Dear Aventine Readers, 

We're checking in on the state of next generation geothermal this week, the rare clean energy source embraced by both our political parties. Projects are going gangbusters, with big investments from big companies. But there's still a lot to figure out, like whether we can make tools that can penetrate miles of rock and withstand extreme heat. A lot is riding on things going right. If the technology works to its fullest capacity, it could unlock more than four times the current power generation capacity of the entire US.

Also in the issue: 

Thanks, as usual, for reading! 

Danielle Mattoon
Executive Director, Aventine

Subscribe

Subscribe to our newsletter and be kept up to date on upcoming Aventine projects

Builders often say you don't know the scale of a job until shovels are in the ground. The geothermal industry, increasingly flush with cash and facing high demand, is now living that reality.

Two years ago, we covered the then nascent next-generation geothermal industry, which aims to drill into hot rock beneath the Earth's surface to generate power. We found an industry poised to enter the mainstream, one emerging from early R&D, with demonstration plants being drilled and funds being raised. Now, geothermal companies are building commercial facilities and fielding a wave of potential partners; one has even gone public. 

Yet as these companies race to keep up with surging interest, particularly from data center providers, they are confronting new challenges: how to push into hotter rock to generate more power, how to scale their projects as efficiently as possible and how to keep the wells they drill pumping out electricity for as long as possible. 

The promise of hot rock

The appeal of next-generation geothermal is its potential to provide power from anywhere in the world. Traditional geothermal taps into naturally occurring wells of hot water beneath the Earth's surface. Unfortunately, such geological conditions are rare. Yet about two miles beneath the earth’s surface, there is rock hot enough to generate power almost anywhere on the planet.

In most places (there are exceptions), rock found at about that depth can reach temperatures between 350 degrees and about 700 degrees Fahrenheit. (We will be using Fahrenheit for all temperatures that follow). Next-gen geothermal projects pump water down to depths where it becomes superheated, bring it back to the surface and then use it to drive turbines to generate electricity. There are competing approaches to doing this. Enhanced geothermal uses hydraulic fracturing, or fracking, to create cracks that give water a route through hot, otherwise impermeable rock between an inlet well and a production well. Closed-loop geothermal uses a similar two-well arrangement, but connects them via a network of narrow channels drilled through the rock that pass water between the hot and cold wells without the rock being fractured. It’s also possible to use a single well, exploiting the natural elasticity of rock to store and release hot water over time.

Whatever the approach, estimates suggest that the technology could unlock more than 5.5 terawatts of power in the US, or more than four times the country's current power generation capacity.

Moving beyond tests

Demand is high. Google, Meta, Amazon and Microsoft have all signed power-purchase agreements with geothermal companies, and financing for the technology nearly doubled between 2024 and 2025.

Gradually, some of the projects generating this excitement are becoming commercial realities. Fervo Energy, which has a deal to provide Google with up to 3 gigawatts of generation capacity, is building what will be the world's largest enhanced geothermal facility in Beaver County, Utah. The plant is designed to deliver 500 megawatts — enough to power roughly half a million homes. The first 100 megawatts of capacity is expected online this year, with maximum capacity expected to be reached by 2028. Fervo, which raised $1.9 billion in an initial public offering in May, has agreed to sell electricity from the plant to Google, Southern California Edison and others.

Sage Geosystems, which hadn't yet completed a commercial demonstration system two years ago, now has a deal to supply Meta with 150 megawatts of power. It is doing the initial groundwork on its first commercial power generation project, which uses a single well, according to Lance Cook, the company’s co-founder and chief scientist. Another company, Eavor, delivered its first electricity to a commercial power grid in December 2025, from its Geretsried facility in Bavaria, Germany — the world's first grid power from a fully closed-loop geothermal system.

Meanwhile, geothermal is the rare clean energy technology with bipartisan support in the US Congress. Most recently, the House passed the Geothermal Energy Advancement Act, which consolidates six separate bills targeting permitting and regulatory bottlenecks on federal land. If the Senate approves the bill, provisions will include a 60-day deadline for approval of drilling permits and environmental exclusions for geothermal, which is on par with those available to oil and gas.

Getting hotter in here

Energy contained in water rises nonlinearly with temperature: Double the water temperature from 390 degrees to 750 degrees and power generated at the surface theoretically increases by up to a factor of 10. The surest way to extract more power from the ground is to drill deeper, where temperatures are hotter. The challenge lies in reaching the heat. At these temperatures, regular drilling equipment borrowed from the oil and gas industry begins to fail, with sensors breaking, drilling fluids becoming unstable and some components melting. So far, the most advanced companies have yet to overcome these issues and hit temperatures anywhere close to the 700 degrees that would increase the power generated by a factor of 10. The highest temperature Fervo reported reaching, for instance, was 555 degrees at the bottom of a 11,200-foot well in Millard County, Utah.

On the flank of Newberry Volcano in Oregon, two startups are experimenting with technologies that could survive the heat. The site, where temperatures exceeding 752 degrees can be found at relatively shallow depths of under 16,400 feet, makes it a useful test bed for pushing the limits of drilling into hotter rock. Mazama Energy, based in Frisco, Texas, is augmenting oil-and-gas drilling tools with cooling systems that inject cooled, pressurized carbon dioxide into the bore hole to counteract the heat. Quaise Energy, an MIT spinout that raised $134 million in new funding just this week, has built a system that uses millimeter waves, a high frequency microwave, generated by a tool adapted from nuclear fusion research, to vaporize rock.

The experiments being conducted at Newberry are in their early stages. Mazama claims to have reached rock that measures 629 degrees. Quaise, which was confined to the laboratory when we last wrote about geothermal, recently began work at the site, according to a company spokesperson. At a separate site in Marble Falls, Texas, Quaise has demonstrated that its drilling technique works, boring a 100-meter hole into granite last summer, with plans to reach one kilometer this year. At Newberry, it will rely on conventional oil and gas drilling for the first stretch, switching to the high-energy microwave system when temperatures climb.

Higher temperatures promise drastically higher power output, but the cost — of drilling deeper and the heat-resistant equipment — will be a major factor in how the technology evolves. Joe Moore, the Principal Investigator, Emeritus at Utah FORGE, a Department of Energy-backed laboratory for developing enhanced geothermal systems, isn't yet sure where the optimal balance between drilling cost and power output will land. "It's a wait-and-see game," he said.

Problems in the ground

Heat and depth aren’t the only challenge.

Another is water availability. In enhanced geothermal, which is by far the most established next-generation technique, water is pumped through fracked cracks that radiate outward, with as much as 20 percent lost in the process. This is a deterrent in regions where water itself is scarce, said Moore, especially as projects scale up. One hope is that rock becomes tighter and less permeable at greater depths, reducing losses. Another is that drilling multiple production wells around a single injection well will create opportunities to recapture some of the water. But ultimately, Moore said, water rights will be a critical factor in how far the technology can scale in certain areas.

A second issue: Wells lose temperature. During 600 days of operation at one of its wells in Nevada, Fervo observed a 2.5 degree decline in the temperature it could generate. That's modest, but Moore said it's "going to be an issue" over the long run, since falling temperatures cap how much power a well produces. The concern, he said, is that operators will just be constantly drilling new wells instead of sticking with wells where the temperature is dropping, which means that costs will remain high — a dynamic that could limit how financially attractive these projects can be.

There are also challenges facing geothermal that have nothing to do with geology. Permitting is still a problem, though the industry hopes it will be eased by anticipated policy changes. More problematic is connecting to the grid. Like every energy project in the US right now, geothermal plants must join grid-connection waitlists stretching up to five years. "The issue isn't how many gigawatts of power you can sign contracts on," said Cook. "It's [the] time to get on the electric grid." Moore suggested there may be ways to work around the problem, including co-locating geothermal plants at the sites of retiring fossil-fuel power plants, where grid connections already exist.

Still, he added, the fundamentals of the technology are proven. Over the past two years, the industry has demonstrated that it can drill to the necessary depths and temperatures to generate power, and that there are customers willing to buy it. The open question now isn't whether next-generation geothermal works, but how far these companies can push it.

Car batteries being used as grid storage. Redwood Materials

Old EV batteries are finally being deployed as grid storage. The idea of giving retired electric vehicle batteries a second life on the grid has been around for more than a decade. But the approach is finally coming to life, reports IEEE Spectrum. June saw a string of announcements: B2U Storage Solutions announced a partnership to repurpose batteries from Waymo's robotaxis; Redwood Materials revealed plans for a 1.5-megawatt storage system built from 100 retired General Motors battery packs; and Moment Energy completed what it says is the world's largest EV battery repurposing facility. The idea is that batteries that are too degraded to use for cars often retain plenty of capacity for stationary storage, for which charging and discharging are slower, more predictable and cause less wear and tear on battery cells. At the same time, it provides an affordable source of storage for grid operators. Until recently, though, there weren't enough retired EV batteries to build a business around. And when they did become available, testing and safely storing them proved technically difficult. Now, better diagnostic tools and battery-management software are making the process practical. The big remaining question is whether the economics make sense. Research from Stanford University suggests that some battery chemistries, particularly those rich in nickel and cobalt, may be worth more as a source of raw materials than as grid storage. Iron-based batteries, on the other hand, may deliver greater value when used as grid storage. The coming wave of commercial projects should begin to reveal where that balance lies. 

Scientists built a cell that grows and divides, from scratch. Researchers from the University of Minnesota claim to have made a leap in synthetic biology, building a cell entirely from non-living components. The result can grow, copy its DNA and divide. The research has been posted as a preprint on bioRxiv and has not yet been peer reviewed, but is the best attempt yet at building a cell from the ground up — potentially ushering in a future in which new synthetic cells could be used for biofuels, drugs and other materials. The main achievement is that every molecular part was specified and made in the lab. That distinguishes it from engineered bacteria, for which edits are made to a cell that has evolutionary ancestry, and from previous top-down attempts to create simpler cells from more complex bacteria. The team combined molecular tools that perform three functions: One replicates DNA, another reads DNA sequences and makes proteins and a third lets the cell's membrane deform, folding in on itself until it divides. But growing and dividing is all these cells do. They are not alive: They have no metabolism and can produce only some of the molecules needed to turn genetic information into proteins. They can grow only because they are fed liposomes that fuse with them to deliver other molecules, like sugars and lipids. The researchers have seen the cells reproduce over five generations, and have identified a form of evolution by seeding a population with subtly different cells and seeing which trait prevails — though, as Quanta points out, the mechanism is a long way from being classed as natural selection. The researchers have founded a nonprofit, Biotic, to develop the technology further. But there is a long, long way to go. "The modern cell is like a [Boeing 787] Dreamliner," lead researcher Kate Adamala told Quanta. "We built a Wright flyer."

Automated laboratories are swinging into action. For decades, many chemists have dreamed of spending less time pipetting and more time thinking. That vision may be inching closer to reality. Chemical and Engineering News reports that a new generation of startups is combining robotics with generative AI to automate significant parts of laboratory research. The concept isn’t new. Previous attempts have struggled to live up to the hype, perhaps most notably Eli Lilly's automated chemistry lab, launched in 2017 with the aim of allowing researchers anywhere in the world to design molecules remotely. The project was ultimately shuttered in 2024 after falling short of expectations. This new wave hopes that the falling cost of robotics combined with increasingly capable large language models can change that. Startups are building automated systems that propose experimental procedures, use robots to perform experiments, analyze the results and then design new experiments in an iterative loop. Several companies are pursuing the vision, including Atinary Technologies in Boston, Radical AI in NYC and Dunia Innovations in Berlin, each targeting different aspects of chemical and materials research. For now, the results are still modest: Today's systems remain highly specialized, performing relatively narrow classes of experiments and struggling with mundane tasks like weighing sticky powders. That means they’re unlikely to replace PhD students any time soon. But they may become valuable research assistants, taking over repetitive experiments for some scientists.

The philosopher inside Google DeepMind AI, from The Guardian
6,800 words, or about 27 minutes

When Iason Gabriel, an Oxford academic who also did crisis work for the UN, joined Google DeepMind in 2017 as an ethicist, it seemed to some at the time like a bit of tech excess. Fast-forward less than a decade, and the role seems almost essential. This story uses Gabriel’s career to trace the evolution of thinking around AI ethics. In particular, it zooms in on the historic tension between the philosophers worried about AI’s short-term threats, like bias and discrimination, and those focused on longer-term threats such as societal collapse and existential risk. Gabriel tried to bridge that divide with a paper in 2020, arguing that alignment isn't just a technical problem of getting machines to follow values but a harder political problem of choosing which values are important. That argument has taken on increased relevance today as governments wrestle with how to harness and regulate the technology. While questions about how individual models behave remain important, they now sit alongside much broader societal concerns, like the fact that AI’s power is concentrated in a handful of companies, that geopolitical competition over developing AI could further destabilize a fraying world order, that the trillion-dollar price tag is warping economies and that AI presents an entirely new and dangerous toolbox for bad actors. So it’s perhaps no surprise that Gabriel now leads a team of ethicists at Google DeepMind tasked with thinking through those questions. Whether those conversations meaningfully influence a company competing in what its founder has called a "ferocious" commercial AI race is another question.

Why American data centers can’t plug in, from Works in Progress
4,500 words, or about 18 minutes

One of the biggest constraints on the AI boom isn't chips, land or even electricity. It's permission to connect to the grid. Across the US, power plants, data centers and other large energy projects face yearslong waits for grid interconnection, the process that determines how and when new infrastructure can be connected to the electricity network. This story argues that much of the problem is bureaucratic rather than physical. Part of the issue is the approval system itself. Grid connections are generally processed on a first-come, first-served basis, allowing speculative projects to clog the queue and delaying more viable ones. Developers often submit multiple applications for the same project as insurance, making the backlog even worse. The technical review process also moves slowly: Grid operators typically analyze projects one at a time, modeling how each would affect electricity flows during periods of peak demand before determining what network upgrades are required. The essay outlines several possible reforms. One is to auction scarce grid capacity so that developers willing to pay the most for rapid access move to the front of the queue. Another is to assess projects in clusters rather than individually, allowing grid operators to process many applications simultaneously while making the system less vulnerable to speculative projects that later drop out. It also suggests allowing some new data centers to connect earlier on the condition that they reduce demand or switch to backup generation when the grid is under stress.

As the world warms, the risk of snakebites is rising, from Grist
2,000 words, or about 8 minutes

Here's one consequence of climate change you probably hadn't considered: venomous snakebites The precise impact of global warming on snake attacks is difficult to quantify, but the evidence is beginning to point in one direction: up. Research from Emory University found that the risk of a snakebite rises by around 6 percent for every degree Celsius increase in daily temperature. Reports from poison centers and snakebite hotlines from the US to Thailand also suggest encounters with venomous snakes are becoming more common. Part of the reason is that rising temperatures are expanding the ranges of some species, while warmer weather is bringing them out of hibernation earlier. In India, for example, the country's four most common venomous snakes are increasingly spotted in places far beyond their typical habitats. Meanwhile, urban expansion — itself often driven by climate-related pressures on agriculture — pushes humans closer to snake habitats. Unfortunately, healthcare systems aren’t really prepared. Antivenoms remain expensive and difficult to manufacture, leading some companies to abandon production altogether. Sanofi Pasteur, for example, stopped producing its Fav-Afrique antivenom in 2015 because it was no longer commercially viable. Recognizing the threat, the World Health Organization has set a goal of halving snakebite deaths and disability by 2030. Its recommendations extend beyond producing more antivenom, including better surveillance, conservation efforts, healthcare worker training and public education. For anyone unlucky enough to suffer a snakebite, those could all be lifesavers.