Axel Heilborn

Swedish chargers were observed available 95.6 percent of the time in Q2 2026. Only 87.4 percent of sessions succeeded. Uptime measures hardware, not outcomes, and the 8.2 point gap between them is where the money leaks. This model decomposes the failed 12.6 percent by cause, prices it, and solves the question operators avoid: when the backend is unreachable, should the charger start the session anyway?

The answer is a closed-form breakeven that is independent of volume, which is why it transfers between operators. Degraded-mode authorization stays profitable until roughly four in ten offline sessions go unpaid. Real rates are nowhere close. The feature is switched off across most of the industry because unsettled sessions appear in the accounts and walked-off drivers do not.

Open the model

Sportlov Saturday, Stockholm to Åre. Set the temperature anywhere from plus twenty to minus twenty-five and it moves your range, your charge speed and whether preconditioning is worth anything. Fifty-nine other cars left the same morning, each deciding exactly what you are deciding.

Charge to 90 everywhere and you finish around fortieth, because the taper holds a stall you barely use. Charge as little as possible and you finish tenth, because you arrive empty at the 75 kW site. The sites run 50, 75, 150 and 250 kW, and a stall or two is out of service somewhere different every run.

Drive it

You are a grid battery. Twice in the next minute the largest unit on the Nordic system trips and the frequency falls. Hold it inside the band using two keys. Sixty seconds.

Do nothing and you get thirty-one seconds of load shedding. A sharp human holds the band about 57 percent of the time. The automated version manages 95. That gap is what the ancillary service market is actually paying for, and it is why response speed, not capacity, is the thing being bought.

Try to hold it

Six million kronor and a 400 kW grid connection. Buy chargers, buy storage, or spend it upgrading the connection, then watch a full day of traffic arrive and see who you turn away.

Of the 68 legal builds, 45 lose money. The ones that win are not the ones with the most hardware. Overbuild the chargers and you own equipment your connection can never feed; upgrade the connection and it eats the budget you needed for chargers. Takes about a minute.

Play it

From September 2026 the Nordic TSOs cap how much FCR-D upward may be procured from static resources, at half the requirement and less when system inertia is low. Enforced at bid selection, with accepted static bids still paid the same marginal price.

It will be read as a tightening, and someone will mark up a revenue assumption on it. That confuses access with scarcity. The volume procured does not change, only who supplies it, and the group receiving the freed volume already oversupplies the requirement fivefold. With days to go, the split between TSOs and the shape of the inertia component are both still unpublished.

Read the analysis

Swedish FCR-D demand is dimensioned against the largest unit that can trip, currently Oskarshamn 3 at 1 450 MW. Svenska kraftnät forecasts no change to that reference incident before 2030. The demand line is therefore flat by construction and indifferent to electrification, to price, and to how many batteries get built.

Against that, prequalified battery capacity alone has reached roughly five times the entire national requirement. The market reads falling prices as saturation, which implies a cycle. The dimensioning method says it is division, which does not reverse. Any BESS model carrying an ancillary revenue line flat past 2027 is wrong by construction rather than by forecast error.

Read the analysis

Sweden added 12.7 percent more charging stations between Q1 and Q2 2026 and session success fell 2.4 points over the same period. Norway grew a third as fast and improved slightly. Build rate and service quality are moving in opposite directions in one market and not the other.

Working out whether that is a commissioning-lag artefact, where new sites underperform until they settle, or something structural about how fast a portfolio can grow before operations stop keeping up. The answer matters for anyone underwriting a buildout plan on assumed uptime.