Editor’s note: Admittedly, we usually spend a lot of time talking about advanced semiconductor manufacturing. This is a common mistake that everyone falls into when it comes to convenience foods, and we’re just as guilty of it as everyone else. The world rightly focuses on the shortage of companies capable of operating at the top, but there are many more half-baked products.
We recently searched for manufacturing capacity data by technology node, and everyone agreed that Bill McLain of IC Insights is the leading expert on the subject. He adheres to one of the strictest models in this matter and fairly charges extra for his reports. This is a must-read for anyone planning a multi-year roadmap for the semester.
Quick Google a search yielded this snippet of IC Insight data, and it tells an important story…
More than 90% of the world’s semiconductor manufacturing capacity operates on a 10nm process or higher. One can argue where to draw the dividing line, but it’s safe to say that the majority of capacities work in the rear.
First, when the world ran out of semiconductors in 2020/2021, most of that shortage occurred in these more mature processes. All of TSMC’s leading customers were able to get most of the capacity they needed on a 7nm process, but it was a real challenge for industrial and automotive customers.
These companies needed prosaic parts like microcontrollers (MCU) and power management ICS (PMICs), and these products are typically manufactured on older assemblies. Today, despite the fact that the lack of supply has translated into excess inventory in many categories, the old products are simply catching up with the pent-up demand of two years ago.
Second, the US government is currently trying to decide how to allocate $52 billion under the CHIPS Act. If the purpose of these funds is simply to bring advanced technology back to the US, then give all the money to Intel. They’re going to pay $7 or $8 billion in dividends to shareholders, and they’re going to continue with their production catch-up plan, which they’re going to have to do anyway.
On the other hand, if the goal is to truly secure the U.S. semi-finished product supply chain, then perhaps the better plan would be to spread that money more widely. Ideally, they will need to spend money to sow many seeds leading to the creation of new companies and for basic research that can then be commercialized by the private sector. Unfortunately, there are still no simple mechanisms for doing this, so another approach is to divide the funds among a wide range of US companies engaged in the production of semi-finished products, provided that they commit to increasing capacity in United States. This applies not only to factories and foundries, but also to tool makers, robotics suppliers and chemical manufacturers – the entire supply chain. Intel should get some of those funds, but not most.
According to the Semiconductor Industry Association, the chip law has had a positive side effect, prompting the private sector to invest about $200 billion in semiconductor manufacturing in the United States.
Finally, these numbers should remind us that the story is broader than TSMC and Samsung. There is still much interesting and important work to be done in advanced foundries.
The most obvious example is Global Foundries. GloFo is not a leader in silicon manufacturing, but it has occupied several very significant “niches” such as silicon on insulator (SOI) and silicon carbide. And while they don’t have nearly a duopoly in that, like TSMC and Samsung with 7nm, they’re getting close on many of their SOI lines. If for some reason the US ever loses access to TSMC, GloFo could become as important a part of the solution as Intel.
Related posts:
