How much solar energy hits the Earth? Short answer: about 173,000 terawatts continuously, roughly 9,000 times the 19 TW or so that humanity uses on average, per NASA's energy budget research. Put differently, Earth absorbs about 660 exajoules of sunlight every 90 minutes, more than the 600-odd EJ world primary energy supply reached in all of 2025 (Energy Institute). That's not hyperbole. Whether your roof gets a usable fraction of it depends on latitude, weather, and a handful of geometric variables most homeowners never think about.
Here's the disparity the global numbers hide. PVGIS (EC JRC) models a 6 kW array at 35 degrees tilt, facing south, with 14% system losses at about 5,670 kWh a year near Newcastle in northern England and about 10,970 kWh a year in Phoenix. Same panels, same tilt. The only variable is where on the planet the array sits.
How Much Solar Power Does Earth Intercept?
Earth intercepts roughly 173,000 terawatts of solar power continuously at the top of the atmosphere (NASA). About 30% reflects back to space off clouds, ice, and bright surfaces (Earth's albedo). The remaining 122,000 TW is absorbed by atmosphere, oceans, and land, driving weather, photosynthesis, and what your panels capture.
The reference at the top of the atmosphere is the solar constant, 1,361 W/m2, measured by NASA's SORCE TIM with under 0.1% variation across the 11-year solar cycle. By sea level under clear skies, absorption and scattering bring direct beam down to about 1,000 W/m2, the STC reference for panel ratings. The absorbed 122,000 TW works out to roughly 3.85 million exajoules a year. World primary energy supply was just over 600 EJ in 2025 (Energy Institute, 2026), so absorbed sunlight exceeds human use by a factor of about 6,000, and the full 173,000 TW arriving at the top of the atmosphere is about 9,000 times the 19 TW humanity averages.
Why doesn't that trivially power the world? Most of the energy hits oceans (71% of the surface) where capture is impractical, and peak solar (midday) doesn't match peak demand (evening), which needs storage and grid we're still building.
How Does Solar Irradiance Vary by Latitude?
Latitude dominates the long-term resource. Equatorial regions get roughly twice the annual insolation of high latitudes because the sun stays high and the atmospheric path is shorter. Tropical zones (within 23.5 degrees) receive 5.5-7 kWh/m2/day; mid-latitudes (35-50 degrees) get 3.5-5; above 60 degrees averages below 3.
| Region | Latitude | Avg daily insolation (kWh/m2) | Peak sun hours |
|---|---|---|---|
| Sahara, Atacama, Outback | 15-25 deg | 6.5-7.5 | 6.5-7.5 |
| Arizona, southern Spain | 30-37 deg | 5.5-6.5 | 5.5-6.5 |
| Mid-US, southern France | 38-45 deg | 4.5-5.5 | 4.5-5.5 |
| UK, southern Germany | 50-55 deg | 2.5-3.5 | 2.5-3.5 |
| Alaska, northern Norway | 60+ deg | 2.0-3.0 | 2.0-3.0 |
Source: NREL NSRDB, EC JRC PVGIS modelled long-term averages
Day length varies too. At 50 degrees north, June days run 16+ hours while December drops to 8; at the equator it stays within 30 minutes of 12 hours all year. So high-latitude systems are summer-heavy while equatorial ones are stable across the calendar. Cloud cover overlays that geometry: the UK at 53 deg N receives roughly 1,000-1,100 kWh/m2/year on a tilted surface (PVGIS gives 1,096 for Manchester at 35 degrees) despite long summer days, because overcast cuts the resource, while Dubai at 25 deg N gets nearly twice the irradiance because clear skies dominate.
What Are Peak Sun Hours and How Are They Calculated?
Peak sun hours convert irradiance into something useful for sizing. One peak sun hour equals one hour at 1,000 W/m2, so a location receiving 5 kWh/m2 in a day has 5 peak sun hours whether the day runs 8 hours or 14. Because panels are rated at 1,000 W/m2 (STC), peak sun hours multiply directly with panel wattage: a 6 kW system in 5 peak sun hours makes roughly 30 kWh/day before losses.
Real systems lose 15-25% to inverter efficiency (96-98% on modern units like the Fronius Primo 8.2-1), wiring (1-3%), soiling (3-10%), temperature derating (5-15%), and shading, so that same system delivers 22-25 kWh/day in practice. NREL's National Solar Radiation Database resolves peak sun hours to 4 km grid cells with hourly time series back to 1998, and pulling your actual location beats a city average: elevation gain in the Sierras or Colorado front range can shift the resource 5-10% through atmospheric clarity. For translating peak sun hours into output, see our solar panel efficiency calculator guide.
How Much Solar Energy Does Your Panel Actually Capture?
A typical 400W monocrystalline panel covers about 1.7-1.95 m2 and converts roughly 21-23% of incoming light. Under STC it makes 400W; over a year in a 5 peak sun hour location it generates roughly 730 kWh, or 580-620 kWh net after 15-20% system losses. Across a 6 kW array (15 panels) that's 8,700-9,300 kWh per year at that 5 peak sun hour midpoint. PVGIS puts the same 6 kW at about 8,100 kWh in Chicago, 7,030 in Seattle and 10,970 in Phoenix. The hardware doesn't change, only the resource does.
The physical ceiling? An ideal single junction tops out at 33.7% (Shockley-Queisser, at the optimal 1.34 eV bandgap). For silicon, Auger recombination pulls the practical limit down to 29.4% (Richter et al., 2013). Real devices are getting close: Fraunhofer ISE's July 2026 Photovoltaics Report lists a record lab silicon cell at 28.1%, the best lab silicon module at 26.0%, and perovskite-silicon tandem cells at 35.2%. Commercial silicon modules are further back: the shipment-weighted average was 22.7% in late 2024, and the best datasheet value was 24.8%. Multi-junction concentrator cells reach 47.6% but cost far too much per square meter for roofs. In real annual conditions terrestrial panels capture maybe 15-20% of surface irradiance, not the datasheet 22-23%, losing the rest to operating temperature above 25 deg C, spectral mismatch against AM 1.5G, and angle-of-incidence when the sun isn't perpendicular. Our best solar panels 2026 guide shows which models sit at the top of that range.
The sun delivers more energy to Earth in 90 minutes than humanity uses in a year. The limit on solar isn't sunlight, it's storage, grid, and economics.
Why Can't Solar Just Power Everything?
If solar input beats human demand 10,000x, why isn't it the whole mix? Intermittency, density, and infrastructure, not resource. Solar drops to zero at night and 10-25% under heavy cloud, so matching demand needs storage or transmission across time zones, which is why Fraunhofer ISE's 2026 report calls batteries with energy management increasingly important. On density, solar averages 5-10 W/m2 after capacity factor (20-25% in good sites) versus 1,000-2,000 for a coal plant footprint. Powering roughly 4,200 TWh of annual US electricity (about 480 GW on average) at that density takes something like 50,000-100,000 km2, about 1% of the lower 48's area and in the range of West Virginia's 62,800 km2 (our arithmetic, not a siting study). And grids built for large dispatchable generators need distribution upgrades for bidirectional flow that lag deployment.
Even so, solar keeps compounding: global PV reached about 2,974 GW at the end of 2025 (Fraunhofer ISE, citing IEA PVPS), up from 2,276 GW a year earlier, with roughly 690 GW added in 2025 alone and PV supplying 8.7% of world electricity. For where solar dominates and lags, our global solar deployment overview has the country breakdown. For what the energy is used for in practice, see how solar power can be used; for the environmental side, see how solar panels help the environment and whether solar is really dirty energy.
What Does the Solar Energy Budget Look Like in Detail?
NASA's planetary energy budget splits the 173,000 TW roughly into 30% reflected to space (Earth's 0.30 albedo), 23% absorbed by the atmosphere (water vapor, ozone, clouds), and 47% absorbed by the surface. That surface absorption (about 81,000 TW) powers photosynthesis, the water cycle, weather, and what panels capture. Photosynthesis captures about 130 TW globally, roughly 0.16% of the surface input, while human civilization runs on about 19 TW, roughly 0.02% of it, mostly fossilized photosynthesis from prior epochs.
Atmospheric absorption shapes the spectrum: water vapor knocks out near-infrared bands, ozone absorbs UV-B and most UV-C. The sea-level spectrum (AM 1.5G) is what silicon panels are designed against. Silicon's 1.12 eV bandgap (about 1,100 nm) means longer-wavelength photons pass through unused and shorter ones waste their extra energy as heat, which is why a panel turns only 20-23% of incoming light into electricity. For more, see our piece on UV light and solar panels.
Summary
Earth catches 173,000 TW of solar power continuously, roughly 9,000 times average global human demand of about 19 TW (just over 600 EJ a year). The sea-level clear-sky reference is 1,000 W/m2, the STC condition for rating every panel made. Practical resource is measured in peak sun hours, from 2.5 in northern Europe to 7+ in equatorial deserts, with NREL's NSRDB giving 4 km US data. A 400W panel in a 5 peak sun hour location makes around 600 kWh/yr after losses. The constraint isn't sunlight, it's storage, transmission, and economics. For converting irradiance into system output, the solar panel efficiency calculator guide covers the math.