Export
What sets the price
Every plant offers power at a price and the grid takes the cheapest first. For thermal plant that offer is roughly its fuel-plus-carbon cost — near zero for wind, solar and nuclear, higher for gas. Stack them cheapest-first and the demand line lands on one plant, the marginal unit; in a pay-as-clear market everyone dispatched is paid that one plant's price, and the cheaper plants pocket the difference (their inframarginal rent). So a typical day is still priced by gas whenever the last megawatt needed comes from a CCGT.
The gas fleet isn't one plant — it's a range of efficiencies. Each band's cost is fuel ÷ efficiency, plus carbon, so the most efficient CCGT sets a lower price than the oldest, and a gas-price spike fans the whole staircase apart. Which gas unit is marginal is what moves the price.
Batteries and pumped storage bid their charging cost — what they paid to fill, divided by round-trip efficiency — which sits below gas, so they're dispatched ahead of gas and push it out of the stack. The shaded rent on a battery is the arbitrage spread it earns off the gas-set price, not a fuel margin. This is a snapshot of storage discharging only; the charging leg and its energy limit over a day aren't drawn. A ~1.8-hour fleet can't hold full power for long — which is why a long, still, cold evening empties it and the oldest gas sets the price. The full charge→discharge cycle is a sibling toy's job.
Wind and solar are sized by what's available right now (the "GW offered"), not their full installed capacity — a still night offers little wind whatever the nameplate. Push availability up and demand down with must-run on and the marginal unit bids below zero: subsidised renewables would rather pay to keep generating than lose their per-MWh support, so the clearing price goes negative — a signal to switch off, not a glitch.
Switch World to a 2035 NESO Future Energy Scenario and the whole stack changes. Two new blocks appear. Gas-CCS captures ~90% of its emissions at an efficiency penalty (plus capture and CO₂ transport-and-storage costs), so it's dearer than efficient gas today but undercuts the fleet as carbon rises — drag the carbon slider and watch it slide down through the gas staircase. And hydrogen-fired plant is a fixed ~£200/MWh zero-carbon backstop: on a still, cold evening in a net-zero world the wind has gone and there's little gas left, so hydrogen — not cheap gas — sets the price. A net-zero grid can even run short of firm capacity on the stillest evenings, which is why it builds hydrogen and storage.
Illustrative structural figures (installed capacities from DUKES/NESO; gas and carbon are representative teaching values) — no live data, ever. The FES worlds are NESO's illustrative 2035 pathways, revised annually — not forecasts, and contested. One uniform-price, day-ahead-style snapshot, not GB's real-time balancing mechanism: flat marginal offers (no unit-commitment or start-up costs), no strategic bidding, imports priced at an assumed neighbour price that can undercut the most efficient gas, no network constraints, and marginal ≠ average carbon. Sibling toys extend the same engine — network constraints turn one price into locational prices, an interconnector couples two markets, the marginal plant's carbon is the carbon of the next megawatt, a price cap opens the capacity-market "missing money" gap, and the storage arbitrage cycle gets its own sketchpad.