The billionaire boss of the chipmaker Nvidia, Jensen Huang, has unveiled new AI technology that he says will help self-driving cars think like humans to navigate more complex situations.
The world’s most valuable company is to roll out the new technology, Alpamayo, which is designed to help self-driving cars handle tricky situations such as sudden roadworks or unusual driver behaviour on the road, rather than just reacting to previous patterns.
Nvidia claims Alpamayo will bring chain-of-thought reasoning to self-driving vehicles, combining what the car sees with language-like reasoning.
In a speech at the annual Consumer Electronics Show in Las Vegas, the Nvidia founder and chief executive said: “The ChatGPT moment for physical AI is here, when machines begin to understand, reason and act in the real world. Robotaxis are among the first to benefit.
“Alpamayo brings reasoning to autonomous vehicles, allowing them to think through rare scenarios, drive safely in complex environments and explain their driving decision. It’s the foundation for safe, scalable autonomy.”
Nvidia has started producing a driverless car powered by its technology, the Mercedes-Benz CLA, in partnership with the German carmaker. It will be launched in the US in the coming months, followed by Europe and Asia. Huang showed a video of the car driving through San Francisco with a passenger sitting behind the steering wheel, their hands in their lap.
“It drives so naturally because it learned directly from human demonstrators,” Huang said, “but in every single scenario … it tells you what it’s going to do, and it reasons about what it’s about to do.”
Huang also said the company’s next generation of chips was in full production, and they could deliver five times the computing power of the company’s previous products when serving up chatbots and other AI apps.
He revealed new details about the chips, which will arrive later this year as Nvidia faces increasing competition from rivals as well as its own customers.
The Vera Rubin platform, made up of six separate Nvidia chips, is expected to debut later this year. The flagship server will contain 72 of the company’s graphics units and 36 of its new central processors.
Huang showed how they could be strung together into “pods” with more than 1,000 Rubin chips and said they could improve the efficiency of generating what are known as “tokens” – the fundamental unit of AI systems – by 10 times.
To get the new performance results, Huang said the Rubin chips used a proprietary kind of data that the company hopes the wider industry will adopt.
While Nvidia still dominates the market for training AI models, it faces far more competition – from traditional rivals such as Advanced Micro Devices as well as customers such as Alphabet’s Google – in delivering the fruits of those products to hundreds of millions of users of chatbots and other technologies.
