Nvidia chief executive Jensen Huang has announced a series of self-driving and robotics developments at the CES technology show in Las Vegas, arguing that artificial intelligence is triggering a fundamental “reset” of the global computing industry.
Speaking during his keynote address, Huang said advances in AI are transforming software, hardware and the physical world.
“Every 10 to 15 years, the computer industry resets,” Huang said. “This time, you no longer program the software — you train it.”
Huang said the shift to AI-native computing is being accompanied by a reinvention of the technology stack, as workloads move from CPUs to GPU-accelerated systems and applications become generative rather than precompiled.
He pointed to recent advances in large language models, reasoning systems and agentic AI as key drivers of the change.
Huang also highlighted the rapid growth of open-source AI models, which he said had accelerated adoption across startups, large enterprises and research institutions.
“Open models have really revolutionised artificial intelligence,” he said, adding that Nvidia is now developing many of its own frontier models in the open, supported by its DGX Cloud supercomputing infrastructure.
Nvidia and self-driving cars
A major focus of the keynote was Nvidia’s progress in autonomous driving, which Huang described as the first large-scale commercial application of what he calls “physical AI” — systems that understand and interact with the real world.
“We started working on self-driving cars eight years ago,” Huang said, explaining that Nvidia entered the sector to understand how AI would reshape the entire computing stack. That effort, he said, required the company to build everything from chips and infrastructure to software models and simulation tools.
Huang formally unveiled Alpamayo, an end-to-end autonomous driving model that Nvidia has open sourced. The system has been developed by a team of several thousand engineers and in partnership with Mercedes-Benz, which began working with Nvidia five years ago.
The company expects its first autonomous vehicle, built with Mercedes-Benz, to launch in the US in the first quarter of this year, followed by Europe in the second quarter and Asia later in the year. Huang said the software would be continuously updated with newer versions of Alpamayo.
Nvidia’s approach to safety, he claimed, relies on reasoning rather than attempting to predefine every possible driving scenario.
“Long tails will be decomposed into quite normal circumstances that the car knows how to deal with,” Huang said.
The system runs two autonomous driving stacks in parallel: Alpamayo and a more traditional, rules-based stack designed to act as a safety fallback.
“If it’s a circumstance that I’m not very confident in, the policy and safety evaluator decides we go back to a simpler, safer guardrail system,” Huang said.
He said the Mercedes-Benz CLA equipped with Nvidia’s technology had already received top safety ratings, and claimed it was the only autonomous system he knew of in which “every single line of code, the chip, the system” was safety certified.
Looking ahead, Huang said autonomous vehicles are likely to be followed by rapid advances in robotics, as the same techniques used in self-driving cars are applied to machines in factories, warehouses and other physical environments.
“This is going to be the first large-scale mainstream physical AI market,” he said. “In the next 10 years, a very large percentage of the world’s cars will be autonomous or highly autonomous.”
Huang stressed that while Nvidia has built a fully integrated autonomous driving stack with Mercedes-Benz, the technology is intended to be used broadly across the automotive and robotics industries.
Nvidia and Siemens
Huang also pointed to Nvidia’s expanding partnership with Siemens as evidence that the same AI systems being developed for self-driving cars are moving into industrial environments.
The companies are working together to apply Nvidia’s AI, simulation and digital twin technologies to factories, infrastructure and industrial automation, allowing machines and production lines to be designed, tested and optimised virtually before being deployed in the real world.
Huang said the collaboration underscored Nvidia’s push beyond data centres and vehicles into what it calls physical AI, as manufacturers seek to use AI to improve efficiency, resilience and automation across global supply chains.