You’re out of free articles.
Log in
To continue reading, log in to your account.
Create a Free Account
To unlock more free articles, please create a free account.
Sign In or Create an Account.
By continuing, you agree to the Terms of Service and acknowledge our Privacy Policy
Welcome to Heatmap
Thank you for registering with Heatmap. Climate change is one of the greatest challenges of our lives, a force reshaping our economy, our politics, and our culture. We hope to be your trusted, friendly, and insightful guide to that transformation. Please enjoy your free articles. You can check your profile here .
subscribe to get Unlimited access
Offer for a Heatmap News Unlimited Access subscription; please note that your subscription will renew automatically unless you cancel prior to renewal. Cancellation takes effect at the end of your current billing period. We will let you know in advance of any price changes. Taxes may apply. Offer terms are subject to change.
Subscribe to get unlimited Access
Hey, you are out of free articles but you are only a few clicks away from full access. Subscribe below and take advantage of our introductory offer.
subscribe to get Unlimited access
Offer for a Heatmap News Unlimited Access subscription; please note that your subscription will renew automatically unless you cancel prior to renewal. Cancellation takes effect at the end of your current billing period. We will let you know in advance of any price changes. Taxes may apply. Offer terms are subject to change.
Create Your Account
Please Enter Your Password
Forgot your password?
Please enter the email address you use for your account so we can send you a link to reset your password:
The all-American EV startup is cutting costs to survive.

America’s most interesting electric-vehicle company is about to have the defining year of its life.
On Wednesday, the company reported that it lost $1.58 billion in the fourth quarter of last year, bringing its net annual losses to $5.4 billion. It announced that it is laying off about 10% of its salaried employees, but — at the same time — promised that it has a plan to achieve a small profit by the end of this year.
Rivian does not seem to be in trouble — not quite yet, at least. But the earnings made clear what electric-vehicle observers have known for a long time: Either the company will emerge from this year poised to be a winner in the EV transition, or it will find itself up against the wall.
That’s partially because Rivian has a stomach-turning number of corporate milestones coming up. Over the next 11 months, it plans to unveil an entirely new line of vehicles, shut down its factory for several weeks for cost-saving upgrades, break ground on a new $5 billion facility in Georgia, and — most importantly — turn a profit for the first time. It also expects to manufacture and deliver roughly another 60,000 vehicles to customers.
Any one of these goals would be difficult to achieve in any environment. But Rivian is going to have to execute all of them during a time defined by “economic and geopolitical uncertainties” and especially high interest rates, its CEO R.J. Scaringe told investors on Wednesday. Since 2021, Rivian’s once robust stockpile of cash has been cut in half to about $7 billion; at its current burn rate, the company will run out of money in a little more than two years.
Although Rivian’s situation is dire, it’s not experiencing anything out of the ordinary. As I’ve written before, the electric truck maker is crossing what commentators sometimes call “the EV valley of death.” This is the challenging point in a company’s life cycle where it has developed a product and scaled it up to production — thereby raising its operating expenses to eye-watering levels — but where its revenue has not yet increased too.
During this vulnerable period, a company essentially burns through its cash on hand in the hope that more customers and serious revenue will soon show up. If those customers don’t arrive, then it either needs to raise more cash … or it runs out of money and goes bankrupt.
It’s a frightening time, but once a company crosses the valley of death, it can reach an idyll. Not so long ago, Tesla found itself in something like Rivian’s position as it prepared to launch the Model 3. Seven years later, it is the most valuable automaker in the world.
Once Rivian’s revenue exceeds its costs, its problems will get easier, or at least more straightforward: Instead of fighting for its survival and watching its cash reserves dwindle, Scaringe will be able to make more strategic trade-offs. Should the company cut costs to expand its profit margin and reward investors, or should it pass the savings along to customers in the form of lower prices, thus growing its market share? Scaringe can’t make these types of decisions until his firm is safely out of the valley.
Claire McDonough, Rivian’s chief financial officer and a former J.P. Morgan director, has a plan for crossing that canyon — an aptly if strangely named “bridge to profitability” that it will attempt to build this year. Rivian’s survival, she said, will depend above all on cutting the unit costs of producing its vehicles, including by using fewer materials to make every car. Other savings will come from making more vehicles faster. That’s what makes the shutdown plan, though it might seem extreme, worth it; McDonough said those improvements alone will get the company about 80% of the way to profitability.
Another 15% will come from marketing more “software-enabled products” to Rivian drivers and by selling air-pollution credits to other carmakers, whose vehicles are not as climate-friendly. This is a tried-and-true technique; Tesla first turned a profit in 2021 by selling regulatory credits needed to comply with federal and California state-level rules to other, dirtier automakers. But that same year, Tesla also debuted an entirely new vehicle: the Model Y crossover, which quickly became its top seller in the United States. Tesla, in other words, finally started to make money by cutting costs, finding new revenue sources, and releasing new products.
New products, however, are becoming a weak point for Rivian. The company says that high interest rates will keep demand for its vehicles flat this year. It expects to make about 60,000 of them, about 20,000 fewer than what it had once anticipated. The Rivian R1S, a three-row S.U.V., has become the company’s flagship; it is selling better and is cheaper to manufacture than Rivian’s pickup, the R1T. It also costs at least $75,000, or nearly $600 a month to lease. The highest-tier models can cost $99,000. Turns out, it’s difficult to sell a lot of $70,000 trucks when even the cheapest new-car loans hover around 6%.
Rivian once had a first-to-market advantage in the electric three-row SUV market, but that may be fizzling out, too. Kia is now selling its own all-electric three-row SUV, the EV9, for $18,000 less than the R1S; in fact, the Kia EV9’s most expensive trim costs $76,000, which is only slightly more than the cheapest R1S. The Kia SUV can also charge faster than the Rivian under ideal conditions. It remains an open question how many rich suburbanites are still interested in buying Rivians, especially now that the Tesla Cybertruck and Ford F-150 Lightning are competing directly with Rivian’s pickup truck.
The company’s hopes, in other words, rest on its next product line: the R2, which it will launch on March 7. We know almost nothing about the R2 line, except that it will probably include an SUV, that it will go on sale in 2026, and that it will fall somewhere in the $45,000 to $55,000 price range. (The median new car transaction in the United States now costs $48,200.) Last year, Scaringe told me that the R2’s timing was perfect because it would fit “beautifully with what we see as this big shift” in the American EV market. In today’s market, he said, “a lot of people ask themselves, Am I gonna get an electric car? Well maybe the next one.” He better hope they’ll start buying that next one in 2026.
Even if they do, Rivian may still have to confront the problem that Tesla has changed the EV market before Rivian could get there. When the first Tesla Model 3s were delivered in 2017, the sedan was instantly one of the best EVs on the market — because it was one of the only EVs on the market. Now every automaker in the world has plans to compete at the Model 3’s price point.
Rivian’s fortunes don’t rest entirely on American consumers; it also sells vans to commercial fleet operators, as well as delivery trucks to Amazon. (Amazon owns about 17% of Rivian.) But that business can be lumpy. Rivian’s vehicle growth slowed down last quarter, for instance, almost entirely because of a near pause in sales to Amazon, which sets up fewer new vehicles in the fourth quarter. If Amazon is willing to bail out Rivian, in other words, it’s not yet clear in the data.
None of this is to say that the company’s outlook is dire. Rivian was always going to find itself at a moment like this, when its expenses exceeded its revenue by such a large amount. The automaker already has devoted fans, and many people — myself included — are interested in the R2 as a potential first EV purchase.
And the company has shown that it can make strides in a single year. Twelve months ago, I had never seen a Rivian on the road before; today, one is regularly parked on my block. The company rocketed from a standing start to become the No. 5 best-selling electric car brand in America last year. What the company has done so far is impressive. But now it must prove that it can be great.
Editor's note: This story has been updated to correctly reflect Rivian's cash burn rate.
Log in
To continue reading, log in to your account.
Create a Free Account
To unlock more free articles, please create a free account.
Plus, the Trump administration appointed a new “beacon of rational thought.”
We got a look at another major tech company’s latest energy and carbon emissions data — and it’s a doozy. On Wednesday, Microsoft released its annual sustainability report, giving us another year’s worth of energy and emissions data for a company that Heatmap’s annual insiders poll once judged to be one of the best hyperscalers for climate change.
The headline: Microsoft’s climate pollution surged last year. Its carbon emissions increased 25% year-over-year, the biggest single-year rise since at least the pandemic. The company emitted the equivalent of 21 million tons of carbon dioxide in 2025, under standard measurement methods. (It emitted slightly less under its own bespoke measurement system, which counts fuel credits and customer energy use differently.)
Electricity, which the company is buying in larger amounts than ever before to power AI data centers, is driving a good share of that increase. In 2024, carbon pollution produced by generating electricity (as well as from making chilled water and steam) was responsible for 2% of Microsoft’s total corporate carbon footprint. In 2025, that same category made up 13% of its overall emissions. The company’s power use rose by more than 24% over the same period.
That means Microsoft’s power use isn’t rising as fast as other companies’. Google’s most recent sustainability report said its own electricity consumption leapt 37% during the same period.
The report suggests, too, that Microsoft is increasingly wary of local fights over data center development — and how water has come to play an outsize role in those battles. The company reports that 2025 was the first year ever that it “replenished” more water on global scales than it withdrew. But “the next phase of our work is increasingly local,” write Brad Smith, the company’s vice chair and president, and Melanie Nakagawa, its chief sustainability officer. That line is clearly in reference to water, specifically — Smith and Nakagawa add that the company hopes to “restore more water to the watersheds where we operate than we withdraw” — but it could also cover the widespread local opposition to data centers that has exploded over the same period.
There’s one more thing to flag about this report: Although it just came out, it covers Microsoft’s 2025 fiscal year, which began in July 2024 and ended more than a year ago. That means it’s inherently an out-of-date view — it shows us what Redmond was doing as the AI and data center boom got underway, but not what it’s doing now. We’ve known for some time that the company is struggling to meet booming AI power demand while maintaining its power commitments; it paused carbon removal buying in April and revised its own clean energy commitments in May.
I should add that Microsoft would prefer that we look at other numbers in the report. First, under its in-house measurement scheme, the company says it released only 20 million tons of carbon pollution over the past year, a figure that appears in its top-line charts. Second, Microsoft estimates that it would have done even more harm to the climate — producing 34 million tons of climate emissions — if not for its corporate policies of buying zero-carbon electricity, using renewable fuels, and improving the energy efficiency and carbon footprint of its XBox game consoles and Surface tablets.
We asked Microsoft for a follow-up interview, but unfortunately they didn’t make anyone available. I’ll be back tomorrow to look at Microsoft’s report in context with other hyperscalers.
Speaking of a sudden rise in gaseous emissions, the Trump administration today named a new leader of the federal government’s marquee in-house climate research office, the U.S. Global Change Research Program. Per Politico, the new top dog is Matthew Wielicki, a UCLA PhD who (1) has a Substack, (2) refers to himself (in the third person) as a “beacon of rational thought” and “professor in exile” on said Substack, and (3) has suggested on X that climate change belongs in the “Department of Imaginary Problems.”
What can I say? Back during President Trump’s first term, his administration tried to bury the publication of the National Climate Assessment by dumping it on a holiday weekend. Now it seems to have taken another strategy. All I can say is, Dr. Wielicki, from one beacon of rational thought to another: I look forward to following your work.
Water pollution in Wyoming has big implications for the future of data center development.
Did a Meta data center introduce a rare, dangerous bacteria into the sewers system of Wyoming’s capitol city? It’s an environmental pollution mystery with an answer that could decide the future of American AI infrastructure development.
Our drama begins in Cheyenne, Wyoming, where the city’s board of public utilities just wrapped up a lengthy investigation into the presence of Cupriavidus gilardii, a potentially lethal bacteria resistant to heavy metals, in the city’s wastewater treatment systems. Apparently, in February, board staff detected the contamination and shut off public access to the city’s water reuse system, a supply of treated non-potable water fed with treated wastewater and used for lawns, athletic fields, and other green spaces. Officials were worried that spraying this water could release into the environment a bacteria found to cause fatal health outcomes in immunocompromised or elderly people who are infected by it.
The board then identified a culprit – Goat Systems LLC, a Delaware-registered firm without a website Meta tasked with overseeing its large $800 million hyperscale project in Cheyenne dubbed Project Cosmo. Goat Systems lost its wastewater disposal permit. The board plans to also fine Goat Systems for violating city code “along with additional fees for our remediation efforts,” board public affairs coordinator Erin Lamb told me in an email. (The only person publicly affiliated with Goat Systems is Pamela Gregorski, an employee for a company that specializes in creating LLCs. Gregorski, who is linked to other LLCs handling Meta projects across the country, did not reply to requests for comment.)
In public comments and statements to me, the board linked the bacteria to water used to flush the Meta data center’s closed-loop cooling system so debris could be removed before the facility was operational. “We were able to connect the Meta data center campus to this through sampling their site,” Lamb said.
This finding led Cheyenne to also indefinitely ban data center projects in the city from ever disposing of “fill-and-flush water” in the sewer system again.
Meta has not denied contamination was found by the city, but says repeated sampling at its project site failed to come up with any evidence confirming they were the source. One can imagine a scenario where the data center and its design played no role in this bacteria showing up, or that city officials erroneously tagged the tech company with responsibility at a time when they’re dealing with political troubles already.
But what is happening in Cheyenne, first reported last week by Wyoming local press, will have consequences for the future of AI infrastructure whether or not Meta was actually even responsible. Right now, all over the country, tech companies are failing to get permits for their data centers because people are worried about water use. These closed-loop data center designs are supposed to address those concerns, letting large hyperscalers contain, cycle, and reuse the water they use for months or even years. A story like this gaining traction in public discourse around data centers will inevitably damage the sector’s public image unless rectified – and fast.
Cheyenne’s claims about the Meta data center being responsible for the bacteria have already metastasized on social media, disseminated through channels often cited by data center opponents on the ground elsewhere in the country. “REPORT: ‘RARE’ BACTERIA DISCHARGED INTO WYOMING WATERSHED LINKED TO DATA CENTER,” reads one post by a Facebook user Izzy Bella that has been shared more than 2,600 times. “Think of this the next time you hear blatant greenwashed lies like ‘closed loop cooling.” This post has been shared by major anti-data center groups on Facebook, including Pennsylvania Data Center Resistance, a social media page for organizing against projects in the Keystone State.
Going solely off what happened in Wyoming, some in the state are concerned the process of cleaning these loops before opening a data center can produce some nasty byproducts. Dr. Jonathan Brand, a civil engineering professor at University of Wyoming, has been studying the data center buildout in Wyoming for years, watching what’s happened in Cheyenne closely, and like me has way more questions than answers.
Usually, Brand said, a company using water in metal-intensive industrial applications – think a metal plating facility – has to test that fluid before it’s dumped into a municipal sewer system. The chain of events spelled out by the board left him “guessing that didn’t happen here,” and he’s worried the bacteria formed within whatever petri dish-like environment was created inside the network of looping pipes before it was flushed.
“The bacterium was the canary they saw, but you could have a lot of residual metals, which is not something we normally test for at a wastewater plant,” he said. “What else was in that discharge? Nobody else has let us know that and they’re probably not going to.”
City officials claim the water was tested before it entered the sewer and was missed, but there’s a trust deficit between locals and the government on what happened. Little of this information was public until a few weeks ago. Cheyenne residents first learned trouble was afoot on June 26, when the board posted a press release “reminding all residential, commercial, and industrial customers that the discharge of hazardous substances into the sanitary sewer system is strictly prohibited.” Nothing was included about data centers at all; all the board said was that the bacteria was dumped by “an industrial user within the system.”
Then Exie Brown, a Cheyenne resident and GOP candidate for state house, blasted a press release out on social media declaring “a credible source with knowledge of the [board] investigation and sampling” told him the “industrial user” was a data center.
I reached out to Brown asking how he learned about this. His answers were cryptic. “I was given a piece of paper with that name of a bacteria on it,” he told me over the phone, declining to name the “very credible source” who told him about the contamination. “That it was released into our waste water system, that it came from a data center, that it was Meta, that they found out in February, and I needed to check into this.” When I asked why the piece of paper, he replied: “Because they [the source] wanted to keep this quiet. Off the phones and stuff.”
City officials deny any malintentions behind the delay and claim they’re learning about all of this at the same pace as the average resident. “We learned here a week or so ago,” Cheyenne mayor Patrick Collins told me in an interview. He added this wouldn’t have stirred as much interest “had it been something else,” referencing the fact it was from a data center.
“As I understand it, the contractor that was building the site was flushing out a closed-loop cooling system, and when they tested the water everything seemed to be fine, but when it was released into our system, bacteria had grown and was released into our wastewater treatment,” Collins said. “It just happened to be a data center. It’s an unfortunate and highly regrettable situation.”
The mayor acknowledged this contamination will make it “a little tougher” to argue for more data centers in the city. There are currently 10 operational data centers in Cheyenne and surrounding Laramie County, according to estimates from pro-business group Cheyenne LEADS, which has said five projects are under construction – including the Meta facility – and at least nine others are “in various stages of planning or due diligence.”
On Monday, the Cheyenne city council will vote on whether to annex land owned by various nearby property owners for more data center deals, including parcels owned by the family of U.S. Senator Cynthia Lummis. Before this event, Cheyenne was incredibly resistant to the anti-data center backlash, handily rejecting proposals to pause development.
Collins thinks Cheyenne will still be open to the tech sector. But the bacteria changed things. “I recognize there’s going to be challenges as we move forward. It’s something we’re going to have to look into. This was a regrettable situation that happened.”
We will see more transparency soon from the Cheyenne city government about the contamination. The board tells me it’s planning a press conference next week where Lamb told me “more information will be made available.”
Francis Brennan, a public affairs manager in the company’s strategic response division, provided me with a statement from an unnamed “Meta spokesperson” claiming that Fortis – the construction company hired by Meta and Goat Systems LLC – was directly handling water disposal on site. After the board “shared that it found a substance in the city’s wastewater” the construction company “began hauling it offsite.” Meta claimed Fortis has not been able to corroborate the presence of this bacteria in comparable water samples.
“Meta is committed to being a good neighbor in Cheyenne, including through the protection of local water resources, and will continue encouraging collaboration between Fortis and the board until this situation is revoked,” the statement read. Meta declined to answer follow-up questions..
Fortis confirmed they were responsible for dumping water on site when the contamination was discovered. They stated they’ve been unable to confirm the presence of the bacteria. In a statement provided to me, the company said: “Immediately upon learning of the issue, we stopped discharging water into the city’s wastewater system. We have since engaged in a thorough investigation that has included ongoing repeat testing by independent environmental specialists and have found no trace of the substance.”
A conversation with Ross Marchard of the Taxpayers Protection Alliance
This week’s conversation is with Ross Marchard, executive director for the Taxpayers Protection Alliance, a center-right advocacy group that focuses on what it sees are onerous policies potentially hindering responsible collection and use of tax dollars. TPA’s position on AI clearly skews pro-free market, as they’ve recently defended Anthropic from Trump administration attacks. TPA also recently took on the mantle of defending data centers from noise complaints, publishing a paper on Tuesday “debunking myths about data centers being excessively noisy.” The paper references various analyses of data centers by state legislators and local regulators to argue that claims the sector is generally noisy are false.
I asked TPA’s executive director to chat with me about why and how the organization will try to quell these fears. The conversation was really interesting so I decided to share it with you in full, sans light editing for clarity and consistency.
What prompted you to write this report?
Obviously, data center projects have been getting so much media attention. With that attention there’s an outsized share of misinformation in coverage of these data center projects, and politicians have irresponsibly spread this misinformation to try and enact moratoria and heavy-handed restrictions on these projects
TPA wanted to get the truth out. Make sure local residents living alongside these data centers have access to all the information they need. Make sure this misinformation is countered.
Before we get into the noise aspect, how is this focusing on “taxpayer protection”?
Sure, well, great case in point is Loudon County. They’ve embraced data centers and look what’s happened, they take in a billion dollars a year in revenue from these data centers and it’s allowed them to lower property taxes. You see a wider pattern across communities. They rake in a tremendous amount of tax revenue and increasingly common well-paying jobs, six-figure blue collar jobs that are a direct result of allowing data centers into communities.
I know you’re based in D.C., near Loudon County. I went to a data center in Sterling, Virginia, in that county, and it was especially noisy. Sort of a worst case scenario on that. Your report talks about misinformation around noise and data center – where is the misinformation happening on this issue?
We saw a recent court case out of New Jersey that alleges data centers generally are as loud as helicopters. Look, anything is possible for a particular project. But what we can say based on our analysis of the data, studies and sound impact assessments, and analyses by state and local governments is that this isn’t the case for the vast majority of data centers.
No use of land is going to be sound-free. I live right on Georgia Avenue in Washington, D.C., so I know noise. But everything we analyzed showed data centers and energy generation on site are going to make some noise but not enough to be harmful to human health. Often it’s no louder than the typical conversation between two people.
Speaking of Loudon County, though, I can point to an example of a project I myself visited that was I’m sure welcomed at first on tax revenue grounds. Now people seem to regret that decision.
As someone trying to address those who are concerned, is it helpful for you to really just call this concern rooted in misinformation? Is this really going to be potent when projects like the one in Sterling exist?
First and foremost, it’s very important to listen to people and their concerns. If folks are living alongside a data center and say they’re hearing loud noises, that warrants investigation. But it’s also important to look at the full array of evidence and we’ve done that. So far, it does not appear to be the case based on the overwhelming amount of evidence that is publicly available that data centers use a lot of water, use inordinate amounts of electricity, or are loud in a way that disrupts human health.
What do you think the policy solutions are to address these noise concerns? How do you listen to people, without going into overgeneralization, as you put it?
People tend to point out the loudest data centers are the ones with on site energy generation. If you ask the operators of data centers and the companies building data centers, they’ll tell you more often than not the reason they’re putting generation on site because the utility permitting process takes far too long. That’s the result not necessarily of utility regulations but state regulations foisted upon utilities. So you have to look at everything from state regulation to grid operation regulation. If you make the process easier for data centers to get hooked up to the grid, you’ll see less on site energy generation, and a lot of the noise complaints will go away.
So from your standpoint, a solution to the noise complaint is that it should be easier to hook up to the grid?
Yes. If you allow data centers to get hooked up to the grid, you’ll see fewer diesel generators and that’ll mean fewer noise complaints.
Now, I want to be clear, the vast majority of data centers with noise complaints – those are usually because of on-site energy generation – are not unduly noisy. If you want to cut down on those complaints, what makes the most sense is to make it easier for data centers to hook up.
Fun question to close: what was the last song you listened to?
“Yellow” by Coldplay.
Are you listening to “Yellow” while you’re writing about data centers?
I listen to the song sometimes when I’m writing about data centers. It’s also a very good somber reflection song, which is a pretty common sentiment amongst millennials.