At some point in 2021, if all goes according to plan, earthmoving equipment will arrive at a scruffy, windswept spot in eastern Idaho and begin excavating a large hole. The patch of sagebrush in question lies not far from the optimistically named Atomic City, Idaho. The town isn’t much to look at today. Fewer than 30 residents remain, and the community’s single gas station no longer sells gas. In its day, though, Atomic City was a thriving little boomtown. Throughout the 1950s and 1960s, the surrounding desert hummed with government-funded nuclear-research projects. The reactors that power nuclear submarines were developed here, as was the first reactor to provide electricity to the civilian power grid.

That boom faded, and most of those reactors went silent. Nonetheless, the Idaho National Laboratory, today under the supervision of the Department of Energy, continues to work on cutting-edge designs. Over the years, INL projects have included exotic reactors that work at unimaginably high temperatures and are cooled by helium gas, liquid salt, or even molten metal, instead of water. For much of the lab’s existence, its researchers hoped that breakthroughs like these would revolutionize nuclear power in the United States. New reactor designs would mean nuclear plants so safe that they could be built right next to cities, and so efficient that they could power our economy with nearly limitless, clean electricity.

That nuclear revolution has been slow in coming. Though nuclear power still provides nearly 20 percent of the electricity in this country, most of the reactors in service today use the conservative, water-cooled designs that became the industry standard by the 1970s. (In fact, many were built in the 1970s.) The country’s nuclear infrastructure is sturdy and reliable but hardly cutting-edge. And it’s not always economically competitive, as fracking drives down natural-gas prices and heavily subsidized wind and solar facilities undercut the rates that nuclear plants can charge their customers. In such an economic climate, it’s no surprise that some existing nuclear facilities are closing and that construction of new plants has crawled almost to a halt.

The work crews expected to arrive in the Idaho desert a year or so from now will be trying something new. They won’t be building the large, domed containment vessels that most of us picture when we think of nuclear plants. Instead, they will first dig a long, concrete-lined trench; imagine a very large, very deep swimming pool. Then, trucks will arrive carrying the components of a pint-size nuclear reactor. In traditional nuclear-plant construction, crews build reactors on-site, a painstaking process that can drag on for years. But this reactor will arrive in Idaho almost entirely assembled, fresh from a Virginia factory. Workers will simply connect the major components. The finished reactor will stand vertically inside a steel containment vessel about 15 feet wide and 75 feet tall. It will look a bit like a high-tech farm silo.

This sort of diminutive, factory-built nuclear reactor is known as a Small Modular Reactor, or SMR. The design planned for the Idaho project is the brainchild of Jose Reyes, a former professor of nuclear engineering at Oregon State University. Reyes left academia in 2007 to launch NuScale, a company dedicated to making SMRs a viable business. It is one of dozens of U.S. startups attempting to build various types of small, next-generation nuclear reactors, but it’s leading the pack in terms of navigating the Nuclear Regulatory Commission’s byzantine licensing process. “We’re on track to be the first SMR to get certified,” Reyes told me.

Alone, a single SMR of this type will generate a modest 60 megawatts of electricity. That is enough to power about 40,000 homes and is a small fraction of the 1,000 megawatts or so that today’s full-size reactors produce. But NuScale’s SMR is not designed to operate alone. When the project is complete, 12 of these modules will stand side by side in the concrete trench, which will be flooded with water as a safety precaution. When the Idaho plant gets connected to the grid—sometime in 2026, its backers hope—it could mark the start of a new era of nuclear power. Advocates envision fleets of small, affordable power plants that could be built quickly near where power is required, and expanded as needed.

If the small-nuclear concept proves out, it could fill some crucial energy niches—and at a crucial time. Democratic candidates are promising various versions of a Green New Deal, which would drastically restrict fossil fuels and, theoretically, replace that energy mostly with wind and solar power. Regardless of who is in the White House, many states are already committed to phasing out the use of coal and natural gas to make electricity. Reality will likely put a crimp in those plans. New research suggests that wind and solar will never fully replace today’s electricity sources. We will always need more reliable sources of energy—ones that don’t fluctuate with the weather.

Even their backers admit that green-energy proposals would be massively expensive. Joe Biden puts the cost of his plan at $1.7 trillion over ten years. Some analysts estimate the total cost of Alexandria Ocasio-Cortez’s Green New Deal at over $50 trillion. Moreover, these plans call for unprecedented regulation of the energy sector, something closer to a command economy than a free market. The potential damage to the U.S. economy, as well as to Americans’ household finances, is difficult to calculate.

But what if there were sources of zero-carbon electricity that didn’t require heavy-handed regulation to make them viable in the marketplace? What if we could produce more power—and do it affordably, with minimal environmental impact? That’s the almost utopian vision that some backers see for the next generation of nuclear power. According to Ted Nordhaus, founder of the “eco-modernist” Breakthrough Institute, SMRs could offer a more decentralized, entrepreneurial approach to reducing CO2 emissions without hobbling the economy. “Will any of them ultimately be a success in the marketplace? You can’t say for sure,” he said. “But I think a bunch of them will get licenses to build test plants.”

SMRs involve a radical rethinking of how to build and operate nuclear plants. Since the early days of atomic power, conventional wisdom has called for making each reactor as large as possible. After all, a reactor is expensive to build and requires a highly skilled workforce to operate. Why not maximize the output from each unit to get the best return on that investment?

Jose Reyes began questioning the bigger-is-better model in 2004, when he spent a year overseas as an advisor to the UN’s International Atomic Energy Agency. “I met folks from Africa, Malaysia, Indonesia,” he said. “They all said the same thing: ‘We need power, but we need it in smaller increments. We can’t afford big reactors.’ ” When he returned to the U.S., Reyes began working on a prototype. His design is a scaled-down version of the kind of water-cooled reactor in common use today—but radically simplified. The entire system operates in a vacuum inside a sealed containment vessel not unlike a thermos bottle. It requires no complex plumbing or pumps. Large nuclear plants demand a constant flow of water to stay cool even when they are shut down. (The Three Mile Island and Fukushima nuclear accidents were partly caused by the failure of their cooling pumps.) But the NuScale reactor is designed to stay cool passively. In the event of malfunction, the water surrounding the containment vessel would safely carry away heat, and even if the tank eventually went dry, the residual heat would dissipate harmlessly into the atmosphere.

Virtually all the companies designing advanced reactors today echo this claim of “walk-away safety.” It’s a key selling point for miniature power plants intended to be sited near communities, inside industrial complexes, and perhaps even on military bases. It is also a key advantage of smaller reactors: because they contain far less fuel, there is far less heat to dissipate if something goes wrong. The classic “China Syndrome” meltdown isn’t possible. The startups are betting that utilities will find it easier to sell the public on these next-generation designs. NuScale already has customers lined up for its proposed plant. One is the federal government, which will buy some electricity to help power the Idaho lab itself. The other is a consortium of utilities in six western states.

Whether other utilities will embrace new nuclear designs is an open question. But one thing is clear: the economic logic behind the traditional approach to nuclear power is in trouble. Recent experience in the U.S. and Europe reveals just how hard it has become for Western countries to build big, capital-intensive nuclear projects. Two reactors under construction at Georgia’s Vogtle power station have encountered massive delays and cost overruns and still aren’t complete after a decade. “These huge new reactors appear to be exhibiting dis-economies of scale,” Jesse Jenkins, an energy analyst at Princeton University, said. “They are the sorts of massive infrastructure projects that we’re just not very good at building in this country.” Jacopo Buongiorno, a professor of nuclear science and engineering at MIT, believes that new modular designs, where most of the unit is crafted in a factory and then bolted together on-site, could change that. “It’s a potential game-changer,” he said. “But not a slam-dunk.”

Building a much smaller reactor in a factory offers some surprisingly big advantages. “We call it ‘the economies of small,’ ” NuScale’s Reyes said. “We can build our containment vessel and reactor components to extremely high tolerances because we’re doing it in a shop, not on location,” he explained. The modular design also means that workers, designers, and managers all gain experience much more quickly. Instead of spending years building a single unit, NuScale hopes eventually to turn out dozens of modules each year. As with any manufactured product, that fast pace should mean rapid improvements in manufacturing efficiency, quality—and cost. “That’s the promise of SMRs,” Jenkins said. “Even though they’re not using radically new technology, they could have radically different economics.”

NuScale leads in the race to get an SMR design approved by regulators partly because its underlying technology resembles today’s commercial reactors. Some other startups are taking more exotic paths. These include reactors that operate at very high temperatures and rely on helium or molten salts as coolants. Many of the proposed designs can run on unconventional types of nuclear fuel, such as spent nuclear pellets or plutonium recycled from retired nuclear weapons. In nuclear nomenclature, these are known as Generation IV reactors. (A quick primer: Generation I denotes primitive designs; Generation II includes most of today’s commercial systems, known as light-water reactors; Generation III describes various enhanced versions of the light-water type, such as the Westinghouse models being built at the Vogtle plant in Georgia.) If any one of these startups makes it to market with a Generation IV reactor, it will be the biggest technical advance in commercial nuclear power since the dawn of the industry.

But while these startup designs sound futuristic, most rely on research done at the Idaho and other national laboratories going back to the 1950s. Oklo, a Sunnyvale, California, startup, is developing a micro-reactor that runs on a metallic fuel. The design is partly modeled on an experimental reactor built by the lab in 1964. “That reactor ran for 30 years,” the company’s cofounder, Caroline Cochran, said. Another California startup, Kairos Power, uses uranium fuel encased in ceramic “pebbles,” a technology also developed at INL. The Maryland-based X-energy is working on a high-temperature reactor that employs helium gas as a coolant. That concept, too, has roots in the Idaho lab. MIT’s Buongiorno believes that this enormous stockpile of existing research will help streamline the process of bringing new designs to market. “Today, the R&D requirements for SMRs and high-temperature reactors are minimal,” he said. “We can build these tomorrow, practically.”

“Because smaller reactors contain far less fuel, there is far less heat to dissipate if something goes wrong.”

The push for new reactor designs isn’t coming just from startups. GE and Hitachi, for example, have teamed up to develop a 300-megawatt modular reactor. Nor is the advanced nuclear movement strictly a U.S. phenomenon. Several Canadian companies are working with that country’s national laboratories to develop demonstration models of their designs. Other programs are under way in Great Britain, Indonesia, China, and elsewhere. By one estimate, more than 100 advanced nuclear projects are in the works worldwide.

There have been some speed bumps: TerraPower, a company backed by Bill Gates, signed a deal to build a demonstration reactor in China, which it hoped to finish by 2022. The Trump administration scuttled that deal in 2018, citing concerns about allowing China access to sensitive U.S. technology. TerraPower is now seeking funding to build a demonstration plant in the U.S., Gates has said. Transatomic Power, a Cambridge, Massachusetts, startup partly funded by tech billionaire Peter Thiel, closed its doors in 2018. Clearly, a career as a nuclear entrepreneur is not for the risk-averse, but Cochran, who launched Oklo after receiving a master’s degree in nuclear engineering from MIT, believes that new reactor designs will open new markets. Oklo’s reactor, which will produce a relatively minuscule 1.5 megawatts of electricity, is intended to operate for decades without refueling, and to fit in a structure about the size of a family home.

The company’s initial market will be off-the-grid locations, such as mining operations or remote Alaskan towns, which today must fly fuel in for their diesel generators. “It’s a very intriguing idea,” says Buongiorno; both Oklo cofounders are his former students. “You aren’t targeting the commodity market; you’re targeting an area where your competition is expensive diesel generation.”