Jensen Huang is Now ‘Too Sacred to Say a Word’ About Quantum

NVIDIA recently held its GTC conference in DC, with participation from all major tech companies worldwide. Apart from its focus on conventional technologies like GPUs and chips, the giant has lately been delving into deeper technologies, marking a move beyond just core AI tech. 

Quantum and robotics have been its two focus areas at all conferences, and this year’s GTC was no exception. 

Huang opened up his keynote by thanking NVIDIA’s lineup of partners. “We couldn’t do what we do without NVIDIA’s ecosystem of partners,” Huang said, nodding to GTC’s reputation once again, as previously mentioned, “the Super Bowl of AI.”

Enter NVQLink

He followed this with announcing significant partnerships and innovations in CPUs, CUDA and telecommunications, and proposed a way to connect quantum and GPU computing as well. Announcing NVIDIA NVQLink, a quantum‑GPU interconnect with latency as low as 4 microseconds, he set new milestones in the industry.

Previously, during an analyst event at CES (Consumer Electronics Show), Huang had suggested that bringing “very useful quantum computers” to market could take decades, citing the need for quantum processors, or qubits, to increase by a factor of 1 million. This comment instantly sent the industry reeling, triggering a massive selloff in the quantum computing sector and erasing approximately $8 billion in market value. 

This time, when asked to comment on quantum, Huang joked that whenever he says ‘quantum,’ stock prices start rising. He continued, “Quantum, quantum, quantum.”

IBM & AMD Bring Proof

The NVQLink development brings in the concept that classical chips and processors could run quantum calculations. To prove this, IBM recently demonstrated that a crucial quantum error-correction algorithm can be executed on mainstream AMD hardware. 

The experiment highlights how conventional processors could play a much larger role in stabilising and scaling quantum systems, an area long considered one of the field’s biggest hurdles.

IBM has been developing algorithms capable of detecting and correcting quantum errors in real time. According to the company’s latest research, those algorithms can now run on a field-programmable gate array (FPGA) chip produced by AMD, a hardware that’s both widely available and affordable.

Arindam Ghosh, professor of physical sciences at the Indian Institute of Science (IISc), Bengaluru, told AIM that such experiments, while significant for testing, are not examples of true quantum computing. 

“What IBM did is essentially run a quantum algorithm on a classical processor that mimics a quantum chip,” he explained. “It’s not by any means quantum computing—the results remain classical.”

Ghosh said that algorithms like the Quantum Approximate Optimisation Algorithm (QAOA) can be run on classical hardware to approximate quantum behaviour, offering insight into algorithm development but not quantum performance itself.

The milestone forms part of IBM’s broader plan to build its next-generation quantum computer, codenamed ‘Starling’, by 2029. 

NVIDIA Supercomputers to Become Hybrid

NVQLink is developed in collaboration with 17 quantum hardware builders and nine US national laboratories, including Berkeley Lab, Los Alamos, Oak Ridge, and MIT’s Lincoln Laboratory. It is designed to shuttle terabytes of data between quantum processors and GPUs thousands of times per second.

“Every NVIDIA GPU supercomputer will eventually become hybrid,” Huang said during the keynote. “NVQLink is the Rosetta Stone that unites classical and quantum computing into one coherent system.”

The point was echoed directly by the ‘AI for Science’ panel, where leaders from Cadence (Anirudh Devgan, president and CEO), Infleqtion (Matt Kinsela, CEO), Lila Sciences (George Church, chief scientist) and Xaira Therapeutics (Marc Tessier-Lavigne, co-founder, chairman, and CEO) described the same hybrid trajectory. 

They argued that pairing quantum processors with high-performance GPUs lets researchers simulate and iterate far faster, outsourcing stabilisation and parts of quantum error correction to classical accelerators rather than waiting for million-qubit breakthroughs. 

Early quantum advantage will first be realised in domains such as sensing and timekeeping, with materials and drug discovery following as logical qubits scale up, as the panel expressed.

The announcement underscores a deeper technical ambition: to make quantum systems usable long before fully error-corrected quantum machines become available. Error correction is the bottleneck that keeps quantum computers experimental, and NVQLink’s low-latency feedback loop could help bridge that gap by outsourcing the correction workload to GPUs.

“The immediate future lies in hybrid quantum–classical systems,” said Ghosh. “High-performance classical computing will take certain jobs, pass them to the quantum processor, and then iterate the results. That’s where real advantage will emerge first.”

He added that IISc has already established a new vertical dedicated to quantum algorithms, with around 15 faculty members exploring areas where quantum advantage could arrive earlier.

The US energy department is also partnering with NVIDIA to deploy seven new AI supercomputers, each integrating NVQLink for quantum research. 

GPUs May Just Not Become Obsolete

At Cypher 2025, a recent AI conference organised by AIM, A S Rajgopal, founder and CEO of NxtGen, described quantum computing as a force that could make GPUs redundant. 

He recalled visiting the IBM Watson Research Centre in New York to explore IBM’s System Two quantum computer. “Quantum is right on the horizon, and it will happen in my lifetime,” Rajgopal said.

“GPUs are not going away—no chance,” Ghosh said firmly. 

“Classical hardware is essential for quantum computers to run. The immediate and even longer-term solution is hybrid, because you still need a classical computer to read what the quantum computer is saying.”

He noted that classical computing will continue to evolve alongside quantum technologies, much as earlier innovations built upon existing systems rather than replacing them outright.

The post Jensen Huang is Now ‘Too Sacred to Say a Word’ About Quantum appeared first on Analytics India Magazine.

Generated by RSStT. The copyright belongs to the original author.

Source