Breakeven utilization: the bar at every retail price

Every charging site has a number that decides it, and the number is not the retail price or the tariff or the capex. It is how many kilowatt-hours each stall dispenses on an average day. Price and tariff and capex set where the bar sits; utilization is what either clears it or does not.

This post puts the bar in one place. For the engine's default eight-stall site on PG&E's BEV-2 rate, with $1.21M of capex financed at 20% down, here is the utilization at which the fifteen-year NPV reaches zero, at every retail price from forty cents to sixty-five.

To isolate the level from the shape, utilization is held flat from month one and the engine's year-over-year growth is switched off. That is not how real sites behave, and the second half of the post puts the ramp back. But a flat level is the clearest way to state what the site has to average over its life.

The bar at $0.45#

NPV crosses zero between 280 and 320. Solving exactly: 308.9 kWh per stall per day, or 902,069 kWh a year across eight stalls. Below that level the site destroys value at a 10% discount rate; above it, each additional 40 kWh per stall per day is worth almost exactly $245,000 of NPV, which works out to about $6,100 per kilowatt-hour-per-stall-per-day. The relationship is close to a straight line because the site's costs barely move: the $27,521 demand subscription is fixed, and the energy charge rises in lockstep with the volume that pays for it.

Note the effective electricity column. The same tariff costs 34.4 cents a kWh at 100 and 26.4 cents at 490, because the fixed demand line is spread over more kilowatt-hours. A low-utilization site pays the most per unit for its power at exactly the moment it can least afford to. What a flat blended rate hides covers that mechanism in detail.

The ladder#

Now change the price and solve again:

At forty cents the site has to run within 8% of what the hardware can physically deliver, every day of the year, just to reach zero. At sixty cents it needs a third of that. Each nickel of price takes a large bite out of the bar: from $0.45 to $0.50 the requirement falls by 74 kWh per stall per day, nearly a quarter.

That is the same asymmetry the two-lever post found from the other direction. Price lands entirely as margin; volume brings its electricity bill and Tesla's network fee with it. So price moves the bar further than a matching amount of volume moves the site.

Putting the ramp back#

Real sites do not open at their steady level. On the engine's Standard ramp the site averages 209 kWh per stall per day in year one, finishes December at 260, and then grows 5% a year to year seven and 3% after that, capped at 490. Against that moving target, the price ladder reads:

The engine also reverse-solves the ramp version of the bar: the year-two utilization at which its internal viability test, a composite of payback, cash position and IRR, stops flagging the site. At $0.45 that is 281 kWh per stall per day; at $0.50, 246; at $0.55, 233; at $0.60, 222; at $0.65, 213. The Standard curve delivers 227 in year two, so at $0.45 the site is short of the bar and at $0.55 it is over it.

Reading the bar for a real address#

The numbers above are one site on one tariff. Three things move them for any other address, and all three come from documents rather than guesses.

The tariff moves the bar most. Under a rate with a heavy demand charge the fixed line is larger, which raises the breakeven level and steepens the effective-cost curve at low volume; under a demand-free EV rate the bar drops. The utility rankings order all 96 rows by effective cost at a reference profile, and the demand-charge tool shows the fixed line for a given subscription.

Capex moves it next. The bar is where fifteen years of margin repays the build; a $900,000 site clears at a lower level than a $1.21M one. The build-cost post sets out where the dollars go.

And the price is the operator's own choice, which is why it is the axis of this post rather than a footnote. What the ladder says is that the question "will enough cars come?" has a different answer at every price, and that the answer at $0.60 is roughly half as demanding as the answer at $0.45. Whether a given corner of a given town delivers 158 or 309 is the diligence the scenario wizard is built around; it computes the bar for the address and tariff entered, and the methodology page documents the growth and cap rules used above.

Engine outputs are model projections under the stated assumptions, not advice or guarantees. Flat-utilization rows hold volume constant for fifteen years with growth disabled, which is a device for locating the bar rather than a forecast of how any site behaves.