Intel has spent the former few years rocking from one gaffe to the next, even having to outsource manufacturing of its latest chips to one of its biggest competitors.
Now, in order to regain its previous glory, the company is betting that it can carry out a series of firm transformations in manufacturing. But she also hopes the rebranding campaign will convince people that she’s not far from the competition following all.
Intel CEO Pat Gelsinger laid out a roadmap for multiple generations of chips at Monday’s event. It includes unused technologies designed to aid the company compete with TSMC, the Taiwanese chip maker that currently makes the most advanced, lofty-performance computer chips, as well as South Korea’s Samsung. The roadmap includes a timeline that will allow executives – and outsiders – to measure Intel’s progress.
In an beforetime sign of success, Intel said Qualcomm and Amazon had agreed to be clients for its unused foundry business, with Intel making chips for other companies Intel said it would start making chips for those companies in 2024. Gelsinger had announced plans for the foundry business in March. , shortly following joining the company where he was previously a Director of Technology. However, in an embarrassing measure of how far behind the company is, Intel is also planning to outsource manufacturing of its more advanced chips to TSMC.
Gelsinger said Intel will adopt a unused naming scheme for coming generations of chips. Currently, chip makers refer to unused processes or “contracts” to make chips using the nanometer scale, with Intel currently using what’s known as a 10nm process and TSMC using what’s called a 5nm process.
The nanometer scale once referred to the actual size of the gate of the transistor, with continued shrinkage to ensure better performance. (A nanometer is one billionth of a meter; a human hair is 50,000 to 100,000 nanometers thick.) Intel co-founder Gordon Moore stated in 1965 that advances in the chip industry could be measured by the aptitude to shrink nearly twice that. transistors on a chip every two years.
But the nanometer scale no longer indicates actual distances on a chip, and Intel and others say their current chips work like those made in TSMC’s seven-nanometer process. It plans to adopt a naming scheme that reflects this, with a unused 10nm version due this year called the Intel 7 that the company says will provide 10 to 15 percent better performance per watt of power. The generations that succeed, which will come in 2023 and 2024, will be called Intel 4 and Intel 3.
“There is forever a interrogate of where marketing ends and where engineering begins, but that is deeply ingrained in engineering reality,” Gelsinger told WIRED prior Monday’s announcement.
“It’s all distinguished but the danger is that they get off their necks and something goes mistaken again.”
Stacy Rasgon, Analyst, Bernstein Research
Stacy Rasgon, an analyst at Bernstein Research, says Gelsinger’s technical roadmap looks promising but will increase pressure on the company to implement. “It’s all distinguished,” he says, “but the danger is that they get off their necks and it goes mistaken again.”
Intel made a series of blunders under its previous leadership. The company has been slow to adapt to the shift to moveable computing, which has seen it lose market share to Arm, which is laying out the blueprints for energy-efficient chips used by companies including Apple, which uses Arm-based chips for the iPhone, iPad and some Macs.
Intel was also surprised by the advent of artificial intelligence. Nvidia, the “Fables” chip company, has capitalized on this trend with specialized chips for AI computations. Nvidia overtook Intel by market capitalization in July 2020.
On the manufacturing side, Intel has been slower than TSMC in adopting the latest method for etching features into silicon, known as extreme ultraviolet (EUV) lithography. The company said Monday that it will ramp up EUV use, sourcing its first next-generation EUV machine from ASML, a Dutch company that is the only EUV machine manufacturer. The initiative would be costly, since the cost of each EUV machine is about $120 million.
But Intel’s comeback plan has national significance and may be of aid. The recent shortage of chips has highlighted the importance of silicon in the global economy. The US government is also concerned that manufacturing of the latest chips, which will be vital to emerging technologies including 5G and artificial intelligence, is concentrated in Asia. So the US government, which is considering a $52 billion investment in the US semiconductor industry, is eager for Intel to regain its supremacy.
“It’s very significant because they (the US company) are the only ones that can manufacture at the highest level,” says Dan Hutchison, CEO of VLSI Research, an analyst firm. “That’s what really worried the Department of Defense.”
Technological changes laid out in the roadmap include a unused transistor design and a unused way to deliver power to a chip, both of which are scheduled to debut in 2024 under the name Intel20A, which stands for angstroms below a nanometer.
The unused transistor design, called the RibbonFET, contains a stack of several strip-shaped channels that pass through the gate. A technology called PowerVia allows power to be delivered to a chip from the base rather than the top on the silicon wafer. Both should allow more performance and efficiency to be pulled out of a piece of silicone. Intel is also developing unused ways to stack components in a chip together that promise better performance and design flexibility.
But Hutcheson says each of these innovations will require a manufacturing learning curve. And while the roadmap looks promising, it does indicate that Intel needs to demonstrate its aptitude to bring it to market. “They need to get some score on the scoreboard and move on,” he says.
More distinguished wired stories