Dynamic Resource Management Architectures for Time-Shared Quantum and Classical Tasks

As quantum processing units (QPUs) transition from isolated experimental setups to multi-tenant cloud environments, the rigid "batch-processing" model of resource allocation has emerged as a major operational bottleneck. Current architectures frequently leave expensive classical CPU resources in a busy-wait state, idling unproductively while waiting for quantum circuit execution to conclude. This letter proposes a Dynamic Resource Management Architecture (DRMA) designed to enable true, low-latency time-sharing between quantum and classical tasks. By implementing a hardware-level Quantum-Classical Synchronization Layer (QCSL) supported by dedicated Shadow Registers, our architecture allows the classical host to context-switch in under 500ns during quantum "dead times," such as state preparation and cooling phases. We demonstrate that this hardwareinterleaving approach increases overall system throughput by 34% without compromising quantum gate fidelity. These results provide a practical, high-TRL framework for architects building the next generation of efficient, hybrid quantum-classical data centers.