Berkeley Lab and NVIDIA Collaboration Accelerates U.S. Leadership in Hybrid Quantum-Classical Computing

Lawrence Berkeley National Laboratory and NVIDIA announced a collaboration in October 2025. The initiative focuses on enabling quantum processing units and graphics processing units to operate together in real time. Goals include shorter delays and greater data throughput in hybrid quantum classical systems. The laboratory team has connected its QubiC control stack to the NVIDIA DGX Spark GPU via the NVQLink platform. This integration provides low latency and high bandwidth GPU QPU communication. Hardware testing is set to conclude in early March. QubiC is an open source system funded by the Department of Energy Office of Science and tested at the Advanced Quantum Testbed. It allows modular replacement of quantum and classical components. The architecture uses a 100 gigabit link for direct data flow from the quantum processing unit to GPU memory. Plans call for integrating Hololink IP into the QubiC gateware. Pre trained neural network models will be deployed on the DGX Spark for readout classification, gate tuning and real time error correction. The effort supports the DOE Genesis Mission to integrate artificial intelligence with high performance computing and quantum technologies.