Quantum annealing
Finding an optimisation problem's best solution by letting a quantum system settle into its lowest energy state.
The problem is encoded as an energy landscape where the deepest valley corresponds to the best solution. The system starts in a simple state and is changed slowly towards the encoded one — go slowly enough and it stays at the bottom. The advantage over classical simulated annealing is tunnelling: the system can pass straight through a barrier rather than having to climb over it.
The method is not universal. An annealer cannot run Shor's algorithm or anything else requiring gates; it is a special-purpose machine for one class of problem. That is why comparing qubit counts between D-Wave's machines and IBM's or Google's is meaningless.