18A Node — Intel will make chips for Microsoft “I want to manufacture every AI chip in the industry,” says Intel CEO Pat Gelsinger.
Michael Acton, Financial Times – Feb 22, 2024 2:35 pm UTC Enlarge / Intel CEO Pat Gelsinger speaks during the “AI Everywhere” event on December 14, 2023, in New York, NY.Intel Corporation reader comments 29
US chip company Intel will make high-end semiconductors for Microsoft, the companies announced, as it seeks to compete with TSMC and Samsung to supply the next generation of silicon used in artificial intelligence for customers around the world.
Chief executive Pat Gelsinger said at a company event on Wednesday that Intel is set to rebuild Western manufacturing at scale, buoyed by geopolitical concerns in Washington about the need to bring leading-edge manufacturing back to the US.
Since Gelsinger took the helm three years ago, Intel has been attempting to reinvent itself as a foundry business, building chips designed by other companies, regaining an edge in making the most advanced semiconductors. Demand for them is soaring, driven by the rise of generative AI.
But Intel needs to convince manufacturing customers that it will treat the chips the same as the ones it designs itself.
Intel has set itself a goal of ensuring that 50 percent of the worlds semiconductors are built in the US and Europe in a decade, compared with 20 percent today. Most global production is concentrated in Asia, particularly in Taiwan.
Intel is becoming maniacally consumer-focused, Gelsinger said. To reflect this, it is reorganizing into two unitsIntel Foundry, which manufactures semiconductors, and Intel Products, involved in their design.
Gelsinger said that Intel was establishing two vibrant new organizations, adding that the company has already inked foundry deals with a lifetime value of $15 billion. Advertisement
Satya Nadella, Microsoft’s chief executive, said the company was supporting Intels bid to become a leading global chip manufacturer. Intel will produce a Microsoft-designed chip using Intels 18A node.
Intels 18A node is a manufacturing process that makes semiconductors smaller and more energy efficient. Taiwans TSMC and South Koreas Samsung are deploying their own technology to produce the next generation of cutting-edge chips to power smartphones, data centers, and artificial intelligence.
Microsoft is one of several companies, including Nvidia and AMD, that is designing the high-end chips that offer the enormous computing power required to roll out AI.
Through our foundry, I want to manufacture every AI chip in the industry, Gelsinger said.
Those include semiconductors based on the architecture of UK chip designer Arm. Arm chief executive Rene Haas appeared at the same event to announce a new partnership with Intel, quipping that, given their history, the two companies made strange bedfellows. Intels designs are based on the X86 architecture, which competes with Arms.
The federal Chips Act, passed in 2022, promises $52.7 billion in funding for US semiconductor research, development, and manufacturing.
Earlier this week the Biden administration announced that California-based GlobalFoundries would receive $1.5 billion for a new facility and manufacturing expansion and modernization at sites in New York and Vermont, to produce automotive, communications, and defense semiconductors.
Gelsinger said that Intels own federal grant under the Chips Act would be announced very soon.
OpenAI, Google, and others building and implementing the large language models behind generative AI require a mind-boggling volume of semiconductors over the coming years, and demand for the most cutting-edge chips is just going to explode, US Commerce Secretary Gina Raimondo said, appearing onscreen at the event.
Intel is the countrys champion chip company, she said.
Intel shares closed 2.4 percent lower on Wednesday.
2024 The Financial Times Ltd. All rights reserved. Not to be redistributed, copied, or modified in any way. reader comments 29 Advertisement Channel Ars Technica ← Previous story Next story → Related Stories Today on Ars
