US chip designer and software specialist Nvidia on Tuesday said that it wants to meet the rapidly growing demand for computing power for artificial intelligence (AI) with a new generation of its chips.
Chief executive officer Jensen Huang said at Nvidia's developer conference GTC in San Jose, California that the company's next-generation Vera Rubin chip was scheduled to be launched in autumn 2026.
Vera Rubin and the further development of the Blackwell system, announced for this year, are expected to drastically reduce the costs of operating AI software compared to previous technology.
Nvidia chips have become a key technology for artificial intelligence.
Industry giants such as Google and Meta, along with AI start-ups like ChatGPT creator OpenAI, rely on Nvidia's chips to train artificial intelligence systems.
This position allowed Nvidia's business to grow explosively in recent years.
Huang also tried to allay investors' concerns that the world could get by with less AI computing power in future - and that expectations of Nvidia's future business could therefore be too high.
He said that the world as a whole is moving towards generating answers from scratch with the help of artificial intelligence, instead of retrieving stored answers. In particular, the new AI models that can build a chain of reasoning step by step to solve a problem are power-hungry.
Overall, you need 100 times more computing capacity than was assumed just a year ago, said Huang.
