By Jane Lanhee Lee
OAKLAND, California (Reuters) – U.S. chip giant Intel Corp said on Wednesday its first semiconductor for data center customers focused on power efficiency, Sierra Forest, would be delivered in the first half of next year, as it outlined a chip release schedule after prior delays.
“It’s been a challenging few years as we had introduced a lot of innovation but also a lot of complexity and our product release dates had pushed out,” Intel Data Center and AI Group head Sandra Rivera told Reuters ahead of an investor event.
Intel still dominates the markets for PC and server processing chips, with a market share greater than 70%, tech research firm IDC has calculated. But that is down from more than 90% in 2017.
Intel’s most powerful fourth-generation Xeon processor for data centers, Sapphire Rapids, had faced delays that gave competitor Advanced Micro Devices Inc time to catch up. But Rivera said Intel’s “roadmap is on track” and was “hitting all of our key engineering milestones”.
She added the fifth-generation Xeon processor, Emerald Rapids, was on track to be delivered at the end of this year and the next-generation Granite Rapids, which customers are testing now, would be delivered next year after the release of Sierra Forest.
Intel said the next power-efficient chip, Clearwater Forest, would come to market in 2025.
Rivera said that Intel was also working on building the Intel Developer Cloud with 256 Xeon chips and 512 Gaudi chips for artificial intelligence (AI) that would be available for AI developers to train and run new models. She said AI startups Hugging Face and Stable Diffusion were already using Intel chips.
(Reporting By Jane Lanhee Lee; Editing by Jamie Freed)
