Intel Process Roadmap 2025 – The beginning of a new era

Intel released a roadmap showing their plans for new process tech and packaging through 2025. The company claims to catch up with industry leader TSMC until 2024 and to regain ‘process performance leadership’ the year after. The most interesting part however was the teaser of their new angstrom-class technology. An angstrom (or ångström to be precise) is a unit of length equal to 0.1 nanometer and even though it is not part of the SI system of units it can still be considered part of the metric system in general. This new unit of measure also brings a change in node names.


The previously named 10nm Enhanced SuperFin will be re-named to Intel 7 and deliver an increase of 10% to 15% performance-per-watt over the Intel 10nm SuperFin (the predecessor of the ‘Enhanced SuperFin’ variant of Intel’s 10nm line). Be cautious not to confuse the new Intel 7 with their Intel 7nm nodes, which will have their name changed to Intel 4. The Intel 4 will provide approximately 20% increase in performance-per-watt compared to the Intel 7 and will start production in the second half of 2022.


A year later, in the second half of 2023, Intel plans on introducing the Intel 3, which is expected to deliver an 18% increase in performance-per-watt over the Intel 4. In 2024 Intel expects to usher in the angstrom era with the Intel 20A and their PoverVia and RibbonFET technologies. PoverVia is a backside power delivery, which eliminates the need for power routing on the front side of the wafer and provides optimized signal routing while reducing droop and lowering noise. RibbonFET is going to be a new transistor architecture delivering faster transistor switching speeds while achieving the same drive current as multiple fins in a smaller footprint, making it an update to their FinFET technology.


Different technology nodes often imply different circuit generations and architectures. In general, the smaller the technology node means the smaller the feature size, producing smaller transistors which are both faster and more power efficient. In the beginning of computer technology, the process node name referred to a number of different features of a transistor including the gate length as well as M1 half-pitch. Given that Intel’s 10nm SuperFin is pretty much on par with the TSMC 7nm, the number itself has lost the exact meaning it once held. It does not correspond to any gate length or half pitch. Nevertheless, the name convention has stuck until now and it is what the leading foundries called their nodes.


Even though since around 2017 node names have been overtaken by marketing using them ambiguously to represent slightly modified processes, the suffix ‘nm’ was always part of the naming process. Thus, causing the size, density, and performance of the transistors among foundries to no longer match between different manufacturers.


While some people see Intel’s new naming scheme as further confusion, it can also be argued to bring some clarity. With this new way of naming nodes, it is clearly just a form of branding, because there is no mention of nanometer or any other measurements. It is undisputed that the previous way of naming was more practical since it allowed to directly compare nodes from various foundries. However, the times when this was the case are long gone and while it used to be that ‘nomen est omen’ nowadays names are nothing but sound and smoke.