A Heterogeneous Low-Latency Interconnect Enabling Large-Scale Qubit Control and Real-Time AI Inference

The rapid scaling of quantum processors to large qubit arrays creates a critical need for a low-latency link among multiple radio frequency system-on-chip (RFSoC) field-programmable gate arrays (FPGAs), as tasks like quantum error correction (QEC) demand coordinated control and computational resources that exceed the capacity of a single device. While the integration of real-time artificial intelligence (AI) offers a promising path for advanced control algorithms and efficient error decoding, it introduces the need for a high-bandwidth, low-latency communication fabric between heterogeneous computing platforms like FPGAs and graphics processing units (GPUs).

To address this challenge, we present the development of an open-source, low-latency data link designed for scalable quantum control systems. Our architecture facilitates the aggregation of multiple RFSoC FPGAs for the coherent control and measurement of large-scale qubit arrays. This system is tightly coupled with local GPUs, enabling real-time QEC decoders, and can interface with specialized AI Engine FPGA hardware for complex, high-dimensional model inference. This work provides a crucial, open-source infrastructure component that bridges the gap between traditional FPGA-based quantum control and emerging AI-accelerated techniques, paving the way for scalable and fault-tolerant quantum computation.