Nvidia CEO Predicts Quantum Computing's Major Breakthrough

Published Date : 12/06/2025 

Nvidia CEO Jensen Huang discusses the upcoming inflection point in quantum computing, unveiling new technologies like CUDA-Q and the Grace Blackwell platform. He envisions a future where quantum and classical computing work together to solve complex problems. 

Nvidia CEO Jensen Huang is turning his attention to the next critical enabler of the artificial intelligence (AI) revolution: quantum computing. “There’s an inflection point happening in quantum computing,” Huang said Wednesday (June 11) at the Viva Tech 2025 conference in Paris. “It is clear now we’re within reach of being able to apply quantum classical computing in areas that can solve some interesting problems in the coming years.”

Huang had famously bet Nvidia’s future on AI and GPUs about a decade before the technology became a household name. At a keynote speech at Viva Tech, which is France’s version of CES, Huang revealed Nvidia is positioning itself to take advantage of the next stop in AI’s timeline. Quantum computing is a class of computers that are magnitudes more powerful than today’s classical computers. Quantum computers can enable faster processing of AI, which is smart software. Quantum computers use quantum bits (qubits) that can process much more data than classical computers, which store data in bits, or ones and zeroes.

Huang expressed a more optimistic view of quantum advancements after comments he made in January tanked some quantum stocks. Back then, he cast doubt on whether useful quantum computers could come online in the next 15 years, according to CNBC. Huang later said he was wrong. At Viva Tech, Huang unveiled CUDA-Q, a new extension of its CUDA platform designed for quantum-classical hybrid computing. CUDA-Q is an open-source hybrid computing platform that lets the hardware and software needed to run quantum computing applications work together.

“For at least the next generation of supercomputers, every single one of them will have a QPU (quantum processing unit) assigned and QPU connected to GPUs,” Huang predicted, describing a future where quantum and classical architectures work together to solve problems once thought unsolvable. Huang said that just as Moore’s Law once predicted exponential growth in classical computing, quantum computing is now poised for a similar trajectory. “I can totally expect 10 times more logical qubits every five years, 100 times more logical qubits every 10 years,” Huang said, citing advances in error correction, robustness, and scalability that are now within reach.

Beyond quantum, Huang also unveiled Nvidia’s next-generation Grace Blackwell platform, which Huang described as a “thinking machine” architected for reasoning and planning. This hardware leap enables the creation and operation of digital twins — digital replicas of physical systems that can be designed and tested virtually before being deployed in the real world. “Because of the scale and the speed by which we can now simulate almost everything, we can turn everything into a digital twin,” Huang said. “Everything physical will be built visually.”

But these AI capabilities need a new type of data center. Huang calls them AI factories. While they may look like traditional data centers, they are not places to merely store and retrieve files. Instead, AI factories process and generate smart tokens (such as groups of words and other data) for AI applications, Huang said, just as car factories manufacture vehicles. “Every country, every society, every company will depend on” intelligence infrastructure, Huang said. “Europe has now awakened to the importance of these AI factories, the importance of the AI infrastructure.”

Huang traced AI’s evolution from perception (computer vision and recognition) to generative AI (multimodal models capable of producing images, text, and other data forms), and now to agentic AI. He defined this new paradigm as intelligence that can perceive, reason, plan, and execute tasks — mirroring the steps of human thought and action. Agentic models enable systems to break down complex problems, research solutions, use tools, and act step by step — moving far beyond simple pattern recognition prevalent in one-step generative AI such as prompting ChatGPT and getting a response. Moreover, agentic AI is not limited to the digital realm, according to Huang. It can be embedded in robots to make them act more intelligently and with context in the physical world.

This vision is underpinned by Nvidia’s Omniverse platform, which provides a simulation environment for developing, training, and deploying both virtual and physical agents. Digital twins powered by Omniverse are being used to simulate factories, warehouses, rail stations, and even fusion reactors before they’re built to increase efficiency and reduce risk. It is being used by companies such as Mercedes Benz and BMW. Huang also announced that Nvidia is partnering with Schneider Electric and French startup Mistral AI to build large-scale AI clouds optimized for open and proprietary models. These clouds are intended to enable both foundational model training and enterprise-grade agentic AI applications.

Nvidia’s industrial AI cloud, launching in Europe, will support real-time design and simulation, from wind tunnels to humanoid robot motion training. Huang described this as the fourth industrial revolution — following steam, electricity, and information — anchored by the convergence of simulation, robotics, and generative AI. With robotics becoming smarter and more customized, Huang sees a day when even mom-and-pop stores and small to medium-sized businesses (SMBs) will have their own robots. These robots will be teachable, reusable, and deployable in industries that could never afford automation before, Huang said. “It was impossible to have that programming capability — until now.” 

Frequently Asked Questions (FAQS): 

Q: What is quantum computing?

A: Quantum computing is a type of computing that uses quantum bits (qubits) instead of classical bits. Qubits can process much more data and perform complex calculations faster than classical computers, making it ideal for solving intricate problems in fields like AI and simulation.

Q: What is CUDA-Q?

A: CUDA-Q is an open-source hybrid computing platform developed by Nvidia. It enables the hardware and software required to run quantum computing applications to work together seamlessly, bridging the gap between quantum and classical computing.

Q: What is the Grace Blackwell platform?

A: The Grace Blackwell platform is Nvidia’s next-generation hardware designed for reasoning and planning. It is a ‘thinking machine’ that supports the creation and operation of digital twins, which are virtual replicas of physical systems used for simulation and testing.

Q: What are digital twins?

A: Digital twins are virtual replicas of physical systems or objects. They are used to simulate and test various scenarios before deploying them in the real world, helping to increase efficiency and reduce risks in industries like manufacturing, logistics, and energy.

Q: What is agentic AI?

A: Agentic AI refers to artificial intelligence that can perceive, reason, plan, and execute tasks. It goes beyond simple pattern recognition and can break down complex problems, research solutions, use tools, and act step by step, mirroring human thought and action. 

More Related Topics :