The semiconductor industry is in a constant struggle to forecast demand as regular chip shortages increase in a troubled market, despite access to massive amounts of economic data. However, this is not the first time the market has been rocked by repeated patterns of boom-bust cycles. Whenever this happens, cross-country skiers promise it won’t happen again thanks to more accurate predictions.
Unfortunately, the failure of these forecasts remains evident. Hence the following questions: how severe is the shortage currently facing founders? How long will it last? Can we trust predictions of a prolonged shortage? Looking at the chip industry from a decades-long perspective, you can imagine that you will probably see the same thing that is usual in this industry: a sharp increase in demand, followed by ‘an equally strong drop’.
Last year, the Semiconductor Industry Association (SIA) released its mid-year 2020 report, and it was grim. “After record sales of $ 468.8 billion in 2018, global sales in 2019 fell 12% to $ 412.3 billion,” the organization said. And to specify that the 2020 sales would be negatively affected by the health crisis, with a modest increase in income expected for 2021. However, the opposite has happened: the chip market has thus returned in force this year, and there is nothing modest about the monthly chip revenue growth of 26% or more, year over year.
An unpredictable production structure
The semiconductor industry is now forecasting a prolonged shortage of chips, as chip revenues skyrocket through 2021 and extend into 2022. However, this prediction is probably as unreliable as previous ones, as it is incredibly difficult to determine the actual demand for chips. It is even more difficult in times of scarcity, as customers order two or three times and try to stock up by all possible means. This creates a certain level of false demand, which disappears as soon as supply increases. It is difficult to unravel the real demand for chips.
This is a big problem for chipmakers because it takes at least six months to make the chips. Once this starts, it is no longer possible to stop it. Chip factories must be kept running near full capacity, in which case operating costs quickly become unsustainable.
In fact, the economics of the semiconductor industry requires factories to continually churn out chips at near capacity. In other words: it is not possible to slow down a production line because demand has fallen. We must continue at all costs. Even if that means ignoring regular revenue … at least, as long as quality forecasts cannot be guaranteed by industry players.