What is the nuclear renaissance and will it happen?

The short answer

The “nuclear renaissance” refers to a shift in policy and corporate stance after two decades of stagnation in OECD countries. The drivers: data center demand growth requiring large-scale firm clean power, decarbonization commitments by states and corporates, and recognition that variable renewables face system-integration limits. The most concrete commitments are reactor restarts (Three Mile Island Unit 1, Palisades) and hyperscaler PPAs.

The IAEA’s SMR Dashboard documents 127 SMR designs in development as of 2024, 51 in pre-licensing or design certification. Microsoft has signed a 20-year PPA with Constellation to restart Three Mile Island Unit 1; Amazon and Google have committed to SMR developers. Vogtle Units 3-4 (Georgia, traditional large reactors) entered service in 2023-24 at a cost of approximately $31.5 Bn for 2.2 GW.

However, no SMR has yet been operationally deployed in OECD countries at commercial scale. The renaissance is, for now, an industry pipeline rather than installed capacity.

→ New to financial education? Financial education hub

What the data shows

Nuclear deployment data from IAEA, NEA, and operator filings provide the picture:

• Active reactors globally 2024: 411 reactors, ~370 GW (IAEA PRIS)
• SMR designs catalogued: 127 (IAEA SMR Dashboard III, 2024), of which 51 in pre-licensing
• SMR commercial operation in OECD: 0 (as of mid-2025)
• Vogtle Units 3-4 cost: $31.5 Bn for 2.2 GW (15+ years construction, multiple overruns)
• Microsoft-Constellation Three Mile Island Unit 1 restart: 20-year PPA, target operation 2028
• EDF Hinkley Point C latest cost estimate: GBP 31-34 Bn (2015 prices) — vs GBP 18 Bn original (2017)
• French nuclear share of electricity 2024: ~67% (vs 70-75% historically)
• South Korea APR1400 export deals: UAE Barakah completed 2024 at relatively low cost vs Western counterparts

The exception that nuances the picture: Russian and Chinese reactor exports have continued throughout the OECD’s two-decade slowdown. China commissioned roughly 4-5 reactors per year in the early 2020s, and Rosatom has built or is building reactors in over a dozen countries. The “nuclear renaissance” is mainly an OECD phenomenon — global capacity has continued to grow incrementally.

→ Dataset: Uranium price history

Why it happens — the macro mechanism

The nuclear renaissance reflects three converging structural pressures.

Channel 1 — Data center demand timing. Hyperscaler AI data center deployments require multi-GW firm power on 2-5 year timelines. Variable renewables cannot deliver firm capacity at this pace, and gas turbines face permitting and emissions concerns. Existing nuclear capacity (reactor restarts, lifetime extensions) is uniquely positioned to fill this gap because the assets exist or existed recently. The Three Mile Island Unit 1 restart announced by Microsoft and Constellation is the highest-profile example, but Palisades (Holtec/Constellation) and the planned restart of Duane Arnold are similar moves.

Channel 2 — SMR economics and the cost-overrun problem. Traditional large reactors face capital cost overruns averaging 100-200% globally over the past two decades, with Vogtle Units 3-4 costing roughly twice their original estimate. SMRs propose to address this through factory production, smaller capital outlays, and lower financing costs. The economic logic is sound, but no OECD SMR has yet operated commercially. Whether the 127 designs in development will actually deliver lower costs at scale depends on first-of-a-kind engineering execution and licensing efficiency. This is the most underappreciated dimension: the renaissance is being underwritten in capital markets and PPAs based on costs that have not yet been demonstrated.

A short note on regulatory uncertainty. The US Nuclear Regulatory Commission has approved one SMR design (NuScale, since cancelled) and is reviewing several others. Regulatory pathways for novel designs (high-temperature gas reactors, molten salt) remain less defined. EDF’s GBP 31-34 Bn revised cost estimate for Hinkley Point C illustrates the persistence of regulatory and construction risk even for established PWR designs.

Channel 3 — Capacity payment and PPA structures. The renaissance is being financed through novel contractual structures: long-dated PPAs (Microsoft-Constellation 20 years), hyperscaler equity investments (Google in Kairos Power, Amazon in X-energy), and government loan guarantees (US DOE Loan Programs Office). Traditional merchant electricity markets cannot support nuclear’s capital structure; the renaissance therefore depends on counterparties willing to underwrite long-dated firm-power contracts. The sustainability of this structure beyond the initial wave is uncertain.

Synthesis by regime: in the post-Three-Mile-Island and post-Fukushima freeze (1979-2010 in much of OECD, post-2011 in Japan and Germany), new builds collapsed and lifetime extensions dominated; in the 2010-2022 cost-overrun regime, large reactor projects in Western markets (Vogtle, Hinkley Point C, Olkiluoto) faced multi-year delays and cost overruns of 100%+; in the 2022-2025 corporate-driven regime, hyperscaler PPAs and SMR pipeline development have shifted the political and financial conversation, but installed OECD SMR capacity remains zero.

The nuclear renaissance is real in the pipeline and the contracts — and not yet real in the megawatts.

→ Framework: Macro-financial regimes

What it means for different economic actors

Savers. Nuclear power, where it operates, provides large-scale firm electricity that stabilizes wholesale prices and reduces gas-price exposure for retail rates. Reactor restarts in regions with high renewable penetration (such as PJM in the US) materially affect grid stability and pricing.

Investors. The renaissance has triggered a re-rating of utilities with existing nuclear capacity (Constellation, EDF, Vistra, KEPCO) and equity investment in SMR developers. Returns depend on PPA execution, regulatory approvals, and ultimately whether SMR cost projections materialize. The capital-intensity and long lead times make nuclear projects sensitive to financing structure — government loans, capacity contracts, and tax credits dramatically alter the economics.

Industrial firms. Hyperscalers and large industrial customers are the principal counterparties of the renaissance. Microsoft, Amazon, Google, Meta, and Oracle have signed PPAs ranging from existing-reactor power to multi-GW SMR portfolios for delivery in the late 2020s and 2030s. The transition from variable-renewable PPAs to nuclear PPAs reflects firm-capacity requirements that exceed what storage alone can deliver at current costs.

A common error is to assume the renaissance will solve the cost-overrun problem inherent to nuclear construction. Vogtle, Hinkley Point C, and Olkiluoto have demonstrated that even mature PWR designs face significant execution risk in OECD jurisdictions. Whether SMRs avoid this depends on whether factory production and standardized designs deliver as promised — an empirical question with no current answer.

Practical observation

Go deeper

Related questions

• Why are utility companies critical to the energy transition?
• What is the levelized cost of energy (LCOE)?
• Why does intermittency matter for renewable electricity?
• How much investment does the energy transition require?

Frequently asked questions

Why are SMRs being promoted as the solution to nuclear’s cost problem?

Small modular reactors (typically 50-300 MW per unit, vs ~1,100 MW for traditional PWR) propose to reduce per-unit capital outlay, enable factory production, and shorten construction timelines. The economic logic is sound: smaller projects face lower financing risk, and standardized designs benefit from learning curves across multiple deployments. However, this remains a hypothesis. Capacity-factor advantages of nuclear (>90%) are preserved at smaller scale, but the per-MW capital cost may not scale down proportionally. The first commercial OECD SMRs (NuScale was cancelled in 2023; X-energy and Kairos Power are in development) will determine whether the cost-out story holds in practice.

Why are hyperscalers willing to underwrite long-dated nuclear PPAs?

AI data center load profiles require firm power 24/7, in volumes that variable renewables cannot reliably deliver at current storage costs. Microsoft, Amazon, Google, Meta, and Oracle face decarbonization commitments while needing massive firm capacity within 5-10 years. Long-dated nuclear PPAs (10-25 years) provide both attributes that combination. The economic logic is hyperscaler-specific: their data center capex is large enough that paying a premium for firm clean power is rational, and their decarbonization commitments are public enough that lower-cost gas alternatives create reputational risk. Whether this dynamic extends to other industrial customers is unclear.

How does the European nuclear picture differ?

EDF’s French fleet (~57 GW) provides about two-thirds of French electricity, having operated continuously since the 1980s build-out. Recent years have seen unplanned outages from corrosion issues and lifetime-extension decisions. France committed in 2022 to building six EPR2 reactors (with option for eight more), but Hinkley Point C cost overruns (now estimated at GBP 31-34 Bn) illustrate execution challenges. Germany shut down its remaining reactors in 2023 despite the gas crisis. The European renaissance is therefore concentrated in France, with mixed signals from the UK, Belgium, Netherlands, and Eastern European states variously committing to new builds or extensions.

Last updated — 21 July 2026