In their quest towards more and more performant computer chips scientists are now stacking silicon layers for increased density. However, silicon has basically reached the end of the road in terms of physics and any tiny improvement affects economics. Further shrinking increases the cost making progress economically unaffordable.
Hence research is now focussing on other materials, and carbon is the leading one.
In the N3XT project researchers at Stanford are trying to use carbon nanotubes. These can push density much further than silicon thus increasing performances up one thousand times.
One of the crucial aspects in todays chip is the mass of wires that are needed to connect the hundreds of millions of transistors and that end up taking most of the space in a chip. Carbon nanotubes are so tiny that can provide connectivity at a fraction of the space needed on silicon. They can also support, being so tiny, different architectures.
In the N3XT project researchers are experimenting with a skyscraper like architecture where processing and storage takes place on different floors, rather than in different buildings as it is the case in silicon based chips (stacking is already used on silicon chips to increase storage density but storage and processing take place in different areas of the chip in a two dimensional arrangement).
This staking of processing and storage makes for shorter path for bits, increases speed and decreases connectivity problems. Carbon nanotubes work like millions of lifts connecting the various floors.
An advantage of carbon over silicon is that it can be manufactured at much lower temperatures. With silicon the stacking of a layer onto another is difficult because the temperature involved is bound to destroy the existing layer. Stacking on silicon is made by creating the different layers separately and then glueing them one on the other and connecting with thousands of wires the two layers. With carbon the various layers can be grown one on the other and connection is achieved as part of this growth resulting in millions of connectors.