Yet there have been Cassandras predicting the demise of Moore ’s Law for decades. So far they have been proven consistently wrong, with new avenues for improvement opening up just when it seemed any hope for further advance was blocked. Until 2004, increases in the clock speed of chips significantly contributed to enhanced performance, the downside being that overheating placed a limit on how far that innovation could persist. In response, manufacturers incorporated more processor ‘cores’ as the primary means for accelerating power, with processors now working on diferent operations in parallel with one another.
It will take similar kinds of innovation to maintain Moore ’s Law, even if it continues to slow down slightly – something which, in his defence, Holt conceded. While within a decade it may become impossible to miniaturise individual transistors any further, simply because of physical limits, adaptations such as 3-D circuitry and quantum computing – both proven concepts – could mean exponential growth continues. Perhaps even beyond the last square of that chessboard.
More than Processing
Because digitisation is a general-purpose phenomenon, it is not just computer chips that have been subject to its incredible powers of transformation. A similar trend is in evidence with internet bandwidth, where user capacity has grown by between 25 and 50 per cent every year since 1983. The same holds true with data storage, which has likewise enjoyed an exponential function in space-to-cost ratio, with a gigabyte of storage falling from around \$200,000 in 1980 to just \$0.03 in 2014.
More than anywhere else, however, it is in storage that progress has visibly started to slow down. Even if Toshiba’s
The Three Disruptions
45