Intel Corp. today unveiled its next-generation AI solutions with the launch of Xeon 6 and Habana Labs Gaudi 3 AI accelerators, in another attempt to capture a major share of the AI chip market. The Xeon 6 is the last word in the servers and supercomputers market, it is a core processor (CPU) for servers, which challenges AMD's rival chip. The Gaudi 3 graphics accelerator was developed in Israel by Habana Labs, an Israeli subsidiary of Intel. According to Intel, Gaudi presents a performance-price ratio 1.8 times higher than Nvidia's H100 processor, working with language models and its market price is estimated at about a third of its Nvidia's new Blackwell platform.
Intel says the Gaudi 3 has been developed with twice the computing power of the Gaudi 2 and double the network bandwidth. The acquisition of Caesarea-based Habana Labs was Intel's great hope in the field of AI. It was acquired in a procedure that some say was too hasty about five years ago for $2 billion, and since then the company has undergone many changes. In fact, Gaudí 3 is the swan song of the Israeli company, which until now operated independently and in recent months was dissolved as an independent company and completely merged into Intel. Habana's founders Dudi Dahan and Ran Halutz left the company to rejoin cofounder and serial chip entrepreneur Avigdor Willenz, to found a new company called Touch, as revealed last month by "Globes."
Intel announced that the Gaudi 3 will be integrated into Dell, Super Micro and Lenovo's servers.
This will be Gaudi's last version in its current format as an independent brand, after it has failed to capture market share compared with the rival graphic processors of Nvidia, and to some degree AMD. The next version will integrate parts of the graphics accelerator of the Israeli processor together with components previously developed within Intel, and "Falcon Shores" software.
The future chip is due to be released next year as a processor focused on running AI models (inference) which, according to the company, will excel at a lower cost compared to rivals and in practice will compete with Nvidia's graphics processor for servers. Intel is aware of the fact that the main difficulty in gaining market share is that software developers have adapted to the Nvidia development environment and therefore promise to provide tools that will help developers move from the existing solutions of Nvidia to those of Intel, such as dedicated sites like GitHub and support platforms like the developer forum of Habana Labs.
How much will the new chip cost?
According to "Forbes", a Gaudi 3-based server is expected to cost about $65,000, a third of the cost of a rival Nvidia product. Analysts say the new chip is expected to bring in $500 million for Intel by the end of the year. The company hopes to break through the glass ceiling and record a significant increase in revenue from AI products this year, including the Lunar Lake processors that provide AI capabilities for PCs. Gaudi has an advantage that Nvidia does not have, which is the ability to work with the popular Ethernet communication standards. Intel is trying to leverage this advantage through alliances with Samsung, Google and HPE. Nvidia, for its part, is trying to move the market to the Infiniband standard of its Israeli company Mellanox.
The Gaudi 3 processor is manufactured by TSMC, which although competing with Intel, produces more and more series of chips made by it. Just a few weeks ago, Intel announced the Lunar Lake AI chips, will also be manufactured outside of its factories. An exception to this is the new Xeon, whose core is manufactured in Intel plants in Ireland and several other components in it, such as the input-output components, are manufactured in Intel plants in Kiryat Gat and Arizona.
Published by Globes, Israel business news - en.globes.co.il - on September 24, 2024.
© Copyright of Globes Publisher Itonut (1983) Ltd., 2024.