Semiconductor companies regularly share "good news" when they announce technology advances or product updates. But when the semiconductor industry as a whole is in the news, it's rarely good. And so it is with recent news cycles that have put a spotlight on the ongoing chip shortage and, lately, the push to expand semiconductor manufacturing capacity.
However, with all the focus on shortage, it's important that we not lose sight of the normal, cyclical nature of the semiconductor market. Those normal dynamics suggest to this analyst that the semiconductor industry is headed for a global glut analogous to the housing crisis of 2008 – and, for those familiar with the semiconductor market, to the previous semiconductor boom-and-bust cycles of the late 90s and early 2000s.
What follows is not a forecast for when a potential glut will hit, and in no way should the below be considered financial advice. Instead, what follows is food for thought for anyone who might be thinking that we are experiencing a "new paradigm" for the semiconductor market.
Part 1: "Everyone get their 'Stimmy'?"
To start, because of the uncertainty fueled by the pandemic and the state of US-China trade relations, a lot of businesses decided in 2020 to double-book semiconductor purchases, or double their inventory, creating enormous pressure on the chip foundries (fabs). How the shortage formed is largely a function of businesses and consumers buying lots of tech products due to the government ordered lockdowns to support work-from-home staff (for businesses) and to keep themselves safe and entertained (for consumers). The faster-than-expected automotive recovery in late 2020 contributed to the chip shortage, too. And then governments the world over, in their infinite wisdom, added fuel to this fire through stimulus packages rolled out during the lockdowns.
Now, as a result of the chip shortage, governments and various industries currently are crying out for more capacity and more chips, and the investment community is listening, as are some governments. The latest news is that some fab companies are looking to expand their operations significantly. For their part, governments are looking to subsidize new investments into more fabs.
The shortage looks like an opportunity to make the big monies quickly, secure market share, and improve national security, and management at TSMC is probably thrilled with the shortage as they can make a lot more profit. The problem is that government subsidies or incentives to get companies to invest in fab expansion, or government encouragement of the investment community to help with fab expansion, are aimed at immediately addressing the chip shortage. But the real issue is not a lack of supply; the real problem is artificially induced demand due to lockdowns and free money. As is typical, when government creates a problem, it then throws money at the problem hoping for a short-term fix while ignoring the longer-term implications – that is, of course, until the long-term consequences do pop up, and then government throws money at that those too.
To the public, this looks like a great thing. The shortage will go away, prices on electronics will come down, and unicorns will fly through the clouds. However, there is a darker side to this.
Part 2: "What about da 'conomy?"
A 2005 academic study argued that semiconductor industry booms and busts are not really a function of demand drying up, but rather more due to surges in inventory levels and fab capacity. In short, overcapacity (as opposed to "under-demand") is what leads to busts: It's just the nature of the semiconductor business cycle that manufacturers build to overcapacity during boom times and then collapse as demand returns to more normal levels.
Fabs are the supply side of semiconductors, the birthing suite. First thing to note is that fabs are expensive and take a long time to get up and running – easily more than two or three years from striking dirt to finished building. Building a fab also is not cheap, requiring billions of dollars upfront just to get the supply chain up and running to build a new plant. Not to mention the cost of the manufacturing tools that must be created before the fab can start producing chips. These manufacturing tools on average account for 70% of the cost of the fab. So, the fixed costs are very high for a fab. It's like building an auto factory in terms of complexity, time, and cost.
Now we get to the crux of the issue today: Right now companies are working to expand fab capacity – a long-term process that will take significant time and investment before production even starts – to try and resolve an immediate-term problem caused by an artificial and sudden uptick in demand that is likely to resolve itself by 2022 as people head back outdoors and back to the office. At the very earliest, the projects to expand fab capacity that are starting today will likely be completed sometime in 2023, just as the shortage starts to go away on its own with current infrastructure and capacity.
Part 3: "Same old same old"
What does this all mean?
It appears that the industry is building toward overcapacity right now. Just when these new fabs will be coming online, demand is likely to be back to more normal levels. The recent demand spike was, in a sense, artificial. The pandemic, lockdowns, and "free money" all drove customers to buy a lot of electronic goods that they probably would not have purchased were it not for these extraordinary events. The chipmakers weren't prepared for this surge because this was not a normal demand cycle; there was no forecast at the beginning of 2020 that could account for a pandemic, lockdowns, and free money.
It's analogous to what happened with the housing crisis of 2008. It started with stimulating consumption/demand in the housing market in the early 2000s via easy money and housing incentives. Consumers were incentivized to borrow at lower rates. As homebuilders saw that consumers were bidding up housing prices, they jumped in and started building more houses. The boom. When rates began to rise, which brought down affordability, consumers stopped bidding up prices, pulled out of the market, and started defaulting on their debts. Consequently, homebuilders who had also borrowed "easy money" were left with incomplete buildings and lots of inventory. The bust.
Of course, housing and semiconductors are not the same thing and the severity of the likely glut will not be known until later. But the pattern is the same: Artificially induced demand via overinvestment or government actions, followed by the bust that is prolonged by lots of overcapacity that takes years to offload or fully utilize in the future.
So, what happens when everyone is vaccinated and heading outdoors again, and demand for tech normalizes to pre-pandemic levels? What happens when hospitals stop buying new healthcare tech as the COVID threat wanes? In the long run semiconductor manufacturers are likely to continue to experience long-term growth and be ever more ubiquitous, but do we really need the chipmakers to be building up all this long-term capacity to solve a short-term problem?
Think about that whenever you hear someone saying that the semiconductor industry has entered a "new paradigm." Maybe it's really just the "same old same old paradigm."
Danny Blodgett, Senior Research Analyst
As a senior market analyst with Omdia’s Semiconductor team, Danny specializes in semiconductor research, automotive market research, and the relationship between the two industries. He works directly with clients and regularly updates the semiconductor team on automotive activity. Read bio
Get ahead in business
Optimize opportunities and make the right decisions with critical data and expert analysis from Omdia.
Sign up to receive regular updates on new trends, insights and findings for your business.