96% of the sun is about to vanish. Britain's solar roofs will shrug.
On 12 August 2026 the deepest eclipse since 1999 crosses the UK. We computed what it actually costs rooftop solar, roof by roof. The answer is almost nothing, and the roofs that do lose something are the ones the industry spent twenty years writing off.
On Wednesday evening the moon parks itself in front of the sun and takes up to 95.8% of it away from Cornwall. It is the deepest eclipse over the UK since 1999. Nothing this good comes back until 2081.
Two million British homes and businesses now have solar panels on the roof, 22.3 GW of the stuff. So: how much electricity does the country lose when someone switches the sun off?
We ran the numbers on our own roofs model rather than waiting for a press release. The honest answer is almost none, and the reason why turns out to be the most interesting thing about the whole event.
Britain loses about one percent of a day
Here is the bit nobody leads with. The eclipse peaks at around 19:06 in Edinburgh and 19:17 in Penzance. Seven in the evening. In August.
By that hour your panels have been working for roughly fourteen hours and have already banked essentially all of the electricity they were ever going to make that day. The sun at maximum eclipse sits between 10 and 15 degrees above the horizon. It is on its way out. The moon is mugging someone who has already spent their wages.
Run it across a whole day and a typical south-facing UK roof loses somewhere around 0.7% to 1.3% of that day's generation. Not 96%. Under two percent.
Take the sun away at one o'clock in June and we would be writing a very different article. Take 96% of it away at ten past seven in August and you have taken an enormous bite out of a very small meal.
The roofs that actually lose something are the ones nobody rates
This is where it gets good.
At maximum eclipse the sun is not in the south. It is almost exactly due west, about nine degrees off it, low over the horizon. So the question of which roofs get hurt is not "which roofs are best", it is "which roofs are still awake at seven in the evening".
A south-facing roof at that moment is running at roughly 6% to 8% of its midday peak. It has clocked off. There is nothing left to take.
A west-facing roof is still at 34% to 40%. It is still working. It is the only one with anything left to lose.
A west-facing roof loses about five times what its south-facing neighbour loses. In Edinburgh a west roof gives up around 5.3% of the day while the south roof across the road gives up 1.3%.
There is a lesson buried in that, and it is not about eclipses. The reason a west-facing roof has something to lose at seven in the evening is exactly the reason it is worth having: it is still generating when you get home. South roofs peak at lunchtime, when the house is empty and you are exporting to the grid for pennies. West roofs run into the evening, when the kettle, the oven and the car charger are all going.
The industry spent two decades telling people that anything other than south was a consolation prize. It was never true, and on Wednesday the sky makes the point for us.
What the grid does about it
Europe is where the real engineering happens. Grid operators across the continent are braced for a clear-sky drop of around 9.7 GW as the shadow crosses Spain, France and Germany, arriving and leaving inside a couple of hours.
The word that matters there is forecastable. This is not a storm. Every operator has known the exact minute for decades, which is the entire difference between an interesting evening and a problem. The UK's own exposure is mostly second-hand: when continental solar drops, our neighbours keep more of their own power and the cheap imports we usually pull across the interconnectors get thinner.
Your lights will be fine.
The honest caveats
Two of them, stated plainly rather than buried.
First, all of the above assumes clear skies. It is August in Britain. There is a decent chance the cloud gets to your roof before the moon does, in which case the eclipse will cost you even less, because you will have had less to lose.
Second, these are modelled generation figures, not meter readings. We compute the sun and moon positions from standard ephemerides and model what lands on a roof at each angle. The eclipse timings are the solid part and check out against published figures to within a couple of minutes. The generation impact is a good estimate, and we would rather say that than pretend otherwise. How we measure.
When you next get a chance
Wednesday's eclipse starts around 18:02 to 18:20 depending on where you are, peaks between 18:59 and 19:17, and is done by about 20:11. The further south and west you are, the more of the sun disappears and the later it happens.
After this one, the British Isles wait until 3 September 2081 for anything comparable, when up to 99% of the sun goes. For actual totality, with the stars out in the middle of the evening, it is 23 September 2090 over Cornwall.
Most of us are getting one shot at this.
One last thing, and it genuinely matters: there is no safe moment to look at this with the naked eye. The eclipse never reaches totality anywhere in the UK, so the sun is never fully covered. Certified eclipse glasses or a proper solar filter, or watch a livestream. Ninety-six percent of the sun is still the sun.
Enough about the theory.
Check your own roof.
Enter your property number and postcode today, and we'll read your roof from the sky to get
a better understanding of whether or not your roof is suitable for solar panels. Free, takes
minutes, and no account necessary.