The growth of the computer industry over the past 50 years has been driven in large part by Moore’s Law. This was the observation by Gordon Moore in 1965 that the number of transistor on a chip doubles every 12-24 months. The semiconductor industry has followed this trend for the past fifty years and not has the number of transistors continued to double each generation, but perhaps more importantly, the cost per transistor has continued to decrease and the power per logic operation has improved too.
Recent announcements by Intel and others is suggesting that this trend is slowing down and is already stretching out (See the Gizmodo article Moore’s Law Stutters at Intel or the Economist article After Moore’s Law). Part of the reason for this is economical. A corollary of Moore’s law is Rock’s Law that states that the cost of a wafer fab to build the new technologies doubles every four years. Today, a state-of-the are wafer fab costs over $10 billion.
But there are other effects that will also limit the size of transistors regardless of how much capital you spend and these are the side affects of quantum mechanics. When the transistors are larger, one could ignore quantum mechanical effects since these were very small in relation to the other physical effects in the transistor. But now that semiconductors are starting to shrink to geometries under 10 nanometers, the transistors channels and gates may only be a dozen or so atoms thick and quantum mechanical effects, like tunneling, can start impacting the reliability of the transistors.
Many technologists think that Moore’s Law will end in 2022-2025 with the 5nm or 7nm generation of process technology. When that happens there will be an increased urgency to move to a radically different type of computer, like a quantum computer, in order to continue the rate of computing performance increases we have enjoyed over the past 50 years. The idea will be to make quantum work for us rather than against us in our quest to continue the progress.