Moore's Law to Die Due to the Expense?

Rob Williams

Staff member
From our front-page news:
Yesterday, we posted about GlobalFoundries' demonstration of silicon built on a 22nm node, and according to some reports, that's just about as low as we're going to be able to go, at least for a while. According to market research company iSuppli, Moore's Law is going to die at 18nm. As you're probably already well-aware, Moore's Law states that the number of transistors will double on an integrated circuit every 1.5 - 2 years.

To date, that seemingly simple theory has held true, so what's going to cause a sudden halt? It's easy to assume that dies are simply going to become too small, and impossible to build, but that's not it at all. And no, it has nothing to do with current leakage. Rather, it appears the reason for the death of Moore's Law is... money. iSuppli states that beyond 18nm, the fabs to produce such chips are going to be incredibly expensive, and given that 32nm fabs currently cost around $4 billion... enough said.

Producing silicon on processes of at or smaller than 18nm isn't impossible, but thanks to the costs, whichever company can produce such products is going to belong to an incredibly exclusive club. Companies like the TSMC and Intel should have no problem stomaching the bill, but up-and-comers like GlobalFoundries may have a hard time with the financials. But the real question is whether or not there will be a true return-of-investment on such expenses. We'll have to wait and see.


"The usable limit for semiconductor process technology will be reached when chip process geometries shrink to be smaller than 20 nm, to 18 nm nodes," said Len Jelinek, director and chief analyst, semiconductor manufacturing, for iSuppli. "At those nodes, the industry will start getting to the point where semiconductor manufacturing tools are too expensive to depreciate with volume production, i.e., their costs will be so high, that the value of their lifetime productivity can never justify it."

Source: TG Daily


Techgage Staff
Staff member
It might be infeasible with current technology, but the question is can they devise new technology and new methods that help bring down the costs of implementation some. Eventually at some point I'd hope we would discover a better, cheaper method of processor/silicon creation than the fabbing industry has been using for the last 40 years.

I do agree with the article though, anyone familiar with the industry knows such a point in time is coming. The number of major silicon manufacturers in the US has dropped by over half since 2000, big giants like Motorola, HP, TI, and many others I never remember the names on have or are going fabless. Still, this means more demand and more silicon to produce and sell by those that will have the current generation fabbing capability. In the very least Intel's Gelsinger said last year he saw a clear way to 10nm fabrication.


The Tech Wizard
I see the distant future of not using silicon but Amorphous, hydrogenated carbon as a base and gateways on the atomic level, even smaller than nano levels. You could run the whole system on a 9 volt battery for a year or more. But costs and development in this state of economy is holding back research.
At one time there was talk of a fluid design, but again development lost funds.


The One, The Only...

<object width="425" height="344"><param name="movie" value=""></param><param name="allowFullScreen" value="true"></param><param name="allowscriptaccess" value="always"></param><embed src="" type="application/x-shockwave-flash" allowscriptaccess="always" allowfullscreen="true" width="425" height="344"></embed></object>

had to. Sorry.


Techgage Staff
Staff member
Heh, not directly related, but this sure as heck got my attention: Link

It just goes to show, if NVIDIA needs to source chips from 5 different major fabs, there is plenty of demand out there between NVIDIA and ATI alone... but honestly those two are only a small portion of the total demand out there.