A new network architecture from Fermilab targets a key challenge in scaling quantum computing systems: low-latency, predictable communication across control hardware.

The U.S. national laboratory’s XCOM system connects multiple quantum instrumentation control kit (QICK) boards in a mesh network, enabling synchronized operation and deterministic communication between distributed control electronics, according to a recent arXiv paper.

As quantum systems grow, particularly in superconducting and spin qubit platforms, control requirements quickly exceed what a single electronics board can handle. Multiboard setups combining RF signal generation, bias control, and readout must be tightly synchronized and able to exchange data rapidly to support operations such as quantum error correction (QEC).

XCOM synchronizes system clocks to within 100 picoseconds without drift and enables all-to-all communication between boards with latency below 185 nanoseconds in current prototypes. The system is designed to scale across a range of qubit technologies and larger hardware configurations.

“The main accomplishments of XCOM are multiboard synchronization and deterministic low-latency message communication,” Gustavo Cancelo, a senior electronics engineer at Fermilab, explained.

Unlike conventional approaches that prioritize throughput, XCOM focuses on deterministic timing, ensuring messages arrive at predictable intervals. This is critical for QEC, where timing variability can limit system performance. The mesh topology also allows any board to communicate directly with any other or broadcast messages across the network.

Cancelo said the system can achieve latency as low as 65 nanoseconds with firmware updates, compared with 185 nanoseconds in the initial prototype, helping meet tight QEC timing requirements.

The architecture also enables some QEC processing to be performed directly on field-programmable gate arrays (FPGAs), including partial syndrome detection and decoding. This reduces reliance on centralized processing and helps mitigate communication bottlenecks.

“Board-to-board communication is certainly a bottleneck. QEC critically depends on that,” Cancelo said.

The system could support other scaling challenges, including multiqubit calibration and crosstalk compensation. The team is also exploring integration with newer FPGA platforms that include built-in AI acceleration capabilities.

The current prototype supports up to five QICK boards, with Fermilab already receiving requests from early users as it looks to scale the system. The researchers plan to refine the hardware design and expand it to larger configurations.

The work underscores the growing importance of classical control infrastructure in quantum computing, where synchronization and communication are emerging as key constraints as systems scale.