Researchers have explored modular quantum computing, a promising approach to scale quantum computers beyond the limitations of monolithic devices. In this paradigm, multiple quantum processing units (QPUs) are interconnected via shared entanglement. The study focused on how errors at module interfaces and within individual QPUs affect fault-tolerant computation, using the rotated surface code for qubit encoding. This work goes beyond the logical-memory benchmark, simulating circuit-level fault-tolerant nonlocal CNOT gates implemented via lattice surgery between QPUs connected by noisy Bell pairs, and analyzing the resulting logical error rates.

The results indicate that interfaces can tolerate noise up to an order of magnitude higher than intra-QPU noise, with only a minor reduction in the fault-tolerance threshold. This suggests unexpected robustness in inter-module connections. Furthermore, an efficient protocol has been developed for the distributed preparation of fault-tolerant logical GHZ states, reducing ancilla overhead, operation time, and nonlocal Bell-pair consumption. Ancilla minimization in this setting is equated to a vertex-cover problem on an associated graph, for which a polynomial-time heuristic algorithm is introduced to find low-overhead solutions.

These findings provide quantitative evidence that distributed quantum error correction can enable scalable, fault-tolerant quantum computation in modular architectures. This advance is crucial for the development of large-scale quantum computers, as it addresses one of the main challenges: managing noise in complex, distributed systems. The ability to effectively interconnect noisy QPUs opens new avenues for the design and implementation of future quantum processors.