Convergence Watch: The Cheapest Nuclear Containment Dome Is a Mile of Rock

In a 9.5-million-patent corpus of US utility grants, exactly two describe the same idea: drop a working nuclear reactor into a hole drilled by an oil rig and let a mile of bedrock do the job that concrete domes used to do. Both belong to the same three-year-old startup. The first issued last November, the second ten days ago. The company plans to take a reactor critical inside a Kansas industrial park on July 4, 2026.

That company is Deep Fission, founded by a UC Berkeley physicist named Richard Muller and his daughter Elizabeth. They raised $30 million in a September 2025 reverse-merger and another $80 million in a February 2026 stock placement at five times the earlier price. They signed a partnership with Endeavour Energy to deliver two gigawatts of power to data centers. The Department of Energy selected them for the Reactor Pilot Program, created by Trump-era executive orders that pulled Deep Fission’s own target from 2029 to 2026. They broke ground at the former Kansas Army Ammunition Plant on December 9, 2025.

What they are actually doing is the most interesting energy convergence of the last decade. They are taking the borehole, a piece of physical infrastructure the petroleum industry has perfected over a century, and turning it into the cheapest nuclear containment vessel ever built.

The two patents

US12469612, issued November 11, 2025, is titled “Heat exchanger in a borehole receives heat through a rock formation from another borehole which includes a nuclear reactor.” The independent claim describes two parallel drillholes punched from the surface “through one or more subterranean formations.” A reactor core sits at the bottom of the first. A heat exchanger sits at the bottom of the second. The two are “thermally coupled” through a slab of rock between them. Primary coolant never leaves the reactor borehole; secondary coolant carries the heat back up the second hole to a surface turbine.

US12614643, issued April 28, 2026, is the simpler variant: reactor and heat exchanger in the same hole. The dependent claims specify “void collectors” — inverted cup-shaped traps that capture vaporized coolant inside the primary loop, an engineering touch suggesting the inventors are taking seriously the question of what happens when water flashes to steam at the bottom of a mile-deep well. They specify primary and secondary coolants are water. They specify pipes running from the surface “to a depth of the first drillhole that is deeper than the depth of a fluid barrier.”

These are not abstract patents. They are a build manual.

Why the rock does the work

A pressurized water reactor, the dominant US nuclear technology since the 1950s, runs at roughly 160 atmospheres and 315°C. Every existing PWR in America surrounds that core with a steel pressure vessel as thick as a man’s torso, then with a reinforced-concrete containment building the size of a parking garage. Those structures exist because if the reactor depressurizes, the water inside flashes to steam. The dome and the vessel are the engineering price you pay for a small, hot, very pressurized object sitting near other valuable objects.

A column of water 1.6 kilometers tall weighs enough to deliver 160 atmospheres of pressure for free. That is the trick. Drop a PWR into the bottom of a mile-deep borehole, fill the hole with water, and ambient hydrostatic pressure matches the operating pressure of the reactor. The pressure differential across the fuel cladding collapses to nearly zero. According to Deep Fission’s NRC white paper and Elizabeth Muller’s interviews with IEEE Spectrum and DataCenter Dynamics, this single substitution removes the need for the steel pressure vessel and the concrete dome. The company estimates the cost reduction at up to 80 percent versus a surface PWR and a build time of six months per reactor.

You are also surrounded, on every side, by a billion tons of solid rock. That is not a metaphor; it is the company’s containment claim to the regulator. The radioactivity, in the words Muller used at a January 2025 event, “stays at the bottom.”

Why this is a convergence story, not a nuclear story

The interesting move here is not that someone wants to build a smaller reactor. Half a dozen well-funded startups want to build smaller reactors: Oklo and Last Energy were selected for the same DOE pilot program, NuScale already has NRC certification for a surface SMR, and the Energy Information Administration counts more than a dozen US microreactor designs in development. What is unique about Deep Fission, and what shows up cleanly in the patent record, is that it is the only US company solving the containment problem by leasing real estate from the petroleum industry.

A 30-inch borehole drilled to 1.6 kilometers is a routine product of the oil and gas service business. Halliburton and Baker Hughes drill thousands of them a year for unrelated purposes. The drilling rigs exist. The casing standards exist. Directional steering, cement bond logging, wireline tools — all of it is commodity infrastructure with a dense supplier network and a regulator (state oil and gas commissions) that has spent a century thinking about keeping what is in the hole away from the water table. Every line item Deep Fission removes from the nuclear bill of materials — the pressure vessel, the dome, the cooling tower, the seismic-rated foundation — already exists in the petroleum supply chain in a form that is two orders of magnitude cheaper.

The convergence is also visible in the inventors’ biographies. Elizabeth Muller spent the previous decade running a different company, also founded with her father, called Deep Isolation. Its single-sentence thesis: the cheapest place to put high-level nuclear waste is at the bottom of a horizontal extension off an oil-style borehole. By the company’s own count it holds 70 issued patents and has signed Amentum to commercialize the concept globally. In April 2025 Deep Isolation and Deep Fission signed a memorandum of understanding to combine their portfolios into a “full-cycle nuclear energy solution” — the same hole that powers the reactor, eventually, becomes the hole that buries the spent fuel.

That arc is the adjacent possible at work. If you can convince a regulator that a borehole holds dangerous things safely for ten thousand years, the leap from “buried waste” to “buried source” is shorter than the leap from “concrete dome” to “no dome.” OpenAlex returns 229 papers with “deep borehole” plus nuclear, waste, disposal, or repository in the title, the bulk from Sandia National Labs and university programs that ran from the late 1980s through the early 2010s. Energy Policy published one in 2014 titled “Can deep boreholes solve America’s nuclear waste problem?” The Mullers read those papers, built a waste company on them, and then asked the obvious next question.

What an R&D director should take from this

Two things matter. First, the cost target. Five to seven cents per kilowatt-hour, if Deep Fission can hold it through the build, competes with combined-cycle gas at scale and demolishes the surface-PWR price curve, which has spent the decade getting worse, not better. Hyperscalers do not need that price for a few flagship sites; they need it for a continent of compute, and the logic of one rig, six months, fifteen megawatts, repeat fits the volume profile of an AI capex cycle.

Second, the regulatory hinge. Kansas state law currently prohibits a private operator from selling power directly to a customer; Muller has acknowledged in KCUR and Lawrence Times coverage that the company is still working through the utility-relationship problem with Evergy. The DOE pilot pathway lets the reactor exist; it does not yet let the reactor sell electrons to a data center across the fence. The whole economic case rests on whether the federal program can carry a commercial power-purchase agreement through to first revenue. That is the variable to watch over the next twelve months. If it works, every oil-services company in Texas is a step away from being a nuclear-services company.

If it doesn’t, Deep Fission has at least already proved a smaller and possibly larger thing: a nuclear reactor and an oil well were always the same hole, drawn at different scales.

Method note

Patent data was drawn from the USPTO bulk grant feed, current through May 5, 2026 (9.58 million issued US utility grants). The “two patents” finding comes from a full-text search for borehole, drillhole, nuclear, and reactor co-occurring in an issued grant; the claims of US12469612 and US12614643 were read in full. Literature counts come from a title-keyword search of OpenAlex (~357 million scholarly works). Funding and project-status figures come from Deep Fission press releases and contemporaneous coverage in IEEE Spectrum, New Atlas, Datacenter Dynamics, World Nuclear News, KCUR, and the American Nuclear Society newswire, dated between January 2025 and April 2026. Deep Isolation’s 70-patent count is the company’s own public claim. Cost and pricing figures (5–7 ¢/kWh, 80 percent capex reduction) are forward-looking targets stated by Deep Fission, not third-party engineering estimates.