June 1[ads1]3 (Reuters) – Advanced Micro Devices Inc ( AMD.O ) on Tuesday provided new details about an artificial intelligence chip that will challenge market leader Nvidia Corp ( NVDA.O ), but the company left out what Wall Street wanted to know – which plans to buy it.
Santa Clara, Calif.-based AMD said the upcoming chip, which will begin trickling out in the third quarter followed by mass production starting in the fourth quarter, will have 192 gigabytes of memory.
It could help technology companies come to grips with the rising costs of providing services similar to ChatGPT, AMD CEO Lisa Su told Reuters in an interview. She was speaking after a keynote presentation in San Francisco where Su demonstrated an AI system on the MI300X chip and wrote a poem about the city.
“The more memory you have, the larger set of models” the chip can handle, Su said. “We have seen in the customer load that it goes much faster. We really think it makes a difference.”
But unlike previous presentations where AMD has talked up a major customer for a new chip, AMD did not say who will adopt the MI300X or a smaller version called the MI300A. The company did not provide details on how much the chip will cost or how it will boost sales.
AMD’s shares have doubled in price since the start of the year and hit a 16-month high earlier on Tuesday, but closed down 3.6% after the presentation of its AI strategy. Nvidia shares ended 3.9% higher at $410.22, making it the first chip maker to close with a market capitalization above $1 trillion.
“I think the lack of a (major customer) saying they want to use the MI300 A or X may have disappointed the Street. They want AMD to say they’ve replaced Nvidia in some design,” said Kevin Krewell, principal analyst at TIRIAS Research.
Nvidia, whose shares have risen 170% so far this year, dominates the AI computing market with an 80% to 95% market share, according to analysts.
Nvidia has few competitors that work at scale. While Intel Corp ( INTC.O ) and several startups such as Cerebras Systems and SambaNova Systems have competing products, Nvidia’s biggest sales threat so far is the internal chip efforts of Alphabet Inc ( GOOGL.O ) Google and Amazon.com ( AMZN.O ) ) cloud device, both of which rent out their custom chips to external developers.
Aside from the AI market, AMD said it has started shipping large volumes of a general-purpose central processing chip called “Bergamo” to companies such as Meta Platforms ( META.O ).
Alexis Black Bjorlin, who oversees data infrastructure at Facebook parent Meta, said the firm has adopted the Bergamo chip, which targets another part of AMD’s data center business that caters to cloud providers and other major chip buyers.
But investors were searching for news about AI. Nvidia’s leadership there has come not only from the chips, but also from more than a decade of providing software tools to AI researchers and learning to anticipate what they will need from chips that take years to design.
AMD on Tuesday delivered updates to its Rocm software, which competes against Nvidia’s Cuda software platform.
Soumith Chintala, a vice president at Meta that helped create open-source artificial intelligence software, said during the presentation that he has been working closely with AMD to make it easier for AI developers to use free tools to switch from the “easy dominant supplier” of AI chips to other offerings such as those from AMD.
“You don’t actually have to do that much work — or almost no work in many cases — to go from one platform to the other,” Chintala said.
But analysts said that just because sophisticated companies like Meta could wring good speeds from AMD chips, there was no promise of wider market traction with less sophisticated buyers.
“People are still not convinced that AMD’s software solution is competitive with Nvidia’s, even though it is competitive on the hardware performance side,” said Anshel Sag, analyst at Moor Insights & Strategy.
Reporting by Stephen Nellis in San Francisco; Editing by Leslie Adler, David Gregorio, Nick Zieminski, and Mark Porter
Our standards: Thomson Reuters Trust Principles.