Supply-chain Reaction

Federal Capitalism is driving the Nuclear Renaissance

Nuclear’s usual pitch rests on four legs, solely related to its output: it’s firm, it lasts, increases national sovereignty, and has a small land and air footprint. Those are good reasons, and they’re all true, but there’s a fifth case that almost nobody makes and has less to do with its output, and it’s arguably the most important one for a country trying to revitalize its industrial base.

A spreadsheet of levelized costs will never capture the complete picture. Building something as complex as a nuclear reactor produces advanced industrial capacity and capabilities that spill over into many different products and industries.

Complexity & Linkage

There’s a school of economic thought, most associated with Ricardo Hausmann and César Hidalgo, that tries to explain why some countries get rich and others don’t, using something more specific than GDP or capital stock. They argue that the complexity of what a country produces is actually the best predictor of a country’s future growth. Not how much it exports, but how hard the stuff is to produce and how many other capabilities that production depends on.

Think of every product a country could make as sitting somewhere in a giant map, the “product space.” Countries that only make simple things (raw commodities, basic textiles) sit on the edges of the map, far from everything else, because those products don’t require much shared infrastructure with other products. Countries near the “core” of the map make things like precision machinery, advanced electronics, and specialty chemicals. Those products all draw on overlapping pools of skilled labor, supplier networks, and institutional know-how. Once a country is near the core, jumping to the next complex product is relatively easy, because most of the underlying capability already exists. If you’re stuck on the periphery, every jump is a leap into capabilities you don’t have.

Albert Hirschman made a related argument decades earlier. He said governments shouldn’t try to develop every sector evenly. They should deliberately invest in the industries with the densest web of backward and forward linkages, what he called “pole” industries, because building one of those triggers a cascade of induced investment in everything that has to exist to support it.

Nuclear power is a great example of a pole industry in energy. Demand for reactors are demand signals for forged pressure vessels, precision welding, specialty metallurgy that can withstand decades of high-stress, high-skilled labor, and many components that need to work without failing for over half a century. None of that capability sits idle once the plant is built. It’s the same capability base that shows up in submarines, in aerospace, and in advanced manufacturing generally. Build the reactor supply chain and you’ve built something with uses far outside the reactor.

Cost is an important variable but treating an economy as a pure cost optimization problem retards the potential expansion of capabilities and capacities. The economy is an evolving organism. The type of technology you choose to build affects the genetic capacity of your entire industrial base. Choose easy, cheap things and you get an economy that’s good at producing easy, cheap things. Choose to build nuclear energy and you get an economy that’s good at the hardest manufacturing problems there are, which happens to be useful for almost everything else with similar levels of complexity you’d want to build next.

The challenge is that this specific brand of industrial capability comes at a high price. If nuclear is indeed a “pole industry” destined to revitalize industrial capacities, the financing structures we build must be resilient enough to endure the brutal math of long timelines and expensive outlays.

The AP1000 Fleet

Standard project finance, which funds most large infrastructure buildouts, runs on three core assumptions. The current state of the US nuclear industry breaks all three.

First, it assumes construction risk can be handed off to a contractor with a balance sheet big enough to absorb overruns. Nuclear EPC contracts run into the range of billions to tens of billions. Contractors do not have balance sheets big enough to backstop overruns at that scale. Vogtle’s two reactors in Georgia were budgeted at $14 billion and came in above $30 billion. The contractors didn’t eat that difference. Ratepayers and shareholders did.

Second, standard project finance assumes cash flows are predictable enough to size debt against. Lenders set a minimum debt service coverage ratio, or DSCR, essentially a cushion requirement: the project’s cash flow has to exceed its debt payments by some margin, so there’s room to absorb a bad year without missing a payment. The size of that required cushion drives how much of the project can be debt versus equity. A $12 billion project where lenders demand a 1.5x cushion needs roughly 29% equity. Relax the requirement to 1.25x and equity drops to 14%, because the project can carry more debt for the same cash flow. That’s the normal tradeoff project finance runs on.

Nuclear breaks it differently. Run a realistic downside scenario, construction delay plus cost overrun plus lower output plus higher rates, all hitting at once, and the coverage ratio a nuclear project would actually produce falls below 1.0. Below 1.0 means the project’s cash flow wouldn’t even cover its debt payments, let alone provide a cushion. In principle you could cure that with enough equity, since less debt means less debt service to cover. In practice, the amount of equity required to keep nuclear’s downside case above water is larger that most private sponsor would be willing to raise for a single project. That’s the real constraint: not that the math is impossible, but that curing it privately would require equity checks no consortium of utilities or developers would be willing to write.

Third, it assumes every risk can be handed to whichever party is best positioned to manage it. Regulatory risk, political risk, and the loss of a domestic manufacturing base after thirty years without building anything, none of those have a private-sector owner that is best positioned to hold those risks. The only party who could plausibly hold those risks and move the needle is the Federal government. And they’re doing just that.

The recent $17.5 billion conditional commitment from the DoE via the Office of Energy Dominance Financing (EDF), finances ten AP1000 reactors through five separate special purpose vehicles (SPVs), two reactors each. Each SPV is 80% funded by a DOE loan, priced at the Treasury rate plus 0.375%, versus the 7 to 9% a commercial lender would charge if one would lend at all. The rest is funded by private sector equity, split evenly between Westinghouse and a utility, each posting $500 million as a letter of credit.

The disbursement mechanism: Westinghouse places purchase orders on behalf of the utility, they submit the invoices to DOE, and get reimbursed pro rata, roughly $80 from DOE and $10 each from Westinghouse and the utility for every $100 of equipment ordered. Money only moves against equipment that’s actually been ordered.

This solves a specific chicken-and-egg problem that has nothing to do with financing rates. Reactor components, containment vessels, steam generators, coolant pumps, take years to manufacture and come from a handful of forges worldwide. No utility board will commit hundreds of millions to components before the project has final approval. But the project can’t get final approval without firm component pricing, and firm pricing requires the order already being placed. Long-lead procurement financing breaks the loop by paying for components before the project has reached that final investment decision, using federal credit to absorb the risk that the components end up stranded.

The most interesting design choice in the whole structure is what EDF’s Julie Kozaraki calls “interchangeable collateral”. Components across all ten reactors are standardized enough that if one utility drops out, another can take its place and use the same equipment. That single fact is what makes an 80% loan-to-value ratio underwritable at all. This spreads risk across a ten-reactor program, so no single project’s failure sinks the collateral value.

The good news for those who get queasy about government lending, is that this program is manufacturing the conditions for its own obsolescence. Every completed SPV becomes a reference for the private markets to price risk. Lenders who couldn’t underwrite nuclear construction risk on their own will have a growing set of federally de-risked precedents to underwrite against. The standardized components lower the entry cost for whatever entity builds reactor number eleven, because the supply chain built for these ten doesn’t disappear afterward. And a DOE conditional commitment functions as a de facto investment-grade rating for a project Moody’s would otherwise call junk, which is exactly the kind of certification effect that gets private capital comfortable entering a space it previously wouldn’t touch.

None of this is proven yet. The $17.5 billion is a conditional commitment, not a closed loan. The power purchase agreements (PPAs) anchoring project revenue are expected to be taken on by hyperscalers. Whether the DOE can execute five, signed deals is still up in the air.

However, none of this works without fuel, and the fuel story has gotten a lot less attention than the reactor story sitting on top of it. In January, DOE put $2.7 billion behind expanding domestic uranium enrichment, including $900 million split across three companies to build out low-enriched (LEU) and high-assay low-enriched uranium (HALEU) supply, plus a smaller award aimed at next-generation enrichment technology. That followed a Fuel Line Pilot Program launched the summer before, five companies selected to build entire domestic fuel production lines from scratch, and a Defense Production Act consortium set up specifically to get fuel-cycle companies coordinating on bottlenecks none of them could solve alone. Enrichment sits about as far upstream as the nuclear supply chain goes, and it’s exactly the kind of atrophied domestic capability the complexity argument points to. The U.S. let this capability go idle for thirty years and ended up dependent on foreign enrichment as a result. Rebuilding it isn’t a side project to the reactor buildout. It’s the input every tier in this piece, conventional and advanced alike, ultimately depends on while rebuilding the human capital and industrial capabilities that improve other domains.

SMR Criticality

The fuel rebuild had a second dimension to it and it’s the one that actually made this summer’s SMR milestones, four SMR companies sustaining a controlled mission reaction (criticality), possible. Advanced reactors mostly run on HALEU, a higher-enrichment fuel the U.S. had almost no domestic supply of, and the same 2025 executive orders that set a July 4th criticality target also directed DOE to release at least 20 metric tons of HALEU from federal stockpiles into a fuel bank for qualifying private projects.

There’s also another financing story underneath this tier of reactors too, and it’s a different mechanism than the AP1000 fleet. EDF’s Title 17 program has a $100 million minimum transaction size, because the due diligence cost of evaluating a deal is too large to justify anything smaller. Most individual SMR or microreactor deployments fall well under that threshold on their own. A single small reactor project simply isn’t big enough to be worth the federal government’s underwriting effort.

The fix is a structure state green banks are already using for other clean energy projects, adapted here for nuclear. A State Energy Financing Institution (SEFI) sets up a bankruptcy-remote special purpose vehicle, isolated from its own balance sheet so a project failure can’t touch the state’s other assets. Individual developers commit their sub-$100-million projects into that shared vehicle’s portfolio. The state posts real equity, at least 20% of total project costs, not a paper guarantee. Once the pooled portfolio clears the $100 million bar, it can apply for EDF financing covering up to 80% of costs. EDF underwrites the whole portfolio with the same rigor as any other Title 17 deal.

It solves a real tension state green banks face. They’re custodians of public money, which makes them cautious, but they’re also mission-driven lenders who want to move projects forward, which makes them want leverage. Taking project debt directly onto their own balance sheet threatens their credit rating. Routing it through an isolated vehicle lets them get the leverage without the balance sheet exposure.

The May 2026 guidance goes further and explicitly allows multiple states to capitalize a single vehicle together, one state holding the primary equity position, others contributing alongside it. That’s the mechanism already proposed for a Mountain West geothermal consortium, and there’s no reason it couldn’t become the template for a multi-state advanced reactor consortium instead, spreading a fleet-scale bet across several state balance sheets instead of asking any one of them to carry it alone.

This domestic consortium and bundling logic is also being applied internationally for export markets. Earlier this week, the United States, Japan, and South Korea signed a memorandum of cooperation to coordinate SMR deployment in other countries, with an initial focus on the Indo-Pacific. The State Department is backing this with over $10 million in new funding for its FIRST program, aimed at technical support and a regional training hub for partner-country workforces. The reasoning behind the MOC:

“The MOC outlines opportunities for our three countries, which have complementary advantages in the civil nuclear field, to encourage mutually beneficial cooperation among their respective nuclear industries. This framework aims to foster fleet deployment models that de-risk project development, achieve economies of scale, catalyze private investment, streamline licensing processes, and optimize supply chains.”

Fusion

If nuclear fission is a pole industry, fusion is the most extreme version of the same bet, because fusion’s secondary supply chain doesn’t exist, yet.

The scientific momentum is real. The U.S. has drawn over $10 billion in cumulative private fusion investment, more than any other country, and is home to the only three fusion companies globally to raise over a billion dollars each. The National Ignition Facility has repeatedly demonstrated net energy gain through inertial confinement, most recently at an 8.6 megajoule yield. Commonwealth Fusion Systems’ SPARC tokamak in Massachusetts is roughly 75% built and expected to begin deuterium-tritium plasma campaigns soon.

None of that changes the fact that fusion needs an industrial base that has never been built. Building a fusion economy means creating specialized robotics capable of handling components after they’ve been activated by neutron bombardment, precision manufacturing tolerances tighter than anything the existing fission industry requires, and materials engineered specifically to survive sustained 14 MeV neutron flux without degrading. Fusion is primarily an advanced manufacturing capacity problem layered on top of a materials science problem, and right now almost none of the supply base for either one exists.

That’s exactly why the DOE’s own strategy shifted this year. The newly reorganized Office of Fusion explicitly moved away from trying to build complete public power plants itself and toward funding shared testbeds instead, projects like the Tritium-Blanket technology Development Platform and the Fusion Prototypical Neutron Source, aimed squarely at the pre-competitive materials and engineering gaps no single private company can solve on its own. It’s the identical logic behind the AP1000 program’s federal backstop, just applied several technology-readiness levels earlier, where the risk isn’t that a reactor might have a cost overrun, but that the entire supporting industry to build one doesn’t exist yet.

Of the three tiers in this piece, fusion is the one with the highest linkage density and the least existing capability to link into. It’s also the one where the payoff, if it lands, looks the most like the kind of pole industry Hirschman was describing: a technology whose secondary supply chain, once built, would be useful for problems well outside fusion itself.

The Linkage Cascade

The three stories running in parallel form a meta-story: industries with dense, technologically advanced linkages rarely get built by markets acting alone, because no single private actor can capture enough of the spillover value to justify funding and building the whole thing. Somebody has to make the first uneven bet, and somebody has to structure the capital stack so no single balance sheet has to hold all the risk of making it.

The AP1000 fleet is building a bootstrapped path from federal financing back toward standard private project finance. The SMR tier is solving a portfolio-aggregation problem so that projects too small to justify federal due diligence individually can still get funded and built, providing the momentum necessary to get to the economies of scale and repetition necessary to attract private capital at scale. Fusion is skipping straight to public investment in industrial capacity that doesn’t exist anywhere yet, because there’s no supply chain to aggregate in the first place.

Many of the capabilities necessary to bring these to fruition are not nuclear specific. They're the same capacities that show up in a variety of advanced production and scientific domains.

Federal Capitalism

Neither Washington nor state governments have ever been completely passive about their own industrial development. What’s changed across American history is the toolkit and sense of urgency. In the nineteenth and early twentieth centuries the government facilitated industrialization through state infrastructure finance, federal land grants to the railroads, tariff protection for infant industries, and entirely new legal structures for organizing capital, the modern corporate charter being one of them. The postwar era ran on a different set of tools: wartime mobilization, defense procurement that created whole technology sectors as a byproduct of buying weapons, and public research funding that built the knowledge base private industry then commercialized. Somewhere around the last half century, Washington slowly stepped back from a national industrial policy, while state governments built increasingly sophisticated, entrepreneurial development programs to fill some of the gap.

What’s emerging since roughly 2022 looks is the resurgence of the traditional model, not a break from it. Neo-Federal Capitalism: the deliberate coordination of private capital, state development programs, and federal fiscal facilities into layered capital stacks that share risk and pull investment toward strategic industrial outcomes. It’s a coordination and capital formation strategy, to unlock private capital that wouldn’t otherwise show up, and to get it moving faster and at a greater scale than any single actor could manage alone.

Federalism as an organizing principle has primarily meant two things: limits and multilateralism. Power, ownership, cost, benefit, and responsibility get divided up and spread across a large number of stakeholders, rather than concentrated on one balance sheet or one level of government. Nobody owns the whole risk, and nobody captures the entire upside.

Going beyond its generation attributes, the complexity and linkage arguments explain why nuclear is worth the bet. Federal Capitalism describes why the institutional machinery making the bet is very much rooted in the American tradition.