Moore predicts the end of his law

EVER SINCE THE beginning of the world (or a little bit after, depending on your views when did world started), there has been statements about laws we cannot break. Earth was centre of the universe, then Sun was centre of the universe etc etc. But there is just one constant in the world of semiconductors: Moore's law.

This law changed quite a bit over the course of years, most notably being modelled to fit Intel's marketing talk - but one cannot argue that the law is even being exceeded by certain parts of the industry, like graphics chips (trannies double there in a period of 12 months, not 18), but it is considered as valid.

In its yesterday's guest keynote, Gordon Moore touched the subject of his cubicle (largest one at Intel) and the longivety of his law. Mr. Moore stated that every law that has ground in physics has its natural end, and in his opinion, Moore's Law has to deal with fundamental limitations to microelectronics. When he asked Stephen Hawking about the "Moore's Law", Stephen said that challenges ahead are maintaining the electrons at the speed of light and atomic nature of the matter itself.

Current chips have five molecular layers in the structure of transistors, and before Hafnium made appearance, a lot of folk thought that was the limit. Now, we could discuss that the physical limit for semico industry is one molecular layer, and then industry will have to invent something new to stay in tune with Moore's law, or the law will become obsolete.

By current projections, we still have some 10-15 years to go. µ