Making electricity: thermal, nuclear, hydro, wind and solar: Solar electricity and the 2025 mix

5.1 A solar cell turns light straight into direct current

A photovoltaic cell is a thin wafer of semiconductor, nearly always very pure silicon, treated so that one layer has a slight excess of electrons and the other a slight shortage. Where the layers meet, a built-in electric field forms. When a photon of sunlight with enough energy is absorbed, it frees an electron, and the field pushes it towards one side while the gap it leaves drifts to the other. Connect the two faces through a wire and a direct current flows. An inverter then converts this to the alternating current used by grids.

Cells are wired together into modules of about 2 square metres, and modules into arrays. In full sun, about 1,000 watts of light fall on each square metre. A typical commercial module now converts around 21 to 23% of that, up from about 10% a few decades ago, so a square metre of panel produces a little over 200 watts at its peak. Physics caps a single silicon junction at about 29%, and an ideal single junction of any material at about 33%; stacking different materials, as in perovskite-on-silicon tandem cells, could push past it.

The history is long and slow. Edmond Becquerel observed the photovoltaic effect in 1839, and Bell Labs made the first practical silicon cell in 1954. For decades cells were too costly for anything but satellites. Smil uses this as a case study in how long it can take a technology to matter: about 60 years passed between the Bell Labs cell and the moment when solar became a measurable share of world electricity in the 2010s.

Sunlight can also make electricity through heat. 'Concentrated solar power' uses mirrors to focus sunlight onto a receiver, heating a fluid to drive a steam turbine. Its advantage is that hot molten salt can be stored cheaply for the evening. But it needs very clear desert skies, is mechanically complex and, at about USD 115 per MWh in IRENA's 2025 data, costs more than twice as much as PV, so it has remained a niche.

• Smil, Numbers Don't Lie: 71 Stories to Help Us Understand the Modern World (2020) · via Smil Mastery Course, M7 'Solar electricity'
• Smil, Energy: A Beginner's Guide (2006) · via Smil Mastery Course, M7 'Solar electricity'
• IRENA, Renewable Power Generation Costs in 2025 (2026) · web

5.2 Solar became cheap through manufacturing scale, but it only works by day

Module prices fell by more than 90% during the 2010s, driven above all by the scale and learning of Chinese factories that now make the large majority of the world's polysilicon, wafers, cells and modules. IRENA's 2025 cost report puts the global weighted-average LCOE of utility-scale solar at about USD 44 per MWh, essentially flat from 2024, with installed costs of about USD 667 per kW, the lowest of any technology. India's figure was about USD 35 per MWh, the lowest of the regions IRENA highlights and slightly below China's USD 36. In sunny places solar is now the cheapest electricity a new plant can make during the day.

The constraint is the daily and seasonal rhythm. Output follows the sun: zero at night, a rising curve through the morning, a peak around noon, a fall towards evening, and less in cloud or haze. Capacity factors run from about 10 to 15% in cloudy northern Europe to 20 to 25% in deserts; much of India sits around 18 to 22%, lower during the monsoon. When many panels feed one grid, midday prices can collapse and evening demand must be met by something else. Grid operators call the resulting shape of net demand the duck curve.

The answer emerging fastest is batteries. IRENA reports that four-hour utility battery systems cost about USD 140 per kWh installed in 2025, nearly 30% less than a year earlier and about 95% less than in 2010, and that at the best sites solar paired with storage can now deliver firm power, available in 95% of hours, for under USD 85 per MWh, down from over USD 100 in 2020. 'Lithium iron phosphate batteries' dominate these projects.

Smil adds a reminder that the best use of sunlight is still the oldest. Photosynthesis grows our food, and passive solar design, daylighting and simple solar water heaters capture sunlight with no conversion losses at all. A rooftop water heater, as the Energy table's 'Solar heat collectors' tile shows, turns well over half of the sunlight it receives into useful heat, three times what a PV panel turns into electricity.

• IRENA, Renewable Power Generation Costs in 2025 (2026) · web
• Smil, Numbers Don't Lie: 71 Stories to Help Us Understand the Modern World (2020) · via Smil Mastery Course, M7 'Solar electricity'
• Ember, Global Electricity Review 2026 (April 2026) · India section, https://ember-energy.org/latest-insights/global-electricity-review-2026/2025-in-review/
• PIB, PM Surya Ghar: Muft Bijli Yojana factsheet (December 2024) · launched 15 February 2024, one crore households by March 2027: https://static.pib.gov.in/WriteReadData/specificdocs/documents/2024/dec/doc2024125465001.pdf

5.3 In 2025 renewables matched coal, and solar and wind covered the growth in demand

Ember's Global Electricity Review 2026 gives the clearest picture of where the world stood in 2025. World demand rose by about 849 TWh, or 2.8%. Solar generation grew by a record 636 TWh, or 30%, to about 2,778 TWh, and wind by 205 TWh. Together they added about 841 TWh, almost exactly the growth in demand, so fossil generation was roughly flat. Renewables of all kinds supplied 33.8% of world electricity and, for the first time, slightly more than coal at 33.0%. Adding nuclear, low-carbon sources reached about 43%.

The IEA's Electricity 2026 report uses slightly different accounting and puts coal at about 34% in 2025, but tells the same story. It expects world electricity demand to grow by about 3.6% a year to 2030, at least two and a half times faster than energy demand overall, as industry, cooling, electric vehicles and data centres electrify. Electricity's share of final energy use is set to rise from about 21% to 24%. Solar is expected to add more than 600 TWh a year, coal's share to fall to about 27% by 2030, and renewables plus nuclear to rise from 42% to about half of all generation.

India is the most important growth market after China. The IEA expects Indian demand to grow by about 6.4% a year, adding more than 570 TWh by 2030, with cooling alone responsible for over a fifth of the increase. In 2025, a mild year, Indian demand grew only 2.4%; renewable output rose a record 98 TWh and fossil generation fell by 3.3%. Coal still supplies about 70% of India's electricity, however (fossil fuels together 73%, according to Ember), so the next decade's question is whether clean supply can keep pace with surging demand.

Smil's reminder applies: electricity is only about a fifth of final energy, so even a clean grid leaves most fuel use in industry, shipping, aviation and heating untouched.

• Ember, Global Electricity Review 2026 (April 2026) · web
• IEA, Electricity 2026 (February 2026) · web
• Smil, How the World Really Works: A Scientist's Guide to Our Past, Present and Future (2022) · via Smil Mastery Course, M7 'Grids, storage and the real cost of electricity'
• Ember, India country page (2026) · fossil fuels 73% of India's generation in 2025, https://ember-energy.org/countries-and-regions/india/

See it in the table

Check your understanding

Why does a solar panel's peak rating overstate the energy it delivers in a year?

In 2025, what roughly covered the growth in world electricity demand?

According to IRENA, what did a four-hour utility battery system cost to install in 2025?