Not only is there a lack of components, the logistics chain is stretched to its limit, at least in the developed markets.

In second quarter reporting the recurring theme was that delivery problems would last well into 2022, in third quarter reporting consensus seems to be that issues will remain into 2023.

We have all seen pictures of the queue of container ships outside the Los Angeles Harbor. Mid-october there was 55 ships idling in the San Pedro Bay, November 10 there were 78 ships waiting. Everybody knows its impossible to get a Playstation 5, that you’ll have to wait until 2023 to get delivery on a BMW iX50, that it can take a year to get a new electric bike and that bitcoin miners are snapping up all the graphics cards in the market. On the other hand you can get a Playstation 5 in the second hand market tomorrow (at double price), if you ordered a Tesla Y in October you’ll get it before Christmas, and you can walk into a store and buy an electric bike (if accept a limited selection of models). On the websites of Dustin and Komplett they have lots of PCs on the shelves, even if the selection of graphics cards is rather limited.

Nor are the same ships lining up in San Pedro Bay in November as in October. The flow of goods through the container port in Los Angeles is higher than in a long time, demonstrated by the number of vessels leaving the port being approximately 50% higher than pre-covid. Despite the component shortage, or rather because of it, days of inventory is at an elevated level for many companies in the semiconductor industry. Logitech, as an example, has around 100 days sales of inventory, historically they have had below 80. Some components can’t be procured anywhere, while others are inventoried in ample numbers, but if there is one critical component lacking, you can’t deliver the finished product. “Everyone” is taking precautions by tactically ordering more than they need, partly because of “rationing” by the suppliers, and partly to build stock to ensure ability to deliver in the face of uncertain lead times for different components. “Everyone” could have sold much more, as the customers say they want much more than the companies can deliver. If you were the only one able to deliver, either because you have wholly unique product or because no one else can increase capacity, that would be true. If not, your customers have probably laid the same tactical orders with your competitors as well.

If a customer in a hypothetical case puts in orders three times real demand with you and your two competitors, and ends up with 25% allocation with all three, the customer has satisfied 75% of his demand. In this case the three competitors could have sold a third more to the customers, while each of them might have a perception they could have sold three times as much. This is of course a very simplified example, but if all suppliers had had twice the amount of goods to sell, I would be surprised if all of them had sold twice as much. At some point in time all inventories will not continue to increase, some might even be reduced, even if many will hold more inventories than historically (the Just-In-Time philosophy had some downsides through the pandemic, e.g., in the auto industry). With time there will be less tactical orders, and some players might stop ordering altogether and reduce stock, at least for some of the components they need.

On top of this, there are large investments being made in production capacity. There is a shortage of production equipment as well, both new and second hand, and establishing new semiconductor fabs can take years. Over the course of a year or to, there should, however, start to come more production capacity in the value chain. Maybe around the same time that stocks are peaking and demand has started to recede somewhat? There are projections from industry associations forecasting close to 40% growth in global demand for semiconductors from 2020 to 2022. With continued tactical ordering and building of inventories that may hold true, but there could be a rude awakening in 2023 (or earlier?) for producers of more generic components. Macro specialists think the demand for goods in the developed countries will fall back over the next quarters following the covid-19 induced boost. Emerging markets might of course contribute more after a weak period, but how fast that could manifest itself is an open question. In general the experts don’t seem to believe in a long lasting economic upturn from where we are now. The covid-19 flare-ups we are seeing could of course delay the consumption of services further and support continued strong demand for goods, but one should think that the effect is weakening.

The big question mark is Taiwan, which headed by TSMC (Taiwan Semiconductor Manufacturing Company) is a critical part of the semiconductor industry value chain. If the tensions in the Taiwan Strait should lead to a reduction in Taiwanese delivery capacity, that would be serious blow to the global market. There are large investments being made in semiconductor fabs in both North-America and Europe, but these are projects that take years to come to fruition. If the Taiwanese capacity remains intact, there are many indications that the component shortage at some point abruptly will change into a situation with over-capacity and a glut of components (at least for certain types of components). As always, timing is the tricky part. How much inventory is there throughout the value chain? When will critical components lacking today become available again? When will delivery times normalize in developed markets (they haven’t increase that much in emerging markets)? How much of the component shortage is driven by tactical orders, both in size and timing as we see large order backlogs for delivery far into the future? It is far from easy to get a complete picture of the situation and outlook throughout the value chain. My impression is that the assumption that will last for more than a year more to a certain degree is driven by the demand perceived by each supplier individually. If this perceived demand turns out to be more exaggerated compared to the real one than assumed, the outlook can change quite suddenly.