Why Nuclear Forecasts Often Miss the Mark
Nuclear power forecasts often miss the mark. Discover the realities behind the numbers and their implications for clean energy's future.
Nuclear power has been projected to grow faster, cheaper, and more broadly than it actually has—not once or twice, but across decades of energy research. A paper in *Energy Research & Social Science* recently put formal structure around what industry insiders have quietly acknowledged for years: nuclear forecasts aren't just occasionally wrong. They're systematically, persistently wrong in the same direction.
That pattern deserves more scrutiny than it typically gets.
The Discrepancy Between Forecasts and Reality
Pull up almost any major energy outlook from the 1970s, 1990s, or 2000s, and you'll find ambitious projections for nuclear's share of global electricity generation. The actual build rates tell a different story. Global nuclear capacity sat at roughly 393 GW in 2023—a figure that has barely budged in three decades despite forecast after forecast predicting significant expansion.
The failure isn't random noise. It's a consistent directional bias toward optimism that has repeated itself across different institutions, different methodologies, and different political eras.
What makes this particularly interesting from an analytical standpoint is that the bias persists even after forecasters have had the opportunity to observe previous misses. You'd expect Bayesian updating—each generation of analysts incorporating the lessons of the last. Instead, the projections kept arriving with similar structures: nuclear as an expanding cornerstone of the grid, costs coming down, construction timelines tightening. Neither happened reliably.
This isn't a criticism unique to nuclear advocates. Government agencies, international bodies, and independent research groups all fell into the same pattern. The *Energy Research & Social Science* paper frames this through the lens of "nuclear imaginaries"—the shared cultural and institutional assumptions that make certain futures seem inevitable even when the evidence repeatedly contradicts them.
Why Nuclear Power Keeps Underperforming Its Projections
The surface-level explanations are familiar: regulatory complexity, public opposition, construction cost overruns. But those explanations are themselves often underspecified in ways that obscure what's actually happening.
Take construction costs. The U.S. nuclear industry hasn't just experienced cost growth—it's experienced what economists call "negative learning." In most industries, costs fall as you build more units and accumulate expertise. In nuclear, costs have largely moved in the opposite direction over time. The Vogtle Units 3 and 4 in Georgia—the first new nuclear reactors completed in the U.S. in roughly three decades—came in at approximately $35 billion, nearly double the original estimates, and years behind schedule. That's not a one-off project management failure. It reflects deep structural issues in how the U.S. builds complex regulated infrastructure.
Political risk is genuinely harder to model than financial risk, and nuclear is more exposed to political risk than almost any other energy technology.
A change in administration, a shift in public sentiment following an incident anywhere in the world (Three Mile Island, Chernobyl, Fukushima each reset the political clock), or a state-level policy reversal can strand capital invested over decade-long build cycles. Forecasters typically handle this by assuming a stable policy environment—which is precisely the assumption that keeps getting violated.
The economics interact with the politics in a compounding way. When costs rise, the political case for nuclear weakens. When the political case weakens, financing costs increase. When financing costs increase, costs rise further. Forecasts that capture the starting conditions of this cycle rarely model where it ends up.
What This Means for Clean Energy Policy and Investment
Here's the non-obvious implication: if your clean energy transition plan relies on nuclear delivering a certain share of firm, dispatchable zero-carbon power by a certain date, the historical track record suggests you should have a contingency.
That's not an anti-nuclear argument—it's a risk management argument. Investors and policymakers who treat nuclear capacity projections as reliable inputs rather than probability distributions are taking on undisclosed tail risk.
The practical consequence plays out in grid planning. Utilities and grid operators that count on nuclear additions that don't materialize have to scramble for replacement capacity—often natural gas, which creates carbon lock-in that can persist for 30-40 years. The gap between the nuclear future that forecasters described and the one that arrived has, in part, contributed to a natural gas buildout that now complicates decarbonization timelines.
For investors specifically, the lesson is about distinguishing between nuclear's theoretical value proposition (firm, low-carbon, high energy density) and its demonstrated execution risk. Those are both real—and they can coexist. The technology works. The industrial system for delivering it at scale and on budget, at least in the Western context, hasn't.
Hydrogen vs. Nuclear: A Parallel Set of Assumptions
The *Energy Research & Social Science* framing about "imaginaries" becomes even more useful when you apply it to hydrogen—because the assumption structures are strikingly similar.
Green hydrogen has attracted enormous policy enthusiasm and capital commitments over the past five years. The narrative is compelling: use excess renewable electricity to produce hydrogen via electrolysis, store it, and deploy it for hard-to-decarbonize applications—steel, shipping, long-duration storage. The forecasts have been bullish. The IEA and others projected significant green hydrogen cost declines driven by electrolyzer manufacturing scale and falling renewable electricity prices.
The actual deployment has been slower and more expensive than anticipated. Electrolyzer costs have not dropped as quickly as projected. The chicken-and-egg problem between hydrogen supply infrastructure and hydrogen demand has proven stickier than models suggested. Several high-profile green hydrogen projects have been delayed or canceled.
The parallel to nuclear isn't that hydrogen is destined to fail—it's that both technologies attract institutional enthusiasm that can outpace honest reckoning with near-term economics and execution complexity.
This is where the concept of energy research "imaginaries" earns its keep analytically. Both nuclear and hydrogen occupy a particular cultural role in energy discourse: they are technologies that can, in principle, solve problems that other technologies can't. That makes them politically and narratively attractive in ways that can distort both forecast construction and policy prioritization.
Where Does Nuclear Go From Here
The case for nuclear hasn't collapsed—it's become more complicated. Small modular reactors (SMRs) represent a genuine attempt to address the root cause of nuclear's cost problem by moving construction from bespoke, on-site megaprojects toward factory-manufactured, standardized units. The theory is sound: factory production enables learning curves, quality control, and supply chain optimization that site-built construction can't replicate.
NuScale, Rolls-Royce, and others are developing SMR designs at various stages of maturity. But SMRs face their own version of the chicken-and-egg problem—manufacturers need orders to drive down costs, but customers need demonstrated costs before placing orders. The first movers carry disproportionate risk.
The policy lever that could actually move the needle isn't a subsidy structure—it's guaranteed offtake and streamlined licensing that reduces the two variables that have historically broken nuclear economics: schedule uncertainty and financing costs.
The U.K.'s Regulated Asset Base model, which allows nuclear developers to collect revenue during construction, is one structural innovation worth watching. It doesn't make nuclear cheap, but it makes the financial risk profile more manageable for private capital. The U.S. has experimented with production tax credits and loan guarantees, with mixed results.
What the historical record of energy research on nuclear forecasts ultimately argues for is epistemic humility baked into planning processes—not abandonment of nuclear as a technology, but honest accounting of the gap between what models project and what industrial systems deliver. That gap exists for solar and wind too (though, notably, in the opposite direction—they've consistently outperformed forecasts). The difference is that solar and wind's positive surprises are relatively easy to absorb. Nuclear's negative surprises are not.
Grid planners, investors, and policymakers would be better served by nuclear projections that range from "this goes well" to "this goes the way Vogtle went"—and by strategies that remain coherent across that entire distribution. The technology is too important to the clean energy growth agenda to be undone by forecasts that keep setting it up to disappoint.
INTERNAL LINK SUGGESTIONS
- [INTERNAL LINK: nuclear forecasts]
- [INTERNAL LINK: clean energy transition]
- [INTERNAL LINK: hydrogen technology]