Great Britain · 206 sites · 7,779 MW
The UK BESS fleet, what it earns and how
Every grid scale battery in Great Britain that was paid by the grid operator in the 364 days to 27 September 2026. That is 206 sites holding 7,779 MW, drawn as 250 circles because a large site trades as several registered units, and it is very nearly the whole working fleet.
Click a circle for one site. The bigger the circle the more megawatts it can push. The darker the colour the more it earned for each kilowatt it has.
Click a circle for one site. The bigger the circle the more megawatts it can push. The darker the colour the more it earned for each kilowatt it has.
Reading the fleet.
What the fleet earned, and what it paid to earn it
£134.8mpaid to discharge
−£109.9mpaid out to charge
£24.9mbalancing, net
£131.3mstanding ready
£61.1mcapacity market
£217.3mthree measured lines
£98.7m balancing and outside the balancing mechanism together, derived from the meters of 150 units: the traded position at the market index, the energy moved under frequency response contracts at the imbalance price; a further £9.8m estimated for 76 unmetered units from their metered peers, and none for 8 held in response most of the time.
A model, not a measurement. Never added to the three lines above. How it is worked out
Over the 364 days to 27 September 2026, across 206 sites holding 7,779 MW. None of it is profit.
The fleet total above is net. A single site on the map is not. A battery that buys energy through a balancing bid sells it in the wholesale market, and nothing here measures that sale, so subtracting the purchase from the balancing offers alone leaves a cost with no matching revenue. Done that way, twenty one sites read as loss making, nineteen of them Scottish and most of them constraint management assets doing exactly what they were built for. Blackhillock paid £4.7m to charge and took £0.3m in balancing offers, because it buys trapped wind in the north and sells it later somewhere this page cannot see. So each site is coloured by income, with what it paid to charge shown beside it rather than taken off it.
The fleet total above is net. A single site on the map is not. A battery that buys energy through a balancing bid sells it in the wholesale market, and nothing here measures that sale, so subtracting the purchase from the balancing offers alone leaves a cost with no matching revenue. Done that way, twenty one sites read as loss making, nineteen of them Scottish and most of them constraint management assets doing exactly what they were built for. Blackhillock paid £4.7m to charge and took £0.3m in balancing offers, because it buys trapped wind in the north and sells it later somewhere this page cannot see. So each site is coloured by income, with what it paid to charge shown beside it rather than taken off it.
A battery earns from four places. Three of them are public
Balancing. The grid operator pays a battery to change what it is doing at short notice, to keep supply and demand matched minute by minute. It works both ways: the fleet was paid £134.8m to discharge and paid out £109.9m to charge, so the balancing line is £24.9m net. Public, per site, and counted here net rather than gross, because the gross figure counts money that went straight back out to buy the energy.
Standing ready. It also pays a battery simply to be available to respond, whether or not it is ever called. Public, per site, and counted here. The figure is what the published auction results imply, award quantity times clearing price times the length of the window, rather than a settled payment, and where a service clears below zero it is a cost rather than income.
The capacity market. A separate auction that pays plant to be there in a shortage, agreed years ahead. Public, per agreement, and now counted here. Together these three came to £217.3m over the 364 days, of which £61.1m is the capacity market across 152 sites.
Wholesale arbitrage. Buying low and selling high. A battery makes money charging when power is cheap and discharging when it is dear. For many sites this is the largest of the four. Nobody publishes it site by site, and until September 2026 we said it could only be modelled. It can be derived: Elexon publishes every unit’s metered output, and meter minus accepted balancing volumes is the position a unit ran outside the balancing mechanism. Part of that position is not trading: a battery holding a frequency response contract moves energy the grid frequency asks of it, and that energy is settled at the imbalance price, not traded at the index. So the expected response energy, contract times measured frequency, is taken out and valued at the imbalance price, and what remains, the traded position, is valued at the market index. Each asset page carries both beside the measured lines. How, and what is wrong with it. For the 150 units with a published meter: £96.6m traded position at the index over the year, −£24.9m for the response energy at the imbalance price across 129 units, £98.7m with the balancing legs. The 76 auction units without a meter carry an estimate from metered peers of the same duration and region, £9.8m at the median, and it is labelled as such. 8 more are paid so much in the response and reserve auctions that they are held in those services most of the time, and carry no estimate: the metered batteries that work that way make nothing from trading.
The capacity market also shows us 36 batteries holding 428 MW of de-rated agreements that neither the balancing data nor the reserve auctions can see at all. They are not on this map, because nothing else about them is published. How this is worked out
There is a second gap, and it is ours. A battery whose balancing registration is held by an aggregator rather than by the site sits in the register under a virtual unit code, declares no capacity of its own and is named after the place rather than the technology. Our fleet does not find those, and £6.8m of balancing offers sat in 48 such units over this window. Not all of them are batteries. We are working through them.
Standing ready. It also pays a battery simply to be available to respond, whether or not it is ever called. Public, per site, and counted here. The figure is what the published auction results imply, award quantity times clearing price times the length of the window, rather than a settled payment, and where a service clears below zero it is a cost rather than income.
The capacity market. A separate auction that pays plant to be there in a shortage, agreed years ahead. Public, per agreement, and now counted here. Together these three came to £217.3m over the 364 days, of which £61.1m is the capacity market across 152 sites.
Wholesale arbitrage. Buying low and selling high. A battery makes money charging when power is cheap and discharging when it is dear. For many sites this is the largest of the four. Nobody publishes it site by site, and until September 2026 we said it could only be modelled. It can be derived: Elexon publishes every unit’s metered output, and meter minus accepted balancing volumes is the position a unit ran outside the balancing mechanism. Part of that position is not trading: a battery holding a frequency response contract moves energy the grid frequency asks of it, and that energy is settled at the imbalance price, not traded at the index. So the expected response energy, contract times measured frequency, is taken out and valued at the imbalance price, and what remains, the traded position, is valued at the market index. Each asset page carries both beside the measured lines. How, and what is wrong with it. For the 150 units with a published meter: £96.6m traded position at the index over the year, −£24.9m for the response energy at the imbalance price across 129 units, £98.7m with the balancing legs. The 76 auction units without a meter carry an estimate from metered peers of the same duration and region, £9.8m at the median, and it is labelled as such. 8 more are paid so much in the response and reserve auctions that they are held in those services most of the time, and carry no estimate: the metered batteries that work that way make nothing from trading.
The capacity market also shows us 36 batteries holding 428 MW of de-rated agreements that neither the balancing data nor the reserve auctions can see at all. They are not on this map, because nothing else about them is published. How this is worked out
There is a second gap, and it is ours. A battery whose balancing registration is held by an aggregator rather than by the site sits in the register under a virtual unit code, declares no capacity of its own and is named after the place rather than the technology. Our fleet does not find those, and £6.8m of balancing offers sat in 48 such units over this window. Not all of them are batteries. We are working through them.
Who owns a battery and who trades it are usually different companies
The market registers name whoever sells into the market. For a battery that is normally an optimiser working under contract, not the owner. We can name the owner for 154 of 250 registered units, and for 123 of them the trader is a different company altogether. Twelve companies trade four or more separate sites between them, which is what an optimiser looks like in the data. A single site company that appears once is what an owner looks like. Both are on every popup.
Every battery has its own page, 206 of them, listed in one place. Each has a permanent address you can send to somebody.
Earnings per kilowatt, from the three public sources
Per kilowatt, so a 20 MW site and a 100 MW site can be compared. None of this is profit. It is income from three of the four routes, before the cost of buying the power, and before anything the site earns from trading.
A handful of sites show a negative figure. That is real rather than a mistake. The auctions to stand ready sometimes clear below zero, and a provider then pays for the right to hold the position.
A handful of sites show a negative figure. That is real rather than a mistake. The auctions to stand ready sometimes clear below zero, and a provider then pays for the right to hold the position.
What the fleet earned from balancing, month by month
Balancing only. Shown as an annual rate per kilowatt so that a part month is not compared with a full one, and part months are drawn paler.
A battery in Scotland charges eleven times as often as it discharges
Every site, ranked by what it earned
Does this cover the whole fleet
Very nearly, and here is the arithmetic rather than a claim.
How we know where each site is
No warranty. Elexon and NESO publish this data as is. Three of the four revenue lines are here: the balancing mechanism, the response and reserve auctions, and the capacity market. The fourth is wholesale trading, and nobody publishes that for individual assets, so no total on this page is an asset's full income. Capacity market income is the cleared price raised for inflation where the auction set a base period, split by month on EMRS's published weights. Registered capacity is Elexon's generation capacity field, is not de-rated and says nothing about duration. Live positions are physical notifications, which are a statement of intent published ahead of time, not metered output.