Why the World’s Most Valuable Company Is Betting on Quantum Computing

Over the past two weeks, NVIDIA—the powerhouse behind and chief beneficiary of the artificial intelligence boom—has moved into quantum computing. “I’m a little surprised they haven’t done it before,” said Richard Shannon, investment analyst at Craig Hallum.

Through its venture capital arm, NVIDIA has joined other backers in funding startups Quantinuum, QuEra, and PsiQuantum, now valued collectively at more than $17 billion. This marks a notable shift for NVIDIA’s CEO, Jensen Huang. In January, Huang dismissed practical quantum computers as being 15 to 20 years away, sparking a selloff in listed quantum firms. But by March he had softened, and in June he suggested the field had reached an “inflection point” that might enable “interesting problems” to be solved in the near future. NVIDIA declined to comment further.

By positioning itself at the heart of the hardware driving AI models, NVIDIA has become the world’s most valuable company, worth $4 trillion. The company designs GPUs (specialized processors for AI workloads), builds CUDA (software allowing GPUs to interconnect), and delivers complete supercomputers—machine-sized units that AI firms are racing to install in their data centers.

Yet quantum computing is unlikely to directly benefit NVIDIA’s AI clientele. “Quantum computing and AI are sort of diametrically opposed,” said Pete Shadbolt, chief science officer at PsiQuantum, one of NVIDIA’s recent investees. AI thrives on massive datasets and pattern recognition, while quantum systems, by contrast, “hate data, and they love precision,” he explained.

Still, advocates argue quantum technology could usher in a fresh computing paradigm. Rather than executing countless simple tasks in parallel like GPUs, quantum computers aim to solve a small number of extremely complex problems. Quantinuum uses ions, PsiQuantum relies on photons, and QuEra employs neutral atoms. All operate under the peculiar rules of quantum mechanics, which allow particles to exist in multiple states simultaneously. This lets quantum machines pursue many branches of a problem at once, offering solutions beyond the reach of classical computers in practical timeframes. That includes breaking modern encryption—something that could take traditional machines millions of years—in just hours, prompting banks to adopt “quantum-resistant” cryptography.

There are also constructive uses. Quantum computers can model quantum mechanical systems, which classical systems cannot, noted Hsin-Yuan Huang, senior researcher at Google Quantum AI. Since quantum mechanics underpins the physical world, this capacity could accelerate the development of new drugs, materials, and chemical processes. “That’s not going to be solvable with just many GPUs. It’s just inherently too hard,” said Huang.

One promising area is greener ammonia production, currently responsible for 2% of global energy use. PsiQuantum is collaborating with Mercedes-Benz to model lithium-ion battery electrolytes, which could speed electric vehicle battery progress, and with pharmaceutical firm Boehringer Ingelheim to study an enzyme linked to human drug metabolism.

For now, though, the lack of sufficiently large quantum machines leaves practical benefits unproven, said Jan Ole Ernst, a Ph.D. researcher at the University of Oxford. Future breakthroughs may emerge as systems scale, but he cautioned that clear applications remain elusive beyond factoring large numbers.

When such systems will mature remains uncertain. Quantum computing timelines, like many in frontier science, often stretch. PsiQuantum, for example, has begun work on facilities in Australia and Illinois, testing cooling infrastructure for its chips. The company now projects a useful-scale computer by 2027—two years later than it predicted in 2021.

Whenever functional quantum machines do arrive, NVIDIA is well placed. “You’ll never be able to run a quantum computer without a ton of classical processing,” Ernst noted. Conventional systems are required for control, error correction, and output analysis. PsiQuantum already uses NVIDIA’s hardware for both pre- and post-quantum processing. NVIDIA also introduced CUDA-Q in 2022 to link quantum and classical computers, and this year launched the NVIDIA Accelerated Quantum Computing Research Center in Boston to “shorten the timeline to useful quantum computing.”

“I think it’s fair to say that they’re doing everything but the quantum computer,” Shadbolt remarked.

That might not hold forever. Shannon believes NVIDIA’s investments will provide insight into which technologies scale most effectively. “I believe, given enough time, it’s a virtual certainty that NVIDIA buys one or more quantum companies,” he said.