The cheapest energy to use is now at lunchtime. Let me explain.

We have all heard it: use energy at night, it’s cheaper. It’s the logic behind Economy 7, which hands you a block of cheap hours overnight, around seven of them, often something like half past midnight to half past seven, though the exact window varies by region. It’s why, if you drive an electric car, your supplier offers a special overnight rate to charge it while you sleep.

And at the level of the tariff, that holds up. The overnight rate genuinely is lower, and if you can push your heaviest use into those hours, you pay less. The dear window is the weekday evening, when everyone gets home and switches on at once. None of that is in dispute.

But step back from the tariff to what sits underneath it: the actual cost of using the grid. That’s two things stacked together, the wholesale energy, and the network charge that carries it down the wires. Pull them apart, and the neat night is cheap story stops matching up.

It turned up in the middle of our own analysis, in the raw numbers behind the market. Every half-hour of every day, Elexon, the organisation that keeps the books for Britain’s electricity system, records the wholesale price of power: the going rate for a unit of electricity at that exact point in the day. We pulled years of it.

What we found surprised us. The cheapest energy of the day is no longer the small hours. It’s the middle of the day. The night still works out cheapest overall, but once you separate the energy from the network charge, that isn’t because the energy is cheap. Let me explain.

The rule is baked into the system

Start with how deep the assumption runs. The day/night split that sits underneath the regulated price cap is 58% day, 42% night. That number was derived from 2017 settlement data, and Ofgem re-confirmed it, unchanged, in its decision of 27 May 2026. It is parked until the market-wide half-hourly settlement programme is far enough along to revisit it, with no date attached.

It is worth being precise about what that split is and is not: it is the consumption split Ofgem feeds into the price cap as its benchmark, not a measurement of how any individual household actually splits its use. But it tells you how the system thinks. The shape of demand it assumes is, in effect, frozen at 2017. The world it describes is eight years old and counting.

One aside on how blunt a single national number is: that same off-peak share runs from about 33% in the East Midlands to about 54% in the South West. "The night" is not one thing across the country.

What the energy actually costs now

Start with the raw energy on its own. On that measure the night is still the cheaper place to be, just. But the gap has been shrinking fast. Back in 2017, the overnight hours came in around 21% below the daytime average. This year, it’s down to about 8%.

And this is not just the post-crisis spike unwinding. Even the calmer pre-pandemic years, 2018 and 2019, sat around 17 to 18% on the same like-for-like measure. So 2026’s 8.2% is well below the historical baseline, not a reversion to it.

“

The night discount is collapsing.

And the energy crown has already changed hands. In 2017, midday power cost more than night power (4.32 versus 3.79 p/kWh). By 2025 that had flipped: midday energy averaged 7.84 p/kWh against the night’s 8.06. The middle of the day is now, on the raw energy, the cheapest window there is.

So on energy alone, the advice has it backwards. If you wanted the cheapest electricity of the day in 2025, you would not have set your timer for 3am. You would have run your load at noon.

So why does the night still win?

Because energy is only half of what it costs to use the grid. The other half is the network charge, the DUoS, levied by the 14 regional distribution companies. It is banded by time of day: a rock-bottom "green" rate overnight, a middling "amber" rate across the day, and a punishing "red" rate in the weekday evening peak.

The gap between those bands is enormous, and it is widening. Averaged across all 14 networks for a domestic connection, the overnight green rate is about 0.24 p/kWh. The evening red rate is 14.29 p/kWh. The overnight charge is under 2% of the peak charge. And the red rate is up almost 60% across the networks’ published charging years, from 2022-23 through to 2027-28 (the later years are rates the DNOs have already published, not our forecast).

Now put the two halves together. Take 2026 energy and 2026-27 network charges and add them up by time of day:

There it is. The night is still the cheapest total, but look at why. It is not the energy: midday energy is cheaper. The night wins entirely on the network charge. Strip the green DUoS rate out and midday is the cheaper hour.

And look at the evening peak. Moving a unit of demand off the weekday peak and into the night saves about 16 p/kWh. More than 85% of that saving is the network charge, not the energy. "Use power at night" has quietly become "avoid the evening network peak." The signal we have trained households on for decades is now carried, almost entirely, by one regulated line item.

And to be fair to the charge, that is exactly what it is for. The rock-bottom overnight rate is deliberate, set to pull demand off the strained evening peak. The signal is working as designed. It is just that it is now doing the work that genuinely cheap energy used to.

They were never the same clock

For years, the two signals agreed. Wholesale energy and the network charge both pointed at the same answer: the night is cheapest. That is why the rule held. What changed is that, around 2025, they stopped agreeing. Line up the raw energy price for the night against the middle of the day, year by year, and you can see the moment the crown changed hands.

They were never really the same signal. The wholesale price tracks the national balance of supply and demand, and it re-prices every half-hour, so when daytime solar floods the system the midday price falls almost at once. The network charge measures something different: the strain on the local wires. Its time bands are set from historical peak data on a slow clock that, by the rules, can only change on 1 April and needs 15 months’ notice. One signal moved with the solar. The other could not. They lined up for years only because the national demand peak and the local network peak both happened to fall in the weekday evening. In 2025 that coincidence broke.

Our read on why

We can show what changed. We are more careful about why, because the honest answer is that our data does not contain it.

Our strong suspicion, and we will call it what it is, a suspicion, is that this is the daytime renewables story finally showing up in the price. More solar and wind on the system in the middle of the day pushes midday wholesale prices down. It is the most natural explanation for midday energy falling below the night. But we hold no generation-mix data, so we are not going to dress a hypothesis up as a finding. If you have the generation data to confirm or kill it, we would genuinely like to see it.

What we think it means

A few things follow, and these are our views rather than numbers.

The cheap, low-carbon energy in the middle of the day is being under-used, because the pricing signal everyone responds to still points at the night. We are steering every new flexible load we have, the EVs, the heat pumps, the home batteries, towards the small hours, on the strength of a signal that is now mostly one network charge rather than genuinely cheap energy. Do that hard enough and for long enough and you risk manufacturing a new peak overnight, in the very window we are all rushing into.

The clear loser in all of this is the weekday evening. Both halves of the cost, energy and network, are highest there, and the network half is rising fastest. That is the window worth designing out of, and it is the one the old "night is cheap" framing barely mentions.

What could fix it is already being built. The country is moving, meter by meter, to settle everyone’s electricity use half-hour by half-hour. That is the machinery you would need to price the genuinely cheapest window of the day, energy and network together, instead of a fixed "night" band set to a 2017 view of demand. The technology exists. The signal has not caught up.

One last guardrail, because it matters: everything above is wholesale market and network cost, not what any household or business actually pays. We have looked at only two layers, the wholesale energy and the distribution (DUoS) charge that carries it down the local wires. Transmission, balancing, policy levies, the supplier’s own costs and VAT all sit on top, and we have left them out, because it is in these two layers that the night-versus-day story really lives. The point is simply this: that story has changed, while the advice built on it has not.