At the 2024 Taipei International Information Technology Show, better known as Computex, Nvidia CEO Jensen Huang highlighted the endless possibilities surrounding artificial intelligence. He unveiled plans for new AI accelerator chips, signaling the company's shift from graphics cards to AI chips.
In March, Nvidia unveiled the $70,000 Blackwell B200 GPU chip, the "world's most powerful AI chip." On Sunday, Huang told the audience about the Rubin AI chip slated for 2026. He said the new chip would use HBM4, the next iteration of the essential high-bandwidth memory.
Huang stressed that companies that fail to embrace AI will be left behind. He said companies can only manage the "computation inflation" of ever-expanding data needs through AI chips.
Combining two processors on a personal computer is already standard practice for increasing computational power, but try this at a data center…
"The more you buy, the more you save. This is the CEO's math. It is not accurate, but it is correct!" the executive told the audience.
Huang's new chip unveiling comes ahead of Monday's cash session in New York. The company's shares are inching closer to the $3 trillion valuation mark.
TechRadar's John Loeffler had a running blog during Computex, commenting on Huang's speech. One big takeaway from Loeffler is Nvidia's transformation from a graphics card company to an AI chip company.
Loeffler continued:
Huang didn't offer many specifics on the new chip that will begin shipping in 2026. However, during an earnings call earlier this year, the exec said the chip giant will design new platforms annually, down from every two years.