In a groundbreaking collaboration, Diraq, QM Technologies Inc., and NVIDIA have achieved the first-ever tight integration of GPUs with silicon quantum processors, deploying a system capable of real-time communication at an unprecedented speed of 3.3 microseconds.
Powered by the NVIDIA DGX Quantum architecture and QM’s OPX1000 hybrid controller, this milestone represents a pivotal moment in quantum computing—shifting the focus from theoretical capability to practical execution.
Breaking Through Bottlenecks in Quantum Computing
At Diraq’s Sydney lab, engineers from Diraq and QM harnessed GPU acceleration to overcome some of the most pressing barriers in scaling quantum computing.
Key Applications Demonstrated:
- Real-time readout enhancement
- Previously possible only via post-processing
- Now achieved instantaneously using GPU-assisted signal analysis
- Automated calibration via machine learning
- Replacing hours of manual tuning with ML models running on NVIDIA GPUs
- Accelerated quantum state initialization
- Faster qubit preparation based on Diraq’s Nature-published algorithms
- Critical to prepare quantum states before they degrade
The system achieved these within just one week of installation—underscoring its immediate impact on research workflows.
Hybrid Architecture: DGX Quantum + OPX1000 + Silicon Qubits
This integration involves:
- NVIDIA Grace Hopper Superchips for AI and GPU performance
- QM’s OPX1000 Controller for low-latency hybrid orchestration
- Diraq’s silicon qubit arrays, uniquely suited for scalable quantum computing
Technical Highlights:
- 3.3-microsecond round-trip latency between QPU and GPU
- Seamless real-time feedback loops to avoid quantum state degradation
- Groundbreaking use of GPU acceleration over FPGAs for complex signal processing
Voices from the Collaboration
Andrew Dzurak, CEO & Founder, Diraq:
“This collaboration pushes the limits of real-time quantum control, highlighting the remarkable integration potential of Diraq’s silicon qubit arrays.”
Itamar Sivan, CEO & Co-Founder, QM:
“Diraq’s success in developing three quantum-GPU applications in one week proves that this technology is now solving real-world problems.”
Tim Costa, Sr. Director of Quantum & CUDA-X, NVIDIA:
“DGX Quantum allows AI advances to accelerate quantum workloads. It is the future of hybrid computing.”
What Makes This a Milestone?
- Hybrid systems are no longer theoretical—they’re now operational and impactful
- GPU-enabled real-time feedback is crucial for error correction and state preservation
- Demonstrates the viability of scalable, silicon-based quantum hardware
- Opens the door to hybrid quantum–AI algorithms, transforming both fields
The Future: CUDA-Q, Real-Time Error Correction & Beyond
What’s Next for DGX Quantum:
- Integration with NVIDIA’s open-source CUDA-Q platform
- Deployment of real-time error-correction protocols using hybrid systems
- Further advancement of quantum–AI algorithm development, enabling new computing paradigms
Upcoming Milestone:
- GTC Paris 2025 will spotlight these results, where Elica Kyoseva, NVIDIA’s Director of Quantum Algorithm Engineering, will outline the future of hybrid algorithm development.
Ushering in the Era of Hybrid Quantum-Classical Systems
The collaboration between Diraq, QM Technologies, and NVIDIA marks a major inflection point in quantum computing. By merging classical GPU power with the precision of quantum processors, this project doesn’t just theorize the future—it builds it.
With real-time integration, scalable hardware, and immediate results, the hybrid quantum–AI future is no longer on the horizon—it’s already here.
