Most computers separate memory and processing and run in regular clock steps. Neuromorphic chips instead borrow from the brain, using many simple units that fire only when needed and keep memory close to computation, which can make certain sensing and pattern tasks far more energy efficient.
These chips suit always-on jobs like detecting a sound or motion on a tiny power budget, where constantly running a normal processor would drain a battery. They are specialised rather than general purpose and need different programming methods, so they remain a research and niche tool, but they point to ways of computing that waste far less energy for the right problems.