His analyses are usually presented on an annual basis – last year, he took a look at the so called merger and acquisition frenzy, concluding the activity was down to the low cost of money and wondering whether the cost of the acquisitions might be met by reducing engineering headcounts and cutting the amount invested in R&D.
This year’s analysis focused on why it is seemingly impossible to predict how the semiconductor market will perform. Despite general agreement that the semiconductor industry is now mature, it continues to suffer seemingly wild booms and slumps. In order to predict the future, Rhines contends, you need also to predict new applications.
In the absence of a crystal ball, Rhines has turned to mathematical modelling – and the Gompertz Curve in particular. This is a mathematical model for a time series in which growth is slowest at the start and the end of a particular period. And the approach – at least for the examples offered by Rhines – fits like a glove.
But one chart he presented contains some frankly astounding data – despite the semiconductor industry now being in its seventh decade, ‘you ain’t seen nothing yet’. Using the Gompertz Curve, Rhines has positioned the cumulative number of transistors produced since the 1960s at the bottom of the curve. If the coefficients used to produce the curve are correct, we are about to see an unbelievable increase in the rate at which they are produced. In the next decade, the number of transistors ever made is likely to multiply by 50 times. Only then will growth begin to slow.
What will drive this huge demand? Rhines offers three main areas – data centres, gateways and IoT nodes, such as sensors, actuators, imagers and transmitters.
And, with his eye on the bottom line, Rhines wonders how the semiconductor industry will make money from things like the IoT when most of the money will be made by those who own the data. Perhaps that will be the topic for next year’s analysis.
