Decentralised energy trading in large-scale power systems is essential for enhancing operational flexibility, resilience, and renewable integration; however, effective coordination of building-level prosumers is hindered by demand–supply uncertainty, network constraints, and the computational complexity of system-wide optimisation. This paper presents a Quantum-AI Digital Twin (QADT) framework that enables scalable, physics-consistent coordination between buildings and electricity markets by embedding high-fidelity digital replicas into the transaction decision layer. The framework integrates data-driven building energy models, probabilistic flexibility forecasting, and decentralised transaction scheduling within a quantum-ready optimisation architecture. It is validated on the IEEE 300-bus system augmented with aggregated building clusters, distributed PV generation, battery storage, and flexible demand resources. Operating in a closed-loop manner, the QADT evaluates energy transactions against power-flow constraints, voltage limits, and building operational requirements prior to execution, ensuring physical feasibility.