But some of the companies seem to have taken steps to ensure access to significant amounts of water. Switch, for instance, took a lead role in developing the effluent pipeline. In addition, Tract, which develops campuses on which third-party data centers can build their own facilities, has said it lined up more than 1,100 acre-feet of water rights, the equivalent of nearly 360 million gallons a year.
Apple, Novva, Switch, Tract, and Vantage didn’t respond to inquiries from MIT Technology Review.
Coming conflicts
The suggestion that companies aren’t straining water supplies when they adopt air cooling is, in many cases, akin to saying they’re not responsible for the greenhouse gas produced through their power use simply because it occurs outside of their facilities. In fact, the additional water used at a power plant to meet the increased electricity needs of air cooling may exceed any gains at the data center, Ren, of UC Riverside, says.
“That’s actually very likely, because it uses a lot more energy,” he adds.
That means that some of the companies developing data centers in and around Storey County may simply hand off their water challenges to other parts of Nevada or neighboring states across the drying American West, depending on where and how the power is generated, Ren says.
Google has said its air-cooled facilities require about 10% more electricity, and its environmental report notes that the Storey County facility is one of its two least-energy-efficient data centers.
Pipes running along Google’s data center campus help the search company cool its servers.
GOOGLE
Some fear there’s also a growing mismatch between what Nevada’s water permits allow, what’s actually in the ground, and what nature will provide as climate conditions shift. Notably, the groundwater committed to all parties from the Tracy Segment basin—a long-fought-over resource that partially supplies the TRI General Improvement District—already exceeds the “perennial yield.” That refers to the maximum amount that can be drawn out every year without depleting the reservoir over the long term.
“If pumping does ultimately exceed the available supply, that means there will be conflict among users,” Roerink, of the Great Basin Water Network, said in an email. “So I have to wonder: Who could be suing whom? Who could be buying out whom? How will the tribe’s rights be defended?”
The Truckee Meadows Water Authority, the community-owned utility that manages the water system for Reno and Sparks, said it is planning carefully for the future and remains confident there will be “sufficient resources for decades to come,” at least within its territory east of the industrial center.
Storey County, the Truckee-Carson Irrigation District, and the State Engineer’s office didn’t respond to questions or accept interview requests.
Open for business
As data center proposals have begun shifting into Northern Nevada’s cities, more local residents and organizations have begun to take notice and express concerns. The regional division of the Sierra Club, for instance, recently sought to overturn the approval of Reno’s first data center, about 20 miles west of the Tahoe Reno Industrial Center.
Olivia Tanager, director of the Sierra Club’s Toiyabe Chapter, says the environmental organization was shocked by the projected electricity demands from data centers highlighted in NV Energy’s filings.
Nevada’s wild horses are a common sight along USA Parkway, the highway cutting through the industrial business park.
EMILY NAJERA
“We have increasing interest in understanding the impact that data centers will have to our climate goals, to our grid as a whole, and certainly to our water resources,” she says. “The demands are extraordinary, and we don’t have that amount of water to toy around with.”
During a city hall hearing in January that stretched late into the evening, she and a line of residents raised concerns about the water, energy, climate, and employment impacts of AI data centers. At the end, though, the city council upheld the planning department’s approval of the project, on a 5-2 vote.
“Welcome to Reno,” Kathleen Taylor, Reno’s vice mayor, said before casting her vote. “We’re open for business.”
Where the river ends
In late March, I walk alongside Chairman Wadsworth, of the Pyramid Lake Paiute Tribe, on the shores of Pyramid Lake, watching a row of fly-fishers in waders cast their lines into the cold waters.
The lake is the largest remnant of Lake Lahontan, an Ice Age inland sea that once stretched across western Nevada and would have submerged present-day Reno. But as the climate warmed, the lapping waters retreated, etching erosional terraces into the mountainsides and exposing tufa deposits around the lake, large formations of porous rock made of calcium-carbonate. That includes the pyramid-shaped island on the eastern shore that inspired the lake’s name.
A lone angler stands along the shores of Pyramid Lake.
In the decades after the US Reclamation Service completed the Derby Dam in 1905, Pyramid Lake declined another 80 feet and nearby Winnemucca Lake dried up entirely.
“We know what happens when water use goes unchecked,” says Wadsworth, gesturing eastward toward the range across the lake, where Winnemucca once filled the next basin over. “Because all we have to do is look over there and see a dry, barren lake bed that used to be full.”
In an earlier interview, Wadsworth acknowledged that the world needs data centers. But he argued they should be spread out across the country, not densely clustered in the middle of the Nevada desert.
Given the fierce competition for resources up to now, he can’t imagine how there could be enough water to meet the demands of data centers, expanding cities, and other growing businesses without straining the limited local supplies that should, by his accounting, flow to Pyramid Lake.
He fears these growing pressures will force the tribe to wage new legal battles to protect their rights and preserve the lake, extending what he refers to as “a century of water wars.”
“We have seen the devastating effects of what happens when you mess with Mother Nature,” Wadsworth says. “Part of our spirit has left us. And that’s why we fight so hard to hold on to what’s left.”
This story is a part of MIT Technology Review’s series “Power Hungry: AI and our energy future,” on the energy demands and carbon costs of the artificial-intelligence revolution.
Measuring the energy used by an AI model is not like evaluating a car’s fuel economy or an appliance’s energy rating. There’s no agreed-upon method or public database of values. There are no regulators who enforce standards, and consumers don’t get the chance to evaluate one model against another.
Despite the fact that billions of dollars are being poured into reshaping energy infrastructure around the needs of AI, no one has settled on a way to quantify AI’s energy usage. Worse, companies are generally unwilling to disclose their own piece of the puzzle. There are also limitations to estimating the emissions associated with that energy demand, because the grid hosts a complicated, ever-changing mix of energy sources.
It’s a big mess, basically. So, that said, here are the many variables, assumptions, and caveats that we used to calculate the consequences of an AI query. (You can see the full results of our investigation here.)
Measuring the energy a model uses
Companies like OpenAI, dealing in “closed-source” models, generally offer access to their systems through an interface where you input a question and receive an answer. What happens in between—which data center in the world processes your request, the energy it takes to do so, and the carbon intensity of the energy sources used—remains a secret, knowable only to the companies. There are few incentives for them to release this information, and so far, most have not.
That’s why, for our analysis, we looked at open-source models. They serve as a very imperfect proxy but the best one we have. (OpenAI, Microsoft, and Google declined to share specifics on how much energy their closed-source models use.)
The best resources for measuring the energy consumption of open-source AI models are AI Energy Score, ML.Energy, and MLPerf Power. The team behind ML.Energy assisted us with our text and image model calculations, and the team behind AI Energy Score helped with our video model calculations.
Text models
AI models use up energy in two phases: when they initially learn from vast amounts of data, called training, and when they respond to queries, called inference. When ChatGPT was launched a few years ago, training was the focus, as tech companies raced to keep up and build ever-bigger models. But now, inference is where the most energy is used.
The most accurate way to understand how much energy an AI model uses in the inference stage is to directly measure the amount of electricity used by the server handling the request. Servers contain all sorts of components—powerful chips called GPUs that do the bulk of the computing, other chips called CPUs, fans to keep everything cool, and more. Researchers typically measure the amount of power the GPU draws and estimate the rest (more on this shortly).
To do this, we turned to PhD candidate Jae-Won Chung and associate professor Mosharaf Chowdhury at the University of Michigan, who lead the ML.Energy project. Once we collected figures for different models’ GPU energy use from their team, we had to estimate how much energy is used for other processes, like cooling. We examined research literature, including a 2024 paper from Microsoft, to understand how much of a server’s total energy demand GPUs are responsible for. It turns out to be about half. So we took the team’s GPU energy estimate and doubled it to get a sense of total energy demands.
The ML.Energy team uses a batch of 500 prompts from a larger dataset to test models. The hardware is kept the same throughout; the GPU is a popular Nvidia chip called the H100. We decided to focus on models of three sizes from the Meta Llama family: small (8 billion parameters), medium (70 billion), and large (405 billion). We also identified a selection of prompts to test. We compared these with the averages for the entire batch of 500 prompts.
Image models
Stable Diffusion 3 from Stability AI is one of the most commonly used open-source image-generating models, so we made it our focus. Though we tested multiple sizes of the text-based Meta Llama model, we focused on one of the most popular sizes of Stable Diffusion 3, with 2 billion parameters.
The team uses a dataset of example prompts to test a model’s energy requirements. Though the energy used by large language models is determined partially by the prompt, this isn’t true for diffusion models. Diffusion models can be programmed to go through a prescribed number of “denoising steps” when they generate an image or video, with each step being an iteration of the algorithm that adds more detail to the image. For a given step count and model, all images generated have the same energy footprint.
The more steps, the higher quality the end result—but the more energy used. Numbers of steps vary by model and application, but 25 is pretty common, and that’s what we used for our standard quality. For higher quality, we used 50 steps.
We mentioned that GPUs are usually responsible for about half of the energy demands of large language model requests. There is not sufficient research to know how this changes for diffusion models that generate images and videos. In the absence of a better estimate, and after consulting with researchers, we opted to stick with this 50% rule of thumb for images and videos too.
Video models
Chung and Chowdhury do test video models, but only ones that generate short, low-quality GIFs. We don’t think the videos these models produce mirror the fidelity of the AI-generated video that many people are used to seeing.
Instead, we turned to Sasha Luccioni, the AI and climate lead at Hugging Face, who directs the AI Energy Score project. She measures the energy used by the GPU during AI requests. We chose two versions of the CogVideoX model to test: an older, lower-quality version and a newer, higher-quality one.
We asked Luccioni to use her tool, called Code Carbon, to test both and measure the results of a batch of video prompts we selected, using the same hardware as our text and image tests to keep as many variables as possible the same. She reported the GPU energy demands, which we again doubled to estimate total energy demands.
Tracing where that energy comes from
After we understand how much energy it takes to respond to a query, we can translate that into the total emissions impact. Doing so requires looking at the power grid from which data centers draw their electricity.
Nailing down the climate impact of the grid can be complicated, because it’s both interconnected and incredibly local. Imagine the grid as a system of connected canals and pools of water. Power plants add water to the canals, and electricity users, or loads, siphon it out. In the US, grid interconnections stretch all the way across the country. So, in a way, we’re all connected, but we can also break the grid up into its component pieces to get a sense for how energy sources vary across the country.
Understanding carbon intensity
The key metric to understand here is called carbon intensity, which is basically a measure of how many grams of carbon dioxide pollution are released for every kilowatt-hour of electricity that’s produced.
To get carbon intensity figures, we reached out to Electricity Maps, a Danish startup company that gathers data on grids around the world. The team collects information from sources including governments and utilities and uses them to publish historical and real-time estimates of the carbon intensity of the grid. You can find more about their methodology here.
The company shared with us historical data from 2024, both for the entire US and for a few key balancing authorities (more on this in a moment). After discussions with Electricity Maps founder Olivier Corradi and other experts, we made a few decisions about which figures we would use in our calculations.
One way to measure carbon intensity is to simply look at all the power plants that are operating on the grid, add up the pollution they’re producing at the moment, and divide that total by the electricity they’re producing. But that doesn’t account for the emissions that are associated with building and tearing down power plants, which can be significant. So we chose to use carbon intensity figures that account for the whole life cycle of a power plant.
We also chose to use the consumption-based carbon intensity of energy rather than production-based. This figure accounts for imports and exports moving between different parts of the grid and best represents the electricity that’s being used, in real time, within a given region.
For most of the calculations you see in the story, we used the average carbon intensity for the US for 2024, according to Electricity Maps, which is 402.49 grams of carbon dioxide equivalent per kilowatt-hour.
Understanding balancing authorities
While understanding the picture across the entire US can be helpful, the grid can look incredibly different in different locations.
One way we can break things up is by looking at balancing authorities. These are independent bodies responsible for grid balancing in a specific region. They operate mostly independently, though there’s a constant movement of electricity between them as well. There are 66 balancing authorities in the US, and we can calculate a carbon intensity for the part of the grid encompassed by a specific balancing authority.
Electricity Maps provided carbon intensity figures for a few key balancing authorities, and we focused on several that play the largest roles in data center operations. ERCOT (which covers most of Texas) and PJM (a cluster of states on the East Coast, including Virginia, Pennsylvania, and New Jersey) are two of the regions with the largest burden of data centers, according to research from the Harvard School of Public Health.
We added CAISO (in California) because it covers the most populated state in the US. CAISO also manages a grid with a significant number of renewable energy sources, making it a good example of how carbon intensity can change drastically depending on the time of day. (In the middle of the day, solar tends to dominate, while natural gas plays a larger role overnight, for example.)
One key caveat here is that we’re not entirely sure where companies tend to send individual AI inference requests. There are clusters of data centers in the regions we chose as examples, but when you use a tech giant’s AI model, your request could be handled by any number of data centers owned or contracted by the company. One reasonable approximation is location: It’s likely that the data center servicing a request is close to where it’s being made, so a request on the West Coast might be most likely to be routed to a data center on that side of the country.
Explaining what we found
To better contextualize our calculations, we introduced a few comparisons people might be more familiar with than kilowatt-hours and grams of carbon dioxide. In a few places, we took the amount of electricity estimated to be used by a model and calculated how long that electricity would be able to power a standard microwave, as well as how far it might take someone on an e-bike.
In the case of the e-bike, we assumed an efficiency of 25 watt-hours per mile, which falls in the range of frequently cited efficiencies for a pedal-assisted bike. For the microwave, we assumed an 800-watt model, which falls within the average range in the US.
We also introduced a comparison to contextualize greenhouse gas emissions: miles driven in a gas-powered car. For this, we used data from the US Environmental Protection Agency, which puts the weighted average fuel economy of vehicles in the US in 2022 at 393 grams of carbon dioxide equivalent per mile.
Predicting how much energy AI will use in the future
After measuring the energy demand of an individual query and the emissions it generated, it was time to estimate how all of this added up to national demand.
There are two ways to do this. In a bottom-up analysis, you estimate how many individual queries there are, calculate the energy demands of each, and add them up to determine the total. For a top-down look, you estimate how much energy all data centers are using by looking at larger trends.
Bottom-up is particularly difficult, because, once again, closed-source companies do not share such information and declined to talk specifics with us. While we can make some educated guesses to give us a picture of what might be happening right now, looking into the future is perhaps better served by taking a top-down approach.
This data is scarce as well. The most important report was published in December by the Lawrence Berkeley National Laboratory, which is funded by the Department of Energy, and the report authors noted that it’s only the third such report released in the last 20 years. Academic climate and energy researchers we spoke with said it’s a major problem that AI is not considered its own economic sector for emissions measurements, and there aren’t rigorous reporting requirements. As a result, it’s difficult to track AI’s climate toll.
Still, we examined the report’s results, compared them with other findings and estimates, and consulted independent experts about the data. While much of the report was about data centers more broadly, we drew out data points that were specific to the future of AI.
Company goals
We wanted to contrast these figures with the amounts of energy that AI companies themselves say they need. To do so, we collected reports by leading tech and AI companies about their plans for energy and data center expansions, as well as the dollar amounts they promised to invest. Where possible, we fact-checked the promises made in these claims. (Meta and Microsoft’s pledges to use more nuclear power, for example, would indeed reduce the carbon emissions of the companies, but it will take years, if not decades, for these additional nuclear plants to come online.)
Requests to companies
We submitted requests to Microsoft, Google, and OpenAI to have data-driven conversations about their models’ energy demands for AI inference. None of the companies made executives or leadership available for on-the-record interviews about their energy usage.
