The Two Horsemen, Riding in Formation

Is There Suddenly an Opening for AI and Climate Sense?

Obviously the story of the moment is the sudden, coordinated bout of cold feet in the AI industry, whose leaders expressed support over the weekend for some kind of effort to slow down the pace of their R&D. This is something that’s never happened in the forty years of the climate fight, and so it shouldn’t be dismissed; in fact, it could conceivably provide an opening for common sense across the entire apocalypse front. So let’s try to game out how responsible political leaders—which means, I tremble to say, Democrats—might try to talk about it.

First, credit where due. The fact there’s an opening at all is owed to the remarkable citizen resistance to data centers that has sprung up in the last year, and also to the courage of a few kids in the AI industry. At Third Act/GrayPAC we’ve been a part of that resistance in place after place—if you want some insight on its roots and future, join us tomorrow night for a conversation with old friend Justin Pearson, the Tennessee state rep who helped kick it off with the fight against Elon Musk’s super-polluting Memphis facility.

Water wastage, noise, land use—the data center fight has often been routed in deeply local concerns. But it has two other dimensions. The first is about energy, both its costs and its pollution. As I’ve noted, one way to understand a data center is that it’s a device for burning natural gas, the latest excuse (after Ukraine and Iran) for the fuel that Big Oil is counting on to keep its business model afloat even as the world turns away from internal combustion cars and crude demand begins to drop. The degree to which this is becoming a climate disaster can hardly be overstated. Here’s the latest, from Bloomberg yesterday

Natural gas consumption for electricity generation at data centers is forecast to increase by 15 billion cubic feet per day in the decade ending in 2035, even when accounting for planned projects that may never be completed. This volume exceeds the current gas consumption of all nations except China, Russia, Iran and the United States, based on data from the US Energy Information Administration. The projection is more than double BloombergNEF's December forecast of 6.9 billion cubic feet per day.

In essence, American data centers are about to become a giant new fossil-fuel burning country, just when we need it least. And if you wonder whether the oil boys have figured this out, don’t. When Josh Shapiro—long a big backer of AI—proposed some limits last week, the reaction went like this:

“We were disappointed in the announcement in that it sends the wrong message to these incredible economic opportunities that are presented for Pennsylvania and Pennsylvania’s economy,” said James D. Welty, president of the Marcellus Shale Coalition…

Marcellus producers are almost solely dependent on demand growth within their own producing region. Artificial intelligence data centers powered by gas-fired electricity are seen by Marcellus gas producers as a key source of that so-called in-basin demand.

But that data center resistance was driven by something else too, that’s been largely ignored—it was a chance to stand up to the oligarchs and this technology they’ve staked their future on. A few years ago Musk and Zuckerberg and the rest had a mixed reputation, admired by some and distrusted by others; now the loathing is much more universal, especially since they’ve allied themselves with the most unpopular president in recent history. They marketed their AI tech on the promise that it would destroy jobs; it’s also proven useful for making naked pictures of anyone you could imagine. They stole books to train it; now it produces a commodity universally reviled as slop.

Oh, and it might kill us all, which brings us back to those kids. Jacob Coxon is a 27-year-old former silver medalist at the International Mathematical Olympiad who worked first at Open AI and then Anthropic, and who resigned last week, kicking off this current storm with his tweeted fear that things were getting out of control. A top scientist at the company backed him up, tweeting “We really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade.”

Which seems like kind of a big deal. Perhaps Bernie Sanders was on to something when, many months ago, he called for a moratorium on this kind of work, a call he’s repeated in recent days. But no one takes Bernie seriously (except the American people, for whom he remains the most popular active politician). So it’s a good sign that there seem to be some starting to follow suit. Jon Ossoff has been nimblest, and not for the first time—he got a start on the messaging by saying

"On day one, Donald Trump rescinded the AI safety order meant to guard against catastrophic risks, including the engineering of bioweapons. Meanwhile, Don Jr's 1789 Capital bought big money stakes in data centers and AI labs ... President Trump is compromised"

And J.B. Pritzker, another 2028 possibility,

“It’s becoming more clear the threat AI poses to humanity, so I’m calling for immediate action from the industry and Washington.”

But as Greg Sargent points out in a prescient piece today, the Dems are mostly playing small ball here, nibbling around the edges instead of biting. I think someone has the chance for a kind of career-defining stance here, one that builds on the growing sense of unease with the overheating world (an unease which is already quite broad and is going to deepen as the budding El Niño takes us ever deeper into the climate future). As we start to look forward to 2028, even Barack Obama—a cautious man—has called for Democrats to be more aggressive. Here’s a stab at what that might look like:

“Elections usually turn on immediate, pocketbook issues—that makes sense, because we all need to eat, and right now the price of groceries is too high, the price of gas is too high, the rent is too high. We know why—tariffs, stupid wars, corruption. We’ve got lots to work on.

But there are also moments when need to think about larger questions, and this is clearly one. AI is not something any of us asked for—just about no one wants a data center in their neighborhood, and no one put up their hand to sacrifice their job to some computer bot. It’s been forced on us by some of the richest people on earth, and it was obnoxious even before it started going on the attack—now, as even its creators say, it’s out of control and offers a very real risk to our health and safety, and maybe even our existence as a species.

So we need to rein it in, with serious government regulation that dramatically slows the pace of development until we can can understand and control it. And that includes making the companies that produce this, and the hedge funds who invest in them, liable for the havoc they create.

But we need to go beyond that. We need to step back and say: what are we doing here, people? Why are we running risks like this? Our civilization, especially in rich countries like this one., is capable of producing the things we need: we’ve never lived longer or better. If we didn’t concentrate wealth in a few hands, if we spread it our more widely, we could provide health care and housing for everyone. Natural human intelligence has produced plenty; the natural human sense of justice can help us figure out how to share it in ways that work for all. Things could be…okay.

So why would we run enormous risks—the ones that come with artificial super intelligence, and the ones that come with rapidly warming the planet? These are the dangers that could attack and impoverish us; no rational society would treat them as casually as the Trump administration has done. Instead we should be hard at work insuring against those risks—by, say, building out cheap and clean renewable energy. And by upgrading the education and the skills of our kids, instead of handing them screens and turning them over to AI teachers.

In some cases this would mean “slowing down progress”; in every case it would mean attacking concentrated wealth and power. But is it really progress? Is there any real reason we need to rush pell-mell into this chaotic and dystopian future. We’re told that medical advances await—but we’ve been making steady progress on this front for decades, with lifespans expanding and more and more diseases becoming manageable. (Or at least we were making progress till this administration decided to upend our scientific research institutions). We’re told that if we don’t rush ahead China will—but that seems more like an argument for negotiating with China about cooperative controls, just as we once negotiated with Russia about the control of nuclear weapons. It’s never easy, it’s always messy, and so far it’s been successful.

If this willingness to stick with the world as we’ve known it—a world where human intelligence prevails, and where keep some ice near the poles—sounds “conservative” then so be it. There are moments when humans need to say ‘enough,’ and this is one. We are fools to endanger all that our ancestors have built in search of some new thing that will enrich a few. I know that the Trump administration wants to rush forward, that he has called AI fears a “hoax,” the exact same word he’s used about the climate crisis. But Trump and his colleagues are greed-seeking missiles, always looking for some way to grift, whether its from AI and crypto or from oil companies and Middle East princelings.

We are told that our destiny is constantly to rush forward—to head for Mars. But actually, one unique human gift is the ability to decide not to do things we’re capable of doing—to say Mars isn’t as important as getting things right on Earth. In fact, we’ve never had a better chance of getting things right. All of a sudden solar panels are the cheapest way to make power, for instance. It wasn’t AI that did that—it was smart people, at work in universities and companies around the world. They’ve given us a real gift, even if it’s going to cost Exxon some profits.

Humans don’t get many chances to make momentous decisions like this. Usually they get made for us. But across the country people have stood up to the data centers, and that’s given us the possibility for real change. Let’s take it.”

In other energy and climate news:

+It’s not just the heat, it’s also the staggering humidity. Andrew Dessler has a typically fine account of how increased air temperature is putting more and more water vapor in the air

A few days ago, news came out that August 2026 was the hottest month in the last 170 years. Yup, climate change strikes again (with an assist from El Niño).

Therefore, it might not be too surprising that August 2026 also set a record for Precipitable Water (PWAT), 27.35 mm, beating the previous record (July 2024) by 0.04 mm¹.

The data I analyzed only goes back to 1979, but given the very strong connection between PWAT and temperature, it is unimaginable that any month in the last few thousand years, maybe 100,000 years, has more PWAT in the atmosphere than August 2026. We are, as the proverb says, living in interesting times.

+Jeva Lange covers a new poll showing that Americans are more interested in saving “the environment” than “the climate.” I will point out that this still leaves a clear majority saying climate extremely or very important, which is also good news; politicians routinely underestimate the potency of this whole complex of issues. And considering that Democrats aren’t currently even willing to talk about climate, it’s not too surprising

The truth is, most Americans are environmentalists, at least if you take a wide definition. In new polling and focus group research conducted by Heatmap News and Embold Research, nearly three in four registered voters (72%) told us “the environment” is extremely or very important to them personally. The environment, in fact, ranked a full 17 points higher than “climate change” in a battery of potential voter issues, despite the latter’s much more obvious existential threat. Even among the subset of people we labeled “skeptics,” for whom climate change was of minimal to no importance, a third (32%) still told us the environment matters to them greatly.

+Painful numbers from Zahra Hirji, Eric Roston, and Lili Pike, who have carefully calculated how much the Trump rollbaks of pollution laws and clean energy funding have damaged America’s effort. Bottom line:

The world’s second-largest emitter after China may now see its contribution to climate change rise by roughly 1 gigaton more than it would have otherwise, in 2040, according to one analysis by Energy Innovation, a nonpartisan energy and policy think tank. That’s the emissions equivalent of the energy used by 134 million houses in a year.

Under this analysis, which attempts to take into account the rescinding of climate controls for current and future coal and gas power plants, US emissions will still drop by 2040, but only to the level they would have roughly been in 2030 had previous policies remained in place.

The two take-home messages: Trump is a climate criminal, and don’t forget to vote Nov. 3

+The Trump administration came at just the wrong moment, because around the rest of the world ever-cheaper solar is driving an ever-faster transition. As the Financial Times lays out in a major essay

The solar revolution made possible by cheap Chinese photovoltaic panels — and the rise of small-scale, individual power generation — is transforming energy in the developing and industrialised world alike. Millions of people in poorer countries now have more reliable access to electricity, often through their own endeavours rather than massive installed infrastructure. And across the globe, consumers are using cleaner energy and bringing down their bills at a time when the cost of living is a constant concern, not least because of the Iran and Ukraine wars.

But solar is also a boon for households. “A home goes from burning two or three kerosene lanterns to a light that is 20 to 40 times brighter, [turned] on every day,” says Anish Thakkar, co-founder of Sun King, a Kenyan business which lends families money to pay for solar and battery installations. In the Philippines, Ember estimates rooftop solar capacity may have almost doubled over the 12 months to April and its calculations suggest residential solar panels can now pay for themselves in just over three years. Meralco, the country’s electricity distributor, said rooftop solar generated 372 gigawatt hours in the first six months of the year.

+For German readers (and to spark imaginations elsewhere)—great climate activists are setting up a series of “human chains” around key cities early next month

The main route, which we intend to fill with human chains, runs from the Federal Constitutional Court in Karlsruhe to the Federal Chancellery in Berlin . We will fill sections of the route together.

In addition, human chains are forming all over the country – between houses, villages, and cities. Everywhere we are showing: We are many.

+Man, cars are expensive. On top of car payments, the average price for maintaining one is almost six grand a year, according to a new report from Sarah Agostino:

Drivers in the U.S. spend an average of $5,851 annually on car ownership, according to new research from Insurify, an insurance comparison site. The biggest piece of that amount — $2,237 — is the average annual cost of full-coverage auto insurance. Gasoline accounts for another $2,126, and maintenance or repairs account for the remaining $1,488. The analysis is based on Insurify’s insurance-quote data, along with AAA estimates for gas and maintenance costs.

The yearly outlay is on top of monthly auto loan payments, which averaged $765 in the second quarter for new cars, according to credit reporting company Experian. For used cars, the average payment was $542.

That electric cargo bike looks a little better all the time

+August tied for the hottest month in human history. You want to know a bad profession in this heat? Bicycle racer. At the Vuelta de Espana, riders were heading up steep hills in temperatures around 110 degrees

Matthew Brennan of England, a three-stage winner at this year’s event, made similar comments after the stage: “When I was on the climb at the start, it was 44C (on his bike computer — 111F). At that point, you start thinking, ‘What are we doing?’”

Brennan criticised the UCI, world cycling’s governing body, for not intervening further, saying riders were “slogging away for their sake”.

“That’s the show we put on, but I think it wouldn’t surprise me if there are a few bodies going home in an ambulance tonight with heatstroke,” the 21-year-old added.

+Liam Denning has a typically insightful column on how Trump’s drive for energy dominance is turning into “energy decadence.”

Yet while US energy riches offer clear advantages, the resort to military operations does indicate the limits of dominance, with earlier sanctions clearly not having delivered in a world of shadow fleets. The subsequent failure of bombing to achieve a quick, decisive outcome has been followed by the US negotiating for a reopening of the Strait of Hormuz — the status quo ante — and now another round of supposedly harsher sanctions, all punctuated by Trump’s frequent declarations of elusive victory. Being No. 1 in oil production appears to have given the US enough confidence to flex its muscles but not enough capacity to ignore the consequences

+Perskovite solar—the highly efficient form of panels that has alwasy been hard to commercialize because of durability problems—is much closer to rooftops, Katie Brigham reports

“Over the better part of a decade, there’s been lots of, for want of a better word, device engineering to engineer the material to just be much more stable,” Sam Stranks, an energy materials professor at the University of Cambridge and co-founder of perovskite startup Swift Solar, told me. That includes swapping out an unstable chemical building block in perovskites called methylammonium with a more stable one called formamidinium, improving the connection point that binds the perovskite layer to the electrode layer, and improving the packaging of perovskite cells to seal them off from air and moisture.

“There’s still work to be done to really get it to the point where you could put it up on a rooftop for 25 years and you know it will last. But the trajectory is very promising,” Stranks said.

Scientists have managed to extend perovskite durability from mere hours to the point where cells and modules are now passing industry-standard tests that suggest they could survive outdoors for five to 10 years. That’s still a ways away from the standard 25-year warranties for silicon solar panels, which typically guarantee that a module will retain 85% to 90% of its original output by year 25. And because these latest perovskites simply haven’t been around that long, scientists have yet to test these more ambitious durability claims in the real world.

Industry experts say there’s no reason continued incremental improvements can’t get perovskites to that 25-year standard relatively soon, however. “If you look at silicon, it’s been 70 years of trial and error,” Scott Wharton, CEO of perovskite startup Tandem PV, told me. “Seventy years of constant tweaking and improvement, whereas perovskites have only had about 12.”

+According to Mal James, writing in Space Daily, a material found in the discarded shells of crabs and shrimps can play a vital role in batteries:

The link runs through a material called chitin, the tough stuff that makes up the shells of crabs, shrimp and lobster. Treat chitin with the right chemistry and you get chitosan, a softer, workable version. The seafood industry produces mountains of shell waste every year, which is exactly the appeal. Hu says that “the most abundant source of chitosan is the exoskeletons of crustaceans, including crabs, shrimps and lobsters, which can be easily obtained from seafood waste.”

So the raw ingredient is not mined or drilled. It is a byproduct of something we already throw away in large amounts. The researchers turned that chitosan into a gel that served as the electrolyte, the medium that lets charge move between the two ends of the battery.

The battery here is a zinc-metal battery, and zinc has a well-known problem. As it charges and discharges, the metal tends to grow uneven, spiky deposits called dendrites, along with unwanted side reactions at its surface. Left unchecked, those spikes can short the cell out. It is one of the main reasons this whole family of batteries has struggled to reach the market, a hurdle other researchers describe as random, spiky growth that can short-circuit a cell.

The team’s gel has zinc ions bound into its structure. That guides the zinc to deposit more evenly, in flat, orderly layers rather than jagged spikes. Even plating is what keeps the anode stable over time, and stability is the whole game.

+Some decent news, as Maine scientists report that the state’s fall foliage may be somewhat resilient to climate change in the years ahead. Let us hope so. I greet the arrival of leaf season with glee each year. This season I’m going to miss most of it, out on the campaign hustings across the country, so I climbed up to a high elevation beaver pond near me the other day, and sure enough the swamp maples had already begun to turn. I hope all of you get out in the still-beautiful world today: