The Nuclear Pipeline: 79 Reactors, 15 Countries, and a World Bank Reversal

 

For the first time in decades, the global nuclear construction pipeline is growing, the world's development bank has reversed a 30-year ideological ban, and the arithmetic is beginning to force its way into policy. Here is what the numbers actually say.

The state of play

As of May 2026, there are 79 reactors under construction in 15 countries, with a combined gross capacity of roughly 86 GW. Another 124 are in advanced planning — funded, approved, or firmly committed — adding a further 110 GW. And 305 more have been proposed, with site designations but uncertain timelines, representing a potential 285 GW. The full pipeline, from shovel in ground to proposal, is approaching 500 GW of new capacity.

To put that in perspective: the entire current global nuclear fleet produces about 400 GW. If half the pipeline were built, it would roughly double world nuclear capacity.

 

The geographic concentration, however, is striking. Most of this activity is happening in Asia — and most of that is China.

China is building half the world's nuclear capacity

China has 39 reactors under construction — nearly half the global total — with a combined capacity of 44 GW. At its current pace, China is expected to reach 100 GW of installed nuclear capacity by around 2030, displacing the United States as the world's largest nuclear power producer.

The technology mix is revealing. China is not simply replicating foreign designs. It is commercializing its own: the Hualong One (a domestically developed pressurized water reactor), the CAP1000 and CAP1400 (derivatives of the AP1000 it licensed from Westinghouse), the CFR600 (a sodium-cooled fast reactor), and the ACP100 — a 125-MW small modular reactor that became the world's first land-based commercial SMR when it connected to the grid in 2024.

In 2025 alone, China began construction on nine new reactors. Over the past decade, 94% of all reactor construction starts worldwide were of Chinese or Russian design. The West, for all its renewed enthusiasm, is mostly talking. China and Russia are mostly building. 

Source: World Nuclear Association, April 7 2026. Ukraine, Japan, Brazil figures include suspended projects.

What is being built, and when

Nearly all reactors currently under construction are large-scale light-water reactors — most above 1,000 MW. The dominant designs fall into three families:

 

Two SMR designs deserve particular attention. The ACP100 in China represents the first commercial land-based SMR anywhere in the world. Russia has two floating RITM-200S reactors (53 MW each) under construction at Cape Nagloynyn, following the already-operational Akademik Lomonosov floating plant. These are not demonstrations; they are commercial projects.

The West, by contrast, has precisely zero SMRs under construction. Canada, the UK, South Korea, and the United States all have SMR programs at various stages of planning and permitting. Several have received regulatory approvals or letters of intent. None have poured concrete.

A word on molten salt reactors. None of the 79 reactors in the WNA's official construction list are MSRs — that technology is operating on a different, longer clock. But it is not standing still. China's TMSR-LF1 at the Shanghai Institute of Applied Physics is the world's only operating liquid-fueled MSR; in October 2024 it became the first reactor anywhere to add thorium to a working molten salt core, and by November 2025 it had successfully bred uranium-233 from thorium — the first experimental confirmation of thorium fuel conversion in a running machine. China's stated next step is a 100-MWth thorium MSR demonstration reactor by 2035, with commercial plants envisioned around 2040. In the United States, Kairos Power has two salt-cooled reactors under active construction in Oak Ridge, Tennessee — Hermes 1 (a 35-MWt non-power demonstration) and Hermes 2 (a 50-MWe commercial-scale plant that broke ground on April 17, 2026, backed by Google and the Tennessee Valley Authority). A technical caveat worth noting: Kairos uses solid TRISO fuel pebbles cooled by fluoride salt, not liquid fuel dissolved in salt — it captures most of MSR's safety and thermal advantages while avoiding the hardest materials chemistry. That is a pragmatic engineering choice, not a failure of ambition, but it does mean the transformative long-term promise of online fuel reprocessing remains for a later generation of machines. Commercial MSRs burning thorium at scale before 2040 seem unlikely anywhere except possibly China. Before 2050, they are plausible — and potentially game-changing.

When will these plants come online?

The construction pipeline is heavily front-loaded — meaning the majority of what is currently being built was supposed to connect to the grid within the next three to four years. That schedule, as anyone who has watched Hinkley Point C will know, is aspirational. Here is the official picture:

 

A word of caution about these dates: They represent the "latest announced or estimated year of grid connection" — a figure set by operators and developers, and historically optimistic. The average construction time for reactors completed globally between 2014 and 2023 was ten years. China has managed as few as six. Western projects have routinely run two to three times over schedule. Hinkley Point C, originally scheduled for 2025, is now projected for 2029 at the earliest.

The planning pipeline: who has serious ambitions?

Beyond what is under construction, 124 additional reactors have cleared the threshold of formal approval or committed funding — the "planned" category in World Nuclear Association terminology. These are mostly expected to be operational within 15 years.

 

Source: World Nuclear Association, April 2026. "Planned" = approvals, funding, or firm commitment in place. "Proposed" = specific site or programme; timing uncertain.

Several stories stand out in this data.

Romania has 8 reactors in the planned column — more than France, Sweden, or the Czech Republic. Two new units at Cernavodă using Canadian CANDU technology have been in planning for decades, but recent US financing interest (via the Export-Import Bank and the Development Finance Corporation) has materially advanced the timeline.

Poland has 7 planned and 22 proposed — one of the more serious pivots in European energy policy. A coal-dependent country, Poland is simultaneously evaluating large AP1000-type plants and SMRs, with both KHNP (South Korea) and Westinghouse competing for the large-reactor contract.

The United States is the most striking entry: zero planned, 25 proposed. The gap between aspiration and commitment tells you something about the regulatory and financial environment for new nuclear in America. Enormous interest; no concrete.

Money: the World Bank reversal and the financing gap

The most consequential development of 2025 was not a new reactor design or a construction start. It was a policy shift at 1818 H Street in Washington, D.C.

In June 2025, the World Bank lifted its longstanding ban on financing nuclear energy, signing a formal partnership agreement with the IAEA to support nuclear deployment in developing countries. The Asian Development Bank followed in November 2025, removing its own parallel exclusion. Two of the most important development finance institutions in the world had, for over three decades, refused to touch nuclear power with borrowed money. That era is now over.

Why this matters: IAEA Director General Rafael Grossi said it plainly: "Nothing could change without the World Bank. It's as simple as that." Access to World Bank financing — low-interest, long-tenor sovereign loans — unlocks the economics of nuclear for dozens of developing countries that cannot otherwise absorb the upfront capital cost of a large reactor.

The initial scope of the Bank's new policy is deliberately narrow: lifetime extensions of existing plants, and small modular reactors in developing nations. Greenfield large-reactor financing is not yet formally on the table. But the architecture is in place, and the direction is clear.

Four countries appear to be the most likely early beneficiaries, based on their existing World Bank relationships and the state of their nuclear programmes: Argentina, India, South Africa, and Ukraine — all seeking lifetime extensions for aging fleets. For new build in Europe, Romania is the leading candidate; it already receives World Bank lending and has a partially built reactor at Cernavodă.

 

Bangladesh and Egypt are an interesting edge case. Both are building Russian VVER-1200 reactors financed largely by Russian state credit — a mechanism that comes with its own geopolitical strings. As Rosatom financing terms become more fraught in the post-2022 environment, both countries have strong incentives to diversify their financing sources. The World Bank is now an option where it was not before.

The investment gap

The financing picture remains insufficient even with the World Bank's reversal. Global annual investment in nuclear reached approximately $75 billion in 2024 — up from a 2017–2023 average of $50 billion. But the IAEA's high-case projection for tripling nuclear capacity by 2050 requires $125 billion per year. The global pledge made at COP28 — to triple nuclear capacity — implies $150 billion annually.

The gap between current investment and what is needed is roughly $75 billion per year. The World Bank's total annual lending across all sectors runs to about $100 billion. Nuclear will be a small slice of that. The signal matters more than the volume, for now — what the Bank's participation does is unlock the conditions under which commercial and multilateral co-financing can flow.

What this tells us about the energy transition

The nuclear pipeline, read alongside the World Bank reversal, tells a coherent story — even if it is not the story that dominates the energy transition discourse.

First, the countries that are actually building nuclear are the ones with the most serious energy needs: China, India, South Korea, Egypt, Bangladesh, Turkey. These are not ideologically driven decisions; they are engineering decisions made by governments that have looked at the arithmetic of energy density, reliability, and carbon intensity and concluded that nuclear belongs in the mix.

Second, Russia and China are winning the export competition for nuclear technology in exactly the way the United States and France once did. The VVER-1200 is in Bangladesh, Turkey, Egypt, Hungary, and China. The Hualong One is in Pakistan and, eventually, elsewhere. The geopolitical implications are not subtle.

Third, the West has rediscovered enthusiasm for nuclear without yet rediscovering the industrial capacity to build it. The United States has 25 proposed reactors and zero planned ones. The UK is building two EPRs a decade behind schedule. France has announced new plants without yet committing to finance them. The gap between declared intent and actual construction start remains wide.

Fourth — and this is the development I find most significant — the financial architecture is finally shifting. The World Bank ban was not merely a financial obstacle; it was a signal to the entire multilateral development community that nuclear was radioactive in the reputational, not just the physical, sense. That signal has been reversed. The Asian Development Bank followed within five months. Others will follow.

A regulatory footnote that may matter more than it seems. In May 2025, President Trump signed Executive Order 14300 directing the NRC to reconsider the Linear No-Threshold model — the 70-year-old assumption that any radiation dose, however small, carries proportional cancer risk — and the ALARA principle (As Low As Reasonably Achievable) that flows from it. The Department of Energy moved faster: Energy Secretary Chris Wright's memo of January 10, 2026 formally eliminated ALARA from DOE's radiation protection framework, and reporting suggests the change was being implemented in internal orders as early as August 2025, before any public announcement. The NRC is now circulating a draft rule under which operators would no longer be required to minimize exposures below legal limits — only to stay beneath those limits. The practical effects on deployment could be significant: ALARA has driven enormous and arguably unnecessary cost in plant design, shielding, and worker scheduling, costs that have made nuclear less competitive than it should be. If the LNT model is replaced by a threshold or hormesis-based standard — as a growing body of radiobiological evidence suggests is warranted — the economics of both new construction and lifetime extensions improve materially. Critics, including the Union of Concerned Scientists, argue the changes lack scientific grounding and procedural legitimacy. That debate will play out in the courts and in the rulemaking docket. But the direction of travel is clear, and it is not unfavorable to nuclear deployment.

The world is waking up to what the arithmetic always said: you cannot decarbonize reliably without high-energy-density, dispatchable, land-efficient power. The pipeline is not perfect — it is weighted toward Russian and Chinese technology, several key projects are years behind schedule, and the financing gap remains large. But the direction is unambiguous. The nuclear renaissance that has been announced repeatedly since 2000 may, this time, be real.

Data sources: World Nuclear Association (April 7, 2026); IEA Global Energy Review 2026; IAEA PRIS database (May 12, 2026); Bulletin of the Atomic Scientists (August 2025); World Economic Forum (October 2025); ANS Nuclear Newswire (July 2025); E&E News (March 2026); NPR (January 2026); Executive Order 14300 (May 2025).

Note: Help of Anthropic's Claude in retrieving the data and assembling graphics is gratefully acknowledged

Nuclear Power, Justice, and the Stories We Choose to Tell

Linda Pentz Gunter’s new book arrives with vigor and moral urgency. Her instinct is right: energy is a human story before it is a technical one. But her conclusions rest on a selective reading of evidence and on an old narrative strain that has hardened into dogma in certain environmental circles. It’s worth pushing back—not because nuclear power is flawless, but because misdiagnosing the problem leads us to misallocate the solutions.

Let’s take her core claims in turn.

1. “Nuclear destroys lives along the uranium chain.”

Historical harms in the uranium cycle are real, especially where regulation was weak and Indigenous communities bore the brunt. But this is not a property of uranium—it’s a property of injustice. Modern uranium mining is far safer and tightly regulated. The per‑kWh externalities from the nuclear fuel cycle are small compared to the vast, diffuse impacts of fossil fuels—and even many metals used in today’s renewable infrastructure.

A typical 1‑GW nuclear reactor uses about 200 metric tons of natural uranium per year, a volume that fits in a single truckload. A comparable annual output from “renewables” demands thousands of times more material—steel, aluminum, copper, concrete, silicon, rare earths—mined from places with far fewer guardrails and far more human cost.

If justice is the lens, we have to apply it consistently.

2. “Nuclear derails climate progress.”

This argument transforms climate timelines into ideology rather than arithmetic. A single cubic mile of oil—my favorite unit—represents roughly the world’s annual oil consumption. Replacing that with intermittent sources alone requires enormous overbuild, land area, and storage. None of this shows up in activist narrative, but the math is unavoidable.

History is clear: France, Sweden, and Ontario—where nuclear was allowed to operate—cut emissions quickly and deeply. Germany, which shut down nuclear early, saw its emissions rise. The atmosphere keeps an honest ledger.

3. “Nuclear provokes war.”

This collapses civilian energy and weapons geopolitics into one undifferentiated fear. Nuclear power does not cause proliferation; geopolitical ambition does. The non‑proliferation regime has been one of the quiet successes of international governance. South Korea, Japan, and dozens of nations operate advanced reactors with no interest in nuclear weapons.

Energy security—local, stable, dispatchable—is a geopolitical stabilizer, not a spark.

4. The missing argument: low energy density is its own form of injustice

Gunter is right to care about environmental justice, but she stops short of examining the land‑hungry reality of the alternatives she champions.

Low energy density sounds like a technical footnote. It is not. It is a justice issue.

A single 1‑GW nuclear plant sits on a few square kilometers. To match it with intermittent renewables requires:

• hundreds of square kilometers of wind, or
• thousands of acres of solar,
• plus transmission corridors, roads, substations, exclusion zones, and multi‑day storage.

Where does all this infrastructure go? Not in affluent suburbs. Not near activist headquarters. It goes in rural landscapes, Indigenous territories, farming regions, open prairies, deserts, mountains—places with less political power and far deeper ecological value.

In the name of “green energy,” we are industrializing vast tracts of land under the moral camouflage of good intentions.

Energy density is not a sin. It is a gift.

5. Radiation risk and the damage of a flawed model

Radiation fear is the emotional engine behind much of anti‑nuclear advocacy. And that fear rests heavily on the Linear No‑Threshold (LNT) model—a simplistic, 1950s-era assumption that every increment of radiation, no matter how small, linearly increases cancer risk.

LNT produces conclusions that border on the absurd. Consider a simple analogy.

A fall from 10 meters might have a 5% fatality risk. LNT implies risk scales perfectly with height:

• 10 m → 5%
• 1 m → 0.5%
• 0.1 m (a curb) → 0.05%

If 1,000 people step off a curb, LNT predicts 0.05% × 1,000 = 0.5 expected fatalities.
Half a death—from curb‑stepping.

This is not physics or biology. It is numerology.

6. Chernobyl: when LNT meets large populations

The infamous projection of 2,000 additional deaths from Chernobyl did not come from observed cases. It came from LNT: a tiny hypothetical incremental cancer risk multiplied across millions of people who received doses comparable to a cross‑country flight or a move to Denver.

The UN Scientific Committee on the Effects of Atomic Radiation (UNSCEAR) later concluded that such small increases are too low to detect above statistical noise.

But the scary number stuck. Fear spreads faster than nuance.

7. Fukushima: when fear becomes the real hazard

Fukushima deepened this mythic structure. The fatalities were caused by the tsunami, not radiation.
The number killed by radiation exposure: zero.

Yet the fear of radiation—driven by LNT thinking—triggered a mass evacuation of over 150,000 people. Vulnerable residents were uprooted from hospitals and nursing homes with little preparation. Social ties were severed. Medical care was disrupted.

Subsequent analyses estimate well over a thousand premature deaths attributable to the evacuation itself.

Radiation didn’t kill these people. Fear did.

Narratives that reinforce radiophobia are not harmless. They have a body count.

8. The subsidy myth

Another favorite talking point of anti‑nuclear activists—though not uniquely Gunter’s—is that nuclear survives only because of subsidies.

Subsidies are not inherently bad. Solar only reached scale because of them—China’s especially. But the narrative that nuclear is uniquely subsidized collapses under actual numbers.

U.S. Federal Subsidies in FY2022 (EIA):
• Solar: $5.6B
• Wind: $4.3B
• Biofuels: $9.1B
• Nuclear: $0.1B
• Oil & gas: ~$2.1B

When normalized by output, wind and solar receive many times more subsidy per kWh than nuclear—and in many years more than oil and gas.

Subsidies scale emerging industries. That’s fine. But selective outrage is not analysis.

9. Signs of progress

There is good news. After decades of treating LNT as untouchable dogma, the U.S. Nuclear Regulatory Commission is stepping away from automatic reliance on it.

And in a major shift, the World Bank—long unwilling to finance nuclear projects—has relaxed its stance. Developing nations can finally pursue the only reliable, dense, low‑carbon baseload option that doesn’t consume vast landscapes.

These changes won’t erase 70 years of radiophobic policy overnight, but they are cracks in the old paradigm—and cracks widen.

A better path forward

If nuclear were invented today, environmentalists would be celebrating it: zero‑carbon, land‑efficient, material‑light, long‑lived, and capable of anchoring a stable grid. Instead, we inherited a narrative frozen in 1979.

We don’t need cheerleading for any technology. We need proportionality. We need consistency. And above all, we need honesty about the trade‑offs.

On that score, nuclear is not the villain. It is the adult in the room.

And the longer we cling to outdated narratives, the more we undermine our climate goals—and the more land, wildlife, and vulnerable communities will bear the avoidable costs of our energy choices.

Marine Carbon Drawdown: Restoring Ocean Abundance While Addressing Climate Change

Readers of this blog know of my advocacy for nuclear power as an essential tool for reducing greenhouse gas emissions. While emissions reductions are necessary, mitigation alone is no longer sufficient. Since the dawn of the industrial age, humanity has added hundreds of gigatons (Gt) of carbon dioxide (CO₂) to the atmosphere. Even under aggressive decarbonization pathways, this accumulated stock of CO₂ will continue to drive warming for decades.

Recognizing this reality, the National Academies of Sciences, Engineering, and Medicine conclude that achieving net-zero—and ultimately net-negative—emissions will require large-scale carbon removal technologies (NASEM, 2021). Carbon drawdown—actively removing CO₂ from the atmosphere—must therefore complement emissions reductions, not replace them.

Among the various approaches for removing CO₂ from the atmosphere, marine carbon dioxide removal (mCDR) stands out for its scale, physical feasibility, and potential co-benefits. The ocean already plays a central role in Earth’s carbon cycle, absorbing roughly 25–30% of anthropogenic CO₂ emissions each year through physical and biological processes (Friedlingstein et al., 2023). At the core of this system is the biological carbon pump, in which phytoplankton convert dissolved CO₂ into organic matter, a portion of which sinks into the deep ocean and is sequestered for centuries or longer while the rest of the biomass becomes the base of the food web for all ocean creatures.

A defining advantage of mCDR is its potential to generate tangible co-benefits. By strengthening the base of the marine food web, enhanced primary productivity can translate into greater fish biomass, more resilient ecosystems, and stronger fisheries—particularly in nutrient-limited regions. This matters politically as much as scientifically. Climate policies that impose visible costs while delivering diffuse or delayed benefits often struggle for public support. In contrast, mCDR offers a pathway that could support fishing industries, strengthen coastal economies, and improve food security, while simultaneously contributing to climate stabilization.

This natural system, however, is constrained. In vast regions of the open ocean, phytoplankton growth is limited not by sunlight, but by the availability of trace nutrients such as iron or silica. This limitation presents an opportunity to amplify the ocean’s natural carbon uptake by selectively adding the missing nutrients in those regions. Could this approach work?

Natural events provide important clues about what is possible. Following the 1991 eruption of Mount Pinatubo, scientists observed a temporary slowdown in the rise of atmospheric CO₂ concentrations recorded in the Keeling Curve (Keeling et al., 1995; Sarmiento, 1993). While part of this effect reflected cooling-induced changes on land, volcanic ash also delivered iron and other micronutrients to surface waters.

Subsequent satellite and oceanographic observations support the conclusion that such nutrient inputs stimulated large phytoplankton blooms, increasing oceanic carbon uptake. Estimates of the resulting enhancement to the global carbon sink reach tens of gigatons of CO₂, demonstrating that the oceans system is capable of large, rapid increases in carbon sequestration under favorable conditions.

Controlled ocean iron fertilization experiments over the past three decades reinforce this lesson. Small additions of iron to high-nutrient, low-chlorophyll (HNLC) regions reliably stimulate phytoplankton blooms (Boyd et al., 2007). These experiments validated the underlying mechanism of nutrient-based mCDR, while also revealing important complexities. Not all carbon fixed during blooms is exported to the deep ocean; much is recycled at the surface through grazing and microbial activity. As a result, early experiments—limited in scale and duration—produced modest sequestration efficiencies relative to initial expectations (Buesseler et al., 2008).

Crucially, these studies were never designed to test large-scale or long-term deployment. More recent research emphasizes precision approaches: combining multiple nutrients, using mineral carriers such as clay, and targeting regions where ocean circulation naturally favors long-term carbon storage (NASEM, 2022). mCDR, in other words, is not about indiscriminate fertilization of the seas. It is about targeted interventions informed by ecology and geography that work with ocean physics rather than against it.

Before industrial whaling, large whale populations played a significant role in ocean nutrient cycling through the so-called “whale pump”, redistributing iron and other nutrients via fecal plumes (Roman & McCarthy, 2010). The near elimination of great whales likely reduced this natural fertilization process across large ocean regions. While whale recovery remains essential, it is necessarily slow. This raises an uncomfortable but important question: should human systems attempt to partially replicate this lost ecological service?

Global shipping fleets offer a compelling answer. Commercial and passenger vessels already traverse most of the world’s oceans on predictable routes. With appropriate safeguards, ships could be equipped to disperse carefully measured nutrients or mineral carriers, mimicking key aspects of the whale pump. This would transform global shipping from being a polluter to an agent for climate and ecosystem restoration.

None of this should be pursued casually. mCDR carries real risks that must be rigorously managed. Among the most prominent concerns is the potential for harmful algal blooms (HABs), which can disrupt marine ecosystems and produce toxins. These blooms occur most often in coastal regions already overloaded with nitrogen and phosphorus from agricultural runoff, underscoring the need for precise calibration and careful site selection.

In conclusion, carbon drawdown is no longer optional—it is indispensable to climate stabilization. The ocean offers one of the few mechanisms capable of operating at the necessary scale. Guided by natural analogues such as volcanic eruptions, informed by decades of ocean research, and implemented with care, mCDR has the potential to deliver both climate stabilization and ecological restoration. At a moment when political feasibility is as important as technical validity, it stands out as a rare solution that aligns environmental urgency with economic and ecological opportunity—helping to restore not only the stability of the climate system, but the abundance of the oceans themselves.

---

References

• Boyd, P. W., et al. (2007). Mesoscale iron enrichment experiments 1993–2005. Science.
  https://www.science.org/doi/10.1126/science.1131669
• Buesseler, K. O., et al. (2008). Ocean iron fertilization—moving forward in a sea of uncertainty. Science. https://www.science.org/doi/10.1126/science.1154305
• Friedlingstein, P., et al. (2023). Global Carbon Budget 2023.
  https://essd.copernicus.org/articles/15/5301/2023/
• Keeling, C. D., et al. (1995). Interannual extremes in the rate of rise of atmospheric CO₂. Nature.
  https://www.nature.com/articles/375666a0
• National Academies of Sciences, Engineering, and Medicine (2021).
  Negative Emissions Technologies and Reliable Sequestration.
  https://nap.nationalacademies.org/catalog/25259
• National Academies of Sciences, Engineering, and Medicine (2022).
  A Research Strategy for Ocean-based Carbon Dioxide Removal.
  https://nap.nationalacademies.org/catalog/26278
• Roman, J., & McCarthy, J. J. (2010). The whale pump. PLoS ONE.
  https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0013255
• Sarmiento, J. L. (1993). Atmospheric CO₂ stalled. Nature.
  https://www.nature.com/articles/365697a0

---