With the semiconductor manufacturers struggling to keep up with exploding demand and a supply chain that has proven to be vulnerable, more chip shortages seem all but inevitable.
Trade tensions prompting stockpiling, a once-in-a-century pandemic and fires, droughts and extreme winter weather hitting manufacturing plants across the globe – the current chip shortage seems the result of a perfect storm. It’s tempting, therefore, to consider it as a freak accident, a one-off. But while this particular set of coinciding circumstances may be unique, it would be foolish to think that supply gluts such as the current one are a thing of the past.
First of all, history shows that major shortages are a common occurrence. Over the last three decades, they happened every five years on average. Usually, they’re precipitated by external factors, such as an economic recession or a natural disaster, but they also ‘just happen.’ Perhaps because it’s been relatively smooth sailing over the past decade – the last supply glut occurred a decade ago after a tsunami in Japan – companies have been lulled to sleep, but there’s no reason to suppose that the historical pattern will change.
On the contrary, pressure on chip suppliers are only rising. One CEO after the other has made clear that the current shortage isn’t merely an after-effect of the Covid year. Underneath, there are massive long-term growth trends across the entire semiconductor spectrum. “We’ve structurally underestimated growth in the industry,” admitted ASML’s Peter Wennink. “This is much more than just a short-term cyclical rebound,” according to Kurt Sievers of NXP. “Demand for semiconductors is no longer about one or two killer applications, but rather an expansive, structural shift in the economy toward digitization and automation,” explained Applied Materials boss Gary Dickerson.
Chip manufacturers are responding with massive spending hikes to add capacity. Consultancy firm Deloitte tallied over 200 billion dollars from 2021 to 2023 in capital expenditures from the three largest players alone. They could reach 400 billion by 2025, and that doesn’t even include the billions governments have committed. These investments are expected to raise wafer capacity across nodes from 83 million 8-inch equivalent wafers in 2020 to 121 million in 2024 – an increase of 46 percent.
It won’t be enough, says Deloitte: demand is growing roughly as quickly (or more) as planned capacity growth. The semiconductor industry will struggle to keep up with the exploding appetite for new applications such as artificial intelligence and machine learning, the intelligent edge and the internet of things.
On top of that, chip manufacturing is highly geographically clustered. Taiwan, in particular, is well on its way to becoming a single point of failure. Distributing manufacturing operations more evenly across the globe, as Western governments have set out to do, would reduce supply disruption risks, but only to a limited extent. A truly robust semiconductor supply chain would have to have complete redundancy built in. This doesn’t seem likely to happen, and, as a result, semiconductor supply will continue to be subject to natural and political disruptions.
There are ways to reduce supply volatility. Chip buyers and distributors could protect themselves by ditching their lean models and embracing a healthy level of stockpiling. And effective communication along tiers would go a long way to break the bullwhip effect, ie the increasingly distorted interpretation of demand as one proceeds upstream along the supply chain.
Nonetheless, it’s obvious that chip supply is vulnerable right now and will remain so for some time. The next unfortuitous combination of events is probably already looming around the corner.
Main image credit: Flickr/Rob Bulmahn