In 2014, Intel introduced its first line of 14nm processors – Broadwell (5th generation). And now, for 7 whole years, nothing has changed: the newest 11th generation CPUs, Rocket Lake, although they received a new architecture, are based on the same 14 nanometers.
At the same time, competitors in the person of TSMC continue to seemingly actively reduce technical processes: a couple of years ago 7 nm were mastered, this year 5 nm, and in a couple of years the release of 3-nm solutions will be launched. However, from a physical point of view, such a large reduction in the technical process does not occur: the key components of the transistor, for example, the same gate, have long fluctuated around 20-40 nanometers, decreasing literally by 1-2 nm per generation.
Where do all these 7 or 5 nm come from? Conventional marketing: companies have long since switched to a three-dimensional arrangement of transistors in several layers, and they continue to count only by area – so it turns out that in this way the transistors “virtually” become smaller. Moreover, each manufacturer thinks differently here, and as a result, 10nm process technology from Intel is similar in key parameters and sometimes even better than 7nm from TSMC.
Comparison of the gate widths of the transistors in 7nm AMD processors and 14nm Intel processors shows that the difference between them is actually minimal: 22 vs 24 nm.
However, for the common user, 7nm sounds better than 10, and Intel understands this. Therefore, according to the resource TechPowerUp, Intel will go into double marketing and will further “reduce” the technical processes of its future solutions in order to “meet industry standards.” Alas, there is no more accurate information about this yet, but it may well be that the 12th generation of the company’s processors (Alder Lake), which will be released in the fall, will “magically” turn from 10-nm solutions into 7- or even 5-nm solutions.