Nuclear Power: From Fission to Firm Electricity

8.1 — Binding energy and E = mc²

A nucleus is held together by the strong nuclear force. Weigh a bound nucleus and it comes out slightly lighter than the sum of the individual protons and neutrons inside it. That missing mass — the mass defect — is the binding energy, converted through E = mc². Because c² is roughly 9 × 10¹⁶, a mass difference far too small to notice corresponds to an energy release far too large to ignore.

Rearranging a nucleus into more tightly bound fragments therefore liberates energy, and the amounts involved are millions of times larger than anything available from rearranging electrons, which is all that chemistry — burning coal, oxidising hydrogen, discharging a battery — can do.

8.2 — Fission and the chain reaction

A U-235 nucleus that absorbs a neutron becomes unstable and splits, typically into two smaller nuclei of unequal mass. The fragments fly apart with enormous kinetic energy, which becomes heat as they collide with surrounding material. Crucially, the split also releases two or three free neutrons.

Those neutrons can trigger further fissions, which release more neutrons, which trigger more fissions. This is the chain reaction. Left to grow unchecked it would escalate at extraordinary speed; suppressed too far it dies out. Holding it exactly steady, continuously, for eighteen months at a time, is the central problem of reactor physics.

8.3 — Criticality: a technical term, not a warning

Few words in energy are more consistently misread than critical. In everyday speech it signals danger. In reactor physics it describes the neutron population, and the normal, desired, steady operating state of every power reactor on Earth is precisely criticality.

A reactor holding 1,000 MW flat for a year is, by definition, critical for that entire year. Bringing it up from 600 MW to 1,000 MW means going briefly supercritical, then returning to critical at the new level. Shutting down means driving it deeply subcritical. The words describe direction of travel, not proximity to disaster.

8.4 — Neutron economy, enrichment and the fuel cycle

Every generation of neutrons faces four possible fates: cause another fission, escape the core entirely, be absorbed by structural material or coolant, or be absorbed deliberately by control systems. Managing that ledger — the neutron economy — determines what fuel a design can run on.

Natural uranium is only about 0.7% U-235; the rest is U-238, which does not readily fission with slow neutrons. A design that loses neutrons freely must compensate by concentrating the fissile fraction — enrichment, typically to 3–5% for a commercial light-water reactor. A design that is miserly with neutrons, usually by using a more effective moderator, can run on natural uranium and skip enrichment entirely.

Note where the vulnerabilities sit. Solar and wind need no ongoing fuel supply at all — once built, the fuel arrives free. Nuclear needs a functioning chain of mining, conversion, enrichment and fabrication for its entire operating life, and every link in that chain is concentrated in a small number of countries. Fuel security for nuclear is a genuine strategic question rather than a commercial procurement one.

8.5 — The reactor is a steam plant with a different furnace

Strip away the reactor island and a nuclear station is architecturally familiar. Heat boils water, steam drives a turbine, the turbine spins a synchronous generator, electricity goes to a step-up transformer and out to the grid. Everything downstream of the heat source is the thermal-plant architecture established in Chapter 1.

The entire technological distinction lies in one row. That is a useful thing to keep in mind when nuclear is discussed as though it were exotic: two-thirds of the plant is conventional heavy engineering that the power industry has built for a century. The exotic part is the box that makes the heat, and the discipline required to guarantee it never stops being cooled.

8.6 — Moderator, coolant and reactor architectures

Neutrons released by fission are fast. Slow neutrons are far more likely to cause further fission in U-235, so most commercial designs use a moderator to slow them down. A coolant carries the heat away. Whether those are the same substance, and what substance it is, defines the reactor architecture — and cascades into fuel requirements, safety behaviour and cost.

The PHWR row is the one worth dwelling on. Heavy water is expensive to produce, but because it is so transparent to neutrons it lets the design run without enrichment at all. A country with uranium but no enrichment capability — or one that does not wish to depend on someone else’s — can build a complete domestic fuel cycle around this choice. A materials decision becomes a sovereignty decision.

8.7 — Control rods, scram and decay heat

Control rods contain neutron-absorbing material — boron, cadmium, hafnium. Inserting them removes neutrons from the economy and reduces reactivity; withdrawing them increases it. A rapid emergency shutdown, universally called a scram, drops the rods fully into the core and halts the chain reaction in seconds.

And then the reactor keeps producing heat.

The fission products accumulated over months of operation are intensely radioactive. They continue decaying regardless of whether any fission is occurring, and that decay releases heat. This decay heat is not a residual trickle. It is the single most important concept in nuclear safety engineering, and it is why “shut down” and “safe and cold” are not the same statement.

8.8 — Defence in depth

Nuclear safety is fundamentally a containment problem: keeping radioactive material where it is. The design philosophy is not to build one very good barrier but several independent ones, so that no single failure — and preferably no plausible combination of failures — produces a release.

The guiding rule is that no single failure should lead directly to a catastrophic release. Redundancy alone is not enough, because identical redundant systems share identical failure modes; hence diversity, where backup systems work on different physical principles and different power sources. And hence the modern emphasis on passive safety, where the ultimate backup depends on nothing but physics.

8.9 — Why nuclear plants are so capital-intensive

Chapter 4 introduced the split between capital-intensive and fuel-intensive generation. Nuclear is the most extreme case of the former in the entire energy system — more so than solar, because on top of a large overnight cost it carries a construction schedule measured in years to more than a decade.

Roughly three-quarters of the delivered cost of a nuclear megawatt-hour traces back to capital and its financing. Fuel is under a tenth. This is the exact mirror image of a gas plant, where fuel dominates and the turbine is comparatively cheap, and it has a direct consequence: nuclear economics track interest rates far more closely than they track uranium prices. A doubling of the uranium price is an irritation. A doubling of the cost of capital is existential.

8.10 — The capacity factor advantage

Against that cost structure sits nuclear’s structural advantage: it runs almost all the time. Output depends on neither sunlight, nor wind speed, nor river flow, nor ambient temperature. The only things that stop a reactor are refuelling and maintenance, both of which are scheduled.

The comparison needs care, though, because the bars in that figure do not all mean the same thing. Nuclear and geothermal run near their technical ceiling. Gas and reservoir hydro sit lower by choice — they are dispatched to follow price and demand, and could run harder if the economics called for it. Only solar and wind sit lower because of a constraint no operator can relax. Reading capacity factor as a quality score misses what each plant is actually for.

8.11 — Nuclear is not perfectly inflexible

A persistent oversimplification holds that nuclear can only run flat out at 100%. It is not true. Modern reactors, particularly the French fleet which has operated in a load-following mode for decades, can and do vary output substantially over a day.

But the ability is not free. Power cycling complicates fuel management, because the neutron flux distribution and the xenon poisoning that follows a power reduction both shift with output. It adds thermal and mechanical cycles to components designed for steady operation. And crucially, because nuclear’s costs are almost entirely fixed, every megawatt-hour not generated is revenue forgone against a cost base that does not shrink. A gas plant that backs off saves fuel; a nuclear plant that backs off saves almost nothing.

So nuclear’s natural system role remains high-utilisation firm generation, with faster and cheaper-to-cycle resources — batteries, gas peakers, hydro, demand response — handling the rapid swings, exactly as the load-curve architecture in Chapter 5 described.

8.12 — Nuclear and grid stability

Like coal and gas plants, and unlike solar and most wind, nuclear turns a large synchronous generator directly coupled to grid frequency. That coupling delivers services that never appear on an energy bill: physical rotating inertia that resists frequency change, reactive power for voltage support, and the short-circuit strength that protection systems need in order to detect and clear faults correctly.

These were discussed in Chapters 2 and 5 as properties of synchronous generation generally. They matter increasingly here because as inverter-based renewables take a larger share of total generation, the pool of machines providing them shrinks. In a system approaching very high renewable penetration, nuclear’s contribution to system stability becomes a distinct value stream from its contribution of energy — and markets that pay only for energy will systematically undervalue it.

8.13 — Nuclear against solar and wind

Read as a scorecard, this table invites the wrong conclusion. These are complementary system components, not substitutes competing for the same job. Solar is the cheapest way to make bulk energy in daylight. Wind is the cheapest way to make bulk energy across multi-day weather systems. Nuclear is one of the few ways to make firm energy that does not care about weather at all. A portfolio question, as Chapter 4 framed it — not a winner-takes-all one.

8.14 — Nuclear for data centres

Large AI and cloud campuses have a demand profile that is unusual in the best possible way for nuclear: very large, very flat, and extremely intolerant of interruption. They do not want annual energy, they want continuous power, and they are willing to sign long contracts to get it.

That has made firm nuclear generation commercially interesting to hyperscale operators through several structures: conventional utility supply, long-term power purchase agreements that underwrite an existing plant’s continued operation, uprates at existing sites, restarts of recently retired units, and co-located or purpose-sited next-generation reactors. The common thread is that a creditworthy counterparty willing to commit for twenty years addresses precisely nuclear’s weakest point — revenue certainty across a very long payback period.

Whether that translates into large volumes of new build depends on the same variable everything else in this chapter depends on: whether anyone can construct reactors on schedule.

8.15 — Small modular reactors

Rather than one enormous bespoke reactor built in place, small modular reactors pursue smaller standardised units manufactured in a factory and assembled on site. The reactor becomes a product with a production run rather than a civil engineering project with a site.

The argued benefits follow from that shift: shorter construction schedules and therefore less interest during construction; lower absolute capital at risk per unit; factory quality control instead of field fabrication; incremental capacity additions matched to demand growth; and siting flexibility for locations that cannot host a gigawatt.

8.16 — The nuclear learning curve problem

Seen this way, SMRs are not really a bet on small reactors. They are a bet that nuclear can be moved out of the megaproject category and into the manufactured-product category, where learning curves operate. That is the actual hypothesis being tested, and it is why the number of units ordered matters far more to SMR economics than the elegance of any particular design.

The counter-evidence is real: the countries that have built nuclear cheaply — France in the 1970s and 80s, South Korea more recently — did so by building the same design repeatedly with a stable supply chain and a stable regulator. Standardisation and repetition were the mechanism, not reactor size. Whether the same benefit is best captured by small factory-built units or large repeated ones is genuinely unsettled.

8.17 — Advanced reactors and nuclear beyond electricity

Beyond conventional water-cooled designs sits a family of advanced concepts using high-temperature gas, molten salt or liquid metal as coolant. Their common goals are higher operating temperatures, better fuel utilisation, and passive safety characteristics that derive from the coolant’s physical properties rather than from added engineering.

That reframing matters. A reactor judged purely as an electricity generator competes against solar plus storage on cost per megawatt-hour, a contest it frequently loses. A reactor judged as a source of continuous high-grade heat competes against burning gas in an industrial furnace, where it has far fewer rivals.

8.18 — Spent fuel, waste and geological disposal

Spent fuel leaves the reactor intensely radioactive and thermally hot. It goes first into a spent fuel pool, where water provides both cooling and shielding for several years while the shorter-lived fission products decay and the heat output falls. It then moves to dry cask storage — passively cooled steel and concrete containers requiring no pumps or power.

High-level waste is remarkable for being simultaneously the smallest and the most demanding waste stream in energy. The entire spent fuel output of a country’s nuclear programme over decades occupies a volume that a coal fleet produces in ash within days. But it must be isolated for timescales that exceed recorded human history.

The engineering consensus on deep geological disposal is broad and has been for decades. The obstacle has consistently been siting and public consent rather than technical feasibility — which is why Finland, having secured local agreement, reached operation of a repository well ahead of far larger nuclear programmes.

8.19 — Reprocessing and proliferation

Spent fuel is not entirely spent. It still contains uranium and the plutonium bred from U-238 during operation. Reprocessing separates these for reuse as fresh fuel, extracting more energy from the same mined uranium and reducing the volume requiring final disposal.

It also separates plutonium, which is precisely the concern. Reprocessing is expensive, chemically complex, and produces its own liquid waste streams — and it puts weapons-usable material into a separated form. The countries that reprocess and those that pursue a once-through cycle have made different judgements about whether the resource benefit justifies the handling burden and the proliferation exposure.

8.20 — India’s three-stage nuclear programme

India has limited high-grade uranium but among the world’s largest thorium resources. That single resource asymmetry shaped one of the most distinctive long-term national energy strategies ever attempted, laid out by Homi Bhabha in the 1950s.

The physics is sound. Thorium is not itself fissile the way U-235 is; it is fertile, meaning it must first be converted into something that is. That conversion requires a neutron source, which is why the thorium stage sits at the end of a chain rather than at the beginning.

Progress has been slower than the original timetable envisaged. Fast breeders have proved difficult and expensive worldwide, not only in India, and commercially scalable thorium power remains unproven anywhere. But as a piece of strategic reasoning — deriving a multi-decade technology roadmap from a national resource endowment — the programme remains without real parallel.

8.21 — The core nuclear mental model

Most public argument about nuclear collapses into two slogans: it is too expensive, or it is clean and firm. Both are true of some projects and false of others, which means neither is useful on its own. The questions that actually discriminate between a project that makes sense and one that does not are these.

• •Physics — how does this reactor generate and control heat, and what happens to it when everything is switched off?
• •Fuel — where does the fuel come from, does it require enrichment, and who controls that step?
• •Reactor — PWR, BWR, PHWR or advanced, and has this exact design been built before?
• •Safety — how is reactivity controlled, and how much of the safety case depends on active systems rather than physics?
• •Cooling — how is decay heat removed with no power, no pumps and no operators, for as long as it takes?
• •Containment — how many independent barriers, and do they fail in independent ways?
• •Economics — overnight cost, cost of capital and, above all, construction schedule. Not the uranium price.
• •Grid — what is firm capacity and synchronous inertia worth in this particular system, and does the market actually pay for either?
• •Waste — is there a funded, sited plan for spent fuel, or an assumption that someone will solve it later?
• •Geopolitics — how secure is the fuel and technology supply chain, and what does it make the buyer dependent on?
• •Scale — is this a repeatable unit or a one-off? This is the question that decides whether costs fall or rise over a programme.

Chapter 9 turns to the oldest source of renewable electricity on Earth, and to the only grid-scale storage technology ever deployed at terawatt-hour scale: hydropower and pumped storage.