EN 日本語
← Back to archive

LCOE

Levelized Cost of Energy · February 17, 2026

Summary

LCOE (Levelized Cost of Energy) is the total cost of building and operating a power plant divided by the total electricity it produces over its lifetime — expressed as dollars per megawatt-hour ($/MWh). It's the standard metric for comparing the cost of different electricity sources: solar vs. wind vs. nuclear vs. gas.

1
Elementary School
Ages 8-10

Imagine you want to sell lemonade. You need to buy a stand, buy lemons and sugar, and pay for cups. If you add up ALL the money you spent, and divide it by how many cups you sold, you get the cost per cup.

LCOE is the same thing, but for electricity! Power plants cost money to build (like buying the stand) and money to run (like buying lemons). LCOE tells us: "How much does each unit of electricity really cost?"

This helps us figure out which way of making electricity is cheapest. Is it cheaper to use sunshine (solar), wind, or burn gas? LCOE tells us the answer!

2
High School
Ages 14-18

Historical context: LCOE emerged in the 1970s during the oil crisis when policymakers needed a way to compare nuclear, coal, and emerging alternatives. Before LCOE, comparisons were messy — how do you compare a nuclear plant (expensive to build, cheap to run) with a gas plant (cheap to build, expensive to run)?

The breakthrough was "levelizing" — spreading all costs over the plant's lifetime and dividing by total energy output. This created a single number for apples-to-apples comparison.

The basic formula:

LCOE = (Capital Costs + O&M Costs + Fuel Costs) ÷ Total Electricity Generated

What goes into LCOE:

  • Capital costs: Building the plant — construction, equipment, land, permits
  • O&M (Operations & Maintenance): Staff, repairs, insurance, administration
  • Fuel costs: Coal, gas, uranium (zero for solar/wind)
  • Financing costs: Interest on loans, return expected by investors
  • Lifetime: How many years the plant operates (20-60 years depending on technology)
2025 LCOE Ranges ($/MWh)

Solar (utility-scale): $24-96 — cheapest in sunny regions

Onshore wind: $24-75 — cheapest in windy areas

Natural gas (combined cycle): $39-101 — depends heavily on gas prices

Nuclear: $131-204 — high upfront cost, but 60+ year life

Coal: $68-166 — increasingly uncompetitive + carbon costs

3
College Undergraduate
Ages 18-22

The full LCOE formula:

LCOE = Σ[(Iₜ + Mₜ + Fₜ) / (1+r)ᵗ] ÷ Σ[Eₜ / (1+r)ᵗ]

Where: I = capital investment, M = O&M costs, F = fuel costs, E = electricity generated, r = discount rate, t = year.

The discount rate matters enormously:

The discount rate reflects the time value of money and project risk. A higher discount rate penalizes capital-intensive projects (nuclear, solar) because their costs are front-loaded. A 3% vs 10% discount rate can change nuclear's LCOE by 50%+.

  • Government projects: 3-5% (low risk, patient capital)
  • Utility projects: 6-8% (regulated returns)
  • Merchant/private: 8-12% (market risk, higher returns needed)

Capacity factor — the hidden variable:

Capacity factor = actual output ÷ theoretical maximum output. This dramatically affects LCOE:

  • Nuclear: 90%+ (runs almost constantly)
  • Natural gas: 40-87% (can dispatch on demand)
  • Wind: 25-45% (wind doesn't always blow)
  • Solar: 15-30% (sun doesn't always shine)

A solar panel in Arizona (30% CF) has lower LCOE than the same panel in Seattle (15% CF).

Why Solar LCOE Dropped 89% (2010-2024)

Solar's LCOE fell from ~$380/MWh to ~$40/MWh through: (1) Module costs dropped 99% via manufacturing scale and Chinese competition, (2) Installation efficiency improved, (3) Financing costs fell as banks got comfortable with solar risk, (4) Panel efficiency increased from 15% to 22%+. This is the fastest cost decline of any energy technology in history.

4
Graduate Student
Advanced degree level

LCOE's limitations — what it misses:

LCOE is useful but incomplete. It treats all electricity as equal, ignoring crucial grid realities:

  • Intermittency: Solar/wind produce power when nature allows, not when demand peaks. LCOE doesn't capture the cost of backup power or storage
  • Location: A wind farm in West Texas needs $billions in transmission lines to reach Dallas. LCOE doesn't include grid integration costs
  • Dispatchability: Gas plants can ramp up in minutes; nuclear takes days. This flexibility has value LCOE ignores
  • System effects: Adding lots of solar depresses midday power prices, reducing the value of the next solar project

LCOS and VALCOE — better metrics:

  • LCOS (Levelized Cost of Storage): Same concept applied to batteries. Includes round-trip efficiency losses, cycle degradation, and depth of discharge limits
  • VALCOE (Value-Adjusted LCOE): LCOE + system integration costs - capacity value. Attempts to capture what electricity is worth to the grid, not just what it costs to produce
  • LCOE+: LCOE + storage + transmission needed to deliver firm power

Learning curves and experience rates:

Technologies follow predictable cost declines as cumulative production increases. Solar has an ~20% learning rate (costs drop 20% for each doubling of cumulative capacity). Nuclear historically had negative learning — costs increased over time due to regulatory ratcheting and first-of-a-kind engineering on each project.

The Duck Curve Problem

California's "duck curve" shows net load (demand minus solar) creating a belly at midday and steep ramp in evening. As solar penetration increases, its LCOE stays low but its market value drops (selling into oversupplied hours). By 2024, California regularly curtails solar — producing electricity that earns $0 or negative prices. LCOE can't capture this value erosion.

5
Expert
Researchers & practitioners

Methodological debates in LCOE calculation:

Social vs. private discount rates: Should LCOE for policy analysis use social discount rates (2-4%) reflecting intergenerational welfare, or private rates (8-12%) reflecting actual financing costs? This choice alone can flip rankings between nuclear and gas.

System boundaries: Do you include decommissioning costs? Waste disposal (critical for nuclear)? Land use opportunity costs? Carbon externalities? Each choice is defensible; few analyses are transparent about assumptions.

Degradation curves: Solar panels degrade 0.5-1% per year; batteries degrade per cycle. Optimistic vs. conservative degradation assumptions compound over 25-year lifetimes.

Marginal vs. average LCOE:

LCOE is an average cost over lifetime. But grid planners need marginal costs — what does the next MWh cost? For renewables, marginal cost is near-zero (no fuel), creating market dynamics where LCOE and market price diverge wildly. A solar farm with $30/MWh LCOE might earn $15/MWh in oversupplied markets.

LCOE in capacity expansion models:

Modern grid planning uses optimization models (NREL's ReEDS, EPRI's US-REGEN) that capture system interactions LCOE misses. These show that optimal portfolios aren't simply "build the lowest LCOE" — reliability constraints, transmission limits, and temporal matching matter enormously.

The emerging consensus on metrics:

  • LCOE: Still useful for tracking technology progress within a category
  • System LCOE: LCOE + integration costs for comparing across categories
  • Avoided cost: What a new generator saves the system — captures value, not just cost
  • Levelized cost of firm power: Cost to deliver 24/7 dispatchable power from any source
The IEA/OECD Controversy

In 2020, IEA published LCOE figures showing solar as cheapest in most markets. Critics noted the analysis used 7% discount rate for OECD countries but 8-9% for developing nations — systematically penalizing capital-intensive renewables in countries that most need cheap clean energy. The choice of discount rate is never neutral; it embeds assumptions about risk, access to capital, and who bears the costs of climate change.

Companies & Tools in This Space

Sources