Deep Fission, Inc.

Overview

Deep Fission is a Berkeley-based nuclear startup developing a small modular reactor designed to operate one mile underground, inside a drilled borehole. The company was founded in 2023 by Liz Muller, who serves as CEO, and her father Richard Muller, who serves as CTO and is an emeritus physics professor at UC Berkeley and a MacArthur Fellow. Both previously co-founded Deep Isolation in 2016, a company focused on disposing of nuclear waste in deep boreholes, which gave the pair direct experience with borehole geology and drilling engineering that now underpins Deep Fission's reactor siting approach. The company exited stealth in 2024, went public in September 2025 through a $30 million reverse merger with a shell company, Surfside Acquisition, and raised an additional $80 million in February 2026. In May 2026 it filed to uplist to the Nasdaq Global Market, initially targeting a $1.66 billion valuation and about $150 million; the IPO ultimately priced on June 18, 2026 at $16.00 per share, the low end of a reduced range, raising roughly $40 million at a market capitalization near $630 million. No reactor has been built. The company began drilling data acquisition wells in Parsons, Kansas in March 2026.

Deep Fission's Gravity reactor is a 15-megawatt pressurized water reactor narrow enough to be lowered into a drilled borehole and designed to run at one mile of depth where water pressure alone maintains the 160 atmospheres the reactor requires, replacing the thick steel pressure vessel that is one of the costliest and most time-consuming components of a conventional nuclear plant. The company claims the approach can reduce nuclear construction cost by 70 to 80 percent relative to surface plants and achieve a levelized cost of electricity between $0.05 and $0.07 per kilowatt-hour. Both claims are theoretical. The Gravity reactor has not been licensed, constructed, or operated. The path from borehole drilling to commercial power generation involves regulatory questions, fuel qualification, and engineering challenges that have not yet been addressed.

Technology & projects

The gravity trick

The insight at the center of Deep Fission's design is worth examining carefully, because it is genuinely clever and the engineering case for it is real, even though the commercial case is unproven. A pressurized water reactor requires high-pressure coolant, roughly 160 atmospheres, to prevent the reactor coolant water from boiling at operating temperature. At the surface, maintaining that pressure requires a reactor pressure vessel: a steel cylinder 20 to 30 centimeters thick, typically around 4 to 5 meters in diameter and 12 meters tall, forged from special low-alloy steel and requiring years to manufacture at one of the few global facilities capable of producing it. The pressure vessel and its associated primary coolant piping represent a major fraction of a conventional nuclear plant's capital cost and schedule.

A column of water one mile tall exerts approximately 160 atmospheres of pressure at its base. If you put a reactor at the bottom of a water-filled borehole one mile deep, the water column provides the required operating pressure through hydrostatics alone, with no engineered pressure vessel required. The reactor still needs structural containment, a tube strong enough to hold its geometry and prevent external rock pressure from crushing it, but this is a much simpler engineering problem than building a pressure vessel rated to 160 atmospheres from the outside. Deep Fission's tube is roughly the diameter of a large oil well casing, manufacturable with existing oilfield pipe supply chains. The same water column provides emergency cooling: if the reactor trips, the water above it absorbs decay heat passively, without pumps or operator action. The surrounding rock provides radiation shielding that a surface plant must achieve with multiple meters of concrete and steel.

The approach borrows from two engineering fields with well-established cost curves. Deep oil and gas wells reach one mile routinely; the drilling industry has developed the rigs, drill bits, and wellbore management techniques to do this in days to weeks at costs that have declined dramatically over decades. Geothermal energy developers drill boreholes to extract heat from the Earth and transfer it to surface systems, working out the heat transfer physics that Deep Fission's cooling approach draws on. Deep Fission is not inventing borehole drilling or PWR physics; it is combining them in a configuration that has not been attempted before for power generation.

The open questions are real. Getting a reactor into a borehole and retrieving it for refueling requires handling fuel assemblies at depth, a process with no prior art in the nuclear industry. The NRC has no existing license framework for a borehole reactor; the entire regulatory pathway must be developed from scratch, including safety analysis methods that account for one-mile-deep operations. Heat extraction from the bottom of a borehole to the surface power systems involves thermal losses along the entire mile-long column. And the 15-megawatt output of the Gravity reactor is small enough that clustering many units is necessary to serve a data center or industrial load at commercial scale, which multiplies the permitting and drilling effort proportionally. Deep Fission's cost and LCOE projections are pre-engineering estimates with no construction data behind them. They may prove accurate; they may not.

Funding

Deep Fission raised a $4 million seed round when it exited stealth in 2024. In September 2025 it closed a $30 million round at $3.00 per share and went public through a reverse merger with a shell company, Surfside Acquisition, with investors including 8VC (the fund co-founded by Joe Lonsdale, who also co-founded Palantir), Deep Future, Wave Function, Ed Eisler of EE Holdings, and Mark Tompkins of Montrose Capital. In February 2026, the company raised an additional $80 million, led by Eisler and Tompkins and joined by Blue Owl Capital, whose real-assets platform invested $20 million and took rights of first offer on power offtake and project financing. Forbes reported roughly $122 million raised in total by April 2026, at a $1 billion post-money valuation. In May 2026, Deep Fission filed to uplist to the Nasdaq Global Market under the ticker FISN, initially seeking about $150 million at a target valuation near $1.66 billion. The offering priced on June 18, 2026 at $16.00 per share, the low end of a reduced range, raising roughly $40 million; the stock has traded below its offer price since. Deep Fission remains pre-revenue, having drilled test wells but not built, licensed, or operated a reactor.

In January 2025, the company signed a strategic partnership with Endeavour Energy, LLC, a US data center developer behind the Edged brand (not the Australian grid operator of the same name), to co-develop up to 2 gigawatts of power for AI data centers, with first reactors targeted for 2029. At 15 megawatts per reactor, 2 gigawatts would require approximately 133 individual units. In June 2026, Deep Fission said its customer pipeline had grown to up to 18.5 gigawatts of letters of intent with data centers, co-developers, and industrial parks, up from 12.5 gigawatts a year earlier. These are non-binding commercial intent agreements rather than financed orders; either side can walk away, and they indicate market interest in the product if it works, not certainty that it will.

Key considerations

Deep Fission is at the earliest stage of any company profiled here: no reactor built, no operating license, no commercial revenue. The borehole PWR concept has first-principles merit, but merit and buildability are different questions. The NRC must develop an entirely new licensing framework for a reactor configuration it has never reviewed, at a depth it has no direct safety analysis experience with. The fuel handling process, inserting and retrieving fuel assemblies one mile underground, has no procedural precedent in U.S. commercial nuclear operations. Heat extraction efficiency over a one-mile borehole depends on thermal resistance calculations that have not been validated at any scale. Each of these is a solvable engineering problem in principle; none of them has been solved yet.

The IPO underscored investor caution: Deep Fission targeted a $1.66 billion valuation but priced at the low end of a reduced range, raising about $40 million, and the stock has traded below its offer price since. The company's S-1 carries a going-concern warning, with an accumulated deficit of roughly $88 million by March 2026 and auditors noting it may need additional capital within a year. For context, TerraPower has raised more than $3.4 billion in total capital, holds a roughly $2 billion DOE cost-share, received an NRC construction permit in March 2026, and began nuclear construction in April 2026, yet its first plant is not expected to generate power until 2031. Deep Fission has raised roughly $122 million privately plus about $40 million from its IPO, drilled test wells in Kansas, and is in early pre-application discussions with the NRC with no license application filed. That gap is not disqualifying, but it is the honest picture of where the company stands.

Sources