The Chip Insider®– Moore’s Law broken?

Summary:

Moore’s Law: How do you know when it is broken? I’ve kept delaying writing this for more than a year, as important issues have gotten in the way. But the timing has become more critical as there is a critical strategic shift that is occurring in the transition from the nanometer to the Angstrom era… The most cited form of Moore’s Law is that achievable density-per-chip doubles with a cadence of one-to-two years, based on Gordon’s 1965 and 1975 papers… Areal density and cost are often overlooked … both of which, in later years Gordon thought were more important … In the 1990s, Gordon said he believed Moore’s Law would fail if the areal cost increased by more than 30% … More recently, areal density gains have slipped. Add to this high inflation in post-COVID. A 4% annual inflation rate equals 8.2% compounded over 2 years. If areal density only grows 7%, cost-per-transistor must rise, breaking Moore’s Law. And 8.2% inflation is optimistic because it is a macroeconomic rate. It does not account for process complexity increases. Hence, the classic version of Moore’s Law is broken… However, there must be something else going on, because designers are moving forward, even as the cost tailwind of Moore’s Law turns into a headwind…

Happenings, Comments, Questions & Answers: Nodal Competitive Advantage and TSMC’s A16.

“If you don’t like what’s being said, change the conversation” – Don Draper, Mad Men

To read the full article, click here.