Intel Corp (INTC)

[Unkwn]

Here's an interesting comment/wrapup of the actual densities of Intel vs foundries.

A feature size, like SRAM cell size, is a much better indicator of actual achieved density than some pitch values, since it is a matter of many factors, how much you can integrate on a given area.

Note that Intel's SRAM size for high-density is 0.0312um2 and for high performance, it is 0.0441um2

GlobalFoundries, which AMD consistently uses for its Ryzen, Polaris and Vega chips, also has a higher density than Intel's. For high-density SRAM, it is 0.0269um2 and for high-performance SRAM it is 0.0353 um2.

What does this mean for AMD? Well, for the first time, Intel can no longer rely on process advancements for higher-efficiency, high-performance chips. Now, it is based on architecture and whose architecture is better. This means that with Zen 2 and Ice Lake, AMD would be the closest it has ever been in more than a decade to not only have a powerful architecture against Intel's Core series but also a process advantage. AMD had to contend with the process advantage for a decade, but now the tides appear to have been turned. Currently, with Zen+ coming out and the 12 nm (which is basically the same 14 nm20 nm process with further enhancements), AMD is still at a disadvantage. The IPC is worse than Kaby Lake's and Intel not only has a process advantage, but it also allows Intel to clock their chips higher than AMD's. (Though Ryzen does give more cores for the money) 2019 will be an interesting year for CPUs.

So, Intel claims to have a real 14nm process and proves that with smaller fin pitches etc, but its SRAM, which consumes much of todays CPU/SoC area, is significantly less dense than that used by AMD on Globalfoundries. I have much doubt that it is different for standard logic, what almost all the rest of the CPUs are made of. In the end, Intel seems to cherry pick dimensions where they have a lead and not talk about overall density, which is what counts, in the end, not nm of features. In the end, Intel may have a lead for some metrics, but loses when it comes to actual density. This is only getting worse at 7nm and Intel won't be able to fight AMD just by process leadership. Not possible for them anymore.

That certainly doesn't mean it's a done fight for AMD. The design is at least as important (I'd say even more important) than a process lead. Intel will have to do lot's more than it did in the past, though, to keep ahead of AMD. Losing its lead in prcoess as well as CPU design will have a horrible impact on the company. We'll see if Intel is paranoid enough this time.