The COVID-19 pandemic has driven tremendous growth in semiconductor sales, but questions abound about when the boom might fade and what happens to the market when it does. The analyst firm IDC has published its analysis of what it expects to happen in 2022 and 2023 as manufacturers make long-term adjustments and new capacity comes online.
The semiconductor market grew by 10.8 percent in 2020, but it’s expected to easily clear that bar in 2021, with 17.3 percent growth. The report notes that growth is being driven by “mobile phones, notebooks, servers, automotive, smart home, gaming, wearables, and Wi-Fi access points,” along with increased memory prices.
IDC expects semiconductor revenue to grow by 128 percent. Mobile phones are expected to grow by 1.28x, while game consoles, smart home, and wearable products will expand by 1.34x, 1.2x, and 1.21x. Revenue in the automotive semiconductor industry should be up by 1.23x, with the additional welcome news that shortages should be substantially alleviated by year’s end. Notebooks and x86 server revenue should increase by 1.12x and 1.25x, respectively.
“The semiconductor content story is intact and not only does it benefit the semiconductor companies, but the unit volume growth in many of the markets that they serve will also continue to drive very good growth for the semiconductor market,” says Mario Morales, IDC’s group VP for enabling technologies and semiconductors.
Unfortunately, there’s also the possibility of a decline in late 2022 or early 2023 depending on what happens with demand. TSMC, Samsung, GlobalFoundries, and Intel have all announced large capacity expansions. The various first and second-tier foundries have increased production at existing facilities and are building all-new fabs around the world to handle anticipated long-term demand.
If currently high levels of demand do not continue, the semiconductor market could find itself with an enormous glut of additional capacity no one needs. The various foundries don’t seem particularly concerned about that outcome at the moment. It’s possible that TSMC, Intel, and Samsung have contingency plans in place that would allow them to finish facilities more slowly over time if demand does not materialize as predicted. Intel’s Pat Gelsinger is on the record as saying he disagrees with the analysts regarding the potential for a near-term downturn and believes the future for silicon production is bright, even with these global capacity increases on the horizon.
Intel is unlikely to hit the pause button quickly; the company was burned in recent memory by its decision to pause the bring-up of Fab 42. This may have made sense when Intel made the decision in 2014, but the company’s supply shortages in 2018 and 2019 were exacerbated by not having enough production capability online. TSMC and Samsung’s behavior is a little harder to predict. But with Intel building fabs in Europe and launching itself into the foundry business, neither rival foundry is going to risk actions that would leave them looking behind their competition.
IDC reports that front-end manufacturing is beginning to meet demand but that “larger issues” and shortages continue to disrupt backend manufacturing. Fab utilization remains near 100 percent and is expected to remain there for the foreseeable future.