Two numbers sit awkwardly side by side. In 2024 Scotland generated about 51.8 TWh of electricity, roughly 1.6 times what it consumed, and sent a net 19.7 TWh south to the rest of the UK — an export the Scottish Government valued at around £1.5 billion. By volume, the "electricity superpower" label is earned. Yet the body that runs the system, the National Energy System Operator (NESO), will not let Scotland's two pivotal power stations, Peterhead and Torness, go offline for maintenance at the same time. A nation that exports more than a third of its output cannot safely lose two machines at once.
That tension is the whole story, and most coverage gets it wrong by collapsing it into a single complaint about an expensive, fragile Scottish grid. There are in fact two distinct problems wearing the same coat. One is a transmission problem — too much wind, not enough wire — which is large, well-documented, temporary and fixable. The other is a system-strength problem — the slow loss of the heavy spinning machines that hold the grid's voltage and frequency steady — which is smaller today, structurally rising, not fixable by wire, and the one that carries the genuine blackout risk. Almost all the money and almost all the attention go to the first. The second is the one to watch — and its shape will be familiar to readers of this site's oil and fuel coverage: a reassuring headline number sitting on top of a structural fault that stays invisible until several things fail at once. This is an electricity story, but it is the same energy-security problem in a different fuel.
The bill everyone quotes
Britain's balancing costs — what NESO pays to keep supply and demand matched in real time — reached about £2.7 billion in the 2024/25 financial year. The largest single driver is transmission constraint: the network physically cannot move all of Scotland's wind south to where the demand sits. Constraint costs alone hit roughly £1.9 billion in 2024/25, about 71% of the total balancing bill, up from 44% just two years earlier. They have nearly trebled since 2020, and NESO's own projection is £4–8 billion a year by 2030 if the network upgrades arrive late.
The mechanism is blunt and expensive. When flows across the Scotland–England interface — the "B6 boundary" — hit their safe limit, NESO pays Scottish wind farms to switch off and pays gas plants in England to switch on in their place. In 2025 that meant roughly £380 million to turn Scottish wind down — estimates vary by source and scope, from about £343 million up — and about £1.08 billion to turn English gas up: close to £1.5 billion to move power the wires could not carry. Northern Scotland alone curtailed about 8.8 TWh of wind that year, meaning only around 61% of the region's possible output actually reached the grid. Scotland's largest offshore farm, Seagreen, delivered only about 30% of its theoretical output, and was paid tens of millions to stand still.
This is a real cost, and a real failure of sequencing — capacity built faster than the network to carry it. But note what kind of problem it is. It is a problem of getting surplus electricity out of Scotland, not of keeping the lights on inside it. And it has a known fix on a known timetable: the Eastern Green Links subsea cables and the B4/B5/B6 reinforcements, all arriving late this decade, are designed precisely to relieve it. The constraint bill is a bridge cost — painful, visible, and shrinking once the wires land.
It is also, crucially, not the thing the energy analyst Kathryn Porter was warning about in Edinburgh this March, even though the two are routinely folded together in the retelling.
The bill nobody quotes
Strip away the congestion, and a second, quieter cost remains — and this one grows as the constraint bill falls.
A power grid needs more than energy. It needs system strength: the inertia and short-circuit power that large synchronous generators provide simply by having massive rotors spinning in lockstep with the network. That spinning mass resists sudden changes in frequency and gives protection systems a strong, clean signal with which to detect and clear faults. Wind turbines and solar farms, connected through inverters, deliver energy but little or none of this stabilising ballast. As the heavy synchronous machines retire, the grid keeps its power but loses its shock absorbers.
In Scotland that ballast is now carried by a startlingly short list. Hunterston B, one of the country's two nuclear stations, closed in January 2022. Torness, the other — about 1.2 GW, generating since 1988 — is scheduled to close around 2028, with EDF reviewing a possible extension towards 2030. The dispatchable backbone otherwise rests on Peterhead's gas plant, itself due to be rebuilt at smaller scale as a carbon-capture unit later this decade. That is the entire synchronous anchor for a grid the size of Scotland's, and it is the reason NESO refuses to let both pillars off at once.
The cost of replacing what those machines do for free is only beginning to appear, which is part of why it escapes scrutiny. NESO's Stability Pathfinder programme has contracted about £323 million for five synchronous condensers and five grid-forming batteries to manufacture short-circuit strength and inertia synthetically in Scotland. A separate inertia-procurement mechanism introduced in early 2024 has saved on the order of £122 million in a single year by buying stability more cleverly — a figure that, read the other way, tells you how much stability now costs to buy at all.
Here is the catch, and it is Porter's strongest point once the rhetoric is set aside. Of the technologies being leaned on, only synchronous condensers and flywheels are physically proven to do this job, because they are mechanically coupled to the grid. Grid-forming inverters and synthetic-inertia systems are promising but, in her words, still amount to "a massive engineering experiment" — they have never been deployed at the scale required to replace multiple large synchronous generators across an entire regional grid. The wires fix the constraint problem on a confident timetable. Nothing fixes the system-strength problem with the same confidence; it is being addressed with a portfolio of part-proven kit and the assumption that it composes.
Figure 1. The transmission-constraint bill (cyan) is real to 2024/25 and follows NESO's own published projection thereafter; its fall after 2030 assumes the network upgrades arrive on schedule. The system-strength bill (amber) is illustrative — no published annual series exists, and it is drawn to convey structure, not to forecast. The point of the chart is the shape, not the year: one bill is a hump that engineering flattens, the other a slope that retirement steepens, and somewhere in the early 2030s the second overtakes the first as the dominant residual cost.
What "didn't work" looks like
The failure mode is not a gradual dimming. It is a discrete event, and the Iberian Peninsula has just handed Europe a fresh anatomy of one.
On 28 April 2025, Spain and Portugal went fully dark; ENTSO-E's final report landed in March 2026. The grid was running on a high renewable share when a disturbance triggered cascading voltage instability. Because most renewable plant there used grid-following inverters, it could not supply the reactive power to arrest the rise, and the few synchronous plants left could not compensate. The system lost synchronism and collapsed, cutting power to around 50 million people, for more than twelve hours in some areas.
Two findings travel directly to Scotland. First, the panel concluded that more inertia alone would not have saved it — the cascade of generator trips collapsed synchronising torque faster than rotating mass could counter. System strength is not just a quantity of spinning metal; it is voltage control, fault response and coordination together. Second, restoration was slowed because black-start capability sat with a handful of conventional synchronous units. A grid that has retired its heavy machines is not only easier to knock over; it is harder to stand back up. ENTSO-E's own chair was careful to name the culprit as "voltage control, regardless of the type of generation" — not renewables as such. That distinction matters, and it is the one a serious argument keeps.
It also reframes the economics. A single dark day in Spain was estimated to have cost between roughly €400 million, on the conservative bank assessments, and €1.6 billion — about 0.1% of GDP — on the main business lobby's figure, with the meat sector alone losing up to €190 million of spoiled stock and one utility booking €135 million in extra costs across the two countries. ENTSO-E notes that retrofitting Spain's inverter fleet for grid-forming capability could run into billions, but stresses that this must be weighed against precisely those losses. Read across to Scotland, the lesson is not that stability spending is expensive. It is that stability spending is cheap insurance against a tail whose single-day cost runs to a year's worth of the spend.
Translated into a severity ladder for Scotland:
- Most likely — the stability tax. Higher balancing costs, more reserve held, more reliance on imports across B6. Money, not darkness. This is near-certain and already underway.
- Under stress — controlled load-shedding. On a windless winter peak with a fault, the system protects itself by disconnecting load: rolling blackouts, managed and temporary.
- The tail — a cascade. The Iberian scenario: a disturbance the thinned system cannot damp, voltage or frequency runs away, regional collapse.
- The tail of the tail — a slow restart. With the synchronous anchors gone, re-energising takes hours rather than minutes. This is where the economic damage concentrates.
Two Scotland-specific aggravators stack on top. The B6 link cuts both ways: Scotland exports south most of the time, but the dangerous condition — cold, still, dark — is exactly when GB-wide margins are tightest and England may have nothing spare to send north. And the anchors are concentrated, not diffuse: this is not a continental grid with dozens of large machines to lean on, but a system already operated on the explicit assumption that both its pillars stay standing.
The policy mismatch
The popular charge — that the Energy Secretary is ignoring all this — is the weakest link in the argument, and worth correcting precisely so the real critique can land.
On new nuclear, Ed Miliband is doing the opposite of ignoring. In October 2025 he asked Great British Energy – Nuclear to assess sites including Torness and Hunterston, and has talked up a "golden age" of nuclear. The binding constraint is not Westminster inattention; it is that planning power over Scottish energy sits at Holyrood, and the SNP government opposes new nuclear, so none will be built in Scotland while it governs. Responsibility for the nuclear veto sits in Edinburgh.
The defensible critique is sharper, and it runs on timescale. Even if every approval were granted tomorrow, a new nuclear station takes well over a decade — it cannot cover the gap that opens when Torness closes around 2028–2030. The pro-nuclear rhetoric, however sincere, does not touch the near-term system-strength hole; and the strategy that does touch it — rapid renewables, leaning on part-proven stability kit, retiring dispatchable plant — is what creates the exposure in the first place. The problem is not that anyone is asleep. It is that the answer on offer is pitched at the wrong horizon, and that both governments can each point at the other while the gap between 2028 and the late 2030s goes unfilled.
What to watch
The honest version of this story is more useful than the alarmist one. Scotland is not running out of electricity; by volume it has a large surplus and a £1.5 billion export business. It is running low on the machines that keep that electricity stable, and it is doing so on a clock set by Torness's closure date.
The constraint bill — the £1.9 billion everyone cites — is a wires problem with a fix in sight. The story underneath it is the system-strength bill: smaller, rising, harder to fix, and attached to a low-probability but high-consequence tail that the Iberian collapse has just made concrete and costable. The risk to track is not any single number but a correlation — a low-wind winter peak, a fault, an unproven stability response and a synchronous-starved restart, all arriving together because they share the same root cause. That correlation is what a naïve reading misses, and it is why the reference event is Iberia 2025, not a simple supply shortfall.
The dates that matter: Torness's confirmed or extended closure (around 2028–2030); the Eastern Green Links and B6 reinforcement commissioning (late this decade); and Coire Glas, the 1.5 GW pumped-storage scheme consented since 2020 but with its investment decision repeatedly slipping and the earliest plausible completion around 2031. Above all, watch whether system-strength procurement scales as fast as the synchronous fleet shrinks. If it lags, the gap is the story — and the bill nobody is quoting becomes the one that matters.
Sources and notes
- Scottish generation, exports and mix (2024): Scottish Government, Energy Statistics for Scotland, Q3 2025 — 51.8 TWh generated, 73.1% renewable, 91.5% low-carbon; net exports 19.7 TWh, indicative value ~£1.5 billion.
- Balancing and constraint costs: NESO, Annual Balancing Costs Report (June 2025) — £2.7 billion total, 2024/25; Modo Energy — constraint £1.9 billion (71% of balancing, up from 44% in 2022/23), £4–8 billion projected by 2030, ~£350 million wind down / >£1 billion gas up in 2025; independent compilations of NESO BOA data — £380 million / £1.08 billion split; Energy Voice / Montel — Scottish curtailment £343 million, Northern Scotland 8.8 TWh / 61% delivered; Seagreen ~30% of theoretical output.
- B6 boundary, Eastern Green Links, network reinforcements: Modo Energy; NESO ETYS.
- Two pivotal stations and the "won't allow both off" constraint: Kathryn Porter, Watt-Logic / Net Zero Watch address, Edinburgh, 9 March 2026.
- Plant specifics: Torness ~1.2 GW, online 1988, closure ~2028 with extension under review (EDF / World Nuclear News / The Ferret); Hunterston B closed January 2022; Peterhead — Peterhead 2 (910 MW CCGT with carbon capture, ~2027) per ClimateXChange.
- System strength: NESO Stability Pathfinder Phase 2 — £323 million, five synchronous condensers and five grid-forming batteries, 11.55 GVA short-circuit level for Scotland and 6.75 GVA·s of inertia for GB, including GB's first grid-forming battery; inertia-procurement mechanism ~£122 million annual saving (NESO, 2025); "massive engineering experiment" — Porter.
- Coire Glas: SSE — up to 1.5 GW / 30 GWh, sub-60-second ramp, consented 2020, investment decision slipped, earliest completion ~2031.
- Iberian blackout: ENTSO-E Expert Panel final report (20 March 2026) — overvoltage-induced cascade, grid-following inverters unable to provide reactive support, inertia alone insufficient, black-start slowed; ~50 million affected, 12+ hours in places. Cost: CaixaBank / Bloomberg ~€400 million; CEOE ~€1.6 billion (~0.1% of GDP); meat sector ~€190 million; Iberdrola ~€135 million combined.
- Politics: New Civil Engineer (October 2025) — Miliband / Great British Energy – Nuclear; SNP planning veto over new nuclear in Scotland.