How Nvidia plans to solve quantum computing's networking problem - SDxCentral

How Nvidia plans to solve quantum computing's networking problem Nvidia's quantum GM discusses NVQLink and why useful quantum computers can't exist without classical AI support October 31, 2025 By Ben Wodecki Have your say Facebook Twitter LinkedIn Reddit Email Share – Nvidia Nvidia CEO Jensen Huang during a panel at CES in January expressed his view as to when quantum computing might become a practical reality: “If you said 15 years for very useful quantum computers, that would probably be on the early side. If you said 30, it’s probably on the late side.”Those comments might seem innocuous to some, but in reality they caused an unwitting cascade upon the share prices of firms hoping to bring quantum technologies to life.The likes of IonQ, Rigetti Computing, and D-Wave Quantum saw substantial losses brought on by the reaction to Huang’s CES panel comments. Nvidia also suffered, with its share price slipping in the days that followed.Those share prices, for the most part, would go on to recover, with Nvidia later becoming the first company in history to be valued above $5 trillion.Huang, however, would walk those comments back just two months later at the company’s annual GTC conference. He went as far as to install a dedicated "Quantum Day" as part of the event, while also revealing plans for a quantum research center in Boston.What a difference two months make. And in the weeks and months since, Nvidia would invest in a host of firms developing quantum technologies, including PsiQuantum, QuEra, and Quantinuum.Nvidia’s about-face culminated this week with the tech giant providing perhaps the biggest piece of quantum news in 2025: unveiling NVQLink, an interconnect method that would allow quantum computers to communicate with classical systems. 28 Oct 2025 Nvidia unveils NVQLink: A way for quantum computers to interconnect with classical systems Breakthrough could enable GPU-based systems to communicate with QPUs with latency as low as four microseconds With the launch of NVQLink, the chip giant has now kicked off what it’s calling the “quantum-GPU computing era.”At the center of that era for Nvidia is Tim Costa, who serves as the company's GM for industrial and computational engineering and quantum. Costa told SDxCentral that there wasn’t a step change in the wake of Huang’s comments and subsequent walk back, with Nvidia instead making steady strides in recent years to grow its quantum offerings as an effective extension of its existing work.“The [quantum] team has been growing consistently and quickly,” Costa said. “If you look at our product release over the years, it was quantum one year, CUDA-Q the next year, [and] it was DGX Quantum after that. [It’s] really just a steady progression of building the different things needed to be built to collaborate with and support the quantum ecosystem.”The Nvidia GM said Huang’s CES comments had “an unfortunate side effect,” but reminded that the CEO got on stage at GTC in March, said he was wrong, and “we moved on.”The quantum-classical bridge Tim Costa, GM for industrial and computational engineering & quantum – Nvidia The culmination of Nvidia’s quantum work over the past few years has resulted in an interconnect technology that would allow operators and engineers to call GPU computations directly from a quantum processor with latency as low as four microseconds.Speaking to SDxCentral shortly before NVQLink was unveiled, Costa said Nvidia views classical GPU-based computers applying AI to quantum devices as “an absolutely essential step milestone on the road to useful quantum computing.”According to Costa, Nvidia believes a functional quantum computer requires the support from an AI supercomputer to handle challenges like simulation, calibration, and overall device management.“Today, error correction is nascent,” Costa said. “Calibration is hand-tuned by physicists, and you simply can't scale that to the size of systems that will be needed for solving real problems of significance.”For Costa and the quantum team at Nvidia, networking represents one of the most critical and complex challenges in making quantum computing practical.The fundamental issue boils down to a delicate balancing act. Quantum processors need to communicate with classical systems fast enough to preserve quantum coherence, while also handling massive data volumes as systems scale to millions of qubits.“Latency definitely matters,” Costa explained. “I have to get the data off and back within the coherence time of the quantum computer.”Different quantum computing modalities have different requirements, but as systems scale, bandwidth becomes equally crucial, with the GM adding: “You can optimize for latency, and you can optimize for bandwidth. It gets really hard if you've got to optimize for both.”This is where NVQLink comes in. The interconnect aims to solve what Costa described as a “pretty significant challenge” – enabling quantum processors to work seamlessly within heterogeneous computing environments that already include CPUs, GPUs, DPUs, and traditional networking infrastructure. NVQLink concept – Nvidia Nvidia's focus extends beyond just physical connections. The tech giant is developing software layers that provide the necessary hooks for different error correction techniques to interact with quantum processing units (QPUs), as well as AI models for calibration that can be effectively served across the quantum-classical interface.“Our effort in this space is focused on what kind of system infrastructure and network interfaces are going to make the most sense both today, but moving forward,” Costa said. “We also need to look at what kinds of software layers we need so that we have the right hooks for different error correction techniques, because it won't be one technique for all systems.”Looking further ahead, Costa told SDxCentral that while NVQLink lays the initial groundwork for quantum-to-classic, the concept of quantum-to-quantum networking will eventually become critical.Earlier this year, a breakthrough in this space saw University of British Columbia researchers create a “universal translator” for quantum computers, allowing them to communicate over a network with virtually no noise.However, for the team at Nvidia, quantum-to-quantum networking is not the immediate focus.“There's going to come a day where quantum networking is the problem,” Costa said. “We’re not there yet, but it’s going to become the problem. Eventually, you’re going to want to string quantum computers together, so the networking actually becomes fairly complex.“You've got classical systems, you've got quantum systems," Costa continued. "You're going to have networking between the classical systems, and networking between the quantum and classical, and then networking the quantum systems. The needs of each of those networks is quite different, and in particular, for quantum-to-quantum, you can’t use classical networks … you're not actually able to use the two quantum computers together to do any kind of useful computing.”Costa suggested that once individual quantum computers scale sufficiently, networking them together will be the next frontier, much like the evolution of classical computing from single nodes to networked clusters. But for now, the quantum-GPU computing era has just begun. Subscribe to HPC & Quantum for regular news round-ups, market reports, and more.