What do you mean precise? What's the cadence of TSMC's latest nodes, since when have they been following this "precise" calendar? By the same token do you not attribute Intel's success in the last decade or so only down to their fabs? At one point AMD was 2.5 nodes behind Intel, and at least 1-1.5 node behind for 10+ years
Since Global Foundries gave up on bleeding edge, which also coincide with zen really taking the fight to Intel with zen 2
. AMD Desktop roadmap have been more reliable than Intel's. Zen 2 had what ? a one month delay ? That's insanely better than intel's 2 years delay (or outright cancelation).
What I was trying to say, is that AMD doesn't give optimistic roadmaps. 3 years ago, AMD said that zen 4 was going to be a 5nm product scheduled to launch in 2022. And that's exactly what happened. In 2018 they said that Vermeer was coming in 2020, on a matured process. That's exactly what happened. To me it sounds like that AMD/TSMC actually know where they are going, and have an efficient communication channel.
Now I never denied that Intel enjoyed a node advantage in the past. But they don't anymore, their fabs are actually a liability, but they keep on giving optimistic roadmaps that they cannot fulfil, and resort to a lot of stop gap, or emergency release. Rocket lake was replaced in less than a year and forced onto a node that it was never meant to be used on. Meteor lake was
two year late, and was ultimately cancelled for the desktop. And now there's rumors that Intel finally yield, and Arrow lake will be a TSMC CPU at 100%, because Intel finally decided to be cautious, an rely on TSMC much higher fabs reliability. Intel 20A isn't doing so hot apparently. Intel new goal to take the leadership back is 18A
