Intel’s Latest Roadmap Shakes Up Past Plans and Ditches Node Nomenclature
Meeting goals sometimes means adjusting and overhauling the original plan. From new technology to updated nomenclature, Intel is trying to do just that with its latest up-to-date roadmap.
It’s no secret that Intel has been feeling pressure from the semiconductor industry in recent years. Between struggles to meet 10 nm and below process node technologies and Apple to ditch its CPUs in favor of in-house silicon, the company has undoubtedly felt the strain.
A past timeline for Intel's 10 nm technology. Image used courtesy of Intel
In an interesting move, Intel has decided to address concerns about the company’s future direction. This week, it announced its latest 5-year roadmap, providing deadlines, expectations, and new technology that will hopefully fuel its momentum.
Since this announcement comes with a lot of news, far too much to thoroughly cover in one article, this article will be part one of three. This article, in specific, will look at what’s being promised in the roadmap and attempt to analyze Intel’s proposed change in nomenclature.
Behind the New Nomenclature
Amongst Intel's setbacks, arguably the most publicized has been its struggles to meet 7 nm processing nodes and below. In an attempt to put this conversation to rest, the company has proposed a change in nomenclature instead of arguing that a more critical measure of IC performance is performance per watt, not simply node size. For this reason, the company plans on ditching the term "nm" in its future device names altogether.
With this in mind, the first immediate step laid out in its roadmap is "Intel 7," which was previously known as "10 nm Enhanced SuperFin." Intel 7 is said to increase performance per watt by up to 15% vs. the 10 nm SuperFin. It also claims that it was introduced this year, with production beginning in the first quarter of 2022.
A visual of Intel’s proposed roadmap. Image used courtesy of Intel
After this comes "Intel 4", which is likened to a 7 nm process node in "old" terminology. Achieved through a full embrace of extreme ultraviolet (EUV) lithography, Intel 4 is expected to bring a 20% performance-per-watt increase along with area improvements. Intel anticipates this technology ready for production in the second half of 2022 and shipping by 2023.
Overview of a EUV system. Image used courtesy of Wurm and Kemp
Following Intel 4 is "Intel 3," which seems to rely on "further FinFET optimizations and increased EUV," the device hopes to improve performance-per-watt by 18% over Intel 4 and should be ready for manufacturing by the second half of 2023.
Beyond these releases, Intel hopes to shift the nomenclature from nm to angstrom completely. The steps beyond Intel 3 are Intel 20A and Intel 18A, which intend to be ready by 2024 and 2025, respectively.
Changing something, even as simple as terminology, can be a major shakeup for a company. In the webinar presented by Intel, there was a lot to unpack.
Unpacking Initial Thoughts
Since there is a lot to unravel, it was difficult to discern what was clever marketing and what was genuine innovation, which can be a tricky game to play.
Concerning the change in nomenclature, Intel appears to see node size as becoming more of a marketing ploy than an accurate metric of device performance, which could have some grains of truth behind it. Despite the overhaul, change in focus, and proposed pacing, Intel is trying to reevaluate and adjust its game plan and future outlook.
The next article will cover Intel's roadmap taking a deeper look at the new processing technologies proposed and how they might help Intel achieve its lofty goals.