Intel Introduces New Node Naming: 7 Nanometer Technologies Now Called “Intel 4” and “Intel 3,” Followed by “Intel 18A” (5 Nm) and “Intel 20A” (5+)

Recent Developments

After the success of Intel’s 14 nm manufacturing process, which was introduced in 2013, the chipmaker revealed a new node—the 10 nm process node—in September 2014. The previous node, the 22 nm node, was announced in 2011. The new node is an important innovation for Intel, as it will allow the company to bring chips with much higher performance down to a smaller size, which will ultimately lead to more efficient chip designs.

So, what’s this about the Intel naming scheme? Well, last year, Intel announced that it would be moving the process of naming its chips from the ternary system of A, B, and C to a binomial (also known as taxonomic or binary) system of 3, 4, 6, and 7. The logic behind this change is simple: the binomial system is easy to tell apart from the ternary system, and the new nomenclature will improve the clarity of Intel’s product names. Moreover, the binomial system will allow Intel to use multiple names for each chip family, resulting in the nomenclature “Intel 4,” “Intel 3,” “Intel 18A,” and “Intel 20A

Intel dubbed the 7nm process it developed with TSMC “Cannonlake,” but it wasn’t the first node it produced. In late 2017, Intel set up a second fab in Arizona to produce high-performance CPUs with a 10nm process. While the chipmaker wasn’t ready to reveal the specs of the first one, Intel did show off a 10nm product using a quad-core processor and a 14nm version based on the company’s 14nm++ process.. Read more about intel 7nm and let us know what you think.

Image credit: Intel

Gelsinger unveiled a brand-new road plan as part of today’s Intel Accelerated event, which aims to reassure investors that Intel is on track to regain its product leadership crown in 2025. Instead of the previous 7 nm and 7+ naming systems, those devices are now referred to as “Intel 4” and “Intel 3,” with future 5 nm and 5+ products being referred to as “Intel 20A” and “Intel 18A,” respectively. The “A” in those names stands for angstrom, a new era of semiconductors that will take advantage of two breakthrough process technologies also announced today by Gelsinger: “RibbonFET, Intel’s first new transistor architecture in over a decade, and PowerVia, an industry-first for backside power delivery.” Intel expects Intel 20A to ramp in 2024, followed by Intel 18A in 2025, according to Intel.

AnandTech image

Based on FinFET transistor improvements, Intel 7 offers a 10% to 15% boost in performance per watt compared to Intel 10nm SuperFin. Intel 7 will be used in products such as Alder Lake for clients in 2021 and Sapphire Rapids for data centers in the first quarter of 2022, according to Intel.
Intel 4 completely adopts EUV lithography, which uses ultra-short wavelength light to print extremely tiny features. Intel 4 will be ready for production in the second part of 2022, with a 20% boost in performance per watt and area improvements, with products launching in 2023, including Meteor Lake for clients and Granite Rapids for data centers.
Intel 3 uses further FinFET optimizations and higher EUV to provide an 18% boost in performance per watt over Intel 4, as well as significant area improvements. In the second part of 2023, Intel 3 will be ready to start producing goods.
With two groundbreaking technologies, RibbonFET and PowerVia, Intel 20A heralds in the angstrom era. Intel’s implementation of a gate-all-around transistor, RibbonFET, will be the company’s first new transistor design since its FinFET breakthrough in 2011. Faster transistor switching rates are achieved while maintaining the same driving current as many fins in a smaller footprint. Intel’s PowerVia is an industry-first implementation of backside power delivery that improves signal transmission by removing the requirement for power routing on the wafer’s front side. In 2024, Intel 20A is anticipated to scale up. The firm is also looking forward to collaborating with Qualcomm on its Intel 20A process technology.
Beyond Intel 20A, Intel 18A is currently in development for early 2025, with RibbonFET improvements that will provide even another significant increase in transistor performance. Intel is also working on defining, building, and deploying next-generation High NA EUV, and expects to get the industry’s first production tool. Beyond the current generation of EUV, Intel is working closely with ASML to ensure the success of this industry innovation.

Image credit: Intel

Traditional nanometer-based process node nomenclature ceased matching the actual gate-length measure in 1997, as the industry has long known. Intel unveiled a new naming system for its process nodes today, establishing a clear and uniform framework that will provide customers with a more realistic picture of process nodes throughout the industry. With the introduction of Intel Foundry Services, this clarity is more essential than ever. “Today’s advancements will not only allow Intel’s product roadmap; they will also be essential for our foundry customers,” added Gelsinger. “There has been a lot of interest in IFS, and I’m excited to announce our first two big clients today. It’s off to the races for IFS!”

Intelligence Community (via AnandTech)