The Innovation Attorney | May 23, 2026

Venture capital deployed more than $2.5 billion in the week of May 17 to 23, 2026, into five emerging technical arenas: agentic infrastructure, physical AI systems, grid-integrated energy technology, software supply chain security, and domain-specific autonomous agents, confirming that the industry has moved past model-layer competition into the industrial build-out of the AI economy.

The transition was predictable, and now it is measurable. Q1 2026 saw $300 billion in global venture investment, representing nearly 70 percent of total capital deployed across all of 2025. That surge was concentrated at the frontier model layer: OpenAI, Anthropic, xAI, and Waymo consumed 65 percent of all global venture dollars in the first quarter. The secondary effect is now visible in the May deal flow. With frontier models funded to a level that no realistic competitive entry can reach, capital is flowing toward the physical, financial, and security layers required to make those models productive.

Where Is the AI Infrastructure Capital Going?

Armada closed a $230 million oversubscribed Series B co-led by Overmatch, BlackRock, and 8090 Industries, reaching a $2 billion valuation, to manufacture modular megawatt-scale data centers that deploy in environments where centralized cloud infrastructure is unavailable: military vessels, offshore oil platforms, and remote mining operations. The Leviathan modular data center is not a product for conventional enterprise use. It is physical compute sovereignty for operators running classified or sensitive AI workloads who cannot route data through shared infrastructure regardless of cost. The simultaneously announced manufacturing agreement with Johnson Controls at a 400,000-square-foot Arizona factory signals that Armada has solved the scale problem that constrains most hardware startups.

GridCARE raised $64 million in a Series A led by Sutter Hill Ventures, with John Doerr and Stanford University alongside National Grid Partners, to compress AI data center grid interconnection timelines from ten years to months. The platform uses physics-based AI to model quadrillions of grid configurations in real time, identifying stranded capacity that traditional analysis misses. The company is currently active on projects delivering power for more than two gigawatts of AI compute capacity across more than a dozen markets. The Sutter Hill lead is the institutional signal: the firm that seeded Snowflake before the enterprise data warehouse was a recognized asset class has categorized power acceleration as equivalent infrastructure.

Unframe raised $50 million in a Series B led by Highland Europe, reaching $100 million in total contract value within twelve months of formal launch with 400 percent net revenue retention, by solving the pilot-to-production gap that stalls most enterprise AI projects. Its platform converts business requirements into production-ready AI applications in days and operates model-agnostically across cloud, on-premises, and managed service deployments. Viktor raised $75 million in a Series A led by Accel, with Slack co-founders Stewart Butterfield and Cal Henderson as angels, to deploy AI agents inside Slack and Microsoft Teams that automate operational work across 3,000 connected enterprise applications. The Butterfield and Henderson participation makes the thesis explicit: agents succeed when they live inside existing interfaces rather than demanding that users change their behavior.

Why Did Physical Intelligence Attract This Much Capital This Week?

Radar reached a $1 billion unicorn valuation on a $170 million Series B by achieving 99 percent retail inventory accuracy through ceiling-mounted RFID sensors, reducing buy-online-pick-up-in-store cancellation rates from 25 percent to 3 percent and shrink by 60 percent at pilot locations. Senkatason raised $110 million led by Sequoia Capital and Paradigm Capital at a $1 billion valuation on $200 million in annual revenue to deliver precision-manufactured hardware components for AI data centers, robotics, and aerospace on demand. Both companies illustrate the hardware-data moat thesis: physical sensors and manufacturing systems generate proprietary operational datasets that pure software competitors cannot replicate without equivalent physical investment.

The physical intelligence category also includes Aboard, which raised $13 million in a Pre-Series A to manufacture EREV-integrated travel trailers with independent energy management and mobile connectivity, and Shatterdome Energy, which raised $3.5 million in pre-seed funding from Crucible Capital to deploy AI-native power trading for data center operators managing grid volatility.

How Did Software Supply Chain Security Produce a Unicorn?

Socket reached a $1 billion valuation on a $60 million Series C led by Thrive Capital, with Andreessen Horowitz and Capital One Ventures, by building behavioral analysis for open-source software dependencies that detects malicious code by examining how packages actually function rather than checking static vulnerability databases. The unicorn valuation is a direct consequence of AI coding tools: as Cursor, GitHub Copilot, and comparable platforms accelerate the rate at which developers incorporate unreviewed open-source packages into production code, the attack surface expands at software development velocity. Socket’s behavioral approach scales with that velocity in a way that human code review cannot. The $1 billion market signal is that the enterprise security community has concluded this is infrastructure, not an option.

What Does This Week’s Deal Flow Predict About 2027?

The seed and pre-seed rounds from this week are the most forward-looking data in the report. Shatterdome Energy’s $3.5 million raise for AI-native power trading, Rely’s $4.5 million raise for AI real estate diligence, Hardline’s $2 million raise for voice-first construction AI, and Cloneable’s $4.6 million raise for expert-knowledge capture in heavy industry are each addressing a vertical where the structural bottleneck is the same: complex, high-stakes workflows that have historically required specialized human expertise and that AI can now automate with traceable, verifiable outputs.

The institutional pattern that precedes large category formation is visible in each of these raises. A credible lead investor, a founding team with direct domain expertise, and a live operational deployment before the raise are present in each case. Rely had George Matelich and David LoBosco, former PropTech operators, and a working product before closing with 2048 Ventures. Shatterdome had integrated power assets into a virtual power plant network before closing with Crucible Capital. These are not PowerPoint companies.

Three questions remain unresolved. First, whether the concentration of capital at the frontier model layer in Q1 2026 begins to reduce availability for infrastructure plays as limited partners reach AI allocation limits. Second, whether the regulatory response to agentic AI operating in financial, healthcare, and legal contexts will create compliance requirements that favor incumbent operators over startup entrants before the startups reach scale. Third, whether physical intelligence companies, which carry hardware cost structures, can achieve the margin profiles that support the valuations the venture market is assigning them.

The next piece in this series examines how Moment’s deployment of AI agents inside institutional investment operations creates a direct intersection with SEC Rule 17a-4, the Investment Advisers Act of 1940, and the emerging framework for fiduciary responsibility in algorithmic financial management. The central question is whether a human portfolio manager who delegates execution to an AI agent retains personal liability for the agent’s decisions, and what the current regulatory text actually says about that delegation.

Bibliography

TechStartups.com. Venture Capital and Startup Funding Roundup, May 19, 2026. https://techstartups.com/2026/05/19/venture-capital-startup-funding-roundup-may-19-2026/

TechStartups.com. Venture Capital and Startup Funding Roundup, May 20, 2026. https://techstartups.com/2026/05/20/venture-capital-startup-funding-roundup-may-20-2026/

TechStartups.com. Venture Capital and Startup Funding Roundup, May 21, 2026. https://techstartups.com/2026/05/21/venture-capital-startup-funding-roundup-may-21-2026/

Crunchbase News. The Week’s 10 Biggest Funding Rounds: Defense Tech Leads With Multiple Large Deals. https://news.crunchbase.com/venture/biggest-funding-rounds-defense-aerospace-ai-fintech/

GlobeNewswire. Rely Raises $4.5M Seed Round. May 19, 2026. https://www.globenewswire.com/news-release/2026/05/19/3297610/

Investment Advisers Act of 1940. 15 U.S.C. ss 80b-1 et seq.

SEC Rule 17a-4. 17 C.F.R. ss 240.17a-4.

The Energy Dominance Financing Program, enacted under the One Big Beautiful Bill Act of July 2025, authorizes DOE to guarantee $250 billion in energy loans through September 2028. It is a direct rebranding of the Energy Infrastructure Reinvestment Program established by the Inflation Reduction Act of 2022. The underlying legal structure is Section 1706 of the Energy Policy Act of 2005. The policy objectives are nearly opposite.

The original program required projects to avoid, reduce, use, or sequester greenhouse gas emissions. The new program replaces that standard with grid reliability, energy security, and domestic resource development. An Interim Final Rule published October 28, 2025, amending 10 C.F.R. Part 609, operationalized the expanded criteria: coal and gas repowering, pipeline replacement, refinery retrofits, nuclear plant expansions, transmission reconductoring, and critical minerals supply chain projects are now eligible.

The shift carries a fiscal asymmetry. The One Big Beautiful Bill repealed more than $8 billion in IRA credit subsidy appropriations across DOE loan programs. The replacement appropriation is $1 billion, estimated to support approximately $200 billion in additional project financing. Whether $1 billion is adequate to underwrite $200 billion in guarantees for assets with documented commodity price risk is a question whose answer will arrive only after the first default.

DOE has already closed three loans totaling $4.1 billion under the program: a $1 billion loan to Constellation Energy to restart the Crane Clean Energy Center at Three Mile Island in Pennsylvania, adding 835 megawatts of baseload nuclear power by 2028; a loan to an AEP subsidiary for transmission reconductoring; and a loan to Wabash Valley Resources to convert a coal plant to fertilizer production. DOE now holds more than $289 billion in available loan authority across all programs, making it the largest energy lender in the world.

For project developers, the practical implication is immediate: coal and gas repowering projects categorically ineligible under the IRA framework can now access federal loan guarantees on the same statutory basis previously reserved for wind and solar.

The unresolved question is credit quality, not program scope. Federal loan guarantees transfer commodity risk, demand risk, and stranded asset risk to the federal balance sheet without eliminating them. Whether the new credit subsidy structure prices that transfer correctly will not be known until the program’s first significant default. The question worth watching: will the $289 billion in available authority produce enough closed loans before September 2028 to determine whether this program delivers on its premise.

Interested in analysis about the intersection of tech, policy and the law? Check out my Substack channel. https://theinnovationattorney.substack.com/

Bibliography

Department of Energy. “Energy Dominance Financing Program.” Office of Energy Dominance Financing. https://www.energy.gov/edf/energy-dominance-financing-program. Accessed May 22, 2026.

Federal Register. “Energy Dominance Financing Amendments.” 90 Fed. Reg. (October 28, 2025). https://www.federalregister.gov/documents/2025/10/28/2025-19675/energy-dominance-financing-amendments.

One Big Beautiful Bill Act, Pub. L. No. 119-____, Section 1706 amendment (2025).

Energy Policy Act of 2005, 42 U.S.C. Section 16516, as amended.

Inflation Reduction Act of 2022, Pub. L. No. 117-169, Section 1706.

Department of Energy. “Letter from Leadership: EDF 2025 Year-in-Review and Looking Forward to 2026.” https://www.energy.gov/edf/articles/letter-leadership-edf-2025-year-review-and-looking-forward-2026. Accessed May 22, 2026.

National Law Review. “What Is DOE’s Energy Dominance Financing Program and Why It Matters.” https://natlawreview.com/article/what-does-energy-dominance-financing-program-and-why-it-matters. Accessed May 22, 2026.

Perkins Coie. “DOE Implements Energy Dominance Financing Through Interim Final Rule.” https://perkinscoie.com/insights/blog/doe-implements-energy-dominance-financing-through-interim-final-rule. Accessed May 22, 2026.

Ankura. “Department of Energy Releases Interim Final Rule for Energy Dominance Financing Program.” https://ankura.com/insights/department-of-energy-releases-interim-final-rule-for-energy-dominance-financing-program. Accessed May 22, 2026.

TD World. “DOE Grants $1 Billion Loan to Restart Constellation’s Crane Clean Energy Center.” https://www.tdworld.com/distributed-energy-resources/news/55331985. Accessed May 22, 2026.

Holland and Knight. “DOE Issues Interim Final Rule Implementing Energy Dominance Financing.” https://www.hklaw.com/en/insights/publications/2025/10/doe-issues-interim-final-rule-implementing-energy-dominance-financing. Accessed May 22, 2026.

Third Way. “Post-OBBBA: What’s Next for the Loan Programs Office.” https://www.thirdway.org/memo/post-obbba-whats-next-for-the-loan-programs-office. Accessed May 22, 2026.

A Critical Analysis of the Deep Fission S-1 Filing and the Engineering Economics of Underground Nuclear Deployment

A. Executive Summary

Deep Fission, Inc., which filed its S-1 registration statement with the SEC on May 20, 2026, seeks up to $156 million to commercialize a pressurized water reactor installed one mile underground, a design that the Nuclear Regulatory Commission has not yet established a licensing framework for and that independent nuclear engineers have identified as technically complex in refueling, heat transfer, and remote monitoring. The company reported no revenue, an operating loss of $21.4 million, and a net loss of $61.0 million for the year ended December 31, 2025. Its non-binding letters of intent total 12.5 gigawatts, a figure that would require approximately 833 to 1,562 individual underground reactor installations at the company’s stated output range of 8 to 15 megawatts per unit. Deep Fission represents a genuinely novel idea: adapting oil and gas borehole drilling and geothermal heat transfer to host a standard pressurized water reactor at depth, using hydrostatic pressure from a mile of water as a natural analog to the engineered containment structures that current NRC regulations require. The contrarian case is not that the physics are wrong. The hydrostatic pressure at 160 atmospheres is real, and the geological shielding from external hazards is real. The contrarian case is that substituting natural geology for inspectable, redundant, engineered barriers inverts the foundational safety logic that NRC regulations have built over five decades, and that the operational and economic consequences of that inversion have not been answered by vendor projections or DOE pilot program timelines.

B. Detailed Findings

What Is Deep Fission and What Does It Propose?

Deep Fission, Inc. was incorporated in Delaware and is headquartered in Berkeley, California. It was founded in 2023 by Elizabeth Muller, who serves as President and Chief Executive Officer, and her father Richard A. Muller, Ph.D., Professor Emeritus at the University of California at Berkeley, who serves as Chief Technology Officer. Muller previously co-founded Deep Isolation, a company developing borehole disposal for nuclear waste. The conceptual origin of Deep Fission traces to a thought experiment during that earlier work: what would happen if fresh uranium fuel were placed underground instead of waste. The answer was no spontaneous criticality from a single fuel assembly, but the conditions for a controlled reaction at depth were present.

The company’s reactor, named Gravity and formally designated DFBR-1 in NRC pre-application documents, is a pressurized water reactor designed to fit inside a 30-inch borehole drilled approximately one mile, or 1.6 kilometers, below grade. The reactor stands approximately 9 meters tall. Two configurations are planned. The initial 2x2 fuel assembly configuration targets up to approximately 8 megawatts electric. A later 3x3 configuration targets up to approximately 15 megawatts electric. Both use standard low-enriched uranium fuel, identical to the fuel used in the 64 pressurized water reactors currently licensed to operate in the United States. The design relies on three established technology domains: pressurized water reactor engineering from the nuclear industry, deep borehole drilling from oil and gas, and heat-to-surface transfer from geothermal energy development. The company claims this combination can reduce overall project costs by 70 to 80 percent compared to conventional nuclear plants and shorten construction to approximately six months from groundbreaking to operation, comprising four weeks of drilling, eight to ten weeks of reactor installation, and two months of commissioning.

What Does the S-1 Filing Reveal About Deep Fission’s Financial Position?

Deep Fission filed its S-1 with the SEC on May 20, 2026, applying to list on the Nasdaq Global Market under the ticker FISN. The offering comprises 6,000,000 shares at a price range of $24.00 to $26.00 per share, with a 900,000-share over-allotment option, for gross proceeds of up to approximately $156 million at the high end of the range. Lead underwriters are William Blair, Stifel, Canaccord Genuity, Benchmark (a StoneX Company), and Seaport Global Securities. Proceeds are designated for engineering, research and development, licensing, and construction of the first pilot reactor.

The company has generated no revenue to date. For the year ended December 31, 2025, Deep Fission reported an operating loss of $21,384,169 and a net loss of $61,002,954. The gap between operating loss and net loss reflects non-cash charges associated with the reverse-merger transaction through which the company completed a restructuring and raised $30 million via private placement in 2025, following an initial $4 million seed round in 2024. The company leases approximately 100 acres at the Great Plains Industrial Park in Parsons, Kansas, where it commenced test drilling for site characterization in March 2026. Strategic partners include Blue Owl for potential offtake and project-level financing, and Endeavour Energy’s Edged data center division under a 2-gigawatt term sheet. Supply chain relationships include Halliburton for drilling, Urenco USA for uranium enrichment, Day and Zimmermann, R-V Industries, and Sparx Engineering for manufacturing and fabrication.

Does the Stated Customer Pipeline Support the Company’s Valuation?

Deep Fission reported in October 2025 that prospective customers had signed non-binding letters of intent representing 12.5 gigawatts of power demand, spanning data center developers, industrial parks, and unnamed strategic partners, with initial sites under consideration in Kansas, Texas, and Utah. Individual agreements range from under 1 gigawatt to 2 gigawatts. Non-binding letters of intent carry no contractual obligation to purchase power or commit capital, and the company has disclosed no binding power purchase agreements. The 12.5 gigawatt figure, taken against the company’s stated output of 8 to 15 megawatts per reactor, implies a requirement of between 833 and 1,562 separate reactor installations to fulfill that pipeline. Each installation requires its own borehole, its own geological characterization, its own NRC authorization, and its own site-specific commissioning. No precedent exists for managing nuclear licensing at that volume. NuScale Power, the first SMR developer to receive NRC design certification under 10 CFR Part 52 in January 2023, pursued a single design across approximately six planned sites and ultimately saw its lead customer, Utah Associated Municipal Power Systems, cancel the project in November 2023 after cost estimates more than doubled. Deep Fission’s pipeline disclosure presents the aggregate letters of intent figure without disclosing the number of counterparties, the terms, or the conditions under which they may be withdrawn.

Why Does Burying a Nuclear Reactor Create Problems That the Surface Equivalent Does Not?

The affirmative case for the underground placement rests on two premises. First, the hydrostatic pressure of approximately 160 atmospheres from a mile-long water column naturally maintains the pressurized conditions required for PWR operation, eliminating the need for above-ground pressure vessels of the scale used in conventional plants. Second, the geological mass above the reactor provides shielding equivalent to the concrete domes and steel containment structures of a conventional plant, while also protecting against aircraft impact, vehicle attack, flooding, and wind events. Both premises rest on established physics and neither is in dispute.

The contrarian case begins with what happens after the reactor starts operating. A pressurized water reactor generates heat through controlled nuclear fission in its core. That heat must travel from the reactor at depth to a turbine at the surface to produce electricity. In a conventional plant, the primary coolant loop and the turbine are separated by a matter of meters. In the Deep Fission design, the primary loop and turbine are separated by a mile of borehole casing. Steam or heated water traveling up a cold borehole loses heat to the surrounding rock formation. The thermal loss over a mile-long conduit is not a rounding error: it is a fundamental thermodynamic penalty that reduces the fraction of reactor thermal output that arrives at the turbine. Deep Fission’s S-1 projects a target levelized cost of electricity of $50 to $70 per megawatt-hour, but no peer-reviewed independent thermal-hydraulic model of the full borehole system has been published to support that projection. The Idaho National Laboratory conducted independent benchmark analysis of Deep Fission’s thermal-hydraulic system behavior under Gateway for Accelerated Innovation in Nuclear program NE-25-36439, but the results have not been publicly disclosed in peer-reviewed form.

What Are the Maintenance and Refueling Realities of a Mile-Deep Reactor?

Every commercial nuclear reactor must be refueled periodically. Conventional PWRs operate for 18 to 24 months between refueling outages. During outage, operators remove spent fuel assemblies from the reactor core, replace them with fresh assemblies, and handle all operations inside a heavily shielded reactor building with established tooling and procedures. The Deep Fission 2x2 configuration targets a refueling interval of approximately six years. The 3x3 configuration targets seven to ten years. When refueling becomes necessary, the irradiated reactor canister must be extracted from the borehole and brought to the surface. A fuel assembly that has operated for six years in a reactor core is intensely radioactive, producing gamma and neutron radiation at levels that require substantial shielding for any personnel in proximity. Extracting a hot, irradiated reactor from a mile of borehole casing, providing adequate shielding for workers during extraction, and then transporting spent fuel through the surface environment to a storage or disposal location is an operations challenge with no existing solution in the nuclear industry. Mary Lou Dunzik-Gougar, nuclear engineering professor and associate dean at Idaho State University, confirmed that the underground placement complicates maintenance and refueling by requiring the reactor to be brought to the surface with alternative shielding for workers. Deep Fission’s MOU with Deep Isolation to explore borehole disposal of spent fuel addresses the endpoint of the fuel cycle, but NRC has not licensed any borehole disposal pathway for commercial spent nuclear fuel in the United States.

How Does the NRC Regulatory Framework Apply to an Underground Reactor?

Nuclear power plant licensing in the United States proceeds under 10 CFR Parts 50 and 52, developed over decades around the assumption that reactors operate in accessible facilities with physical containment structures subject to inspection, testing, and direct maintenance. NRC’s containment design requirements under 10 CFR Part 50, Appendix A, General Design Criteria, specify that reactor containment must be designed to maintain its functional integrity throughout the design life of the plant. No NRC regulation currently defines natural geology as a qualifying substitute for engineered containment. Deep Fission’s own pre-application regulatory engagement plan, docketed at NRC ADAMS accession number ML24135A163, acknowledges that compliance with NRC requirements concerning monitoring and visual inspection will be challenging due to deep borehole emplacement, and that additional guidance from NRC on remotely operated nuclear reactor designs will be required as an early matter of discussion. NRC acknowledged this challenge in a 2025 pre-application interaction noted at ML25017A406. The agency has not yet established that natural hydrostatic pressure and geological mass provide reasonable assurance of containment integrity equivalent to the inspectable steel and concrete structures required under current regulation.

Beyond containment, NRC regulations require physical access for periodic inspection of reactor components, ability to manually operate certain safety systems, and proximity of qualified operators to the reactor. Remote operation of a nuclear reactor from the surface through a mile of cabling or wireless communication introduces failure points and latency considerations that current NRC regulations on operator response time and manual actuation were not written to accommodate. Deep Fission is participating in the DOE Reactor Pilot Program, which operates under a separate DOE authorization framework distinct from NRC licensing. Achieving criticality under DOE authorization does not satisfy NRC licensing requirements for commercial operation. The company’s S-1 discloses that it intends to pursue DOE authorization for its pilot reactor while preparing for future NRC licensing, a sequencing that acknowledges the NRC pathway remains open.

Is the Power Output Matched to the Stated Market?

The primary commercial target disclosed in the S-1 is hyperscale data centers and other large power users. A hyperscale data center facility in 2025 and 2026 typically requires between 100 megawatts and 500 megawatts or more of continuous power. The Gravity reactor at 3x3 configuration produces approximately 15 megawatts electric. To serve a 100-megawatt data center with Deep Fission reactors would require approximately seven reactors at minimum, each in its own borehole, each requiring its own geological characterization, NRC authorization, and commissioning. The 2-gigawatt term sheet with Endeavour Energy’s Edged division would, at the 3x3 configuration, require approximately 133 reactor installations. Deep Fission’s response is that multiple reactors can be clustered at a single site: a block of ten reaches 150 megawatts, and larger groupings could reach 1.5 gigawatts at a single site. That claim has not been validated by any independent engineering study, and no regulatory framework exists for operating 100 interconnected underground nuclear reactors on a single site as a unified power plant.

What Happened to the July 4, 2026 Criticality Target?

In December 2025, Deep Fission announced that, pending DOE authorization, the company aimed to complete construction of its first reactor and achieve criticality by July 4, 2026. In March 2026, KCUR reported that the company acknowledged the Kansas underground reactor might not happen on that timeline. The drilling activity at Parsons, Kansas, as of March 2026, was characterized as data acquisition drilling to gather geological, hydrological, and thermal data, a preliminary step that precedes, not follows, commitment to a reactor installation location. The shift from a criticality announcement to a site characterization drilling program in the span of three months reflects a realistic adjustment to the engineering requirements of borehole siting, but it also means the company is seeking up to $156 million through its IPO without a confirmed pilot site, without a finalized reactor design approved by NRC or DOE, and without any revenue or commercial operations.

What Do the Cost Projections Actually Show?

Deep Fission’s S-1 discloses estimated capital cost of approximately $152 million for a first-of-a-kind 2x2 reactor producing approximately 8 megawatts electric. At that figure, the capital cost per kilowatt electric is approximately $19,000, which compares unfavorably to conventional large nuclear plants and to the $9,000 to $13,000 per kilowatt range that NuScale projected for its 77-megawatt module before costs escalated further. The nth-of-a-kind 3x3 configuration projects capital cost of approximately $84 million for 15 megawatts electric, or approximately $5,600 per kilowatt electric. That projection assumes successful serial manufacturing, standardized geological conditions across sites, and regulatory approval pathways that do not yet exist. No commercial nuclear SMR in the United States has yet progressed from first-of-a-kind to nth-of-a-kind cost curves. The 70 to 80 percent cost reduction claim relative to conventional nuclear plants is referenced against greenfield large plant construction costs, which are an unfavorable comparison point given that both conventional plants and Deep Fission’s reactor have no operating commercial track record.

C. Legal and Regulatory Implications

NRC Licensing Pathway and the Absence of a Framework for Geology as Containment

Commercial nuclear power plants in the United States operate under licenses issued pursuant to the Atomic Energy Act of 1954, as amended, and NRC regulations at Title 10 of the Code of Federal Regulations. Two principal pathways apply to new reactor licensing: 10 CFR Part 50, which involves a construction permit followed by an operating license, and 10 CFR Part 52, which allows combined construction and operating licenses. Neither pathway includes provisions for reactors emplaced in deep boreholes, nor do the General Design Criteria in Appendix A to 10 CFR Part 50 define standards for geological containment as a substitute for engineered containment structures. Deep Fission’s pre-application engagement with NRC, documented in ADAMS at ML24135A163 and ML25017A406, confirms that supplemental regulatory guidance will be required. The timeline for developing new NRC guidance, running from formal initiation through public comment to final rulemaking, typically requires several years. Deep Fission’s projected commercial construction timeline of 2027 to 2028 for early sites assumes regulatory clarity that does not presently exist.

Price-Anderson Nuclear Industries Indemnity Act and Liability Allocation

The Price-Anderson Nuclear Industries Indemnity Act, codified at 42 U.S.C. Section 2210, establishes the liability framework for nuclear incidents at licensed commercial facilities. Price-Anderson requires operators to carry the maximum amount of commercially available private insurance and provides a pooled indemnity layer funded by assessments on all licensed commercial reactor operators. As of 2025, the total coverage available under Price-Anderson exceeds $13 billion per incident. A Deep Fission reactor operating under DOE pilot authorization rather than NRC commercial license would operate outside the standard Price-Anderson indemnity structure applicable to commercial licensees. The liability classification for a novel underground reactor operating under a DOE pilot program raises questions that the S-1 does not fully address and that prospective investors and counterparties should assess independently.

State Law Constraints: Kansas Direct-Sales Prohibition

Deep Fission’s primary commercial model targets direct power sales to hyperscale data centers and industrial users. Kansas law, as reported in March 2026 coverage of the Parsons project, includes provisions that limit a private company’s ability to sell electricity directly to end-use customers, requiring instead that power flow through regulated utilities. This constraint materially affects the company’s stated business model at its primary pilot site and would require either a regulatory change at the state level or a restructuring of the commercial arrangement through a utility intermediary. The S-1 does not fully resolve this conflict, and the regulatory environment in Texas and Utah, the other named candidate states, would require independent analysis before conclusions can be drawn about the feasibility of direct-sale commercial operations.

Securities Disclosure Considerations

A company filing an S-1 under the Securities Act of 1933 is required to provide full and fair disclosure of all material risks to prospective investors. The risk factors section of the Deep Fission S-1 addresses the pre-revenue status, the absence of NRC licensing precedent, and the dependence on DOE pilot authorization. Investors evaluating the offering should note that non-binding letters of intent for 12.5 gigawatts are presented alongside a financial profile that includes a $61 million net loss for 2025, no revenue, and a company founded three years before the filing. The gap between the pipeline disclosure and the financial reality is not unusual for early-stage technology companies, but in the nuclear sector, where development cycles span decades and regulatory costs are substantial, the distance between letter-of-intent announcements and commercial operations is historically longer and more expensive than projections made at the S-1 stage.

D. Open Questions

Several consequential issues remain without definitive answers as of the date of this report.

First: the NRC has not established a regulatory framework for natural geology as a substitute for engineered containment. What timeline does NRC anticipate for developing supplemental guidance on remotely operated underground reactors, and will that guidance be completed before Deep Fission requires a commercial operating license?

Second: the thermal-hydraulic performance of a mile-long steam conduit has not been independently validated in peer-reviewed literature. What is the actual thermal efficiency of the Gravity reactor system from core output to turbine input, and how does that efficiency change across different geological formations with varying thermal conductivity?

Third: the refueling and spent fuel management pathway for a borehole-emplaced reactor has no regulatory precedent. How does NRC intend to classify and license the extraction of an irradiated reactor canister from a mile-deep borehole, and what shielding and transportation requirements will apply?

Fourth: Kansas law as currently written appears to prohibit the direct commercial power sales model that Deep Fission’s primary business case depends on. Does the company anticipate a legislative change, a utility intermediary structure, or a different primary deployment state?

Fifth: the company’s target of first criticality by July 4, 2026 was effectively suspended by March 2026. What is the revised timeline for the pilot reactor, and what specific milestones does DOE require before authorizing the Kansas facility to proceed from data acquisition drilling to reactor installation?

E. Source List

1. Deep Fission, Inc., Form S-1, SEC EDGAR Accession No. 0001104659-26-064074 (May 20, 2026) Primary regulatory filing; source of financial data, corporate structure, offering terms, and business description.

2. NRC ADAMS ML24135A163: Deep Fission Regulatory Engagement Plan (2024) Official NRC pre-application document acknowledging regulatory gaps for borehole-emplaced reactors.

3. NRC ADAMS ML25017A406: Deep Fission Pre-Application Interaction Summary (2025) NRC documentation of supplemental guidance requirements for remote operation and monitoring.

4. NRC ADAMS ML24172A286: Deep Fission Reactor Design White Paper (2024) Technical description of Gravity reactor dimensions and design specifications.

5. Idaho National Laboratory, GAIN Program NE-25-36439: Independent Benchmark of Deep Fission Thermal-Hydraulic System Behavior DOE-funded independent analysis of borehole reactor thermal performance; results not yet publicly disclosed in peer-reviewed form.

6. Shannon Cuthrell, ‘Deep Fission Plans to Sink Nuclear Reactors Deep Underground,’ IEEE Spectrum (November 20, 2025) Primary-source interviews with Deep Fission CEO and independent nuclear engineering faculty; technical and regulatory analysis.

7. KCUR, ‘Nuclear company now says Kansas reactor buried 1 mile underground might not happen’ (March 25, 2026) Reports timeline shift from criticality announcement to site characterization; identifies Kansas direct-sales law constraint.

8. Renaissance Capital IPO Research: Deep Fission S-1 Summary (May 2026) Summary of IPO structure, use of proceeds, and key financial metrics from the S-1 filing.

9. NRC Reactor Pilot Program: Deep Fission Pre-Application Activities, nrc.gov (2025) Official NRC tracking of Deep Fission’s pre-application engagement status.

10. Atomic Energy Act of 1954, as amended, 42 U.S.C. Section 2011 et seq.; Price-Anderson Nuclear Industries Indemnity Act, 42 U.S.C. Section 2210 Governing statutes for nuclear licensing and liability in the United States.

F. Bibliography

Atomic Energy Act of 1954, Pub. L. No. 83-703, 68 Stat. 919 (codified as amended at 42 U.S.C. Sections 2011-2297h-13).

Deep Fission, Inc. Form S-1 Registration Statement Under the Securities Act of 1933. SEC Accession No. 0001104659-26-064074. Filed May 20, 2026. https://www.sec.gov/Archives/edgar/data/1918102/000110465926064074/0001104659-26-064074-index.htm

Deep Fission, Inc. Regulatory Engagement Plan. NRC ADAMS Accession No. ML24135A163. 2024.

Deep Fission, Inc. Pre-Application Interaction Summary. NRC ADAMS Accession No. ML25017A406. 2025.

Deep Fission, Inc. Reactor Design White Paper. NRC ADAMS Accession No. ML24172A286. 2024.

Deep Fission, Inc. ‘Deep Fission Unveils the Name of Its Deep Borehole Reactor: Gravity.’ Business Wire. November 18, 2025. https://www.businesswire.com/news/home/20251118989916/en/Deep-Fission-Unveils-the-Name-of-Its-Deep-Borehole-Reactor-Gravity

Deep Fission, Inc. ‘Deep Fission Expands Customer Pipeline to 12.5 Gigawatts.’ October 2025. https://www.deepfission.com

Deep Fission, Inc. ‘Deep Fission and Deep Isolation Sign MOU on Managing Spent Fuel.’ April 2025. https://deepfission.com

Cuthrell, Shannon. ‘Deep Fission Plans to Sink Nuclear Reactors Deep Underground.’ IEEE Spectrum. November 20, 2025. https://spectrum.ieee.org/underground-nuclear-reactor-deep-fission

KCUR. ‘Nuclear company now says Kansas reactor buried 1 mile underground might not happen.’ March 25, 2026. https://www.kcur.org/environment-agriculture/2026-03-25/nuclear-company-now-says-kansas-reactor-buried-1-mile-underground-might-not-happen

Nuclear Regulatory Commission. ‘Deep Fission Pre-Application Activities.’ nrc.gov. 2025. https://www.nrc.gov/reactors/new-reactors/advanced/who-were-working-with/pre-application-activities/deep-fission

Price-Anderson Nuclear Industries Indemnity Act, 42 U.S.C. Section 2210 (2018).

Renaissance Capital. ‘Small modular nuclear reactor developer Deep Fission files for a $150 million IPO.’ May 2026. https://www.renaissancecapital.com/IPO-Center/News/119212/Small-modular-nuclear-reactor-developer-Deep-Fission-files-for-a-$150-milli

10 CFR Part 50, Appendix A, General Design Criteria for Nuclear Power Plants. U.S. Nuclear Regulatory Commission.

10 CFR Part 52, Licenses, Certifications, and Approvals for Nuclear Power Plants. U.S. Nuclear Regulatory Commission.

Idaho National Laboratory. Independent Verification and Benchmark of Deep Fission’s Deep Borehole Microreactor Thermal Hydraulic System Behavior. GAIN Program NE-25-36439. 2025. https://gain.inl.gov/ne-25-36439/

World Nuclear News. ‘Deep Fission Unveils Underground Reactor Concept.’ 2024. https://www.world-nuclear-news.org/articles/deep-fission-unveils-underground-reactor-concept

Interested in analysis about the intersection of tech, policy and the law? Check out my Substack channel. https://theinnovationattorney.substack.com/

Executive Summary

Quick Answer
In late March 2026, the U.S. Nuclear Regulatory Commission (NRC) finalized Part 53 and proposed Part 57, establishing a transformative, repeatable licensing pathway for advanced nuclear reactors and microreactors. Mandated by the ADVANCE Act of 2024, this update shifts the NRC’s mission statement to explicitly include enabling commercial industry. However, while regulatory hurdles have changed fundamentally, a critical market gap remains: regulatory approval does not equal financed construction. Project developers must still navigate substantial capital, offtake, and supply chain barriers before achieving commercial scale.

The Historical Gridlock: Why 2026 is Different

For thirty-five years, developers of non-light-water advanced reactors encountered a systemic, structural brick wall. Every innovative nuclear design had to be evaluated under regulatory frameworks originally engineered for legacy, light-water technologies that the new reactors did not use. Worse, these rules were administered by an agency whose historic mission statement omitted any mandate to support or enable commercial development. This mismatch made repeatable, standardized deployment economically irrational.

Two monumental regulatory shifts in early 2026 completely dismantled this paradigm:

· NRC Part 53 & Part 57 Framework: The NRC finalized Part 53, introducing the first dedicated advanced reactor licensing pathway since 1989. Simultaneously, the agency proposed Part 57, a dedicated framework tailored specifically to scale microreactors efficiently.

· The ADVANCE Act Mission Shift: Codified via the ADVANCE Act of 2024, the NRC’s governing statement now explicitly includes enabling industry progress alongside safety, forcing a historic cultural and operational pivot.

These are not minor, incremental compliance updates. Licensing reform and accelerating commercial deployment are moving in tandem for the first time in over forty years.

Commercial Milestones Accelerating in 2026

The tangible impact of this regulatory modernization is already evident across the energy sector:

· Near-Term Criticality: The Department of Energy’s (DOE) Office of Nuclear Energy project and expect multiple advanced reactors to achieve criticality by July 4, 2026.

· Global SMR Deployment: International commercial activity is surging. In May 2026, GE Vernova and Hitachi finalized a memorandum of understanding (MOU) to actively identify commercial deployment opportunities for the BWRX-300 small modular reactor (SMR) across Southeast Asia.

The Hidden Market Gap: Regulatory Approval vs. Financed Construction

Despite historic structural breakthroughs, global financial markets have fundamentally failed to price the significant operational distance between achieving regulatory approval and initiating financed construction. Resolving the licensing pathway is a critical milestones, but it represents only one phase of commercial execution.

Strategic Insight for Investors:
For institutional investors, venture funds, and energy developers, reading the NRC’s Part 53 framework as a guarantee of commercial success is a mistake. It is a necessary but not sufficient condition for market scale.

To capture true market value, energy stakeholders must solve three unresolved commercial bottlenecks:

1. The Project Finance Question: Securing multi-billion-dollar project capital remains a high hurdle for unproven first-of-a-kind (FOAK) advanced reactors.

2. The Offtake Agreement Question: Commercial scalability requires long-term power purchase agreements (PPAs) or industrial offtake commitments to guarantee revenue streams.

3. The Fuel and Component Supply Chain: High-Assay Low-Enriched Uranium (HALEU) supply lines and standardized specialized component manufacturing are not yet fully operational at scale.

The Multi-Billion Dollar Question

The ultimate market winners will not be those who simply obtain a license under Part 53. The true leaders will be the developers, partners, and financial institutions that successfully close these remaining infrastructure gaps on an accelerated timeline.

References & Source Materials

Roll Call. (May 13, 2026). “Nuclear Regulatory Commission updates processes to meet new demands.” https://rollcall.com/2026/05/13/nuclear-regulatory-commission-updates-processes-to-meet-new-demands/

American Nuclear Society. (May 2026). “Industry Update: May 2026.” https://www.ans.org/news/article-7957/industry-updatemay-2026/

National Conference of State Legislatures. (May 2026). “News Reactor: May 2026.” https://www.ncsl.org/state-legislatures-news/details/news-reactor-may-2026