By Ian King
Intel Corp. rolls out a new version of its artificial intelligence chip, aiming to compete with Nvidia Corp. in one of the fastest growing sectors of the semiconductor industry.
The updated processor, called Gaudi 3, will be widely available in the third quarter, Intel said at a company event Tuesday. The chip is designed to improve performance in two key areas: helping train AI systems – a process that involves bombarding them with data – and running the finished software.
Growing demand for AI services has caused tech companies to flock to these so-called accelerator chips, but Nvidia has reaped most of the benefits. Previous versions of Gaudi failed to achieve the market share gains that Intel hoped for, CEO Pat Gelsinger said. He hopes the new model will have a greater impact.
But challenging Nvidia won’t be easy. The runaway success of the company’s H100 accelerator helped more than double its revenues and boost its market valuation to more than $2 trillion. And now Nvidia is looking to extend its lead with a recently announced chip platform called Blackwell. Systems based on this product will be available later this year, the company announced in March.
According to Intel’s assessment, Gaudi 3 will be faster and more energy efficient than the H100. It will train certain types of AI models 1.7 times faster and be 1.5 times better at running software, the chipmaker claims. The product will be roughly equal to Nvidia’s new H200, Intel said, with slightly better performance in some areas and a little behind in others.
Intel, based in Santa Clara, California, said it could not provide comparisons to Nvidia’s upcoming Blackwell line until those products were publicly available. Intel’s rival, Advanced Micro Devices Inc. — its longtime rival in personal computer processors — is also getting into the business. It unveiled a line of accelerators called MI300 in December.
Intel’s Gelsinger said it’s not just trying to catch up with Nvidia. He expects AI to bring a greater windfall to the industry, especially as the technology spreads beyond its current concentration in the data centers of companies such as Microsoft Corp. and Google from Alphabet Inc. Personal computers, mobile phones and networking equipment will require chips that can handle AI tasks and provide users with instant feedback – something not always possible with battery packs. remote servers.
When Intel reported its fourth-quarter results in late January, Gelsinger said it was increasing Gaudi’s supply to meet growing orders and that the company had a 2024 “pipeline” of “more than $2 billion.” dollars and growing.” The broader market for enterprise spending on generative AI equipment will grow from $40 billion in 2024 to $151 billion in 2027, Intel said, citing market research.
But that only highlights Nvidia’s lead. The company had data center revenue of more than $47 billion in the 12 months ended January. For the current fiscal year, that total will exceed $95 billion, according to analyst estimates.
First publication: April 9, 2024 | 9:31 p.m. STI
