Intel launches new AI chip Gaudi 3 to challenge Nvidia

Intel is launching a new version of its artificial intelligence chip, aiming to challenge Nvidia in one of the fastest-growing areas of the semiconductor industry.

The updated processor, called Gaudi 3, will be widely available in the third quarter, Intel said at a company event on Tuesday. The chip is designed to improve performance in two key areas: helping train artificial intelligence systems (a process that involves bombarding them with data) and running finished software.

Booming demand for artificial intelligence services has tech companies rushing to buy these so-called accelerator chips, but Nvidia is already seeing most of the benefits. CEO Pat Gelsinger said early versions of Gaudi failed to achieve the market share gains Intel had hoped for. He expects the new model to have a greater impact.

“They’re a good competitor, but people want an alternative,” he said of Nvidia on Tuesday. “The world needs more suppliers and we are very committed to providing that choice.”

Intel shares rose less than 1% to $38 as of 1:33 pm in New York on Tuesday, reversing earlier losses. As of Monday’s close, the stock price has fallen 24% this year.

Gelsinger declined to disclose pricing but said his chips would be “well below” the cost of Nvidia’s current and future chips. He said they will provide “very good” total cost of ownership.

Challenging Nvidia is not easy. The huge success of the company’s H100 accelerator has more than doubled revenue and taken its market valuation to over $2 trillion (approximately Rs. 1,66,39,120 million). Now, Nvidia is looking to solidify its lead with a just-released chip platform called Blackwell. Systems based on the product will be available later this year, the company said in March.

According to Intel’s assessment, Gaudi 3 will be faster and more power-efficient than H100. The chipmaker claims it will train certain types of artificial intelligence models 1.7 times faster and run software 1.5 times faster. Intel says the product will be roughly on par with Nvidia’s newer H200, performing slightly better in some areas and slightly worse in others.

Santa Clara, Calif.-based Intel said it couldn’t provide comparisons with Nvidia’s upcoming Blackwell series until the products are publicly available. Intel’s rival Advanced Micro Devices, its longtime rival in PC processors, is also moving into the space. It launched a series of accelerators called MI300 in December.

Intel’s Gelsinger said he doesn’t just want to catch up with Nvidia. He expects artificial intelligence to provide an even bigger windfall for the industry, especially as the technology spreads beyond its current concentration in data centers at companies like Microsoft and Alphabet Inc’s Google. PCs, mobile phones and network equipment will need chips that can handle artificial intelligence tasks and provide users with instant feedback – something that remote server farms are not always able to do.

When Intel reported fourth-quarter results in late January, Gelsinger said he was increasing Gaudi supply to meet growing orders and that the company’s “pipeline” for 2024 was “over $2 billion and counting.” increase”. Citing market research, Intel said broader enterprise spending on generative AI devices will increase from $40 billion (approximately Rs 3,32,784 crore) in 2024 to $151 billion (approximately Rs 1,256,387 crore) in 2027.

But this just highlights Nvidia’s lead. The company’s data center revenue exceeded $47 billion (approximately Rs 3,91,061 crore) in the 12 months to January. Analysts estimate that this total will exceed $95 billion (approximately Rs 7,90,442 crore) during the current financial year.

© 2024 Bloomberg

(Except for the headline, this story has not been edited by NDTV staff and is published from a syndicated feed.)