For one it highlights the not-so-apparent fragility of the semiconductor industry. As a complex network with nodes spread all across the world the supply chain works seamlessly at most times hiding the underlying absence of redundancy.
William Shih of HBS has a beautiful analogy to describe this system:
This structure is best thought of as a transcontinental relay race with hidden hurdles
The Economist chart illustrates these hurdles. For several narrowly defined markets in the semiconductor manufacturing process, there is just one dominant player.
What this shows is that just like the chip design process, the semiconductor supply chain also has several layers of abstraction. As a chip designer, you don’t need to know the intricate details of a transistor; abstracting it out as a switch can work as well. Similarly, downstream firms in the supply chain may only know their direct suppliers and need not worry about the upstream interactions at all. That is so until something goes wrong.
Like it has now when the US has chosen to use these choke points to constrain China. Given that this conflict is strategic and not economic, any concessions China agrees too are unlikely to change the US stance. The policy instruments might change but the goal will stay.
Which brings me to the next point: what are some of the choke points that the US can exploit in the future? I have always thought of Taiwan, particularly TSMC, as one such critical chokepoint because of its commanding dominance in cutting-edge chip fabrication and IC packaging. Some figures to illustrate this:
Within individual industry segments, Taiwan leads in IC packaging and testing with 50 percent of the global market, and in foundry manufacturing it has a commanding 70 percent market share. Further, Taiwan accounts for 20 percent of the world’s IC design value and is the number-four memory chip maker with 10 percent of the market. [TheNewsLens, 2018]
For now, TSMC’s exports to Huawei haven’t been affected as its supplies are below the American intellectual property threshold for the ban to become applicable. But this could well change if the US wants to ratchet up the pressure or if relations between Taiwan and China go downhill.
For the semiconductor ecosystem itself, this friction will translate into significant deadweight losses. The chip fabrication market has been consolidating for quite some time now:
from 2002 to 2016, the number of semiconductor manufacturers with a leading-edge fab has fallen from 25 to just 4: Intel, TSMC , Samsung and GlobalFoundries. And GlobalFoundries announced in 2018 that they would not pursue development of the next node. [Thompson and Spanuth, SSRN]
It might well be that the ongoing geopolitics rather than technological limitations will halt the development of faster and smaller chips. Political friction will put further strains on this already consolidated industry and the development of leading-edge nodes (below 1nm) will slow down.
SemiconPolitics is here.