What the IAEA’s Sixth SMR Forecast Upgrade Means for U.S. Investors

The agency that tracks global nuclear capacity has just raised its small modular reactor forecast for the sixth year running, and this time it projects SMRs at up to 28% of all new nuclear additions by 2060, with North America leading the pack at roughly 60%. Six consecutive revisions in the same direction is not a rounding error. It is a pattern, and the pattern itself is the signal worth interrogating.

The revision landed in mid-September 2026, the same week the U.S. Nuclear Regulatory Commission (NRC) cleared expedited early site-work at the Palisades Energy Center in Michigan, where Holtec International is pushing toward two SMR-300 units targeting 2030 operation. Institutional projection and regulatory action arriving together is what turns a long-range chart into a live investment question.

This piece lays out where the SMR opportunity actually stands in September 2026: what the International Atomic Energy Agency (IAEA) numbers do and do not imply for near-term U.S. deployment, and which variables will decide whether the high-growth scenario materialises or reverts toward a far more cautious baseline.

What the IAEA’s tripled capacity forecast actually signals

The headline number is a full tripling. The IAEA’s high-growth case now projects global nuclear capacity reaching 1,284 GW(e) by 2060, roughly 3.4 times the 377 GW(e) recorded in 2025. Of the new capacity added in that scenario, SMRs account for 28%, or about 284 GW(e).

The more revealing shift sits at the bottom of the range. Even the low-growth case, which sees global capacity reaching only 696 GW(e) by 2060, now puts SMRs at 23% of new additions, around 120 GW(e).

The single most underreported figure in this forecast is not the tripling headline. It is the low-case SMR share jumping from 5% to 23%. The agency has stopped treating SMRs as a fringe outcome even in its pessimistic scenario.

Six upward revisions in a row is the analytical point here. This is not a one-year outlier; it is a sustained repositioning by the world’s leading nuclear authority. When the IAEA’s General Conference in Vienna extended projections to 2060 for the first time this edition, it did so while lifting SMR assumptions across every scenario it publishes.

The IAEA’s sustained upward revisions reflect a broader institutional shift in how the nuclear sector evaluates SMR technology fundamentals, including pressurised-water designs, passive safety systems, and factory-fabrication models that distinguish small modular units from traditional large-reactor construction economics.

Set against that optimism is the International Energy Agency’s (IEA) far cooler read. Under current policies, the IEA sees only about 40 GW of SMRs globally by 2050, treating the IAEA’s high-growth figures as a frontier scenario rather than a central expectation.

The gap between roughly 40 GW and 284 GW is where the investment risk actually lives. For U.S. investors weighing SMR-exposed equities and project timelines, the revision pattern is genuinely bullish, but the spread tells you the opportunity is policy-contingent. The direction of travel is your reason to look; the policy dependency is your hedge.

Palisades and the deployment gap between forecast and ground truth

In September 2026, Palisades is the most advanced SMR site in the United States, and understanding exactly what has been approved there tells you how early in the cycle this all sits.

On 28 August 2026, the NRC granted Holtec International an exemption authorising the installation of permanent support-of-excavation (SOE) and cutoff-wall systems, the engineering needed to stabilise and prepare the ground. The commission was explicit that this is not a construction permit and does not predetermine decisions on the Limited Work Authorization (LWA) or full construction permits.

Read plainly, this is site preparation, not construction commencement. The two SMR-300 pressurised-water units planned at Palisades, together expected to generate roughly 600-680 MW(e), are targeting 2030 commercial operation, with Holtec additionally planning up to four units at Oyster Creek in New Jersey aiming for 2036.

Pull back from Michigan and the deployment base looks thin. Russia and China are currently the only countries with functioning, grid-integrated SMR installations. Every Western project remains in design, licensing, or early site-work phases.

Here is where global SMR deployment actually stands:

• Operational (grid-integrated): Russia and China
• Advanced site-work: United States, Palisades (SOE and cutoff-wall approved)
• GDA / licensing stage: United Kingdom, Rolls-Royce SMR
• Design-approved but undeployed: United States, NuScale US460

The IAEA’s forecast, in other words, rests on a foundation that is still being poured. For investors, the distinction between a regulatory milestone and commercial progress is the entire game. Palisades is the leading edge of U.S. deployment, and its current stage clarifies just how far the pipeline is from bankable, operating megawatts.

Why U.S. deployment timelines have slipped before

The clearest domestic case study in slippage is NuScale Power’s cancelled Carbon Free Power Project (CFPP) in Idaho. Costs escalated from approximately US$3.6 billion for a 720 MW plant to US$9.3 billion for a scaled-down 462 MW plant before the project was cancelled in November 2023.

The NRC did approve NuScale’s US460 light-water design on 29 May 2025, a genuine regulatory marker. But a design approval is not a deployment path, and that approval has not yet translated into a confirmed commercial project.

The read for U.S. investors is simple. When a design clears the regulator, that resets the licensing risk but leaves financing and offtake risk fully intact. Current timelines should be discounted accordingly.

The cost competitiveness question that determines everything

The cost debate is not a verdict; it is a genuine tension, and both sides of it are supported by real numbers.

Start with the barriers. First-of-a-kind (FOAK) capital costs are estimated at US$6,000-9,000 per kW, and FOAK SMR levelised cost of electricity (LCOE), the all-in cost of generating a megawatt-hour over a plant’s life, is put at roughly US$80-120 per MWh initially. Some Wood Mackenzie analysis has flagged that figure potentially reaching US$200 per MWh, several times the roughly US$40-60 per MWh of utility-scale solar. David Schlissel, Director of Resource Planning Analysis at IEEFA, has argued that actual early-project experience does not yet support the economic competitiveness claims made for the technology.

Now the pathway. Proponents point to learning-curve effects: as standardised units repeat, costs are projected to fall toward US$3,000 per kW by around Unit 50. The IEA’s own modelling sees SMR LCOE in 2040 at roughly US$110 per MWh in the U.S. and US$130 per MWh in the EU, about 20% above large reactors. Genuine competitiveness is generally pegged to costs falling below roughly US$4,500-5,000 per kW.

The most tractable lever in that equation is the cost of money.

Financing is the swing variable. A weighted average cost of capital (WACC) around 12% is a primary barrier to SMR economics, whereas concessional finance at 3-5% could roughly halve the cost of SMR power.

That is the crux for U.S. investors. The competitiveness case is not really a question of whether costs can fall, but whether the policy environment and financing structures align fast enough to compete with renewables-plus-storage before the demand window closes. Early SMR investment is therefore a bet on policy continuity and cheap capital as much as on the technology itself, and the learning curve only pays out after a meaningful number of units are actually built.

The JEK2 experience in Slovenia reinforces the same dynamic: financing costs in nuclear projects can exceed the reactor’s overnight construction cost when commercial WACC applies to multi-decade build timelines, a pattern that U.S. SMR developers will need to avoid if they are to compete with renewables-plus-storage on a levelised basis.

How AI data center demand is reshaping the SMR investment case

The demand side is what changes the calculus. A single hyperscale AI data center is estimated to consume electricity equivalent to about 50,000 residential homes, with training and inference clusters requiring hundreds of megawatts of continuous power per campus.

AI data center energy demand is scaling faster than most grid operators projected three years ago, with hyperscale campuses now routinely requesting dedicated interconnection agreements rather than drawing from shared utility supply, a structural shift that makes the always-on baseload profile of SMRs distinctly attractive to developers who cannot tolerate intermittency.

That is a very specific kind of demand: firm, low-carbon, always-on baseload that renewables alone struggle to guarantee. It is also demand attached to buyers willing to sign long-term offtake agreements to lock it in, which is precisely what SMR project finance has lacked.

The capital is already moving. The clearest signals are the corporate agreements struck across 2024-2026.

The distinction in that final column matters. The Microsoft-Constellation deal is a restart of the Crane Clean Energy Center at Three Mile Island, an existing large reactor that can support AI baseload before 2030. The Google-Kairos and Amazon-X-energy agreements are dedicated SMR development, and those are 2030s stories.

That is the timeline tension in a nutshell:

1. AI infrastructure scales on 2-5 year planning cycles.
2. SMR licensing and construction runs closer to 5-7 years and often longer.
3. The first operational SMR-powered data centers are projected around 2030.
4. Broad commercial maturity for the integration sits in the early-to-mid 2030s.

There is a physical constraint underneath the schedule too: SMR concepts require cooling water measured in tens of millions of gallons per day, adding siting, environmental, and permitting complexity wherever developers try to co-locate reactors with dense computing campuses.

For investors, the takeaway is balanced. Tech-sector capital gives the IAEA’s high-growth forecast a demand-side anchor it previously lacked, and the commitments signed in this window will heavily influence which developers reach commercial deployment first. But most of that money is pointed at the 2030s, not the back half of this decade.

Where the SMR opportunity sits for U.S. investors right now

Pull the four threads together and a single decision-frame emerges. The IAEA forecast supplies directional momentum, Palisades supplies the deployment reality, the cost curve supplies the economic tension, and AI demand supplies the missing buyer. The question is which of those forces holds.

For the U.S. specifically, the IAEA’s high-case 60% North America SMR share is the ceiling the market is being sized against. The IEA’s roughly 40 GW global baseline is the floor if policy momentum stalls. Holtec’s 2030 target at Palisades is the nearest commercial proof-of-concept that will tell you which end of that range is realistic.

Three conditions would support the high-growth scenario:

• Policy continuity on nuclear permitting and financing support
• Strong FOAK cost performance at lead projects such as Palisades
• Tech-sector offtake agreements converting into bankable project finance

Three risks would push outcomes toward the IEA baseline:

• Cost escalation or schedule slippage repeating the NuScale CFPP pattern
• Financing staying at commercial WACC without concessional support
• Corporate commitments stalling before they reach final investment decisions

The honest question in September 2026 is not whether SMRs will be built in the United States. It is whether the first wave of commercial deployments establishes the cost and performance foundation that makes the IAEA’s high-growth case achievable within this decade. The Palisades exemption and the forecast revision arriving in the same week show U.S. deployment has moved from policy ambition into regulatory process, without yet reaching commercial certainty. Investors now have, for the first time, concrete milestones to track it against.

For investors assessing the full risk stack behind SMR project timelines, our dedicated guide to SMR critical mineral supply chains covers the uranium enrichment, zirconium, and specialty steel dependencies that could constrain first-wave U.S. deployments independently of licensing or financing progress.

This article is for informational purposes only and should not be considered financial advice. Investors should conduct their own research and consult with financial professionals before making investment decisions. Past performance does not guarantee future results. Financial projections are subject to market conditions and various risk factors, and forward-looking statements are speculative and subject to change based on policy and market developments.