Engineers at UC Davis built a Stirling engine that generates power at night. One side of the engine sits on the ground, absorbing ambient heat — roughly 300 Kelvin, the temperature of the surface. The other side faces a panel that radiates infrared photons upward through the atmosphere and into space. The effective temperature of the cold side is not the air temperature but the radiative temperature of the night sky, which in clear, dry conditions can be tens of degrees below ambient. The engine runs on the temperature difference.
After a year of testing, the device produced 400 milliwatts of mechanical power per square meter of radiating panel. This is a tiny number — roughly a thousandth of what a solar panel produces in daylight. But the comparison to solar panels is the wrong frame. Solar panels harvest incoming radiation. This engine harvests outgoing radiation. They are complementary, not competitive. A solar panel stops producing at sunset. The night engine starts.
The deeper point is about what counts as a resource. Standard energy thinking treats the environment as a source of heat — burn fuel, capture sunlight, tap geothermal gradients. The night engine treats the environment as a source of cold. Outer space is 2.7 Kelvin. The atmosphere is partially transparent to thermal infrared in certain wavelength bands, allowing surfaces on Earth to radiate directly to space without heating the intervening air. The cold is already there. It has always been there. It was never hidden — anyone who has felt a clear night cool faster than a cloudy one has experienced the effect. What was missing was not the physics but the engineering decision to build a heat engine with space as the cold reservoir.
The Stirling engine is a natural choice for the application. It operates on any temperature difference, requires no internal combustion or working fluid phase change, and can be made from simple components — a piston, a flywheel, a sealed cylinder. The challenge is not the engine but the radiator: the panel must emit efficiently in the atmospheric transparency window (8-13 micrometers) while minimizing absorption of downwelling atmospheric radiation. The panel is an antenna tuned to a specific band, aimed at a reservoir three Kelvin above absolute zero.
The practical limitations are significant. The power density is low. The system works best in arid regions with clear skies — humidity blocks the transparency window, reducing the effective temperature difference. Cloud cover eliminates the gradient entirely. These are the same conditions that favor solar power during the day, which means the night engine is geographically complementary with solar rather than offering access to new locations.
But the structural insight survives the practical limitations. Every heat engine needs two reservoirs. For most of industrial history, the engineering effort went into the hot side — hotter fires, more concentrated sunlight, deeper geothermal wells. The cold side was treated as given: the atmosphere, a river, the ocean. The night engine reverses this. The hot side is given — the ground is warm. The engineering effort goes into accessing a colder cold reservoir than the local environment provides. Space is the coldest reservoir accessible from Earth's surface. Using it requires not energy input but wavelength selection — choosing the right infrared band to radiate through the atmosphere without being absorbed.
The resource was always overhead. The engineering was looking down.