What a flat blended electricity rate hides

Almost every quick Supercharger pro forma prices electricity the same way: one blended number, usually somewhere around $0.30 per kWh, multiplied by annual throughput. It is an understandable simplification. Utility tariffs are long, and a single rate makes the arithmetic fit on a napkin.

It is also the assumption most likely to be wrong by a margin that changes the answer. Here is what the filed tariffs say when the same site is priced in every territory in the model's library.

The reference site#

To isolate the tariff, everything else is held constant: eight V4 stalls at 235 kWh per stall per day — the model's default ramp at month six — which is about 57,200 kWh dispensed per month. Energy is priced at the model's default time-of-use mix, grossed up 12% for charging losses, and the territory's filed demand charge is added. The total divided by kWh actually dispensed gives an all-in effective cost per kWh sold.

Only the tariff changes between rows. Every other input is identical.

The spread is 8.5×#

The cheapest and most expensive filed tariffs in the library are a factor of 8.5 apart for the same physical site delivering the same energy. That is not a modelling artifact; it is what the filings say.

Three consequences follow directly.

A $0.30 assumption sits below the median. The library's median effective cost is $0.345/kWh, and roughly three in five filed tariffs price above $0.30. A napkin figure of $0.30 is therefore not a neutral middle estimate — for most territories it is optimistic.

The dollar error is large. For this site, dispensing about 686,000 kWh a year, the gap between a flat $0.30 assumption and the actual filed tariff runs from about $121,000 too high in the cheapest territory to about $519,000 too low in the most expensive one. On a project whose all-in build cost is near $1.26M, a half-million-dollar annual error in a single operating line is decisive.

The error is not symmetric. Because the expensive tail is far longer than the cheap tail, a flat assumption understates cost more often and by more than it overstates it.

Why the variance is so large#

Filed energy rates differ several-fold between utilities, but energy is not where most of the spread comes from. The demand charge is.

A Supercharger draws enormous power in short bursts. Utilities bill that capacity separately from consumption, and they file that second charge in structurally different shapes — subscription blocks, a flat $/kW on monthly peak, tiers keyed to load factor, or a per-kW rate capped by throughput. Because the same physical site produces the same peak in every territory, a territory's demand structure lands almost entirely on the bottom line.

In the library, the demand component exceeds 40% of the total electricity bill in 67 of the 96 priced tariffs — most of them. In the most expensive territories it exceeds 60%. And in 17 tariffs it is zero, because those utilities either file no demand charge for this class or have suspended it.

A blended $/kWh cannot express any of that. It collapses a cost driven by peak kW into a cost expressed per kWh, and the ratio between those two quantities is exactly what varies most between sites. Two sites dispensing identical annual energy on identical hardware can face demand bills an order of magnitude apart, and no single blended rate will describe both.

The load-factor trap#

There is a second-order effect worth naming, because it inverts the intuition that more utilization is always better on the cost side.

Demand charges are driven by peak kW; energy charges by kWh. A lightly used site has a low load factor — a big peak spread over few kilowatt-hours — so its demand charge divided across each kWh sold is very high. As utilization rises, the same demand charge spreads over more energy and the effective cost per kWh falls, sometimes steeply.

This means the effective $/kWh is not a property of the tariff alone. It is a property of the tariff and the utilization, and a rate that looks unaffordable at 15% utilization can be workable at 35%. Some utilities have written this relationship into their filings directly: the Massachusetts EV rates tier the demand charge by annual load factor, and Florida Power & Light's GSLD-1EV caps billed kW at kWh purchased divided by 75 hours. Both are explicit attempts to keep early-life charging sites from being priced out by a structure designed for steady industrial load.

A flat blended rate hides this entirely, which matters most in exactly the years when a new site is ramping and least able to absorb a surprise.

What the model does instead#

Rather than a blended rate, the engine carries the filed structure for each territory: the time-of-use energy rates, the demand structure in its own shape, the loss factor between metered and dispensed kWh, and annual escalation on both energy and capacity. It re-derives the demand line every month of year one and every year after, so the load-factor effect appears in the cash flow rather than being assumed away.

The utility rankings tool publishes the full ordering for every filed tariff, the demand-charge explorer shows the structure behind any single territory, and demand charges, explained for Supercharger hosts covers the mechanics in more depth. Every rate carries its filing source and effective date on the methodology page.

Figures here are model outputs for one reference site under stated assumptions, not quotes or advice. Filed tariffs change; verify current rates against the cited filings before relying on any number.