I have to take their 7nm claim with a GIANT grain of salt.
consider, in 2006 they "announced" a 350GHz room-temperature SiGe CPU. ten years later where is it? nowhere to be found.
in 2010, they "announced" the "World's Fastest CPU" -- 5.2GHz on PD-SOI at 45nm. But again, the marketing claim was totally bogus. anyone who bought one of their mainframes with that chip inside is contractually forbidden from actually benchmarking it. anonymous tests pegged it at about 650MIPS, about as powerful as a 1990's Pentium chip.
so ...
buried in the fine print of today's IBM "news" (read: marketing) is the gem: "IBM is not saying when the 7nm technology will be commercially available".
... no, of course not! because it won't be.
there's a big litho hurdle there.
and an even bigger one once you factor in the need to obtain manufacturing yield that's high enough to actually make any money instead of just bleed it all over the plant floor.
in comparison, Intel has had enough trouble with 14nm that they shoved their 10nm timeline out to 2018 because they can't actually do it.
And you expect me to believe IBM beat them to 7nm today?
If this is anything other than a "hey look! we made some xenon atoms dance!" nonsense, I'll eat my hat.
it's just another Rommety-era marketing operation to distract everyone from how completely irrelevant IBM has become, especially to fundamental research.
Sorry, IBM ... I call BS.
GLAL,
R.
