Nuclear power should be central to net-zero energy transition, not marginalised

The grid reliability case for nuclear is straightforward: it provides dispatchable firm power, meaning it runs when the grid needs it, not when the wind blows or the sun shines. At high penetration levels of variable renewables, the marginal cost of balancing the grid increases sharply — you need either massive storage (currently expensive and geographically constrained) or firm dispatchable generation. Nuclear is the lowest-carbon firm power source we have at scale.

France's nuclear fleet is the empirical case. France has the lowest per-capita CO2 emissions from electricity in Europe, the lowest consumer electricity prices of any comparable economy, and has avoided the energy security crisis that hit Germany (which closed its remaining reactors) far harder. The French case doesn't require believing in nuclear's future cost competitiveness — the existing fleet is fully amortised and produces some of the cheapest electricity on the European grid.

The LCOE comparison is misleading because it doesn't capture system costs. The LCOE of solar at 5% grid penetration is genuinely cheap; at 80% penetration, the marginal value of additional solar collapses because it all arrives at the same time. The 'true cost' of solar-dominated grids includes the cost of the firm capacity you need to back it up. Nuclear's LCOE looks bad; its value to a high-renewable grid as a firm-power backstop is much higher than the LCOE calculation captures.

The conditional is: if we want net zero by 2050 in systems with limited hydro and interconnection (which describes most of the US, most of India, most of sub-Saharan Africa), nuclear is the most proven large-scale firm zero-carbon option. Advanced reactor designs (SMRs, HTGRs) are in demonstration now. The question isn't whether to rebuild French nuclear in Texas — it's whether to write off the possibility of cheaper, smaller nuclear entirely at a moment when climate timelines are not forgiving of wrong bets.

The cost trajectory is the decisive argument against nuclear centrality. Solar LCOE fell 90% in a decade; wind fell 70%. Nuclear costs have gone up, not down — the Learning curve runs backwards for nuclear. Hinkley Point C went from £18bn to £46bn. Vogtle Units 3 and 4 cost $35bn against an initial $14bn estimate. Meanwhile battery storage costs are following the solar trajectory. Betting on nuclear as a central pillar means betting on a cost trajectory that has never materialised.

SMR optimism has been with us since the early 2000s. NuScale's SMR — the furthest-along project in the US — was just cancelled because costs came in at $89/MWh and rising, against declining renewables plus storage. 'Advanced reactors are coming' has been the response to cost overruns for 20 years. At some point the forecasted nuclear renaissance has to make contact with an actual project that comes in on time and budget. Until it does, treating nuclear as central involves betting on a cost reduction that hasn't happened.

The system cost argument for nuclear assumes we can't build the storage and interconnection to balance high-renewable grids. But storage technology is on a learning curve and interconnection is fundamentally a political/regulatory problem. Investing in nuclear at $10-15bn per GW is locking in a costly solution to a problem that cheaper solutions may solve before the plant comes online (10-15 years from permit to first power). We'd be building plants to solve a problem that doesn't exist when they open.

The France case is for the existing fleet — which I agree should be preserved and life-extended. The debate is about central pillar of new build. South Korea has the best construction track record for new nuclear (APR-1400 came in on time and near budget). But South Korea has a highly centralised industrial base, a small number of utilities, and a deeply experienced construction workforce. These conditions don't transfer to the US, UK, or most European countries attempting their first new plants in decades.