Price per watt for every major energy technology — solar, wind, nuclear, batteries, coal and gas — from Lazard, BloombergNEF, NREL, IEA and EIA. Verified numbers, no vibes. Hover any bar for the source.
What it costs to build one watt of capacity — not the electricity it produces (that's LCOE, covered below). Batteries are priced per kWh of storage, the industry standard, in the second chart.
Bars show the published range (tick = low end). "Est." = estimated from related builds; ranges reflect different sources and regions.
Batteries are bought by energy stored, not power output — so $/kWh is the honest unit. Pack prices per BloombergNEF's 2025 survey unless noted.
Good at: cheap, safe, long-lived. The workhorse of grid storage and standard-range EVs. No cobalt or nickel.
Bad at: energy density — heavier and bulkier than NMC for the same range; hates extreme cold.
Best for: home batteries, grid storage, buses, budget EVs.
Good at: packing the most range into the least weight — the long-range EV chemistry.
Bad at: price ($128/kWh packs), shorter life, thermal runaway risk, and a cobalt supply chain with real human-rights problems.
Best for: long-range EVs where every kilogram counts.
Good at: raw-material cost — sodium is ~1,000× more abundant than lithium and needs no cobalt or nickel. Charges at −30°C without damage.
Bad at: not yet cheaper than LFP at scale (about 13% more per kWh in 2025); lower energy density (160–175 Wh/kg).
Best for: stationary storage, cold climates, budget city EVs. The one to watch this decade.
Good at: being dirt cheap upfront and 99% recyclable — the most recycled product on Earth.
Bad at: everything else: heavy, short-lived, toxic if mishandled, terrible for deep cycling.
Best for: starter batteries, backup power. Not a grid-storage future.
Good at: duration — store energy in liquid tanks, so adding hours of storage just means bigger tanks. Decades of life, non-flammable.
Bad at: price ($250–400/kWh est.) and bulk — terrible energy density, vanadium prices swing wildly.
Best for: 8–12 hour grid storage, microgrids, places that need decades of daily cycling.
Good at: the promise: 2× the energy density of lithium-ion with no flammable liquid electrolyte.
Bad at: existing affordably. Still in pilot production; timelines have slipped for a decade.
Best for: someday: premium EVs and aviation. Don't plan a grid around it yet.
Good at: being the cheapest new power on Earth — Lazard puts unsubsidized LCOE at $40–98/MWh (2026).
Bad at: night. Needs storage or complementary generation for round-the-clock power.
Good at: cutting a business's power bill with zero land use.
Bad at: soft costs — permits, sales and labor dominate the price, not the panels.
Good at: household resilience and, in sunny states, beating retail electricity prices.
Bad at: US soft costs are the world's highest — two-thirds of the system price is paperwork, sales and labor.
PERC was the 2010s standard (~21% efficient). TOPCon is today's mainstream (~23–25%). HJT pushes higher (~25–26%) at a premium. CdTe thin-film (First Solar) skips silicon entirely — great in heat, slightly lower efficiency. The panel is now the cheapest part of any system.
Good at: the lowest LCOE of any new generation — Lazard: $37–86/MWh unsubsidized. Turbines now top 5+ MW on land.
Bad at: needing windy sites and transmission to cities; local opposition to views and noise.
Best for: windy plains and ridgelines — the backbone of cheap grids.
Good at: massive, steady output — 42% average capacity factor vs ~34% onshore; 15 MW turbines; near coastal demand.
Bad at: capital cost — foundations, cables, vessels and ports roughly triple the installed cost vs onshore.
Best for: coastal megacities with poor onshore wind (US Northeast, UK, East Asia).
Good at: 90%+ capacity factor for 60–80 years — the densest baseload on Earth. Marginal operating cost is tiny ($29–38/MWh).
Bad at: construction cost in the West — Vogtle ran ~$15/W; Olkiluoto ~$7.50/W. South Korea and China build for far less.
Best for: countries that can build on schedule — then it's cheap power for generations.
Good at: factory assembly, smaller grids, replacing coal plants site-for-site.
Bad at: first-of-a-kind cost — MIT estimates $14,000/kW; the NuScale UAMPS project hit ~$20,000/kW and was cancelled. Factory savings are unproven.
Best for: the 2030s, if fleets actually get built and learning curves kick in.
Molten-salt, sodium-cooled fast, and high-temperature gas reactors promise to burn existing waste and run hotter (industrial heat, hydrogen). All still pre-commercial — price claims are vendor aspirations, not data. Treat accordingly.
Good at: cheap, fast to build, flexible — the grid's current workhorse.
Bad at: fuel exposure — gas price spikes flow straight to power bills; still emits ~0.4 t CO₂/MWh. Lazard LCOE: $48–109/MWh.
Good at: firing up in minutes for demand spikes.
Bad at: the most expensive electricity on the grid — Lazard LCOE $149–251/MWh. Batteries are eating this job.
Good at: nothing new — its only virtue is existing plants' sunk cost.
Bad at: cost (Lazard: $71–173/MWh new-build), pollution, and economics — virtually no new coal is being built in the US because wind and solar undercut it.
Fuel reality: a coal plant's lifetime fuel bill dwarfs its build cost. "Cheap coal power" was always about ignoring the fuel and the ash.
No lectures — just the mechanics. Here's what the numbers say about replacing fossil baseload, and what it means for the communities built around it.
New wind and solar are simply cheaper than new coal or gas in most of the US — Lazard's 2025 numbers put unsubsidized onshore wind at $37–86/MWh and utility solar at $38–78/MWh, against $71–173/MWh for new coal and $48–109/MWh for new gas combined-cycle. That's why 2024–2025 US capacity additions were overwhelmingly wind, solar and batteries. Nobody is choosing coal on economics anymore; the debate is only about how fast the old plants retire.
The grid doesn't need a single baseload replacement — it needs firm capacity, which comes from a portfolio: overbuilt wind and solar (cheap enough to overbuild), 4–8 hour batteries for daily peaks, longer-duration storage and flow batteries for multi-day stretches, existing nuclear kept running (its marginal cost is just $29–38/MWh), and gas plants kept as rarely-used backup while their run-hours collapse. Texas — the wind capital of America — already runs this playbook.
A retired coal plant is a grid connection, water rights, rail access, and a trained workforce sitting idle. The fastest transitions reuse the site: solar or solar-plus-storage on mine land, battery factories in the turbine halls, and — where communities want it — small modular reactors on the same footprint, reusing the switchyard and cooling. The infrastructure is the asset; the fuel was the liability.
Wind turbine technician is one of the fastest-growing jobs in the US, and solar installers aren't far behind — but a coal electrician in West Virginia can't teleport to a Texas wind farm. Honest transition policy means: wage insurance during retraining, union apprenticeship pipelines into clean-energy trades, and siting new manufacturing (battery plants, tower factories) in fossil communities, not just near cheap land. The electrons are easy; the livelihoods need planning.
Two bottlenecks decide the pace: transmission (the US needs roughly double its grid to move windy/sunny power to cities — see our bottlenecks page) and interconnection queues (years-long waits to plug in). Building generation without wires is like drilling wells without pipes. Permitting reform for transmission is the single highest-leverage climate policy available.