In the last few months, I have had some very interesting conversations with executives knee-deep dealing with the supply chain shortages for their company’s procurement. The narrative about the shortage of chips has concentrated on the leading edge manufacturing node, meaning 7nm, 5nm, etc. But from the conversations I have had, this is not the biggest issue impacting our industry shortage, nor is it their worry about where semiconductor manufacturers are investing their money. It turns out, the semiconductor’s biggest issue is at the trailing edge.
The Trailing Edge and Legacy Nodes
The leading edge gets all the attention because it is the most exciting. The leading edge powers the supercomputers in the cloud, our desks and laps, and our pockets. But computing devices are not just made up of leading-edge microprocessors. The vast majority of other components are made up of legacy nodes and quite often many chips on the trailing edge.
What became clear in these conversations is most of the fabs in the media like TSMC, Samsung, Intel, and even Global Foundries, to a degree, are not relevant in the trailing edge semiconductor manufacturing process. Most of the companies making these chips largely on 90nm process and larger are located in China. Generally, these chips are a commodity, which is why most big fabs do not invest much in the trailing edge. Yet, most modern digital devices run at least a handful of chips built on the trailing edge. So while not nearly as sexy as the leading edge, chips made on the legacy processes are still as important to the manufacture of computing devices.
No End in Sight
Sadly, this likely means there is no end in sight for the capacity shortage. The fear alone of semiconductor drought has caused most of the big technology firms to stockpile both chips and POs with suppliers. Unfortunately, this only exacerbates the demand and delay for the foreseeable future.
In discussions on the matter, even an easing of demand is not necessarily light at the end of the tunnel. This situation has highlighted the steep challenge of predicting demand and the fine line companies procurement divisions have to walk to manage such a diverse supply chain of components. However, there are worries that companies are now hyper-aware of the delicate supply chain balance and may move forward with a new strategy and philosophy of procurement that includes more advance purchasing and volume guarantees.
If, in the end, supply chain management and procurement logistics go through a strategy and philosophy change, it could mean a prolonged challenge to get core components in a timely fashion. The discussions of strategy happening in the supply chain are uncharted waters, which is fascinating in its own right.
A Point on Semiconductor Supply Chain Nationalization
I have written in the past the role the semiconductor supply chain plays in a national security discussion for a nation-state. This was the basis of my thesis around the US needing to invest more in semiconductor manufacturing for both the competitive need of its companies and its own national security. And while that is true and needed, it is likely entirely impossible.
Even if there was a leading node manufacturer owned by the US company and operated on US soil, said the company would still be subject to a global supply chain of parts required to make silicon. For example, things like wafers, lithography machines, etc., are generally purchased from companies outside of the US.
I make this point simply to say that for all the points we make about the need for local manufacturing of Silicon, the reality is we can’t escape the global supply chain of semiconductor manufacturing that make it near impossible for a nation like the US to have every aspect of that supply chain operations within its borders.
Our takeaway, then, is first and foremost that this semiconductor supply chain shortage has no end in sight. Second, manufacturers of mass-scale technology are securing their orders in bulk and creating a wait-in-line scenario for everyone else, exasperating the shortage. And lastly, even an emphasis on domestic manufacturing of silicon can’t alone solve this problem for companies that operate on its soil.