Matthias SpeicherIn short
A petrol station isn’t a multi-layer revenue model at all. It’s a business with two legs: a thin margin on fuel, a healthy one on the shop, and only together do they make the money the oil groups are actually after. A charging park can’t copy that shop. On a greenfield site the frequency isn’t there, and at a motorway services the side business belongs to somebody else. Its money has to come from the electricity itself and from the infrastructure that moves it: trading, grid services, shaving the peak. None of it runs on its own. What decides the return is the operating, not the installing.
The previous part argued that a charging park doesn’t add up if you run the numbers like a petrol station. No shop, no group cross-subsidy, no second leg made of coffee and snacks. It ended on a promise: that electricity carries more than the thin margin on the kilowatt-hour, and that the real revenue layers sit elsewhere. This is what those layers look like.
What the petrol station really earns
Worth one look back, because the contrast does the work. Almost nobody at a petrol station makes money on fuel. In the US, fuel sales account for around two thirds of revenue and still only a little over a third of the profit. Net margin at the pump runs at one to two percent, a few cents a litre, and the card fee takes a bite out of that straight away.
The money is made indoors. Coffee, sandwiches, drinks, cigarettes, plus the car wash, the franchised food counter, the lottery terminal. More than sixty percent of the profit is made in the shop, on categories carrying far fatter margins. Both businesses stand on their own, the fuel side solidly, the shop clearly better. But only together do they turn a decent low-margin operation into the highly profitable one the groups want. Fuel brings the people in, the shop brings the profit. That still isn’t a multi-layer revenue model. It’s two businesses sharing a forecourt and doing each other a favour.
Why the charging park can’t copy that
The second leg, the shop, doesn’t transplant. On a greenfield site the frequency isn’t there. Ten, twenty, fifty sessions a day won’t carry a shop that a motorway petrol station fills effortlessly with a thousand customers.
And where the frequency does exist, at a motorway services, the charge point operator has no access to the money. The food service, the shop, the services building itself usually belong to whoever was there first. The operator gets the parking bay, not the shelf. The revenue from the dwell time flows to someone else, and so even at a high-traffic location the operator is back to selling nothing but electricity. And the space isn’t free. In the best spots, at existing services and truck stops, there’s often more than rent to pay: a share of the charging park’s revenue on top, or a rent coupled to that revenue. Which squeezes the economics one more time.
That reads like a dead end. It’s the point where it gets interesting. Because electricity is a different product from petrol, and a revenue model rests on that difference which the petrol station never structurally had.
The kilowatt-hour is only the surface
Petrol is stock. It gets bought wholesale, it sits in the tank, it gets sold once, then it’s gone. One litre, one transaction, one customer. Electricity moves differently, and the infrastructure a charging park needs anyway, the grid connection, the buffer storage, the site, the screens, can earn in several markets at once. Not the same electricity several times over, that would be nonsense. But the same connection, the same battery, the same site, working in more than one value chain.
The most visible layer is the charging fee, the sale of the kilowatt-hour. It’s also the thinnest. In a market with more than a thousand providers, the price per kilowatt-hour converges on whatever everyone can offer. And there’s a distinction worth holding on to: only the electricity itself is a commodity. The charging session is already a service, one that somebody designs, runs and answers for. Should a pass-through model ever arrive, where the driver buys electricity from their own supplier, that service doesn’t evaporate. It has to be invoiced as what it is, the provision and operation of the charge point. Sell nothing but the kilowatt-hour and you’re competing on price, a race the customer wins and the operator loses.
What the buffer storage makes possible
Underneath the charging fee sits the buffer storage, and it opens several layers at once. The first is arbitrage. Buy electricity when it’s cheap, hold it, sell it on or push it into the charge point when purchasing is expensive. That’s particularly interesting for operators who can’t or won’t commit to large fixed volumes. The big providers mostly hold firm supply contracts. Those give certainty, but they often stop the operator taking profitable advantage of the price signals coming off the exchange.
The second layer is grid services. Storage can offer the network operator balancing capacity, the readiness to absorb or release power within seconds to hold the frequency steady. That pays, in the mature storage markets somewhere in the region of 30’000 to 80’000 dollars per megawatt per year, part of it simply for standing by. And here the charging park holds an advantage: its grid connection is usually strong enough to move serious power in a short window, which is exactly what grid services need. Those figures are approximations, heavily market-dependent, and they shrink as soon as a lot of storage chases the same service. Which is happening now. In the more mature markets the earnings are tipping away from balancing towards arbitrage, because the oversupply is diluting what the grid services pay.
The third layer is the least glamorous and often the one that matters most to a charging park: peak shaving. A fast-charging site draws enormous connection capacity, and it pays demand charges on precisely that, regardless of how much energy actually flows. Storage clips the load peaks, brings down the maximum drawn and cuts the network costs directly. That isn’t extra income, it’s an avoided cost, and at the demand charges of a large charging park it quickly outweighs a good deal of trading revenue. Storage can only ever do one of these at a time, so the control system decides afresh in every interval which layer pays best right now.
None of which makes a battery a self-runner, and nobody should still be under that illusion in 2026. Pure arbitrage operation stopped adding up a while ago, degradation has to be priced in, and it takes a deep feel for power trading and its dynamics. Installing isn’t even half the job. The operating is the business. Then there’s the connection capacity itself. It’s scarce, and plenty of sites never get the power they need because the network is congested where they want to build. At which point storage isn’t a revenue source at all, it’s the precondition for offering high charging power behind a limited connection.
Looking ahead
So far the charging park hasn’t sold a single coffee or run a single advert. It’s already earning in several places, purely off the electricity and the infrastructure that moves it, through the market, through the grid, through the peak it doesn’t draw.
The other half sits above the electricity: generating your own, the dwell time, holding on to the customer, and the layer that gets underrated most consistently, the data. That’s the second part.
Sources
Petrol station economics (revenue and profit split, net fuel margin, share of in-store profit): National Association of Convenience Stores (NACS), convenience.org; Toast POS, pos.toasttab.com; Thunder Said Energy, thundersaidenergy.com. Figures relate to the US market.
Storage revenues, revenue stacking, peak shaving and demand charge management, the shift from balancing services to arbitrage under market saturation: Modo Energy, modoenergy.com; Energy IB, ibinterviewquestions.com; BCLP, bclplaw.com; EticaAG, eticaag.com. Revenue ranges per MW per year are market-dependent approximations and have been trending down.
Margin compression on the charging price and flexibility as a second income stream for charge point operators: Codibly, codibly.com; AMPECO, ampeco.com. The EU framework for flexibility (Network Code on Demand Response, ACER proposal 2025, national implementation expected around 2027) should be checked against the current state before publication.