How Financialization Can Complement Reindustrialization

The United States has the most sophisticated private financial system in history, but it seemingly cannot finance critical and strategic assets. While American financial markets excel at pricing liquid, standardized assets with limited payback periods, they starve those that are illiquid, high-complexity, and long-duration—precisely the assets that increase American strength in fields like semiconductors, shipbuilding, and energy.

Nuclear power is an instructive case for understanding how financing and risk management, rather than the factors more commonly blamed, are the rate-limiting constraints on strategic industrial investment. The recent history of U.S. nuclear construction is routinely narrated as a story about excessive regulation, insufficient technological innovation, or the failure to adopt a state ownership model. The most recent builds in Georgia point to a different set of challenges.

The Nuclear Regulatory Commission (NRC) swiftly approved 195 license amendment requests for Vogtle’s third and fourth reactors and stood up a dedicated project office, showing that regulation was not the binding constraint.¹ The AP1000 was chosen over a proven GE-Hitachi design already built on time and on budget in Japan; if anything, the reach for next-generation innovation introduced underappreciated first-of-a-kind (FOAK) risks that a more mature financing system would have forced the industry to price honestly before breaking ground.²

The absence of robust risk management turned Vogtle into a financial debacle, leaving investment and financing appetite for the industry in an uncomfortable state of purgatory. Vogtle carried critical lessons in failure: no standardized practices for assessing whether a design, supply chain, and workforce were construction-ready; fixed-price contracts that concentrated risk in firms without the capitalization to bear it, producing a cascade of distress from Shaw to CB&I to Westinghouse; and no commitment mechanism across a multi-build sequence for sustaining construction continuity and scaling up the learning curve (and down the cost curve). Instead, every setback at Vogtle proved to be an existential crisis.³ The fear of financing new nuclear power is entirely understandable, and it stems from the absence of the appropriate financial and risk allocation infrastructure; the kind of financial infrastructure required for nuclear power may be understood in relation to what the United States has for housing finance and homebuilders.

Homebuilding has the characteristics of an asset class that the American system should struggle to finance: a heterogeneous, capital-intensive, physical product with uncertain payback horizons, subject to local building codes and zoning regulations. Yet the American homebuilding industry is a juggernaut: purely in terms of residential floorspace relative to population, the United States has delivered far more housing at scale than our peers.⁴

This success doesn’t fall neatly into traditional economic development tropes. There was no persistent deregulatory agenda at the heart of the postwar homebuilding boom. Nor was there a state-owned enterprise building homes by the millions. The real story is that public financing infrastructure enabled private sector investment. Iterative construction of that infrastructure—originating with the Federal Housing Administration and Reconstruction Finance Corporation and maturing through Fannie Mae and Freddie Mac—drove greater risk sharing, standardized underwriting and risk management practices, created markets for tradable securities, and made private capital flow where it otherwise would not. For all the criticism these institutions attracted during the 2008 financial crisis, the sector stands as a model of how private investment can be catalyzed at scale.

The key lesson is that the absence of well-developed underwriting standards, risk allocation mechanisms, and a robust fleet-scale plan left every Vogtle participant underprepared, undercoordinated, and vulnerable to cascading delays and overruns. When each new project must reinvent risk sharing from scratch rather than execute broadly understood practices, private capital will rationally stay away. These practices must be developed, codified, and disseminated until they become market convention.

To harness the urgency for strategic industrial investment into a self-sustaining force, public financing infrastructure must be built that enables deployment of private capital toward strategic objectives. The conventional debate frames financialization and industrialization as antagonists; in fact, the former can enable the latter. The public sector’s critical role is not as a substitute for markets but as a coordinator of market infrastructure and a backstop for financial risk that the private sector cannot bear alone. The ultimate purpose of this public role is to make the construction and delivery of strategic assets a well-understood, repeatable practice, one that private financiers can evaluate with confidence, and with a declining scale of public support over time.

The Paradox of American Financial Sophistication

The American financial system is extraordinarily good at what it does. It prices risk on standardized, liquid assets with established secondary markets. It allocates capital to software, services, and consumer credit with remarkable efficiency. AI and life sciences together have accounted for roughly two-thirds to three-quarters of venture capital investment in recent years, driven increasingly by AI, which alone captured nearly half of all venture capital dollars in 2024 and nearly two-thirds in 2025.⁵ One must scroll far down the list of the largest venture exits to find a company that is not an internet business. Meanwhile, legacy industrial giants, from General Electric to Boeing, can be fairly accused of losing core technological capabilities over the past several decades in pursuit of shiny quarterly numbers for shareholders.⁶

What the American financial system does not do well is finance long-duration, capital-intensive physical assets at scale, not unless there is preexisting infrastructure to make those assets attractive to private capital. Traditional financial institutions struggle to underwrite capital-intensive manufacturing because most private investors, banks or otherwise, are increasingly accustomed to less capital-intensive sectors like software. The prevailing investor and management preference, especially outside of the AI boom, has been to transition toward asset-light business models that reduce capital intensity and maximize risk-adjusted returns. It is a rational response to financial incentives, but it nevertheless can pull talent and attention away from the physical assets the country needs.

The status quo still fails to catalyze the requisite volume of strategically valuable industrial investments. Oil and gas, a notoriously capital-intensive business, made real breakthroughs in the performance of production through the 2010s, only for those breakthroughs to be paired with a corresponding period of overinvestment and poor financial performance. Tesla investors are willing to own shares at an unusually high premium, backing Elon Musk’s thinking (and there are real capabilities and gigafactories to show for that premium), yet the lessons to extrapolate from such an idiosyncratic example are understandably limited. Firms needing large amounts of capital for physical production find themselves in a difficult middle ground: they require far more than what venture capital typically funds, but the complexity associated with cutting-edge investment comes with a steep cost of capital that is difficult to square with commercial viability.⁷

This frustration is bipartisan and increasingly urgent. On the left, senators call for a pro-worker trade agenda paired with proactive industrial policy that revitalizes communities and reestablishes the United States as a manufacturing leader.⁸ On the right, legislators push for local content requirements to boost domestic production of critical goods.⁹ The diagnoses proliferate: too much regulation choking energy development,¹⁰ not enough public ownership of power generation,¹¹ and finance as a parasitic force.¹²

Each of these explanations captures something real, but each also misses a key constraint. The problem is not that American finance is too sophisticated; it is that its sophistication has been deployed in sectors where standardization was achieved decades ago—housing, consumer credit, corporate debt—while the sectors that matter most for strategic competitiveness remain stuck in bespoke, illiquid arrangements.

The United States has built capital-intensive physical things before and recently. While perhaps not as capital-intensive as strategic industrial investments, homebuilding is nevertheless an expensive and complicated business. Homebuilding also suffers from highly decentralized supply chains, heterogeneous local regulations, and relatively long payback periods. Yet the U.S. homebuilding industry has also delivered the highest residential floorspace outcomes on a per capita basis: 1.4 million privately owned housing units built per year undersells the scale of U.S. residential construction.¹³ American homes are more than twice as large as homes of peer economies, and adjusted qualitatively for amenity and feature availability, even further ahead.¹⁴ Americans generally enjoy among the most spacious and affordable housing in the developed world, with housing expenditure consuming a comparatively low share of disposable income by OECD standards.¹⁵

Homebuilding is scalable in the United States not because the private market figured it out unaided, nor because the government builds houses, but because the public sector spent decades constructing the financial infrastructure that made private capital comfortable with the asset class. If the United States wants to build the next generation of strategically critical and competitive industrial assets, and to do so in a scalable manner that avoids one-off and ad hoc efforts, it first needs to build comparable institutional infrastructure.

The Problem with Nuclear Capital Formation

Whether nuclear power can be viable in the United States is particularly a question of risk management and financing conditions, rather than reactor design, construction technique, or the regulatory environment. Fixed investment costs far outweigh variable costs, and the full timeline of the most recently completed reactors in Georgia ran well over a decade. The cost of various sources of generation is highly sensitive to the cost of capital as a result. At 9 percent cost of capital, financing costs can represent two-thirds of the final cost of nuclear power, versus less than one-third at 3 percent.¹⁶

The difference between a bankable project and an unbankable one is not predominantly about engineering or the regulatory environment, but whether the financial infrastructure exists to assess, distribute, and price risk in a manner that constrains capital costs.¹⁷ Assets lasting multiple decades and generations, sixty to eighty years in the case of a nuclear reactor, mean that the full economic value of the investment accrues over a timeframe that exceeds the planning horizon of virtually every private financial institution. Commercial bank investment committees typically evaluate a five-year horizon, and even managers of large long-duration asset pools are rarely equipped to evaluate risks at the early design and construction phases of such projects.¹⁸ The financial risks of this profile are compounded by political and regulatory uncertainty, illiquidity, and the near total absence of secondary markets in which the associated debt can be traded.

The Vogtle Units 3 and 4 in Georgia, the only new nuclear construction completed in the United States in a generation, loom largest over any serious proposal for new nuclear builds. The dominant narratives for why nuclear fails to get built point to an overbearing NRC, insufficient technological innovation, or the absence of a state-owned enterprise model.¹⁹ Each narrative captures a grain of truth about the multidecade struggles of the American nuclear industry. None of them get to the root causes of Vogtle’s financial failure, which remains the most recent and relevant episode constraining the investment and financing of new nuclear generation in the United States.

Start with the NRC criticism: despite concerns about the regulatory burden, the Commission’s approach to Vogtle was accommodating. The NRC completed 222 licensing actions for Units 3 and 4, including 195 license amendment requests, an extraordinary level of responsiveness for a project built on an incomplete design.²⁰ It established a dedicated Vogtle Project Office.²¹ The NRC was uniquely accommodative throughout the process and was far from a binding constraint in Vogtle’s construction challenges.

Nor was the problem a lack of technological ambition: if anything, the emphasis on innovation was counterproductive. The AP1000 was selected over the Advanced Boiling Water Reactor, a GE-Hitachi design already built in Japan in under forty months, on budget, and licensed in the United States.²² The AP1000 was a generational leap: the first fully modularized nuclear island with 342 separate modules and passive safety systems representing genuine engineering advances.²³ But the supply chain for those modules had never been built. As one veteran nuclear engineer observed, “We keep on doing the same thing, we keep on building an unbuilt reactor thinking that these new designs will just fix everything that ever happened.”²⁴ The pattern of reaching for the next-generation design rather than building on proven experience is a symptom of an industry lacking the financial infrastructure to sustain iterative deployment.

What doomed Vogtle was underdeveloped risk management and a contracting architecture that concentrated risk instead of distributing it. Construction began with an incomplete reactor design. As late as 2014, Westinghouse acknowledged only 88 percent of construction blueprints were finalized, contradicting earlier claims of 95 percent readiness.²⁵ The supply chain was assembled for the first time and onsite module fabrication suffered persistent quality problems. Components had to be sourced from dozens of global suppliers. The workforce lacked recent nuclear construction experience after a decades-long hiatus.²⁶ The project’s fixed-price engineering, procurement, and construction (EPC) contracts compounded these problems, forcing contractors to bear enormous downside risk for a scope of work not fully defined, amounting to a structure engineered to produce disputes.

The result was a chain of corporate casualties. Shaw Group, an oil and gas services firm with no nuclear experience, acquired the nuclear EPC contractor Stone & Webster and took on construction responsibilities it was not equipped to manage.²⁷ Chicago Bridge & Iron (CB&I) acquired Shaw in 2012 and quickly recognized the scale of the inherited liabilities.²⁸ CB&I then sold Stone & Webster to Westinghouse in 2016, essentially paying Westinghouse to take the problem off its books.²⁹ This made Westinghouse simultaneously the reactor designer, intellectual property holder, and primary construction contractor for its own first-of-a-kind project. This arrangement lasted months before Westinghouse brought in Fluor as construction manager.

But by March 2017, Westinghouse had filed for Chapter 11 bankruptcy, attributing the decision primarily to the fixed-price contracts for the Vogtle and V.C. Summer projects.³⁰ Toshiba, the parent company of Westinghouse, subsequently recorded approximately $6.3 billion in write-downs and was later forced into its own restructuring.³¹ Southern Nuclear took over project management for Vogtle, replaced Fluor with Bechtel, and required an additional seven years and approximately $15 billion beyond the original estimate to complete the two units.³² The V.C. Summer project to build an AP1000 in South Carolina during this same time period faced an even worse fate, with the build being abandoned and the utility executives later facing criminal prosecution.³³

The failures associated with Vogtle and the AP1000 were foreseeable given the nature of first-of-a-kind projects. Projections were vulnerable to inaccurate anticipation, and the practices to price and distribute risk within adverse scenarios were naïve at best. But the fixed-price contract, intended to protect project owners against cost overruns, instead concentrated risk within entities such as Shaw, CB&I, and Westinghouse, none of which possessed sufficient capitalization or expertise to absorb these risks. In a mature risk allocation framework, these factors would have been identified and addressed prior to the disbursement of funds.

Three key deficiencies in financial infrastructure contributed to the inevitability of this outcome. First, the absence of standardized underwriting practices resulted in a lack of shared benchmarks for evaluating reactor design readiness, supply chain maturity, and contractor capitalization. As a result, participants assessed these risks inconsistently, leading to a wide range of evaluations that obscured the actual exposure. Second, there was no established mechanism to allocate construction-phase risk among parties capable of managing it. Consequently, risk was concentrated with the holder of the EPC contract at any given time, rather than being distributed according to capacity.

Third, there was no committed fleet program or institutional framework to ensure that a first-of-a-kind build would be capturing lessons and engaged in continuous construction to move down the cost curve. As a result, nuclear power generation was no more investable after Vogtle Units 3 and 4 were built than it was before; in many ways, it only further cemented its dubious reputation as a private investment opportunity. As the Chinese experience building the AP1000 has shown, subsequent builds can be completed with lower costs and timelines, critical developments for any potential future private investor base for financing new nuclear construction at scale. But progress in China also hinged on a commitment to fleet-scale construction across multiple series of builds, centralized lesson transfer, and retention of experienced developer entities. Without a credible public policy commitment to subsequent builds, learnings are less likely to be captured and made available to investors who might otherwise be interested in financing future builds. Each cost overrun and financial shock associated with Vogtle posed an existential threat to the entire industry, whereas a fleet-scale approach may have made these challenges more manageable and enabled future successes. The good news is that the United States does have experience addressing these challenges. Homebuilding may be viewed as a relatively simple industry nowadays, but when plagued with the challenges of physical and financial complexity in the 1930s, the federal government developed institutional solutions that have withstood the test of time. Those solutions, which hinged on the creation of government-sponsored enterprises (GSEs), carry broader lessons for the nuclear sector and beyond.

The GSE Origin Story

Before the New Deal, housing finance in the United States suffered from the structural deficiencies that cripple so much of industrial finance today. The three gaps that doomed Vogtle—the absence of standardized risk assessment, the inability to distribute risk, and the lack of scalable capital flow—had close analogues in the pre-Depression mortgage market. And each was solved by an institutional intervention that, taken together, transformed housing from an illiquid, locally financed, high-risk asset into the most financeable asset class in the United States.

Before the 1930s, mortgages were primarily short-term instruments, typically ten years or less, with required down payments of roughly half the home’s purchase price. Most were structured as balloon loans: borrowers made interest-only payments then owed the entire principal at maturity. Refinancing depended on whatever the lender chose to offer, if they chose to offer anything at all.³⁴ Mortgage rates carried spreads of roughly 200 basis points over high-grade corporate bonds (compared to 50 basis points since the mid-1980s), and regional disparities in borrowing costs could exceed 2 to 4 percentage points.³⁵ These elevated interest rates resulted from the bespoke and localized nature of each loan, combined with a highly opaque market structure.³⁶ The absence of standardized terminology prevented meaningful comparison between mortgages. Consequently, homebuilders faced significant uncertainty, as the availability of buyers depended entirely on the lending capacity of local banks, which disappeared rapidly during the Great Depression.

Established in 1934, the Federal Housing Administration (FHA) advanced mortgage market reform by insuring new mortgages originated by private lenders, rather than purchasing distressed loans. This insurance eliminated lender default risk in exchange for compliance with standardized underwriting criteria. The FHA did not provide direct loans; instead, it set standards for private lenders and absorbed credit risk, enabling longer loan terms, lower down payments, and fixed interest rates.³⁷ Long-term conventional fixed-rate mortgages with an embedded right to refinance, now a foundational element of the American housing market, were a policy innovation that enhanced affordability and prevented the refinancing crises that had previously harmed borrowers. For homebuilders, FHA insurance fundamentally altered risk calculations, ensuring that qualified buyers could reliably access financing on standardized terms. This reduced the uncertainty that had previously made homebuilding highly dependent on fluctuating local credit conditions.

The onset of the Great Depression precipitated the collapse of the pre-FHA mortgage industry. Thousands of banks failed, and those that survived declined to renew maturing mortgages. By 1933, housing starts had declined to approximately ninety-three thousand units, representing only one-tenth of previous peak levels of 937,000 units in 1925. Foreclosures occurred at a rate of about one thousand homes per day—primarily because banks refused to renew notes rather than due to missed payments by borrowers.³⁸ The construction sector suffered severe losses, with building trades idled, supply chains disrupted, and overall capacity diminished. The balloon payment mortgage structure concentrated risk within the borrower-lender relationship, which proved unsustainable during economic downturns.

The federal response commenced with the establishment of the Home Owners’ Loan Corporation (HOLC) in 1933, which acquired distressed mortgages from banks and refinanced them under new terms.³⁹ The HOLC introduced the fully amortizing, long-term, fixed-rate mortgage, a product intended primarily to eliminate the refinancing risk that had caused widespread financial distress, rather than directly promote homeownership. By restoring favorable credit conditions for mortgage origination, the HOLC provided relief to both borrowers and banks, thereby reviving a pipeline of housing demand that had nearly disappeared.

As economic historians Richard K. Green and Susan M. Wachter have documented, the combined efforts of the HOLC and FHA “represented a piece of early ‘financial engineering’ that allowed illiquid financial institutions to become liquid again.”⁴⁰ The more fundamental innovation, however, was the FHA’s insurance mechanism, which shifted default risk from individual lenders to a public balance sheet with sufficient capitalization and a countercyclical mandate. The Depression revealed a structural reality: institutions that retain mortgage risk on their own balance sheets tend to behave procyclically, expanding lending during booms and contracting precisely when countercyclical investment is most needed. The HOLC, FHA, and the institutional framework they established ensured that a public entity could absorb the tail risk that private institutions were unable to manage.

This lesson applies beyond the housing sector: any capital-intensive industry characterized by extended construction timelines and cyclical demand necessitates a financing structure in which an entity is explicitly designated to hold risk during economic downturns.

Despite the FHA’s standardization of underwriting and redistribution of default risk, mortgage lending continued to be limited by the availability of local bank deposits.⁴¹ Banks in capital-rich cities often possessed surplus funds with insufficient borrowers, while banks in rapidly growing suburbs faced high demand but lacked adequate capital. No mechanism existed to transfer capital efficiently from areas of abundance to those with greater need.

Fannie Mae, established in 1938 as a subsidiary of the Reconstruction Finance Corporation, addressed this issue by purchasing FHA-insured mortgages from originating banks, thereby replenishing their capital and enabling further lending. The primary objective was to strengthen the private mortgage market rather than supplant it. Importantly, Fannie Mae financed these purchases by issuing bonds at par, enabling a broader range of investors to confidently invest in mortgages originated in areas with limited local capital. This innovation created, for the first time, a mechanism for the geographic redistribution of mortgage funding.⁴² During the 1970s, Fannie Mae provided assistance primarily by buying mortgages while Freddie Mac, created in 1970, concentrated on securitizing them.

By the 1980s, the pooling of mortgages into tradable securities had expanded the potential holders dramatically. Investors who purchase mortgage-backed securities assume prepayment risk rather than credit risk.⁴³ The result was that housing, despite being a physically heterogeneous product built by fragmented local industries on idiosyncratic plots of land, became the most liquid and financeable real asset class in the United States. By doing the work of defining what a conforming mortgage looked like, the GSEs made individual loans fungible—interchangeable in secondary markets—and that attracted institutional capital at scale.⁴⁴ Because the GSE guarantee removed default risk from the securitized product, investors assumed the risk tied to general interest rate volatility, more so than outright default risk, and the resulting inflow of global capital pushed national mortgage rates down.⁴⁵ From the homebuilder’s perspective, the conforming framework meant something concrete: if they built to the market the GSEs were serving, they could be confident their buyers would have financing on predictable terms.

The system’s track record speaks for itself. The FHA’s mortgage insurance (in conjunction with other similar programs such as those at Veterans Affairs) made home loans so desirable that lenders actively sought projects to bankroll and developers found themselves with a virtually unlimited supply of capital, which enabled construction on a scale virtually unknown in U.S. history.⁴⁶ The homeownership rate rose from 43.6 percent in 1940 to over 60 percent by 1960, the fastest expansion in any nation’s history.⁴⁷ By the end of 2019, Fannie Mae and Freddie Mac were jointly responsible for approximately $6.68 trillion, or 61.6 percent, of U.S. residential mortgages outstanding.⁴⁸

This system financed the suburbanization of America, arguably the largest physical construction transformation in human history, not through central planning but through public financing infrastructure that channeled private capital. These institutions were powerful enough to produce adverse distributional consequences too: FHA guidelines severely limited the access of mortgages to African Americans, with only 2 percent of the $120 billion in new housing subsidized between 1934 and 1962 going to non-white borrowers, entrenching persistent lending patterns across generations.⁴⁹ But these shortcomings are ultimately separable from the question of whether the financing mechanism works. The mechanism works. The question is who benefits from how standards are set, and the design of those standards is a policy choice with distributional consequences that require prudence.

Crisis and Backlash

Despite its benefits, the GSE model has gained a bad reputation, stemming from the role of the GSEs in the Global Financial Crisis of 2008. The GSE model, whereby public institutions could standardize underwriting, guarantee certain types of risk, and create secondary markets for long-duration assets, had merit. But it failed when the benefits were privatized while the risks were silently socialized on the balance sheet of the federal government.

The privatization began in 1968, when the Johnson administration converted Fannie Mae from a government agency into a shareholder-owned corporation—primarily to move its debt off the federal balance sheet during fiscal pressure from Vietnam and the Great Society. Fannie’s biggest advantage in its new form was that investors continued to see it as virtually an arm of the government, meaning it could borrow at interest rates almost as low as those paid by the U.S. Treasury. Freddie Mac, created in 1970, was structured the same way.

For the next four decades, both entities operated in a twilight zone: nominally private corporations with an implicit government guarantee. This arrangement was the worst of both worlds. The implicit guarantee gave the GSEs access to cheap funding, which they used not only to support mortgage markets but to grow enormous retained portfolios of mortgage securities, essentially leveraged bets on interest rates funded by the implicit subsidy of government-backed borrowing. Private shareholders captured the upside. Market discipline over management was lacking because investors assumed that the government would never let them fail. They assumed, again correctly, that the social costs threatened by GSE failure would compel a government bailout.⁵⁰ This implicit guarantee created a textbook moral hazard: it was simultaneously the source of the GSEs’ market power and the mechanism by which risk could be mispriced. Implicit guarantees create liquidity, but they also create liquidity problems if investors become concerned about GSE insolvency, which is precisely what happened in the summer of 2008.⁵¹ The GSEs’ own prospectuses explicitly disclaimed any government guarantee; market participants simply did not believe the disclaimer.

When above normal mortgage defaults eroded the GSEs’ capital below statutory minimums and spikes in short-term borrowing rates produced losses on their retained portfolios, the government stepped in—exactly as the RFC model had always assumed it would need to, but without having priced for it or maintained the institutional capacity to manage it. In September 2008, the Federal Housing Finance Agency placed both Fannie Mae and Freddie Mac into conservatorship, assuming the powers of management, board, and shareholder. The U.S. Treasury provided financial support through Senior Preferred Stock Purchase Agreements, ultimately injecting approximately $190 billion.⁵² The public backstop that was never supposed to exist had materialized, at enormous cost, because it had never been honestly acknowledged, properly priced, or prudently governed.

The dominant postcrisis narrative, particularly from the political Right, drew a straightforward conclusion: the GSEs were inherently dangerous creatures of government meddling, and the solution was to eliminate them. Peter Wallison’s dissent from the Financial Crisis Inquiry Commission argued that the crisis was caused by government housing policies rather than failures of private sector risk management.⁵³ The Heritage Foundation went further, arguing that the GSE institutional model had failed and that policymakers should move toward a mortgage market without Fannie Mae and Freddie Mac entirely.⁵⁴

This diagnosis is understandable but flawed in its proposed solution. The core issue was the dishonest structure of the involvement. The implicit guarantee allowed private actors to capture the benefits of public backing while the public bore the risks. The solution is not to withdraw the public role but to make it explicit, properly priced, and transparently governed.

As Georgetown Law professor Adam Levitin explained in testimony before the Senate in 2011, a government guarantee is inevitable in the housing finance market. Five previous attempts at purely private mortgage securitization in the United States failed due to investors’ inability to manage credit risk in securitization, and no housing finance market in the developed world exists without either an explicit or implicit government guarantee of at least catastrophic risk. The fundamental choice is between an implicit, unpriced, and ungoverned guarantee and an explicit, well-structured, and appropriately priced guarantee.⁵⁵

The post-conservatorship record supports Levitin’s assessment. The GSEs under government control have arguably performed better than they ever did as quasi-private entities, shifting from portfolio-driven risk-taking to a guarantee-fee model that has generated tens of billions annually and returned more to the Treasury than the bailouts they received. Although conservatorship was intended as a temporary measure,⁵⁶ it is entering its seventeenth year. This confirms that the model is working (or at least that the alternatives appear perilous).⁵⁷ Housing finance continued to function through the pandemic and subsequent interest rate shocks precisely because the infrastructure held: the standardized GSE framework enabled hundreds of thousands of mortgages to enter temporary forbearance during the Covid-19 economic shock, a coordinated response that would have been impossible in a fragmented, purely private market. An estimated 535,737 households were saved from foreclosure because of the loss mitigation strategies the framework made possible.⁵⁸ The GSEs today support roughly half of all U.S. mortgage originations, providing standardized underwriting, consistent pricing, and a reliable secondary market.⁵⁹

The right lesson from 2008 is that public financing infrastructure requires honest public governance: explicit guarantees that are priced and managed, transparent standards, and institutional accountability. The RFC’s designers understood something that the privatizers of the 1960s and the deregulators of the 2000s forgot: that when tail risks emerge at business cycle troughs, private markets will retreat from the riskiest and longest-duration assets, and public risk absorption becomes critical to avoiding systemic underinvestment. The public backstop is the load-bearing element of the system rather than a vestigial subsidy to be shed. Nuclear power and other strategic sectors would clearly benefit from a corresponding public financing infrastructure. The better question is how such infrastructure can be designed and maintained from the outset, with the transparency and governance that housing finance returned to only after the pain of the 2008 financial crisis.

From Housing to Industry: Adapting the GSE Logic to Nuclear Power and Beyond

Nuclear power shares many of the characteristics that made housing finance dysfunctional before the New Deal’s interventions: high upfront capital intensity, long construction timelines, long asset lives, lack of through-the-cycle support for continuous builds, and bespoke one-off financing arrangements. But nuclear also presents additional challenges that housing does not: the technological complexity is simply orders of magnitude higher, especially if strategically valuable FOAK builds are pursued again with a new reactor design.

The first units of a nascent reactor design are dramatically more expensive and risky than “nth-of-a-kind” (NOAK) units, but there is no path to NOAK economics without financing the FOAK builds. At NOAK cost, the AP1000 should be able to deliver electricity at roughly $66 per megawatt-hour, competitive with the current national average. But the realized FOAK cost is far higher, particularly during the thirty-year financing period.⁶⁰ The long-run economics are sound, but no individual private actor can bear the front-loaded costs and risks of getting there.

China’s experience demonstrates that this coordination failure is solvable. Their first AP1000 builds—the same reactor design as Vogtle, built from transferred Westinghouse intellectual property—took over nine years from first concrete to commercial operation, comparable to Vogtle’s timeline. But China kept building. Subsequent AP1000-derivative builds have come down to approximately five years.⁶¹ Where the United States was compelled to cut its losses, such as in the case of Westinghouse’s bankruptcy or V.C. Summer’s abandonment, China absorbed the FOAK costs and unlocked the learning curve.

It is worth noting how much learning itself depends on confidence that future builds will be pursued; there is limited incentive to internalize lessons from a painful first build if the next one is unlikely to happen. The Chinese solution, state-owned developers financed by state banks at rates as low as 1.4 percent, is not available to the United States, and forfeits the upsides of private sector innovation and risk management discipline. The question is whether the United States can achieve the same functional outcome through market infrastructure rather than state-directed credit. Critically, financial standardization and industrial standardization are symbiotic: standardized financing terms incentivize developers to converge on proven designs, mature supply chains, and repeatable construction practices because those are the projects that qualify for conforming treatment. Conversely, industrial standardization produces the consistent cost and schedule data that underwriters need to develop reliable risk models. Each reinforces the other.

The housing analogy has clear limits, but it nevertheless offers a concrete agenda for institution-building. Just as the FHA defined what a conforming mortgage looks like, success in the development of nuclear power generation would ultimately have to involve standardized underwriting practices for reactor projects; these would likely run, at least initially, through the Office of Energy Dominance Financing (EDF)—formerly the Loan Programs Office (LPO)—at the Department of Energy (DOE). The dimensions of conformance might include reactor design certification status, supply chain and workforce readiness, EPC contractor track record and capitalization, offtake agreement structure and creditworthiness, construction milestone insurance, and capital commitments from other creditors and shareholders. A project meeting these criteria would be “conforming,” that is, eligible for purchase, guarantee, or securitization by a public entity on transparent and consistently priced terms.

A public entity could then purchase or guarantee conforming project finance instruments, setting the stage for a future tradable asset class that attracts institutional capital. The precise structure of these instruments, whether project bonds, securitized products, or some other novel financial product, cannot be determined by the public sector alone. It must be developed with involvement and iteration by the private sector and their assessment of risk-adjusted returns. Limited public resources can be deployed in the most targeted fashion. Mortgage-backed securities succeeded because of the conforming standards that rendered the underlying loans comparable and predictable. Standardized term sheets could ultimately reduce transaction costs, lower the cost of capital, and create a feedback loop where each successful project makes the next one easier to finance, needing a smaller government guarantee, and driving costs down a learning curve rather than reinventing the financing wheel each time.

Standardized term sheets could ultimately reduce transaction costs, lower the cost of capital, and create a feedback loop where each successful project makes the next one easier to finance. This feedback loop could also lower the size of the government guarantee required for subsequent builds if underwriting standards and risk allocation mechanisms do not need to be constantly reinvented. Elements of this infrastructure already exist in fragmented form. The EDF conducts technological due diligence more rigorously than private lending institutions, and its loan commitments are viewed positively by the financial community precisely because of this technical expertise.⁶² Investment tax credits for nuclear facilities have been maintained through 2032 by the One Big Beautiful Bill Act of 2025.⁶³ But these interventions lack the systematic, standardizing character that made the GSE model transformative for housing. LPO deals have generally taken on a bespoke character across the previous four administrations, with opaque terms that are negotiated in an ad hoc manner. Investors and industry participants are left less likely to learn systematically from the government’s risk assessments and practices.

The FHA’s purpose was to publish standards so that private lenders could conform to them. The LPO has historically operated more like a sophisticated one-off lender than market infrastructure. This one-off approach also forfeits the symbiotic link between financial and industrial standardization: when every deal is bespoke, borrowers receive no template to conform to, and the resulting project-to-project variation makes it harder to develop the consistent performance data that would support standardized financial products. The substantial technical diligence capabilities of the LPO go underutilized when the lessons from each transaction are not codified into replicable standards that the broader market can adopt. The gap between where public nuclear finance stands today and where it needs to be is primarily a gap in the establishment of standardized risk management practices and institutional architecture.

The Trump administration’s October 2025 announcement of an $80 billion strategic partnership with Westinghouse, Brookfield, and Cameco represents a potentially significant departure from this pattern.⁶⁴ The partnership framework envisions fleet-scale deployment of AP1000 reactors—not a single bespoke project but a program to purchase components for eight to ten gigawatts of new capacity through what amounts to a bulk order, allowing Westinghouse to invest in and scale its production capacity.⁶⁵ The Department of Energy has begun meeting with four to five potential utility and developer partners, each of whom would commit to purchasing parts for two new reactors. If executed, this would be the largest coordinated nuclear procurement in the United States since the original wave of construction in the 1960s and 1970s.

The fleet-scale approach is promising precisely because it begins to address the standardization gap this article identifies. A coordinated multi­site build using a single certified design creates the conditions under which underwriting practices, risk allocation frameworks, and financing structures can be developed once and applied repeatedly and iteratively. The disbursement mechanism for the $80 billion is still unclear, however, and the relationship between this partnership and the EDF’s existing loan authority has not been defined. Furthermore, there is no public framework for how private investors would participate alongside public capital.⁶⁶

There’s an opportunity to identify and publicize robust underwriting standards and financing structures emerging from this fleet build, thereby making the United States well positioned to support nuclear construction on a continuous basis. The ambition is laudable; the institutional design that converts ambition into durable market infrastructure is what remains to be built.⁶⁷

This is not a call for the government to build reactors, pick technologies, or operate power plants. It is not a call for the kind of state-directed credit that characterizes Chinese nuclear development. It is a call for the government to extend what it’s done successfully for housing and other sectors: set standards that make the investment proposition understandable and attractive to private capital, provide an explicit backstop that makes institutional investors comfortable with the asset class, and, if successful over time, create the secondary market infrastructure that generates liquidity.⁶⁸ The public sector’s core advantage is bearing the tail risks and solving the coordination failures that private actors are poorly suited to resolving on their own.

The fear that this kind of public financing infrastructure crowds out private investment is exactly backward. Before the GSEs, mortgage markets were local, fragmented, and starved of capital. The GSE infrastructure did not divert capital from other investments; it created a multitrillion-dollar asset class that attracted entirely new pools of capital. Prior to the development of the modern secondary mortgage market, moving funds from capital-rich areas to capital-poor areas was difficult and expensive. The thirty-year mortgage expanded the investable universe rather than shrinking it.

The gap between where public nuclear finance stands today and where it needs to be is not primarily a gap in funding; it is a gap in standardization and institutional architecture.⁶⁹ La Caisse, a Quebec-based pension manager, recently committed £1.7 billion to Sizewell C under the UK’s Regulated Asset Base framework, explicitly citing the model’s predictability for long-term investors.⁷⁰ That commitment was brought about through some substantial derisking measures that gave investors comfort with the unique risks associated with Sizewell C and created alignment between the cash flows of nuclear power generation with institutional investors’ mandates. The challenge will be in creating a sustainable flywheel: identifying systematic approaches to mitigating risk at the private investment and holistic project level, without undermining the push to deliver nuclear power generation with increasing cost competitiveness and independent economic value over time.

Building the Market That Can Build the Plants

The United States has ample capital, financial expertise, and engineering talent. What it lacks is the market infrastructure to connect these strengths to the physical assets the country needs. This is a design failure rather than a resource constraint, and there is historical precedent for fixing it.

The Westinghouse fleet partnership worth $80 billion, the chips Act investments, and emerging bipartisan support for defense-industrial base financing all signal recognition that something must change. But public funds, however large, are finite. Market infrastructure is not; it can help generate compounding returns as each project builds on the standards, instruments, and knowledge established by its predecessors. The GSE model proved the difference: the RFC, FHA, and Fannie Mae did not plan or build homes. They built the market that made private homebuilding financeable at a scale that purely private arrangements never achieved.

What is needed now is the technical financial infrastructure to do the same for strategic industrial investment, starting with standardized underwriting frameworks for nuclear and other clean firm power generation, model term sheets defining what a “conforming” project looks like, securities structures that can attract institutional capital, and governance arrangements that avoid the implicit guarantee trap that brought the housing GSEs to crisis. This is not work that any single organization can do. It requires capital markets practitioners, utilities, construction engineers, safety regulators, and policymakers working in concert.

The United States does not have decades to build this infrastructure for the industrial assets it needs now. But it has the advantage of knowing what the destination looks like because it has built this kind of market before.

This article originally appeared in American Affairs Volume X, Number 1 (Spring 2026): 136–54.

---

Notes