Quantum Annealing

last updated Tue Jun 02 2026 00:00:00 GMT+0000 (Coordinated Universal Time)
Combinatorial OptimisationStochastic / Ising MachinesCoherent Ising MachineThermodynamic ComputingProbabilistic ComputingPhysics-Native ComputeQuantum Ann…

Quantum annealing is a metaheuristic optimisation technique that exploits quantum mechanical tunnelling to escape local minima when searching the energy landscape of a combinatorial problem. It maps a problem onto an Ising Hamiltonian — a network of binary spins with pairwise couplings — and evolves the system from an initial quantum superposition toward its ground state by slowly reducing a transverse magnetic field (the quantum fluctuation term). The goal is for the system to settle in, or near, the lowest-energy configuration, which corresponds to the optimal or near-optimal solution.

The physics distinguishing it from classical simulated annealing: tunnelling allows the system to pass through energy barriers rather than over them, which in theory enables faster escape from local minima at low temperature. Whether this translates to a practical advantage on real problem instances remains the central open question — benchmark evidence is mixed and highly instance-dependent.

The dominant hardware platform is the superconducting flux-qubit architecture pioneered by D Wave Systems (Advantage system: ~5,000 qubits with Pegasus graph topology). Qilimanjaro Quantum Tech is developing coherent quantum annealing processors. Classical digital annealers — CMOS or FPGA chips that emulate Ising dynamics without quantum coherence — are built by Fujitsu Digital Annealer Division and others; they reach thousands of spins at room temperature, making hardware comparison with true quantum annealers non-trivial.

In the KB, quantum annealing sits within the Extreme Low Power Compute thesis as the “Ising/annealing” substrate vertical — investable where there is a demonstrated energy-per-operation advantage on a specific customer problem class, not as a general-purpose compute bet. Its closest conceptual siblings are Stochastic / Ising Machines (classical probabilistic emulation) and Coherent Ising Machine (photonic or optical-parametric Ising machines).

Frontier

Related concepts

Frontier questions