Turning Heat Fluctuations Into Computation: The Promise of Thermodynamic Machines
Thermodynamic computing seeks to harness ambient thermal noise and energy flow as a computational resource, using either equilibrium dynamics or driven nonequilibrium trajectories to solve problems with far less energy and heat than conventional digital hardware. Early silicon-based prototypes and startup efforts show potential for tasks like matrix inversion and generative AI workloads, offering a path that could complement or compete with quantum approaches and providing new insights into how information processing occurs in biology.