Intel spent the last few years, wandering from one wrong step to another, and even producing its latest chips had to be left to one of its biggest rivals.
Now, in order to regain its former glory, the company is betting that it can make a series of awkward production shifts. But he also hopes that the rebranding campaign will convince people that it is not far behind the competition.
Intel CEO Pat Gelsinger unveiled a roadmap for several generations of chips at Monday’s event. It includes new technologies designed to help the company compete TSMC, a Taiwanese chip manufacturer that currently makes the most advanced computers and high performance chips, as well as Samsung in South Korea. The roadmap includes a timeline that will allow executives – and outsiders – to measure Intel’s progress.
In the early signs of success, Intel said Qualcomm i Amazon agreed to be customers for its new foundry business, where Intel will produce chips for other companies; Intel said it would start producing chips for those companies in 2024. Gelsinger announced plans for the foundry in March, shortly after joining a company that once held a CTO. However, to the embarrassing extent of the company falling behind, Intel also plans to transfer production of its most advanced chips to TSMC.
Gelsinger said Intel will adopt a new naming scheme for upcoming generations of chips. Currently, chipmakers are calling for new chip-making processes or “nodes” using the nanometer scale, with Intel currently using what is known as the 10-nanometer process, and TSMC using what it calls the 5-nanometer process.
The nanometer scale once referred to the actual size of the transistor gate, with continuous shrinkage guaranteeing better performance. (A nanometer is a billion meters; human hair is 50,000 to 100,000 nanometers thick.) One of Intel’s founders, Gordon Moore, famously said In 1965, this progress in chip production could be measured by the ability to assemble approximately twice as many transistors per chip every two years.
But the nanometer scale no longer refers to the actual distances on the chip, and Intel and others say its current chips work like those made in TSMC’s 7-nanometer process. It plans to adopt a naming scheme that reflects that, with a new version of its 10 nanometers to be unveiled this year, dubbed the “Intel 7,” which the company says will deliver 10 to 15 percent better performance per watt of power. Generations after that, which will come in 2023 and 2024, will be called “Intel 4” and “Intel 3.”
“There’s always the question of where marketing ends and where engineering starts, but this is very deeply grounded in engineering reality,” Gelsinger told WIRED ahead of Monday’s announcement.
Stacy Rasgon, an analyst at Bernstein Research, says the technical plan presented by Gelsinger seems promising, but will increase pressure on the company’s execution. “This is all great, but the danger will be that they stretch their necks and go wrong again,” he says.
Intel made a series of mistakes under its previous leadership. The company was slowly adjusting to the transition to mobile computing, which caused it to lose market share Arm, which designs energy-efficient chips used by companies, including Apple, which uses hand-based chips for the iPhone, iPad and some Macs.
Intel is also surprised by the increase artificial intelligence. Nvidia, The “fabless” chip company took advantage of this trend with chips specializing in AI calculations. Nvidia overtook Intel market capitalization in July 2020.
On the production side, Intel has been slower than TSMC to adopt the latest method to corrode characteristics into silicon, known as extreme ultraviolet lithography (EUV). On Monday, the company said it would step up the use of EUV and secured the first next-generation EUV machine from ASML, a Dutch company that is the only manufacturer of EUV machines. The initiative will be expensive, because each EUV machine costs around 120 million dollars.