NVIDIA NVQLink Brings Classical Supercomputing and QPUs Together to Usher in the Quantum-GPU Era
NVIDIA NVQLink
On October 28, 2025, NVIDIA unveiled NVIDIA NVQLink, an open system architecture intended to tightly couple GPU computing capabilities with quantum processors (QPUs). The launch of NVQLink, a fast interconnect, creates a first-of-its-kind system that can connect quantum devices to traditional CPU and GPU-based systems. At NVIDIA’s GTC conference in Washington, D.C., this groundbreaking event was unveiled to create accelerated quantum supercomputers.
NVQLink is the “Rosetta Stone connecting quantum and classical supercomputers,” bringing them together into a single, cohesive system that marks the beginning of the quantum-GPU computing era, according to Jensen Huang, founder and CEO of NVIDIA. According to Huang, all NVIDIA GPU scientific supercomputers will soon be hybrid machines that are closely connected to quantum processors, increasing processing potential.
You can also read BTQ and ICTK Sign $15 Million Deal To Develop QCIM Chip
Overcoming Quantum’s Scaling Roadblock: The Need for High-Speed Control
The intrinsic difficulty of scaling quantum hardware is the primary reason for NVQLink’s requirement. The information units known as qubits, which allow quantum computers to process data in novel ways, are fragile and prone to mistakes. Complex control algorithms, calibration, and advanced quantum error correction (QEC) are necessary for quantum systems to function well and provide significant applications.
A “symbiotic relationship” between classical and quantum computing was explained by Tim Costa, general manager of NVIDIA’s engineering, semiconductor, and quantum business areas. The process of making a quantum computer functional entails a significant computational difficulty pertaining to optimal control, calibration, and error correction, even while QPUs unleash the potential to tackle problems that classical systems cannot. Costa underlined that without using a massive AI supercomputer to solve these control problems, there would not be a practical quantum computer.
Large volumes of AI computation must be delivered via an incredibly demanding low-latency, high-throughput connection to the traditional supercomputer for this control process, especially as the size of the quantum processors grows. The intricacy of quantum signals has historically made it difficult to effectively connect these different systems, which has been a significant obstacle to quantum advancement. This vital connection environment was specifically provided by NVQLink’s engineering.
Architectural Integration and Performance Specs
As quantum researchers grow their gear, NVQLink offers a uniform, turnkey solution designed to address the major integration difficulties they encounter. Scientists and engineers can call GPU calculations straight from a quantum processor, the fast link.
Researchers and developers may design and test hybrid applications that smoothly use CPUs and GPUs in addition to quantum processors the architecture the NVIDIA CUDA-Q software platform. The extensible libraries required for QEC are supported by CUDA-Q. The interconnect offering supports all of the main QPU modalities, such as trapped-ion, photonic, neutral atoms, and superconducting systems, and is made to be interoperable.
You can also read How Quantum Investments Are Transforming Cybersecurity
The real-time capabilities required for extensive mistake correction are highlighted in the specifications:
• Minimum GPU-QPU Latency: As low as 4.0 microseconds (round trip, FPGA to GPU to FPGA).
• Maximum GPU-QPU Throughput: Up to 400 Gb/s.
• Leading-Edge AI Performance: 40 PFLOPS (FP4) with sparsity.
From initial calibration to complete fault tolerance, the platform is designed to speed up the quantum workflow. NVQLink accelerates the following important workloads:
QPU Calibration: To achieve zero QPU downtime and enhance the fidelity of quantum operations, real-time calibration necessitates a tight computational coupling.
QEC Decoding: Converting a noisy QPU into a logical, functional QPU by speeding up QEC decoding with low latency and high-throughput computation.
Logical Orchestration: This is crucial for advanced QEC protocols because it enables the execution of complex logical programs through dynamic routing and just-in-time compilation.
A Collaborative Bridge to Fault-Tolerant Computing
Under the direction of researchers from top supercomputing centers at nine U.S. national laboratories, NVQLink was established openly and cooperatively. The Department of Energy’s Oak Ridge National Laboratory, Pacific Northwest National Laboratory, Sandia National Laboratories, Los Alamos National Laboratory, MIT Lincoln Laboratory, Fermi Laboratory, Lawrence Berkeley National Laboratory (Berkeley Lab), and Brookhaven National Laboratory are among the participating labs.
Five control system manufacturers and 17 QPU builders are among the many hardware providers supported by the design. IQM Quantum Computers, Alice & Bob, IonQ, Pasqal, Quantinuum, QuEra, and Rigetti are just a few of the numerous partners of QPU. Zurich Instruments, Qblox, QubiC, Quantum Machines, and Keysight Technologies are partners in control systems.
Government authorities see this partnership as strategically significant. “It must build the bridge to the next era of computing: accelerated quantum supercomputing” to “maintain America’s leadership in high-performance computing,” stressed U.S. Secretary of Energy Chris Wright.
He pointed out that NVQLink offers the vital technology required to bring together top-tier GPU supercomputers and cutting-edge quantum processors, and that close cooperation between national labs, startups, and industry partners, such as NVIDIA, is essential to this endeavor.
NVQLink was cited by numerous partners as a crucial component for future scalability when they swiftly announced their acceptance. Pasqal pointed out that by enabling improved logical architectures, integrating with NVQLink will expedite their roadmaps toward Fault-Tolerant Quantum Computing (FTQC) and is a significant step toward utility-scale quantum computing.
This was confirmed by Alice and Bob, who said that NVQLink addresses important FTQC stack layers, a definite indication that fault-tolerant quantum computers are getting close to commercial maturity. Additionally, Quantum Machines emphasized how the open architecture expands their microsecond-latency control solution, allowing for real-time data interchange and deterministic feedback.
You can also read SpeQtre: UK-Singapore Quantum Satellite Mission Nears Launch
