NVIDIA Accelerated Quantum Research Centre to Advance Quantum Computing
Based in Boston The most difficult problems in quantum computing can be solved with the help of NVAQC’s strong tools.
In order to create accelerated quantum supercomputers that can tackle some of the most challenging issues in the world, quantum computers will eventually merge with AI supercomputers as they advance.
Developing new applications, facilitating advancements in quantum error correction and device control, and helping uncover discoveries essential to operating future quantum hardware all depend on the integration of quantum processing units (QPUs) into AI supercomputers.
Following today’s announcement at the NVIDIA GTC global AI conference, these advancements will take place at the NVIDIA Accelerated Quantum Research Centre, or NVAQC. In addition to the NVIDIA Quantum-2 InfiniBand networking technology and an NVIDIA GB200 NVL72 system, the facility will feature a supercomputer with 576 NVIDIA Blackwell GPUs for quantum computing research.
“To scale quantum computing to next-generation devices, the NVAQC leverages long-sought and much-needed tools,” stated Tim Costa, senior director of NVIDIA’s computer-aided engineering, quantum, and CUDA-X. “The centre will be used for training and deploying AI models for quantum, tight integration of quantum processors, and large-scale simulations of quantum algorithms and hardware.”
Together with academic partners from the Harvard Quantum Initiative and the Engineering Quantum Systems group at the MIT Centre for Quantum Engineering, leading companies in quantum computing such as Quantinuum, QuEra, and Quantum Machines will collaborate on projects with NVIDIA at the centre to investigate how AI supercomputing can hasten the development of quantum computing.
The MIT Centre for Quantum Engineering’s director and leader of the EQuS group, William Oliver, a professor of physics and electrical engineering and computer science, said, “The NVAQC is a powerful tool that will be instrumental in ushering in the next generation of research across the entire quantum ecosystem.” In order to achieve practical quantum computing, NVIDIA is an essential partner.
The NVAQC has already the potential to significantly impact a number of important quantum computing problems.
Safeguarding Qubits Through AI Supercomputing
Qubit-to-qubit exchanges have two sides. To be controlled and measured, qubits must interact with their environment, but these same interactions can also introduce noise, or undesired disruptions, which reduces the precision of quantum computations. Controlling the ensuing noise is essential for the operation of quantum algorithms.
By encoding noiseless, logical qubits among several noisy, physical qubits, quantum error correction offers an answer. The outputs of repeated measurements on these noisy qubits can be processed to detect, monitor, and fix qubit problems without erasing the sensitive quantum data required for a calculation.
Decoding is the process of determining where mistakes happened and what fixes to use. To keep noise from getting out of hand, a traditional computer must complete the very challenging task of decoding in a certain amount of time.
Investigating the potential of AI supercomputing to speed up decoding will be a primary objective of the NVAQC. Collocating quantum hardware within the centre will enable the creation of parallelised, AI-enhanced, low-latency decoders that run on NVIDIA GB200 Grace Blackwell Superchips.
The NVAQC will also address further quantum error correcting difficulties. Through rigorous simulations of intricate quantum circuits, QuEra and NVIDIA will collaborate to expedite the search for new and enhanced quantum error correction codes.
“The NVAQC will be a crucial instrument for identifying, evaluating, and improving new quantum error correction codes and decoders that can help the entire industry get closer to practical quantum computing,” stated Mikhail Lukin, a codirector of the Harvard Quantum Initiative and a Joshua and Beth Friedman University Professor at Harvard.
Creating Utilisation for Enhanced Quantum Supercomputers
Since most practical quantum algorithms utilise both classical and quantum computing resources equally, an accelerated quantum supercomputer that integrates both types of hardware is ultimately needed.
Quantum computations, for instance, frequently require the output of classical supercomputers. Research on creating and refining these hybrid algorithms is made possible by the heterogeneous compute infrastructure that the NVAQC offers.
Through work with Quantinuum, the NVAQC will also investigate new AI-based compilation approaches that could speed up the execution of all quantum algorithms. Using the NVIDIA CUDA-Q platform, Quantinuum will provide its hardware and emulators, expanding on its prior integration work with NVIDIA. CUDA-Q users can now test out Quantinuum’s System H1 QPU emulation and hardware for ninety days.
Integration of QPUs
In order to run practical quantum hardware, one of the fundamental obstacles still standing is the integration of quantum hardware with AI supercomputing.
An integration of this kind may have very high requirements. The ability to transfer data from millions of qubits between quantum and classical hardware at ultralow latencies is necessary for the decoding needed by quantum error correction.
NVIDIA and Quantum Machines will collaborate at the NVAQC to create and refine novel controller technologies that enable fast, high-bandwidth interfaces between GB200 superchips and quantum computers.
Integrating quantum and classical hardware requires a platform that enables developers and researchers to swiftly switch between these two different computing paradigms within a single application. Researchers will be able to utilise the quantum-classical integration of the NVAQC using the NVIDIA CUDA-Q platform.
Building on technologies such as CUDA-Q and NVIDIA DGX Quantum, a reference architecture for combining quantum and classical hardware, the NVAQC is expected to serve as a hub for next-generation advancements in quantum computing, accelerating the development of qubits into significant quantum computers.
