It’s been clear for a while that the major world powers are going to need to do something about securing their own technological future in a world of competing nation-states (“The resurgence of national tech wars,” newsletter #53). One of the things that has come out of the global pandemic, beyond the generic need to have a close look at supply chains (“Markets finally notice coronavirus,” newsletter #19), is the specific problem we have with the semiconductor industry. Mostly, when we think of chips, we think of CPUs, or memory chips — but chips are in everything. This has come to the fore as auto-manufacturing plants have turned idle because of chip shortages. That immediate problem is one of just-in-time manufacturing, and the impact of a global pandemic that shut down … everything; but this shortage is definitely being viewed as a national security issue by the U.S.
There are two parts to the national-security question: who has the chip technology, and who does the manufacturing. Right now, the U.S. has the technological lead in R&D, design, and processing technology — and that lead has garnered U.S. companies 45% of global market share. But production is a different story:
While U.S. semiconductor companies are making the lion’s share of revenue, the vast majority of production happens overseas. Worse, from a supply-chain fragility perspective, just five “fabs” (semiconductor fabrication plants) around the world represent 54% of global capacity. A U.S. congressional research report (from which I excerpted the data for the above chart), published in October, does a deep dive on the current state of the semiconductor industry, both from a commercial and defense perspective. It’s against this backdrop that President Biden moved quickly to support the $37 billion in federal subsidies for the U.S. semiconductor manufacturing industry. But what, exactly, will $37 billion do?
We get a hint at the answer by looking at Micron’s actions this week. Micron is the only remaining manufacturer with a significant production supply in the U.S. They’re the #3 producer in the world, by capacity, at 9.3%. This week they announced they’re exiting one of their minor businesses — an alternative memory technology dubbed 3D XPoint — and will be shutting down and selling a Utah-based fab for production of these chips, because there wasn’t enough demand for the chips. If the U.S. (or any country) wants to maintain technology and production leadership, and wants to control their own supply chain, then by definition, they will have to support excess capacity. Micron estimates that underutilisation of the Utah fab was costing it around $400 million per year. Moreover, this is a plant dedicated to small production of a new technology. Tim Arcuri of UBS estimates that if someone bought the plant, the replacement of equipment alone to enable it to produce some other chip would exceed $3 billion. If a single, small plant takes billions of dollars in equipment and costs half-a-billion a year to run under capacity, all of a sudden that $37 billion doesn’t look so meaningful. Certainly not “control your own destiny” meaningful.
But if the U.S. wants to maintain the lead it has in intellectual property (never mind national security and supply chain issues), it has to do something, because China isn’t standing still. The chart above shows that China manufactures about the same percentage of silicon that the U.S. does, but there are two key differences. First, China’s production is primarily tied to low-end chips; they are generally understood to be years behind in fabrication technology, and in high-end logic-circuit design. Second, while they represent 12% of the world’s production capacity, they consume (either directly, or for inclusion in components that will be on-sold) 60% of global semiconductor production. From their perspective, worse, only about 15% of the semiconductors that are directly used internally are produced in China, and around 50% of that is produced in China, but by foreign-owned corporations with foreign technology. They have been concerned about this deficiency for a while, and established two campaigns to resolve it. In 2014 they set a goal of “becoming a global leader in all segments of the semiconductor industry by 2030;” and as part of the “Made in China 2025” initiative they set the goal of “achieving the know-how with regards to advanced semiconductor manufacturing as a vital component of China’s future economy and society”.
Today, they’re meaningfully behind on both of those goals. However, this week Xi Jinping came out with gold goals of reducing China’s tech dependency on the West. This has renewed urgency, because it looks as though President Biden is going to move forward with President Trump’s rules aimed at China’s technology threats. Mind you, it’s easy to think this is easier said than done, but when Xi talks people start moving, and let’s not forget that in the last two decades technologically-backwards China has sent people into space, built their own aircraft carrier, and developed their own stealth fighters. China is “only” committing roughly the same amount of money that Biden is aiming for, but they may end up being far more effective with that money. The details of their strategy will come out over the next week or so, as the various ministries release their five-year plan, but you can bet that they’ll be aiming for a leap-frog fast-follower implementation, where all of that money is going directly to copying the most successful and interesting current endeavours, without having to go through the intervening 30 years of internal innovation. To make things extra-exciting from a geopolitical perspective, let us also not forget that China thinks Taiwan is part of China, and guess where the world’s largest semiconductor manufacturer is homed?