Nvidia CEO Jensen Huang seems to have changed his tune about quantum computing.
Speaking at Nvidia’s first GTC Paris event this week, he said, “Quantum computing is reaching an inflection point.”
Not that Huang was ever bearish on the subject of quantum computing–after all, his company has partnered with a large portion of the quantum ecosystem since its earliest days–but that simple statement is a far cry from what Huang said back in January when he suggested that “very useful” quantum computers were likely 20 years away. The quantum sector and the stock market to some degree overreacted to that forecast, and Huang in March invited his quantum partners to the company’s GTC Spring to brow-beat him a little about his earlier statement, and talk about why the quantum future looks so bright.
Now, Huang sounds full-on bullish, particularly when it comes to the integration of quantum and classical computing resources:
“It is clear now we're within reach of being able to apply quantum computing and quantum-classical computing in areas that can solve some interesting problems in the coming years,” he said in Paris. “This is a really exciting time. We've been working with all of the supercomputing centers, and it's very clear now that over the next several years, or at least, with the next generation of supercomputers, every single one of them will have a QPU [quantum processing unit] assigned and connected to GPUs. The QPU will do quantum computing, of course, and the GPUs would be used for pre-processing, for control, for error correction, which will be intensely computationally-intensive, post processing and such between the two architectures. Just as we accelerated the CPU, there will be a QPU working with the GPU to enable the next generation of computing.”
Along with Huang’s keynote speech, Nvidia issued a press release describing five quantum computing-related workloads that its latest GPU-driven powerhouse–the Blackwell NVL72 GB200–is supporting, including improving error correction, developing better quantum algorithms, designing low-noise-qubits, generating quantum training data, and exploring hybrid applications.
Nvidia partners, such as France-based Alice&Bob, accompanied Nvidia’s release with their own statements. Alice&Bob announced the ongoing integration of the Nvidia CUDA-Q platform for hybrid quantum-classical computing with its own Dynamiqs libraries, a combination which the French firm claimed “outperforms the most widely used libraries today, accelerating the simulation of complex quantum dynamics by up to 75x, on early benchmarks.”
“Simulation is a critical step in the development of useful quantum processors, allowing us to understand how these complex quantum systems behave,” said Théau Peronnin, CEO of Alice & Bob. “Thanks to the integration with Nvidia CUDA-Q, Dynamiqs can now run these simulations even faster, speeding up the development of our QPUs.”
Meanwhile, underscoring Huang’s point about supercomputers incorporating a hybrid quantum-classical approach, Ansys announced it is using CUDA-Q running on the Danish Center for AI’s (DCAI) Gefion supercomputer in Copenhagen to perform GPU-accelerated simulations of quantum algorithms applicable to fluid dynamics applications.
“CUDA-Q’s GPU-accelerated simulations have allowed us to study quantum applications in the regimes where we can really begin to see their effects,” said Prith Banerjee, chief technology officer of Ansys. “Working with Nvidia and DCAI, we’re expanding the role of quantum computing in engineering disciplines like computational fluid dynamics.”
“We’re seeing how CUDA-Q can unlock hybrid quantum-classical computing for researchers using Gefion,” said DCAI CEO Nadia Carlsten, who was a key exec and Alphabet quantum spin-off SandboxAQ before taking the helm at DCAI last year. “Partnering with Nvidia and Ansys has allowed us to drive the convergence of quantum technologies and AI supercomputing.”
Also this week, quantum computing firm Diraq said it is using the Nvidia DGX Quantum reference architecture to connect its spins-in-silicon qubits to Nvidia GPUs.