By dropping a nuclear reactor 1.6 kilometers (1 mile) underground, Deep Fission aims to use the weight of a billion tons of rock and water as a natural containment system comparable to concrete domes and cooling towers. With the fission reaction occurring far below the surface, steam can safely circulate in a closed loop to generate power.
The California-based startup announced in October that prospective customers had signed non-binding letters of intent for 12.5 gigawatts of power involving data center developers, industrial parks, and other (mostly undisclosed) strategic partners, with initial sites under consideration in Kansas, Texas, and Utah. Individual agreements range from under 1 GW to 2 GW, including a previously announced 2 GW deal with Endeavour’s Edged division, a company building high-efficiency data centers that don’t rely on water to cool servers.
Deep Fission’s small modular reactor (SMR), called Gravity, is designed to stand 9 meters tall while remaining slim enough to fit inside a borehole roughly three-quarters of a meter wide. The company says its modular approach allows multiple 15-megawatt reactors to be clustered on a single site: A block of 10 would total 150 MW, and Deep Fission claims that larger groupings could scale to 1.5 GW.
Deep Fission claims that using geological depth as containment could make nuclear energy cheaper, safer, and deployable in months at a fraction of a conventional plant’s footprint. Still, independent experts say the underground design introduces its own uncertainties, both regulatory and practical.
Innovative Nuclear Reactor Design
“We are unique in that we’ve combined three existing mature technologies in a way that nobody had ever thought of before,” says Liz Muller, founder of Deep Fission. The same oil and gas drilling techniques that reliably reach kilometer-deep wells can be adapted to host nuclear reactors, while using steam to transfer heat to the surface for power generation follows geothermal methods. Locating the reactors under a deep water column subjects them to roughly 160 atmospheres of pressure—the same conditions maintained inside a conventional nuclear reactor—which forms a natural seal to keep any radioactive coolant or steam contained at depth, preventing leaks from reaching the surface.
Deep Fission is one of several SMR companies participating in the U.S. Department of Energy’s Reactor Pilot Program, which targets initial criticality—producing a self-sustaining chain reaction for the first time—in July 2026 through an expedited authorization process. While some industry players like Oklo and Kairos Power are designing novel reactors, Deep Fission selected a standard pressurized water reactor (PWR) using readily available low-enriched uranium fuel, similar to 64 of the 95 reactors licensed to operate in the U.S. today.
The idea of bringing PWRs underground stems from Muller’s earlier work at Deep Isolation, a company she founded to develop borehole disposal for nuclear waste containment. She remembers that one day, a customer raised a hypothetical scenario: If someone accidentally put fresh uranium fuel underground instead of waste, could it start a chain reaction due to the pressure? “The answer is no,” says Muller, “a single fuel assembly a mile underground will not go critical. But, as we were doing those calculations, we recognized that you actually have the conditions you want for a nuclear reactor when you’re in a borehole a mile underground.”
Deep Fission’s reactor would operate at the bottom of a 1.6-kilometer-deep borehole, sending steam upward to power a turbine.Deep Fission
Deep Fission’s safety case draws on oil and gas drilling precedent. “The oil and gas industry has shown how to protect the water table,” Muller says, referring to the upper groundwater zone where water saturates rock and soil. “They have really nasty stuff coming out of their borehole. All that’s coming out of our borehole is clean water, so all the radioactivity stays at the bottom, and it’s just clean hot water coming up, as it would with geothermal.”
Underground siting removes many potential dangers for surface reactors, such as aircraft impacts, vehicle collisions, tornadoes, hurricanes, and flooding. Muller argues that even a worst-case scenario could cause economic losses to the reactor or borehole, but wouldn’t impact humans or the environment. If an earthquake ever disrupted the site, “you seal it off at the bottom of the borehole, plug up the borehole, and you have your waste in safe disposal,” she says.
Construction takes about six months, including four weeks of drilling, eight to 10 weeks of installation to lower the canister into the hole, and another two months for commissioning, which involves system tests, inspections, and initial low-power runs to confirm the reactor behaves safely as designed. For waste management, Deep Fission is eyeing deep geological disposal in the very borehole systems they deploy for their reactors. The company signed a memorandum of understanding with Deep Isolation in April to explore the licensing and use of its deep repository technology.
Regulatory Challenges for Underground Reactors
While the company expects to license its reactor under existing Nuclear Regulatory Commission (NRC) rules, the agency will likely need supplemental review guidance to evaluate a design that’s installed underground, reliant on surrounding geology for containment, and monitored remotely—areas that current reactor regulations don’t yet cover in detail.
Independent experts see both promise and challenges. Leslie Dewan, a nuclear engineer with experience designing molten salt-based SMRs, says placing the reactor deep underground “changes the parameters for shielding, containment, and site design in a way that could simplify surface infrastructure and streamline construction,” especially if Deep Fission can deliver on security, footprint, and cost. The company claims it can cut overall costs by 70 to 80 percent compared with full-scale nuclear plants. Its projected levelized cost of electricity of US $50 to 70 per megawatt-hour—the average price to cover a plant’s lifetime costs—is lower than the estimated averages for other SMRs.
But Dewan cautions that the concept still faces technical unknowns. “A design that relies on the surrounding geology for safety and containment needs to demonstrate a deep understanding of subsurface behavior, including the stability of the rock formations, groundwater movement, heat transfer, and long-term site stability,” Dewan says. “There are also operational considerations around monitoring, access, and decommissioning. But none of these are necessarily showstoppers: They’re all areas that can be addressed through rigorous engineering and thoughtful planning.”
Each Gravity unit is designed to generate 15 MW of power, though an array of multiple units can meet gigawatt-scale power demands.Deep Fission
The underground configuration could also complicate operations. Mary Lou Dunzik-Gougar, a nuclear engineering professor and associate dean of the Idaho State University College of Science and Engineering, calls the use of a conventional PWR “solid,” but says that placing the reactor so far underground seems less promising. “While the underground placement does provide shielding from the radiation being produced, maintenance and refueling of the reactor will be complicated by the need to bring the reactor to the surface and to provide alternative shielding for workers,” she says.
Sending water to the reactor and steam to the surface for power production also requires plumbing that could fail or lose heat, says Dunzik-Gougar. Operating and controlling the reactor remotely from the surface means extensive cabling or wireless communication, again creating points of potential failure. Licensing may also prove difficult given the unusual configuration and distance between the reactor, its operators, maintenance staff, and supporting equipment installed above-ground.
Deep Fission’s early regulatory filings acknowledge these challenges, noting that the deep borehole complicates compliance around monitoring and visual inspection, and that additional guidance will be required for remote operation.
Deep Fission’s Expansion Plans
Deep Fission exited stealth mode in 2024 with a $4 million funding round. This year, after new federal executive orders accelerated the DOE’s advanced-reactor timelines—moving Deep Fission’s own target from 2029 to 2026—the company completed a reverse-merger transaction and raised $30 million through a private placement.
Deep Fission expects to own and operate at least its first reactor, but may potentially pursue a different business model thereafter. “There is tremendous demand right now for electricity. By having a flexible model, we can grow faster than if we wanted to build, own and operate all of our facilities ourselves,” says Muller. “We do recognize an advantage there, but in terms of international expansion and bandwidth to build many reactors, I think having a licensing model or even selling the reactors is an area where we want to be flexible.”
The company aims to finalize its reactor design and confirm the pilot site in the coming months. Muller says the plan is to drill the borehole, lower the canister, load the fuel, and bring the reactor to criticality underground in 2026. Sites in Utah, Texas, and Kansas are among the leading candidates for the first commercial-scale projects, which could begin construction in 2027 or 2028, depending on the speed of DOE and NRC approvals. Deep Fission expects to start manufacturing components for the first unit in 2026 and does not anticipate major bottlenecks aside from typical long-lead items.
“Trying to hit the 4 July, 2026 target is going to be extremely challenging, not just for us but for everybody, but what I feel really good about is our ability to build this pilot reactor in 2026,” Muller says. “Had the DOE pilot program not been there, we would’ve been on a significantly slower timeframe.”
