What does electricity cost to make?

Split a plant’s costs, discount the future and build a levelized cost

Lesson 7 of Energy Systems, Step by Step: split a power plant’s cost into building, upkeep and fuel, see why money later is worth less, build a levelized cost of electricity (LCOE) from EIA and ATB inputs, rank what moves it with a tornado chart, and find what LCOE leaves out: when the energy arrives.
  • Analyst · L2
  • Lesson 7 of 14
  • 15 min
  • After lesson 5

Illustrative LCOEs from EIA plant assumptions and U.S. fleet averages (real 2025 dollars, $/MWh)

Plant Building the plant Fixed upkeep Fuel Upkeep per MWh LCOE
Onshore wind farm 33.44 11.68 0.00 0.00 45.11
Solar farm 40.82 11.23 0.00 0.00 52.05
Gas combined-cycle plant 11.61 2.47 23.97 3.57 41.61
Gas peaker 56.21 6.03 35.20 4.17 101.60

With JavaScript on, you work through this in eight steps: predict, split the cost, discount, blend a WACC, find the price that pays back, use the calculator, read a tornado chart, and see when the energy arrives.

EIA and ATB inputs in real 2025 dollars, with no taxes, credits or grid costs, so not the agencies' own LCOEs; Prairie Town's plants are illustrative. Sources and dates: where the inputs come from.

Need a refresher? Toolbox K1 ($ per kW against $ per MWh) and K2 (fuel cost per MWh from a heat rate).

What you just did, as diagrams

Four costs
Building the plant, fixed upkeep, fuel and upkeep per MWh; wind and solar are mostly building cost, gas plants mostly fuel.
Discounting
Money later is worth less than money now, so future costs and sales are divided by (1 + r) for every year of waiting.
WACC
The blended rate a project pays its lenders and owners; it is the discount rate of the plant.
Capital recovery factor
The share of the building cost to repay each year so that it is paid off, with interest, over the plant's life.
LCOE
The steady price per MWh that repays every cost over the life: yearly costs divided by yearly MWh, plus fuel and upkeep per MWh.
Tornado chart
Each input swung from low to high, one at a time, to rank what moves the result over the chosen ranges.
LCOE's blind spot
It gives every MWh the same worth, whatever the hour; what one more MWh saves depends on the hour it arrives.
Go deeper · AnalystThe LCOE formula and the capital recovery factor

Paying off a building cost \(C\) ($/kW) with \(n\) equal yearly payments \(A\) at a real rate \(r\) means the payments’ present value equals \(C\):

\[C = \sum_{t=1}^{n} \frac{A}{(1+r)^t} = A\,\frac{1-(1+r)^{-n}}{r},\]

\[\text{so}\quad A = \frac{r}{1-(1+r)^{-n}}\, C = \mathrm{CRF}\cdot C .\]

When \(r = 0\) the capital recovery factor is simply \(1/n\).

Need a refresher? Toolbox K3 (discounting and the capital recovery factor).

The LCOE is the steady price at which the plant’s discounted sales repay everything, so its net present value is zero.

Each kW makes \(E = \mathrm{CF} \times 8{,}760 / 1{,}000\) MWh a year, with the capacity factor (lesson 5) as a fraction: 33.8% is 0.338.

\[\mathrm{LCOE} = \frac{\mathrm{CRF}\cdot C + \mathrm{FOM}}{E} + \text{fuel} + \mathrm{VOM}\]

Here fuel per MWh = heat rate (MMBtu per MWh) × gas price ($ per MMBtu).

Check it on the wind farm:

  • \(\mathrm{CRF}(4\%, 30) = 0.0578301\);
  • yearly cost: \(0.0578301 \times 1{,}712 + 34.57 = 133.58\) $/kW;
  • yearly output: \(E = 0.338 \times 8.76 = 2.961\) MWh per kW;
  • LCOE: \(133.58 / 2.961 = \$45.11\) per MWh.

The 2025 ATB writes the same equation with a fixed charge rate, \(\mathrm{FCR} = \mathrm{CRF} \times\) a factor for taxes and depreciation; here that factor is 1.

The WACC with taxes: loans’ share \(d\) at rate \(r_d\), owners’ money at \(r_e\) and tax rate \(\tau\) give \(d\,r_d(1-\tau) + (1-d)\,r_e\). Step 4 left the tax out.

A nominal rate \(n\) becomes real with \((1+n) = (1+r)(1+i)\), for inflation \(i\).

Go deeper · ModelerWhere LCOE's pieces go in a planning model

A capacity-expansion model can account for these costs directly, instead of comparing LCOEs.

  • It charges each kW it builds a yearly \(\mathrm{CRF}\cdot C + \mathrm{FOM}\), and every MWh its fuel and upkeep.

  • It chooses what to build and how to run it, hour by hour, in one optimization, so step 8’s timing problem is inside the model.

  • Capacity expansion with representative periods (M3, in the Modeler tier) uses exactly this yearly charge.

  • Techno-economic analysis inside an optimization model (M11) adds taxes, credits and a boundary choice.

  • Year-by-year cash flows, NPV, IRR and payback for one project are Analyst lesson A1.

Go deeperWhat this calculator leaves out
  • Taxes, depreciation and tax credits: the agencies include them, so their LCOEs differ from these.
  • Interest paid during construction, and the time a plant takes to build.
  • Grid connection, new transmission, and the backup or balancing a variable plant may need.
  • Output that falls as equipment ages, and costs that differ by region.
  • Any change in costs over time: every input is one fixed number that you can move.
  • Step 8 values each MWh at its hour's running cost of the last plant needed. A plant's worth also includes the capacity it offers at the peak.
  • That hourly value is for one more MWh, not what a whole farm saves: growing solar from 200 to 400 MW cuts the day's running cost by $56,460, or $31.38 per added MWh.
  • LCOE here is the cost per MWh a plant can make. When some of it is spilled, the cost per MWh actually used is higher.

Check yourself

Myth or fact? “The plant with the lowest LCOE is always the cheapest choice for the grid.”

  • Myth
  • Fact
Answer

Myth. LCOE leaves out when and where the MWh arrive: on Prairie Town's summer peak-demand day, a wind MWh was worth $56.35 on average and a solar MWh $50.27.

Myth or fact? “The discount rate is a technical detail that barely changes the answer.”

  • Myth
  • Fact
Answer

Myth. From 4% to 7% real, the wind farm's LCOE rises from $45.11 to $58.27 per MWh, by 29.2%.

Myth or fact? “A tornado chart shows how likely each outcome is.”

  • Myth
  • Fact
Answer

Myth. It shows how far the result moves over ranges someone chose, one input at a time; other ranges can change the ranking.

Myth or fact? “With the same building cost, a windier site gives a lower LCOE.”

  • Myth
  • Fact
Answer

Fact. The same yearly cost is spread over more MWh: at a 43% capacity factor instead of 33.8%, wind's LCOE falls from $45.11 to $35.46.

A plant's yearly repayment plus fixed upkeep is $130 per kW, and each kW makes 2.628 MWh a year (a 30% capacity factor). With no fuel and zero upkeep per MWh, what is its LCOE?

  • $13.00/MWh
  • $39.00/MWh
  • $49.47/MWh
  • $130.00/MWh
Answer

$49.47/MWh: 130 ÷ 2.628 = 49.47. Yearly cost per kW divided by yearly MWh per kW.

At 7% a year, what is $100 paid in 10 years worth today?

  • $93.00
  • $70.00
  • $50.83
  • $7.00
Answer

$50.83: 100 ÷ 1.07¹⁰ = 50.83. Each year of waiting divides by 1.07.

Numbers, sources and what’s next

Numbers to remember

Number What it means Source
$1,712/kW Building cost of a new onshore wind farm, before interest during construction (2025 dollars). EIA, AEO2026 assumptions, Table 3 (PDF) · April 2026
$1,032/kW Building cost of a new gas combined-cycle plant (multi-shaft design), on the same basis. EIA, AEO2026 assumptions, Table 3 (PDF) · April 2026
33.8% · 24.0% Average capacity factors of U.S. utility-scale wind and solar photovoltaic generators in 2025. EIA, Electric Power Annual, Table 4.08.B · final 2025 data, October 2026
$3.85/MMBtu Average cost of natural gas to U.S. power plants in 2025, per MMBtu; it was $7.21 in 2022 (in 2022 dollars). EIA, Electric Power Annual, Table 7.4 · final 2025 data, October 2026
30 years The cost recovery period behind both the 2025 ATB's and EIA's levelized costs. NLR, 2025 ATB · September 2026; EIA, AEO2026 levelized costs (PDF) · April 2026

Next on the starter path

Lesson 8 · Who runs, and what sets the price?

In the real world

A levelized cost and a system's cost can move in opposite directions.

A higher hydrogen price makes each kilogram worth more, so the optimizer builds a larger electrolyzer to capture more surplus wind.

The extra capacity runs fewer full-load hours, so the cost per kilogram, the levelized cost of hydrogen, goes up.

Yet the added hydrogen revenue lowers the net cost of the whole system.

The cost per kilogram and the system's cost move in opposite directions.

The full case: case file CF2 (when the cost per kilogram and the cost of the system disagree).

Case file CF7 (what waiting for a grid connection costs, and who pays).

Where the inputs come from

Input Source
Building cost, upkeep, heat rates (2025 dollars) EIA, AEO2026 assumptions, Table 3 (PDF) · April 2026
Capacity factors, 2025 EIA, Electric Power Annual, Tables 4.08.A and 4.08.B · final 2025 data, October 2026
Gas cost, 2025 EIA, Electric Power Annual, Table 7.4 · final 2025 data, October 2026
Rate and 30-year life, case without tax credits NLR, 2025 ATB · September 2026
Prairie Town's summer peak-demand day, 29 July 2025 Southwest Power Pool shapes, EIA-930 (accessed 2 November 2026); the town's plants and running costs are illustrative

Sources and further reading

Spotted an error, or stuck? Write to hlli@ksu.edu. Corrections are made in place and dated here.

Text licensed CC BY 4.0, code MIT: teachers and students may reuse and adapt it, with credit to Honglin Li.