AMD says Meta is using its cloud chip as it rolls out AI strategy update

(Reuters) – Advanced Micro Devices (AMD.O) said on Tuesday that companies such as Facebook parent Meta Platforms are ordering high volumes of a new central processor, as investors were anxious for updates on a different new chip that is designed specifically for artificial intelligence systems like ChatGPT. Chief Executive Lisa Su said AMD has started shipping its “Bergamo” central processor. Alexis Black Bjorlin, who oversees computing infrastructure at Meta, (META.O), said the firm has adopted it. They spoke at an event in San Francisco where AMD was set to introduce its new AI chip.

The update came as AMD planned to reveal new details about an AI “superchip” that analysts believe will be a strong challenger to Nvidia Corp (NVDA.O), whose chips dominate the fast-growing artificial intelligence market.

Analysts expect fresh details about a chip called the MI300, AMD’s most advanced graphics processing unit (GPU), the category of chips that companies like OpenAI use to develop products such as ChatGPT.

Nvidia dominates the AI computing market with 80% to 95% of market share, according to analysts. Last month, Nvidia’s market capitalization briefly touched $1 trillion after the company said it expected a jump in revenue after it secured new chip supplies to meet surging demand.

Nvidia has few competitors working at a large 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 at Alphabet Inc’s (GOOGL.O) Google and’s (AMZN.O) cloud unit, both of which rent their custom chips to outside developers.

AMD’s new chip won’t change that right away. Su told investors during an earnings call last month that the MI300 will start generating sales in the fourth quarter, but “will be more meaningful in 2024.”

As well, Nvidia’s lead has come not just from its chips, but from more than a decade of providing software tools to AI researchers and learning to anticipate what they will need in chips that take years to design.

Kevin Krewell, principal analyst at TIRIAS Research, said AMD will have to catch up on both software and research trends.

“They have to connect with the research community to understand what the market is doing and what they need to do, well ahead of when something becomes important,” Krewell said.

Was it worth reading? Let us know.