The Anchor

Nuclear power is not an alternative to the electrical energy transition. It is the thing that holds it together.

There is a conceptual error at the heart of a lot of energy debates, and it goes like this: baseload power is essential, renewables cannot provide it, therefore, we need nuclear (or gas, or coal) to keep the lights on.

The error is not in the conclusion. It is in the premise.

Baseload is not a law of physics. It is a historical artefact of combustion. The concept emerged from energy systems organised around coal, oil, and gas plants that were designed to run continuously because starting and stopping them was costly, slow, and wasteful. In those systems, reliability was equated with constant output, demand was treated as fixed, and supply was engineered to match it with large, inflexible units running at high utilisation.

This made sense in a fuel-based world. Coal plants achieved their best economics at steady state. Nuclear plants were designed around the same principle. Gas turbines offered flexibility but at a cost premium. The entire system assumed that supply should be constant and demand predictable — an assumption that reflected the limitations of combustion, not the requirements of physics.

Electrified systems behave differently. Wind turbines and solar panels do not need to run continuously to be economically viable; their fuel is free. Electric motors and power electronics respond in milliseconds. Storage can absorb and release energy without fuel handling or thermal losses. In this context, baseload loses its organising role.

Reliability is no longer delivered by a small number of inflexible plants running at full tilt, but by the coordinated behaviour of many assets responding to system conditions. What matters is not constant output, but the ability to meet demand across time.

The persistence of baseload thinking reflects a bigger conceptual mistake: equating reliability with constant output. In modern electricity systems, reliability is better measured by outcomes — frequency stability, loss-of-load probability, and recovery time — than by operating profiles.

So far, so good. The renewables-plus-storage crowd gets this part right. Where they go wrong is in assuming the anchor is no longer needed.

The anchor problem

Strip away the ideology on both sides — the nuclear maximalists who want to build reactors for everything, and the 100%-renewable absolutists who insist storage alone can do the job — and you arrive at a practical question: what holds the grid together while flexibility scales?

Because flexibility is scaling. Batteries are being deployed at 130 GW per year globally. Demand response programmes are turning millions of heat pumps, EV chargers, and industrial loads into grid participants. Software platforms such as Kraken are orchestrating distributed assets in real time. The electrical system I described in my last piece — “End to End” — is being assembled, piece by piece.

But it is not assembled yet. And even when it is, it will need something underneath it. Something firm.

Consider what “firm” means in practice. A four-hour battery is excellent for shifting solar generation into the evening peak. It is not designed for a five-day winter anticyclone over northern Europe when wind generation drops to 5% of capacity across the entire continent, and solar output is negligible. Germany experienced exactly this in January 2024 — a “Dunkelflaute” that lasted nearly a week. France’s nuclear fleet covered the gap, exporting power across the interconnectors while gas plants ramped everywhere else.

Firm power is not about normal conditions. It is about the conditions that stress the system. The tail risk. The week when everything that can go wrong does go wrong, and the grid still needs to function.

Nuclear delivers that. Not because it runs all the time — increasingly, it doesn’t — but because it can. A nuclear plant has a capacity factor above 92%, the highest of any generation technology. It does not depend on weather, does not transit a chokepoint, and does not need a gas pipeline from a country that might decide to turn it off.

France is already showing the future

If you want to see what nuclear flexibility looks like in practice, look at France.

EDF’s nuclear fleet — 56 reactors, roughly 61 GW — now swings 6 GW per day in response to solar generation on the French grid. In 2022, that daily swing was 1.5 GW. Nuclear modulation volumes rose 17% in 2025. Seven out of every ten curtailed gigawatt-hours were driven by economics rather than safety or technical constraints. When midday solar pushes wholesale prices towards zero, EDF ramps nuclear down. When the sun sets, and demand climbs, it ramps back up.

This is not a reactor from a 1970s textbook. This is a fleet that flexes with the market. EDF distributes the modulation cycles across units — Cattenom 2 curtailed nearly 1.4 TWh in 2025 alone — concentrating wear where it is least costly and preserving asset life across the portfolio.

The old framing said nuclear was rigid and renewables were flexible. The reality is that nuclear is learning to dance, and the music is being set by solar.

France expects 350-370 TWh of nuclear output in 2026 and 2027. That is roughly 65-70% of the country’s electricity. But the character of that output is changing. It is no longer a flat baseload. It is firm, flexible, and market-responsive — exactly the anchor that a high-renewables grid needs.

China understands this better than anyone

China has 58 operating reactors and 33 under construction. In April 2025 alone, the State Council approved 10 new units across five projects — over 200 billion yuan in investment. The country’s nuclear capacity is on track to reach 110 GWe by 2030 and 150 GWe by 2035.

But here is the part that Western commentators consistently miss: China is not building nuclear as an alternative to renewables. It is building nuclear as the counterweight that allows renewables to dominate.

China installed more solar in 2024 than Europe's entire existing solar capacity. It has the world’s largest wind fleet. Its battery storage deployment is accelerating faster than in any other country. Nuclear is not the headline in China’s energy transition — it is the footnote that makes the headline possible.

China’s newer reactor designs are smaller, more modular, and built for combined heat and power integration. They provide low-carbon, dispatchable power and heat in applications where electrification alone remains insufficient. They offer inertia and frequency stability — the electrical equivalent of the flywheel effect that spinning turbines provide. And they anchor regions less suited to large-scale renewable deployment.

China did not “solve” nuclear power. It made it boring, repeatable, and cheap — and in doing so, integrated it quietly into the world’s most significant electrical energy transition. Infrastructure matters more than innovation theatre. Skills, factories, and repetition determine outcomes. Standardisation is a political choice, and when systems are allowed to repeat, cost, safety, and speed improve together.

The key lesson: electrification is institutional before it is technological. Financing structures, governance capacity, and workforce depth shape results more than reactor designs.

How big is the anchor?

The answer depends on where you are.

A country’s nuclear requirement is not a fixed percentage of its electricity mix. It is a function of its seasonal resource profile — how much renewable output varies across the year, and what is available to fill the gaps.

Consider latitude. In northern Europe, winter solar generation drops to 10–15% of summer output. A country like the UK, at 51–58°N, loses most of its solar contribution precisely when demand is highest — dark, cold months when heating loads peak and daylight hours shrink below eight. Wind is stronger in winter on average, but also more volatile. A week-long anticyclone — the Dunkelflaute that hit Germany in January 2024 — can suppress both solar and wind simultaneously across the entire continent.

The further north you go, the wider the seasonal gap, and the larger the firm anchor needs to be.

Countries with significant hydro resources can offset some of this. Norway generates over 90% of its electricity from hydropower and needs no nuclear anchor at all. But Norway’s geography is exceptional. Most countries do not have deep fjords and mountain reservoirs to draw on. Pumped hydro storage helps where the terrain permits it, but the global pipeline remains limited relative to need.

This is why nuclear’s share of the mix varies so dramatically across countries planning for 2050.

China currently generates roughly 5% of its electricity from nuclear. By 2050, most projections place that figure between 15% and 28%, with installed capacity reaching 400–554 GWe. China’s nuclear share stays relatively modest because of the sheer scale of its renewable deployment — the anchor is large in absolute terms, but the system it sits within is enormous. China also has significant hydro resources, particularly in the south-west, which absorb some of the seasonal variability.

France is the opposite case. Nuclear currently provides 65–70% of French electricity, and the government’s target is to maintain 50% or above through 2050. Independent analysts suggest 30–40% may be more realistic as renewables scale, but even at the lower end, France retains the largest nuclear share of any major economy. The reason is partly institutional — France built its fleet in the 1970s and 1980s and has the workforce, the supply chain, and the political consensus to maintain it. But it is also geographical. Northern France has a winter solar profile similar to southern England, and while the country has good wind resources, it lacks Norway’s hydro advantage although it could further develop it’s Alpine pumped hydro. The nuclear fleet fills the seasonal gap that storage and interconnection alone cannot close.

The UK targets 25% nuclear by 2050, up from roughly 15% today, with 24 GWe of planned capacity including Hinkley Point C and Sizewell C. Britain sits at higher latitudes than most of continental Europe, with winter solar output that drops to a fraction of summer levels. It has good offshore wind but faces the same Dunkelflaute risk as its neighbours. Without a firm anchor, the UK’s path to a decarbonised grid relies on gas backup or massive overbuild of storage — both expensive and politically fragile.

Then there is Spain, which illustrates the other end of the spectrum. Spain currently generates about 20% of its electricity from nuclear, but official policy targets a full phase-out by 2035. The logic, at least on paper, is that Spain’s solar resource is among the strongest in Europe — Andalucía and Extremadura receive nearly twice the annual irradiance of northern France. Winter solar output drops less steeply at 36–43°N, and Spain’s wind profile is more consistent than northern Europe’s. The country is betting that solar, wind, batteries, and interconnection can do the job without a nuclear anchor.

Whether that bet holds is another question. Spain’s grid still depends on gas for flexibility, and the political consensus behind the nuclear phase-out is narrower than it appears. One severe Dunkelflaute or a prolonged gas supply disruption could shift the calculation. But the point stands: a country with strong solar at lower latitudes and less seasonal variability needs a smaller anchor than one at 55°N with eight hours of winter daylight.

Geography does not determine energy policy, but it constrains it. The size of the anchor is not ideological. It is physical.

The demand shock that didn’t exist before

Every previous nuclear cycle was driven primarily by government policy, utility procurement, and post-oil-shock energy security. This cycle has a new driver that dwarfs them all: digital infrastructure.

Data centres consumed an estimated 460 TWh of electricity in 2022. By 2026, that figure is expected to exceed 1,000 TWh — more than a third of all the electricity generated by the world’s nuclear fleet last year. And the trajectory is steepening, not flattening, as AI inference scales.

What makes this demand different from anything that came before is its character. A data centre does not want intermittent power. It does not want power that varies with the weather, the season, or the time of day. It wants 24/7 baseload — or more precisely, 24/7 firm power — delivered continuously with five-nines availability, at a price that doesn’t spike whenever a cold front crosses the continent.

Big Tech understands this. Microsoft has committed $16 billion to restart Three Mile Island’s Unit 1 — renamed the Christopher M. Crane Clean Energy Centre — under a 20-year power purchase agreement targeting 835 MW. The site is 80% staffed and targeting a 2027 restart. It will be the first time a retired nuclear reactor in the United States has been brought back to life to serve a single corporate client.

Amazon is investing $20 billion at Susquehanna nuclear plant in Pennsylvania, converting it into an AI campus drawing hundreds of megawatts from an existing reactor. Google and Kairos Power signed the first corporate SMR fleet deal — 500 MW by 2030, half a dozen reactors. In total, Big Tech has signed over 10 GW of new US nuclear capacity in the past year.

These are not expressions of environmental preference. They are hard-nosed procurement decisions driven by a single calculation: what is the cheapest way to guarantee uninterrupted power at a massive scale? The answer, for workloads that cannot tolerate intermittency, is nuclear.

The supply problem beneath the surface

And this is where the story turns from constructive to concerning.

The uranium market is not prepared for what is coming. After Fukushima in 2011, the industry entered a prolonged depression. Prices remained below the cost of production for most of the next decade. Mines were therefore mothballed, exploration budgets were cut, and the entire supply pipeline was allowed to atrophy. The spot price hit $18 per pound in mid-2016. At that level, nobody invests in new capacity. Nobody explores, and nobody trains the next generation of engineers.

What we see now is the consequence. Prices have recovered — the spot hit $106 in January 2026 before pulling back to $88 by late March — but the supply infrastructure they are supposed to activate no longer responds in the same way. Bringing a new uranium mine from discovery to production is not a matter of months. It is a 10-15 year process involving permitting, environmental review, financing, construction, and commissioning. That lag is what turns a normal cycle into a structural imbalance.

The numbers tell the story starkly. In 2025, demand for uranium was 182 million pounds, while mining supply was 176 million pounds — a deficit of 3.3%. By 2030, demand reaches 245 million pounds against a supply of 185 million — a 24.5% deficit. By 2040, demand is projected to reach 397 million pounds, while supply is projected to be 201 million pounds. A 49.4% deficit. That is not a gap that can be closed by marginal increases in output.

Layered on top of this physical supply constraint is a geopolitical one that the market is still underestimating. Russia, through Rosatom, controls approximately 46% of global uranium conversion capacity and 65% of commercial enrichment services. For high-assay low-enriched uranium — the fuel required for many advanced reactor designs, including SMRs — Russia is currently the only commercial supplier on earth.

The United States has zero commercial enrichment capacity online today. The first domestic facility is licensed for 2027. The Department of Energy announced $2.7 billion in funding in January 2026 to expand enrichment over the next decade, spread across three companies at roughly $900 million each. The target is to reduce Russian separative work units to 10-15% of US needs by 2028. Even if that target is met, it leaves a multi-year window of dependence amid intensifying geopolitical friction and a shadow war between Russia and Europe.

When demand is structural, and supply is constrained, the adjustment mechanism becomes price. Based on how the fundamentals are developing, it is hard to argue that adjustment is anywhere near complete.

Anchoring the E-Flip

I wrote in my book that the end of baseload thinking does not eliminate the role of nuclear power — it reframes it. Nuclear is no longer positioned as the backbone of continuous supply, but as a provider of firm, low-carbon energy delivered over long time horizons, particularly in systems seeking to minimise fuel imports and emissions.

That reframing is now being validated in three places simultaneously.

In France, it looks like a 56-reactor fleet learning to flex with solar, swinging gigawatts daily and earning revenue from market signals rather than running flat out on cost-plus contracts. In China, it looks like 33 reactors are under construction, sized precisely to complement the world’s largest renewable build-out, with standardised designs, state financing, and institutional patience. In the United States, it looks like the world’s largest technology companies are spending tens of billions to secure nuclear power for AI workloads that cannot tolerate interruption.

Three geographies. Three different institutional models. One conclusion: the electrical energy system needs an anchor, and nuclear is the strongest candidate.

This does not mean nuclear replaces renewables. The economics of solar and wind are overwhelming and will only strengthen. It does not mean nuclear is easy — the Western track record on cost and schedule is poor, and the uranium supply chain carries real geopolitical risk. And it does not mean every country needs nuclear — some will build all-renewable grids with long-duration storage and interconnection.

But for systems that need to electrify fast, at scale, with firm power that doesn’t depend on weather or geopolitics, nuclear is not optional. It is the counterweight that lets flexibility scale. It is the thing that runs when everything else can’t.

Nuclear allows countries to electrify faster without destabilising the grid, while renewables, storage, and flexibility scale around it. It is neither maximalist nor ideological. It is sized to the role it plays — not the role it once played in fossil-dominated systems.

The E-Flip is from 80% molecules to 80% electrons. Nuclear ensures the electrons are always there.

Gradually, then suddenly.

end

Nadim Chaudhry is the author of ElectroState: How the Electrification E-Flip, China, Geopolitics will Reorder the Global Economy, examining the global transition from fossil fuels to electrification through geopolitical and systems lenses.

ElectroState.com offer premium intelligence services and consulting for the electrical energy transition