Maybe Moore's Law is Not Dead

Moore’s Law, at least based on current technology, is slowing, most would agree. But some argue semiconductor technology could jump to a new curve, and thus sustain progress, if new designs, materials and approaches prove to be commercially viable.

So one cannot meaningfully assess whether Moore's Law is dead without specifying the conditions. Chips based on silicon will reach a physical limit. But other materials might replace silicon. Also, chip architectures can change. In other words, "how" chips and systems conduct computing can change.

And then there are tweaks such as creating specialized chips for specific purposes, or applying more-clever programming.

Also, many are working on ways to improve performance by using either custom designs or more-clever programming. In the past, nobody really bothered too much with those approaches because fundamental progress at the physical layer was prodigious.

In the past, silicon technology has driven Moore’s Law because components became smaller. But silicon-based approaches are getting near a physical limit of atomic size. “A Skylake transistor is around 100 atoms across, and the fewer atoms you have, the harder it becomes to store and manipulate electronic 1s and 0s,” scientists say.

So a growing constraint is the cost of manufacturing chips that are ever-smaller. That is one reason why GlobalFoundries decided to exit production of advanced processors, for cost reasons.

In that sense, commercial profit now is becoming a big issue. Intel recently has said that issues 10-nanometer fabrication would delay its shift to seven-nanometer production, for example. And some predict potential revenues for even-smaller chips will fail to cover investment costs. That is why GlobalFoundries got out of the business of making advanced processors.

Smaller transistors now need trickier designs and extra materials. And as chips get harder to make, fabs get ever more expensive.

In the future, to get Moore’s Law back on track, manufacturers will have to rely on new architectures and new materials. There is hope that human ingenuity can succeed. What matters is the economics of computing--its cost versus performance--rather than simply being a matter of physics and manufacturing costs.

source: The Economist