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:
Sweltering heat and earlier sunsets are a big problem for solar power.

The biggest problem in renewable energy goes by a few names.
Classically, it’s called the “duck curve,” which shows the relationship between solar generation and how much power the rest of the grid uses during the day. At the bottom of the curve, typically around midday, solar can sometimes generate 100% of the power demanded from the grid. But as the sun traces its arc towards the horizon, solar power generation falls and then quickly goes to zero as the sun sets. Often, temperatures and electricity use remains high, especially as people come home from work and start using home appliances, meaning non-solar sources of power must quickly come on line to fill in the gap.
It’s not a coincidence that utilities and grid operators tend to ask consumers to conserve in the later afternoon or evening. The phenomenon is classically associated with solar-heavy California, but it has come to Texas as well, where it goes by the name of the “Armadillo Curve” or the “Dead Armadillo Curve.”
But the relation between the sun and the Earth doesn’t just create darkness and light on daily scales but on annual ones as well. Yes, I know this isn’t breaking news, but it’s important, especially as the power system and climate are changing.
Right now we might be in the neck of the annual duck curve.
In case you haven’t noticed, the sun is setting earlier and it’s still really hot out. Kids are going back to school while much of the country is still facing summer temperatures.
In New York City this week, high temperatures are forecast to get into the 90s, while the sun will set before 7:30; in Washington, D.C., forecast highs are over 100 later this week with a sunset just past 7:30; in Houston, daytime highs to get over 100 later this week, with the sun setting before 7:40; and in Los Angeles , sunsets are at around 7:15 with expected daytime highs in the 90s this weekend.
And Septembers are only getting hotter. Septembers 2021 and 2022 were tied for the fifth hottest on record for the last 143 years, according to the National Oceanic and Atmospheric Administration; 10 of the hottest Septembers have occurred since 2012. The warmest was in 2020.
These higher temperatures mean prolonged periods of high electricity usage, even as one resource — solar — becomes less potent. This matters because, at least in the United States, we tend to organize our lives — and our electricity usage — around the clock, not the sun.
As the sun is setting earlier, our high electricity usage stretches longer compared to the length of the solar day, exacerbating the duck curve dynamics inherent to solar power. A dishwasher that runs when the sun’s still up in July is pulling the same power from the grid as one that runs during fall’s early twilight. The saving grace of shorter days in a grid that uses solar power is supposed to be that air conditioning usage goes down, but that doesn’t happen when summer temperatures persist past Labor Day.
If hot Septembers and even Octobers become the norm, grid conditions could tighten up both during and across the days, with higher cost, less reliable power or increased usage of fossil fuels to fill in the gap.
These longer, hotter summers can make operating electric grids more difficult. ERCOT, the electricity market that covers the vast majority of Texas, restricts power plants from having planned outages between May 15 and September 15 for maintenance. While still in the summer restriction window, ERCOT on Tuesday issued an alert for later this week, warning of “forecasted higher temperatures, higher electrical demand, and the potential for lower reserves.” If ERCOT extends its restrictions on outages for maintenance, there should be more unplanned outages, making power scarcer, meaning higher prices and a greater possibility of blackouts.
Not every country sees peak electricity usage in late summer. In New England, peak electricity demand tends to hit in July. In the sprawling PJM Interconnection last year, the electricity market that spans from the Chicago area to Virginia, demand peaks tended to be in June or August. In New York, peak demand is often in July.
But summer peaks are later in the year in two the country's largest electricity markets: California and Texas.
The Texas energy market had hit its peak day in July in 2022, but it moved out to August this year. And Texas is already bursting through its September demand records. It reached over 78,000 megawatts in just the first week of this month, well over its previous record of 72,370 megawatts, which it set in 2021.
And California hit its power demand record last September amidst a heat wave that covered much of the western United States.
It’s not just there being literally fewer hours of sunlight that drags down solar production later in the year, but also the lower angle of the sun. “As the sun gets lower in the sky we see solar production numbers will drop,” Joshua Rhodes, a senior research scientist at the University of Texas, told me.
“As the sun is lower in the sky it’s up fewer hours ... the photons are coming in at a steeper angle, the panels are not going to get as much light. Even when the sun is at the highest point of the day, the panels are not getting the same level of irradiance as when the sun is at the highest point of the day [at other times of year].”
The best angle for solar panels can change around 15 degrees a year, depending on the year and solar panels are more efficient when they can track the sun during the day. Most homeowners who install solar panels won’t have tracking technology, while utility-scale solar developers are more likely to. This means that a state like Texas, whose renewable mix is more focused on large solar arrays, could see less dramatic drop-offs in solar power throughout the day or throughout the year than a state like California, which has more residential solar.
A roof-mounted four kilowatt-hour solar PV system with standard specs where I grew up in Northern California would get 7.45 kilowatts-hours per meter squared per day in July, generating 689 kilowatt-hours of power, according to National Renewable Energy Laboratory PVWatts tool; in September, solar radiation would drop down to 6.6 kilowatt-hours per meter squared per day and 587 kilowatt-hours per month.
This admittedly basic math suggests it's possible California could struggle this month — and in future Septembers — with meeting electricity demand.
In the past 10 years, California’s annual load peak has occurred in September five times, with the peak loads in 2022 and 2021 occurring on September 6 and 8 respectively.
This year has been, so far, not particularly stressful for the Golden State’s grid thanks to some good luck — no region-wide, prolonged heat waves that max out California’s grid and make imports scarce, mild temperatures on the coasts where the state’s population is concentrated, no major wildfires, and plentiful hydro power thanks to massive snowfall this past winter — as well as massive deployment of batteries across the grid. The batteries especially can help alleviate these duck curve dynamics, as they essentially redistribute power from the sunniest part of the day to the evenings.
While Texas set several new records this year in electricity usage, California has stayed well short of its 52,000 megawatt record last September. California set records for solar power in June and July, with almost 16,000 megawatts, while total demand over 40,000 megawatts.
“While we haven’t seen substantial stress on the grid this summer, we haven’t been fully tested. If we got the kind of west-wide heat we experienced in September 2022, we could need to tap into the state’s emergency or strategic reserves again,” Anne Gonzales, a spokesperson for the California Independent System Operator, told me in an email.
Log in
To continue reading, log in to your account.
Create a Free Account
To unlock more free articles, please create a free account.
Though the tech giant did not say its purchasing pause is permanent, the change will have lasting ripple effects.
What does an industry do when it’s lost 80% of its annual demand?
The carbon removal business is trying to figure that out.
For the past few years, Microsoft has been the buyer of first and last resort for any company that sought to pull carbon dioxide from the atmosphere. In order to achieve an aggressive internal climate goal, the software company purchased more than 70 million metric tons of carbon removal credits, 40 times more than anyone else.
Now, it’s pulling back. Microsoft has informed suppliers and partners that it is pausing carbon removal buying, Heatmap reported last week. Bloomberg and Carbon Herald soon followed. The news has rippled through the nascent industry, convincing executives and investors that lean years may be on the way after a period of rapid growth.
“For a lot of these companies, their business model was, ‘And then Microsoft buys,’” said Julio Friedmann, the chief scientist at Carbon Direct, a company that advises and consults with companies — including, yes, Microsoft — on their carbon management projects, in an interview. “It changes their business model significantly if Microsoft does not buy.”
Microsoft told me this week that it has not ended the purchasing program. It still aims to become carbon negative by 2030, meaning that it must remove more climate pollution from the atmosphere than it produces in that year, according to its website. Its ultimate goal is to eliminate all 45 years of its historic carbon emissions from electricity use by 2050.
“At times, we may adjust the pace or volume of our carbon removal procurement as we continue to refine our approach toward sustainability goals,” Melanie Nakagawa, Microsoft’s chief sustainability officer, said in a statement. “Any adjustments we make are part of our disciplined approach — not a change in ambition.”
Yet even a partial pullback will alter the industry. Over the past five years, carbon removal companies have raised more than $3.6 billion, according to the independent data tracker CDR.fyi. Startups have invested that money into research and equipment, expecting that voluntary corporate buyers — and, eventually, governments — will pay to clean up carbon dioxide in the air.
Although many companies have implicitly promised to buy carbon removal credits — they’re all but implied in any commitment to “net zero” — nobody bought more than Microsoft. The software company purchased 45 million tons of carbon removal last year alone, according to its own data.
The next biggest buyer of carbon removal credits — Frontier, a coalition of large companies led by the payments processing firm Stripe — has bought 1.8 million tons total since launching in 2022.
With such an outsize footprint, Microsoft’s carbon removal team became the de facto regulator for the early industry — setting prices, analyzing projects, and publishing in-house standards for public consumption.
It bought from virtually every kind of carbon removal company, purchasing from large-scale, factory-style facilities that use industrial equipment to suck carbon from the air, as well as smaller and more natural solutions that rely on photosynthesis. One of its largest deals was with the city-owned utility for Stockholm, Sweden, which is building a facility to capture the carbon released when plant matter is burned for energy.
That it would some day stop buying shouldn’t be seen as a surprise, Hannah Bebbington, the head of deployment at the carbon-removal purchasing coalition Frontier, told me. “It will be inevitable for any corporate buyer in the space,” she said. “Corporate budgets are finite.”
Frontier’s members include Google, McKinsey, and Shopify. The coalition remains “open for business,” she said. “We are always open to new buyers joining Frontier.”
But Frontier — and, certainly, Microsoft — understands that the real point of voluntary purchasing programs is to prime the pump for government policy. That’s both because governments play a central role in spurring along new technologies — and because, when you get down to it, governments already handle disposal for a number of different kinds of waste, and carbon dioxide in the air is just another kind of waste. (On a per ton basis, carbon removal may already be price-competitive with municipal trash pickup.)
“The end game here is government support in the long-term period,” Bebbington said. “We will need a robust set of policies around the world that provide permanent demand for high-quality, durable CDR funds.”
“The voluntary market plays a critical role right now, but it won’t scale, and we don’t expect it will scale to the size of the problem,” she added.
Only a handful of companies had the size and scale to sell carbon credits to Microsoft, which tended to place orders in the millions of tons, Jack Andreasen Cavanaugh, a researcher at the Center on Global Energy Policy at Columbia University, told me on a recent episode of Heatmap’s podcast, Shift Key. Those companies will now be competing with fledgling firms for a market that’s 80% smaller than it used to be.
“Fundamentally, what it will mean is just an acceleration of something that was going to happen anyway, which is consolidation and bankruptcies or dissolutions,” Cavanaugh told me. “This was always going to happen at this moment because we don’t have supportive policy.”
Friedmann agreed with the dour outlook. “We will see the best companies and the best projects make it. But a lot of companies will fail, and a lot of projects will fail,” he told me.
To some degree, Microsoft planned for that eventuality in its purchase scheme. The company signed long-term offtake contracts with companies to “pay on delivery,” meaning that it will only pay once tons are actually shown to be durably dealt with. That arrangement will protect Microsoft’s shareholders if companies or technologies fail, but means that it could conceivably keep paying out carbon removal firms for the next 10 years, Noah Deich, a former Biden administration energy official, told me.
The pause, in other words, spells an end to new dealmaking, but it does not stop the flow of revenue to carbon removal companies that have already signed contracts with Microsoft. “The big question now is not who will the next buyer be in 2026,”’ Deich said. “It is who is actually going to deliver credits and do so at scale, at cost, and on time.”
Deich, who ran the Energy Department’s carbon management programs, added that Microsoft has been as important to building the carbon removal industry as Germany was to creating the modern solar industry. That country’s feed-in tariff, which started in 2000, is credited with driving so much demand for solar panels that it spurred a worldwide wave of factory construction and manufacturing innovation.
“The idea that a software company could single-handedly make the market for a climate technology makes about as much sense as the country of Germany — with the same annual solar insolation as Alaska — making the market for solar photovoltaic panels,” Deich said, referencing the comparatively low amount of sunlight that it receives. “But they did it. Climate policy seems to defy Occam’s razor a lot, and this is a great example of that.”
History also shows what could happen if the government fails to step up. In the 1980s, the U.S. government — which had up to that point been the world’s No. 1 developer of solar panel technology — ended its advance purchase program. Many American solar firms sold their patents and intellectual property to Japanese companies.
Those sales led to something of a lost decade for solar research worldwide and ultimately paved the way for East Asian manufacturing companies — first in Japan, and then in China — to dominate the solar trade, Deich said. If the U.S. government doesn’t step up soon, then the same thing could happen to carbon removal.
The climate math still relied upon by global governments to guide their national emissions targets assumes that carbon removal technology will exist and be able to scale rapidly in the future. The Intergovernmental Panel on Climate Change says that many outcomes where the world holds global temperatures to 1.5 or 2 degrees Celsius by the end of the century will involve some degree of “overshoot,” where carbon removal is used to remove excess carbon from the atmosphere.
By one estimate, the world will need to remove 7 billion to 9 billion tons of carbon from the atmosphere by the middle of the century in order to hold to Paris Agreement goals. You could argue that any scenario where the world meets “net zero” will require some amount of carbon removal because the word “net” implies humanity will be cleaning up residual emissions with technology. (Climate analysts sometimes distinguish “net zero” pathways from the even-more-difficult “real zero” pathway for this reason.)
Whether humanity has the technologies that it needs to eliminate emissions then will depend on what governments do now, Deich said. After all, the 2050s are closer to today than the 1980s are.
“It’s up to policymakers whether they want to make the relatively tiny investments in technology that make sure we can have net-zero 2050 and not net-zero 2080,” Deich said.
Congress has historically supported carbon removal more than other climate-critical technologies. The bipartisan infrastructure law of 2022 funded a new network of industrial hubs specializing in direct air capture technology, and previous budget bills created new first-of-a-kind purchasing programs for carbon removal credits. Even the Republican-authored One Big Beautiful Bill Act preserved tax incentives for some carbon removal technologies.
But the Trump administration has been far more equivocal about those programs. The Department of Energy initially declined to spend some funds authorized for carbon removal schemes, and in some cases redirected the funds — potentially illegally — to other purposes. (Carbon removal advocates got good news on Wednesday when the Energy Department reinstated $1.2 billion in grants to the direct air capture hubs.)
Those freezes and reallocations fit into the Trump administration’s broader war on federal climate policy. In part, Trump officials have seemed reluctant to signal that carbon might be a public problem — or something that needs to be “removed” or “managed” — in the first place.
Other countries have started preliminary carbon management programs — Norway, the United Kingdom, and Canada — have launched pilots in recent years. The European carbon market will also soon publish rules guiding how carbon removal credits can be used to offset pollution.
But in the absence of a large-scale federal program in the U.S., lean years are likely coming, observers said.
“I am optimistic that [carbon removal] will continue to scale, but not like it was,” Friedmann said. “Microsoft is a symptom of something that was coming.”
“The need for carbon removal has not changed,” he added.
What happens when one of energy’s oldest bottlenecks meets its newest demand driver?
Often the biggest impediment to building renewable energy projects or data center infrastructure isn’t getting government approvals, it’s overcoming local opposition. When it comes to the transmission that connects energy to the grid, however, companies and politicians of all stripes are used to being most concerned about those at the top – the politicians and regulators at every level who can’t seem to get their acts together.
What will happen when the fiery fights on each end of the wire meet the broken, unplanned spaghetti monster of grid development our country struggles with today? Nothing great.
The transmission fights of the data center boom have only just begun. Utilities will have to spend lots of money on getting energy from Point A to Point B – at least $500 billion over the next five years, to be precise. That’s according to a survey of earnings information published by think tank Power Lines on Tuesday, which found roughly half of all utility infrastructure spending will go toward the grid.
But big wires aren’t very popular. When Heatmap polled various types of energy projects last September, we found that self-identified Democrats and Republicans were mostly neutral on large-scale power lines. Independent voters, though? Transmission was their second least preferred technology, ranking below only coal power.
Making matters far more complex, grid planning is spread out across decision-makers. At the regional level, governance is split into 10 areas overseen by regional transmission organizations, known as RTOs, or independent system operators, known as ISOs. RTOs and ISOs plan transmission projects, often proposing infrastructure to keep the grid resilient and functional. These bodies are also tasked with planning the future of their own grids, or at least they are supposed to – many observers have decried RTOs and ISOs as outmoded and slow to respond. Utilities and electricity co-ops also do this planning at various scales. And each of these bodies must navigate federal regulators and permitting processes, utility commissions for each state they touch, on top of the usual raft of local authorities.
The mid-Atlantic region is overseen by PJM Interconnection, a body now under pressure from state governors in the territory to ensure the data center boom doesn’t unnecessarily drive up costs for consumers. The irony, though, is that these governors are going to be under incredible pressure to have their states act against individual transmission projects in ways that will eventually undercut affordability.
Virginia, for instance – known now as Data Center Alley – is flanked by states that are politically diverse. West Virginia is now a Republican stronghold, but was long a Democratic bastion. Maryland had a Republican governor only a few years ago. Virginia and Pennsylvania regularly change party control. These dynamics are among the many drivers behind the opposition against the Piedmont Reliability Project, which would run from a nuclear plant in Pennsylvania to northern Virginia, cutting across spans of Maryland farmland ripe for land use conflict. The timeline for this project is currently unclear due to administrative delays.
Another major fight is brewing with NextEra’s Mid-Atlantic Resiliency Link, or MARL project. Spanning four states – and therefore four utility commissions – the MARL was approved by PJM Interconnection to meet rising electricity demand across West Virginia, Virginia, Maryland and Pennsylvania. It still requires approval from each state utility commission, however. Potentially affected residents in West Virginia are hopping mad about the project, and state Democratic lawmakers are urging the utility commission to reject it.
In West Virginia, as well as Virginia and Maryland, NextEra has applied for a certificate of public convenience and necessity to build the MARL project, a permit that opponents have claimed would grant it the authority to exercise eminent domain. (NextEra has said it will do what it can to work well with landowners. The company did not respond to a request for comment.)
“The biggest problem facing transmission is that there’s so many problems facing transmission,” said Liza Reed, director of climate and energy at the Niskanen Center, a policy think tank. “You have multiple layers of approval you have to go through for a line that is going to provide broader benefits in reliability and resilience across the system.”
Hyperlocal fracases certainly do matter. Reed explained to me that “often folks who are approving the line at the state or local level are looking at the benefits they’re receiving – and that’s one of the barriers transmission can have.” That is, when one state utility commission looks at a power line project, they’re essentially forced to evaluate the costs and benefits from just a portion of it.
She pointed to the example of a Transource line proposed by PJM almost 10 years ago to send excess capacity from Pennsylvania to Maryland. It wasn’t delayed by protests over the line itself – the Pennsylvania Public Utilities Commission opposed the project because it thought the result would be net higher electricity bills for folks in the Keystone State. That’s despite whatever benefits would come from selling the electricity to Maryland and consumer benefits for their southern neighbors. The lesson: Whoever feels they’re getting the raw end of the line will likely try to stop it, and there’s little to nothing anyone else can do to stop them.
These hyperlocal fears about projects with broader regional benefits can be easy targets for conservation-focused environmental advocates. Not only could they take your land, the argument goes, they’re also branching out to states with dirtier forms of energy that could pollute your air.
“We do need more energy infrastructure to move renewable energy,” said Julie Bolthouse, director of land use for the Virginia conservation group Piedmont Environmental Council, after I asked her why she’s opposing lots of the transmission in Virginia. “This is pulling away from that investment. This is eating up all of our utility funding. All of our money is going to these massive transmission lines to give this incredible amount of power to data centers in Virginia when it could be used to invest in solar, to invest in transmission for renewables we can use. Instead it’s delivering gas and coal from West Virginia and the Ohio River Valley.”
Daniel Palken of Arnold Ventures, who previously worked on major pieces of transmission reform legislation in the U.S. Senate, said when asked if local opposition was a bigger problem than macro permitting issues: “I do not think local opposition is the main thing holding up transmission.”
But then he texted me to clarify. “What’s unique about transmission is that in order for local opposition to even matter, there has to be a functional planning process that gets transmission lines to the starting line. And right now, only about half the country has functional regional planning, and none of the country has functional interregional planning.”
It’s challenging to fathom a solution to such a fragmented, nauseating puzzle. One solution could be in Congress, where climate hawks and transmission reform champions want to empower the Federal Energy Regulatory Commission to have primacy over transmission line approvals, as it has over gas pipelines. This would at the very least contain any conflicts over transmission lines to one deciding body.
“It’s an old saw: Depending on the issue, I’ll tell you that I’m supportive of states’ rights,” Representative Sean Casten told me last December. “[I]t makes no sense that if you want to build a gas pipeline across multiple states in the U.S., you go to FERC and they are the sole permitting authority and they decide whether or not you get a permit. If you go to the same corridor and build an electric transmission that has less to worry about because there’s no chance of leaks, you have a different permitting body every time you cross a state line.”
Another solution could come from the tech sector thinking fast on its feet. Google for example is investing in “advanced” transmission projects like reconductoring, which the company says will allow it to increase the capacity of existing power lines. Microsoft is also experimenting with smaller superconductor lines they claim deliver the same amount of power than traditional wires.
But this space is evolving and in its infancy. “Getting into the business of transmission development is very complicated and takes a lot of time. That’s why we’ve seen data centers trying a lot of different tactics,” Reed said. “I think there’s a lot of interest, but turning that into specific projects and solutions is still to come. I think it’s also made harder by how highly local these decisions are.”
Plus more of the week’s biggest development fights.
1. Franklin County, Maine – The fate of the first statewide data center ban hinges on whether a governor running for a Democratic Senate nomination is willing to veto over a single town’s project.
2. Jerome County, Idaho – The county home to the now-defunct Lava Ridge wind farm just restricted solar energy, too.
3. Shelby County, Tennessee - The NAACP has joined with environmentalists to sue one of Elon Musk’s data centers in Memphis, claiming it is illegally operating more than two dozen gas turbines.
4. Richland County, Ohio - This Ohio county is going to vote in a few weeks on a ballot initiative that would overturn its solar and wind ban. I am less optimistic about it than many other energy nerds I’ve seen chattering the past week.
5. Racine County, Wisconsin – I close this week’s Hotspots with a bonus request: Please listen to this data center noise.