"Everything that moves will be robotic someday and it will be soon."

In a wide-ranging interview with Cleo Abram for the “Huge Conversations” series, NVIDIA CEO Jensen Huang traced the arc of modern computing from 90s video games to the current AI explosion, outlining a future driven by accelerated computing, digital biology, and ubiquitous robotics.

Reflecting on his company’s journey, Huang declared, “We found ourselves being one of the most important technology companies in the world and potentially ever.” He discussed the core beliefs that guided NVIDIA through decades of high-stakes bets, culminating in a new era of AI that he believes will empower humanity and reshape every industry.

From Gaming’s ‘Flywheel’ to a Computing ‘Time Machine’

The conversation began by revisiting the origins of the GPU in the early 1990s. Huang explained the foundational insight that drove NVIDIA’s creation: the realization that computer programs are a mix of sequential tasks, best handled by a CPU, and parallelizable tasks, which could be massively accelerated. “The proper computer, the perfect computer, is one that could do sequential processing and parallel processing, not just one or the other,” Huang stated. “That was the big observation.”

NVIDIA chose video games as its initial market, a decision Huang described as strategic. “We had the good observation that video games has potential to be the largest market for entertainment ever,” he said. This large market would provide the R&D budget necessary to fund complex technological development. “That flywheel between technology and market and greater technology was really the flywheel that got NVIDIA to become one of the most important technology companies in the world. It was all because of video games.”

This new form of computing, he argued, acts as a “time machine.” He recounted a quantum chemistry scientist telling him, “Jensen, because of NVIDIA’s work, I can do my life’s work in my lifetime. That’s time travel.” This ability to accelerate discovery, whether in scientific research or simulating a self-driving car in a virtual city, became a central theme of the company’s mission.

The CUDA Bet and the ‘AlexNet’ Moment

As researchers began “tricking” GPUs to solve non-graphics problems, Huang’s team saw an opportunity to unlock their full potential. This led to the creation of CUDA, a platform that made it dramatically easier for programmers to harness the power of parallel processing. The decision, Huang said, was born from a “soup” of external inspiration from medical imaging, internal needs for more dynamic game physics, and a core belief. “Fundamentally the reason why I was certain that CUDA was going to be successful… was because fundamentally our GPU was going to be the highest volume parallel processors built in the world,” he explained.

This bet paid off spectacularly in 2012 with the emergence of AlexNet, a neural network trained on NVIDIA GPUs that shattered records in an image recognition competition. For Huang, this was a pivotal moment. “All of a sudden, computer vision is solved. All of a sudden speech recognition is solved… These incredible problems associated with intelligence… all of a sudden one after another get solved every couple of years. It’s incredible.”

Seeing the potential, NVIDIA asked a critical question: “How far can it go? And if it could go to the limits of what we think it could go… what would it mean for the computer industry?” The conclusion was that deep learning could reshape everything, prompting NVIDIA to “reinvent the entire computing stack… we’ve reinvented computing as we know it.”

This led to a decade of intense investment, long before the current AI boom became mainstream. When asked what that period felt like, Huang was pragmatic. “It probably felt like today,” he remarked. The key was sticking to foundational principles. “At some point you have to believe something,” he stated. “We invested, you know, tens of billions of dollars before it really happened. And yeah, it was 10 long years. But it was fun along the way.”

The Future of Physical AI: Robots Trained in a Multiverse

Looking ahead, Huang focused heavily on physical AI and robotics, delivering one of the interview’s most striking predictions. “Cleo, everything that moves will be robotic someday and it will be soon,” he declared. “The idea that you’ll be pushing around a lawn mower is already kind of silly… Every car is going to be robotic. Humanoid robots, the technology necessary to make it possible, is just around the corner.”

The key to this leap, he explained, is a new paradigm for training. Instead of relying on slow and risky real-world practice, robots will learn in vast, physically-accurate digital worlds. He detailed NVIDIA’s strategy, centered on two platforms: Omniverse, a physics-based simulator, and Cosmos, a new “world foundation model.”

He drew an analogy to large language models. A model like ChatGPT can hallucinate until it is “grounded” by factual data, like a PDF or search results. In the physical world, Huang explained, Cosmos is the generative world model that understands concepts like gravity and object permanence. Omniverse provides the “ground truth” through its physics simulator. “Using the simulator to ground or to condition Cosmos, we can now generate an infinite number of stories of the future. And they’re grounded on physical truth.”

This technology will allow robots to learn from countless simulated scenarios, preparing them for the physical world. For Huang, this leads to a future where we are all accompanied by personal AI assistants. “A future where you’re just surrounded by robots is for certain,” he said. “I think the idea that we’ll have our own R2-D2 for our entire life and it grows up with us, that’s a certainty now.”

AI Safety, Energy Limits, and The Next Big Bets

When asked about the risks, Huang acknowledged a spectrum of challenges from bias and hallucination to impersonation and the engineering complexities of safety-critical systems like self-driving cars. He stressed the need for a multi-layered approach, akin to aviation safety with its redundant systems, pilots, and air traffic control.

The ultimate physical constraint, he noted, is energy. He highlighted NVIDIA’s focus on efficiency, holding up a model of a modern AI supercomputer. The first DGX system, delivered to OpenAI in 2016 for $250,000, required 10,000 times more power than a modern, more performant desktop version. “Eight years later we’ve increased the energy efficiency of computing by 10,000 times,” he said. “And that’s essential because we want to create more intelligent systems.”

Huang also addressed the debate between creating specialized chips versus general-purpose, programmable ones. He firmly sided with flexibility, arguing against “burning” a specific algorithm like the transformer model into hardware. “Transformer is a stepping stone towards evolutions of transformers that are barely recognizable as a transformer years from now,” he argued. “We believe in the richness of innovation.”

His current bets extend beyond robotics into two other transformative fields:

• Digital Biology: “So that we can understand the language of molecules and understand the language of cells… if we can learn that, and we can predict it, then all of a sudden our ability to have a digital twin of the human is plausible.”
• Climate Science: Creating high-resolution models to “predict the regional climates, the weather patterns within a kilometer above your head,” with profound implications for managing climate change.

Becoming Superhuman

Ultimately, Huang’s vision is one of human empowerment. When asked for his advice to the public, he was unequivocal: “Go get yourself an AI tutor right away.” He argued that learning to prompt and interact with AI is the critical skill for the next generation, just as his generation had to learn to use computers.

“The next generation… has to ask… ‘How can I use AI to do my job better?’ That is start and finish I think for everybody,” he said. He dismissed fears of being replaced, comparing AI to being surrounded by brilliant experts. “It actually empowers me and gives me the confidence to go tackle more and more ambitious things,” he explained. “We’re going to become super humans, not because we have super [powers], we’re going to become super humans because we have super AIs.”