Back in October we wrote a piece about shortages causing disruptions to the supply chain. Inadvertently, this turned out to be one of most predictive pieces, and now this topic is all over the news. We could supply some links but the stories are everywhere ranging from small stories about individual plants shuttering to massive losses at auto companies to the US President issuing executive orders to supposedly solve the problem. We think it is probably time to call the peak of these shortages. It is unclear how much longer they will last, but they will end. And while this year’s shortages are more acute than any in memory, this is not the first time the chip industry has faced shortages. There is no reason to think this time will end any differently.
The current shortage is driven by two things: Covid and manufacturing is hard. Last year, factories in China started shutting down early in the year. By the time they started re-opening later in the year, factories everywhere else had shut down, or at least drastically scaled back. So many Chinese factories did not scale production up in a hurry. The result was reduced output everywhere. Then early this year it became clear that pandemic was under control in China and vaccines were coming soon everywhere, and that demand for many electronic devices was stronger than ever. Companies across the supply chain suddenly had to race to catch up, but often found their suppliers were not back to producing at full scale. This caused wrinkles to spread across the industry. Finally, it appears now that shipping has become a bottleneck with air freight prices sky high (pun intended) and US ports still dealing with scaled back working conditions. You could probably also throw in the US-China Trade War, but let’s not go down that particular rabbit hole now, just chalk it up to further complicating the situation.
The best example of this, and one we commonly hear, is the proverbial $2 electric motor which prevents completion of a $50,000 car. Now multiply this by 10,000 companies all hindered in their ability to ship because of some shortage from someone else’s factory.
Our October post was inspired by the situation with one of our clients. They had just finished manufacturing their chip at TSMC but could not get the chip back because they needed to place the die in a plastic package. Chip packaging is often seen as the most boring, taken for granted, corner of the chip industry. Labor intensive and not terribly profitable. But a shortage there of one of their feeder components (like a $0.10 part) prevented them from shipping. We were also working with another company that saw lead times for parts for its new product go from 4 weeks to 27 weeks, which was threatening to cripple their launch and their ability to ship to a major customer.
In the ensuing months, as conditions worsened, chip customers have responded in all the normal ways, as have the suppliers. The problem is that restarting a manufacturing line is not as simple as flipping a switch. It takes time to restart, and especially in China, long closures meant the loss of staff, with weeks required to train replacements. Machines have to be retooled and cleaned, inventories restocked etc. For their part, customers have taken a whole host of measures to ensure supply. The tech press is full of stories of companies doing everything they can do jump ahead of the line. Suppliers are putting customers “On Allocation”, meaning only sending out a portion of orders, with a rank ordering of customers.
The dirty secret of chip industry practices is double ordering. Maybe a company buying a 100,000 parts from a supplier is not enough to get the supplier’s attention, but 200,000 makes the buyer a top priority customer. So the customer orders 200,000 parts, vaulting them to first in line. You can see where this is headed. Whenever the shortages ease, the customer cancels half their order. And since chips diminish in value quickly (thank you Moore’s Law), the supplier is stuck with inventory of a part that it probably now has to discount. We have not reached that point in the cycle, but we think customers are so desperate that they are not only double ordering but likely triple ordering. When the bubble bursts, the problem is going to be that much larger.
Now some people are arguing that these shortages will persist into 2022. We think that is unlikely. Anecdotally, both of our clients who were suffering supply shortages in October found that their suppliers resolved their shortages far faster than feared, with parts available much sooner than the worst case scenarios. More broadly, many of the current problems seem very temporary in nature. For instance, today there was a story about a plant in Texas that shut down because of the clod weather spell there last month. The factory itself was not damaged, so there are few reasons why it will not be up and running in a week or two. The same goes for a host of other cases. Suppliers are busy adding capacity (but not too much because they know their customers are double ordering) and while factories cannot be turned back on overnight, the delay is measured in weeks not months.
Finally, there is the question of what we can do to ensure this does not happen again. The short answer is nothing. The world just went though a huge shock but everyone can still buy all the iPhones they want. We do not mean to belittle the real suffering that the pandemic caused, that is very real, but the supply chain was at most a temporary casualty. The electronics industry has proven fairly resilient.
Photo by Mick Haupt on Unsplash
Thanks for the excellent article and a number of cogent points regarding the intricacies of the industry. Just curious where the shortage easing situation you are seeing comes mainly from? Is it industry-wide, or say autos, smartphone, PC, server etc. Love your articles
I think the shortages are everywhere. If I make devices and my chip vendor says they are out of stock then I say its my chip vendor causing the problem. But maybe the chip vendor is short of substrate or some other input. And that vendor is short of some chemical or workers. Ripples everywhere. Part of the reason we hear so much about autos is that they have a lot of components so there is a big shortage surface. But also, autos consume something like 4% of semis, so they are way back in the line.
Thanks for your kind words and thanks for reading.
Thank you, that provides much more context than a generic news article talking about the shortage. In this article, you actually mentioned this shortage is “easing”. Just wondering where is this improvement from before coming from within the industry?
Pingback: Start Up No.1502: Russia’s vaccine disinformation campaign, Biden gets antitrustful, Huawei slashes smartphone orders, empty Olympics?, and more | The Overspill: when there's more that I want to say·
Pingback: Unfolded: #9 – Robin Grainger Consulting·
Recent comments from Gelsinger on shortages having a material impact in to 2023 seem to be accompanied by others echoing similar predictions. I suppose that may be partially a rationalization of the new strategy, but to say that they are unlikely to persist to 2022 is quite the deviation. Not saying you are wrong, but I’d be curious as to how you feel this article has aged over the past month and a half. Have you seen more anecdotal evidence either way?
I think Gelsinger is speaking to Intel-specific problems. Porting their products to TSMC leading edge process is going to take a lot of time.
But everyone I have spoken to since writing this argument says the same thing. Conditions everywhere else are bad now, but they are no longer getting worse. I don’t know when the shortages will end but I’m fairly confident we are past the the peak. If I had to guess, I’d still say shortages are over by end of 21.
Pingback: The global shortage of microchips: understanding the complexity of those damn things – Gregory Bufithis·
“Now some people are arguing that these shortages will persist into 2022. We think that is unlikely.” – Sadly, that prediction aged like milk. As of mid 2022, Microprocessors, FPGAs, and many automotive-rated components are now on 1 to 2 year waiting lists without any real guarantee that they will be available even then.
It depends on the sector. Wireless and PC are seeing inventory levels rising. Automotive is still tight, but even there are signs that we are past the peak.