Written on BA 904 flying London to Frankfurt and dispatched to silicon.com later the same week via a public wi-fi connection at 34Mbps.
In the early stages of the industrial revolution, every town and village had a blacksmith and an iron foundry. Design, manufacture and support also started as local activities but then much later became geographically concentrated by specialisation and to be closer to production lines.
Farm machinery and automobiles originated in thousands of small manufacturing shops across Europe and the US. Later, radio sets and TVs enjoyed similar distributed origins. But these products were crude, unreliable and expensive, and the invention of the production line and quality control resulted in centralisation and huge improvements in products, their supply and support.
Over the decades, production lines have been honed and improved. Competition has ensured the innovators survive and the laggards fall by the wayside. The net result is that now all our basic commodities are produced by a handful of suppliers, while numerous system integrators satisfy our desire for good design and customisation.
It is the system integrators that assemble our cars and electronic goods from standard components produced by a handful of specialised plants. Typically, there are no more than five to 10 major suppliers of, say, batteries, LCD displays, keyboards, internal combustion engines, or electrical and transmission systems.
So far there seems to be some immutable law that says big is beautiful – with concentrations of resources doing a better job, more efficiently, than any distributed option we have so far come up with. This law holds true for power generation, food production and city living. All do far more for far less than distributed alternatives.
And, in the struggle to provide more to growing numbers of people using fewer and fewer materials, concentration still seems to be the only viable option.
The latest industry to follow this trend is the chip and integrated circuit sector, with foundries and production facilities now costing over $2bn to set up. Here the exponential pace of processor innovation has reduced chip feature sizes from…
…10,000 nanometres to less than 100nm in 30 years, with some advanced technologies now at less than 30nm. To give you some perspective, it’s worth noting that a common virus is of the order of 50nm to 150nm in diameter.
Within the next five years chip feature sizes will fall below 10nm, and 3D integration is likely to become the norm. We will probably also see a degree of programmability and self-repair.
So how many companies will be in the running to take part in this vital industry? It looks like being somewhere between five and 10 suppliers, with Japan, Korea, Taiwan and the US being the key players. Major collaborations between the Japan, Korea, the UK and the US are being organised to tackle the formidable design, material and manufacturing problems at 10nm and below.
And, unlike the battery and display industries, it looks as though production facilities will be sufficiently distributed across the face of the planet to avoid the disruptive effects of earthquakes, hurricanes, and other natural disasters. However, probably fewer than five major suppliers will be operating at the sub-10nm level in the endgame.
So what happens beyond this current phase of chip building? Rafts of new technologies are lining up to move into the same area and continue the feature size shrinkage down toward the 1nm barrier and below. Exactly how this will pan out no one knows. We can nonetheless expect a winner or two, probably starting life in a large number of distributed shops, gradually migrating to the big few. In fact, very much a repeat of what has gone before.
Will this silicon cycle continue ad infinitum? Unlikely. Somewhere in the sub-10nm region we can expect a gradual migration to a new paradigm involving the integration of inorganics and organics, leading to programmability and self-organisation.
That will be a real game-changer that might just result in some significant redistribution of manufacture and supply at a local level.
At the same time, we might just see technologies that break the established efficiency barrier between distributed and concentrated production. After all, the organic world we live in achieved that condition more than two billion years ago.