EU Wants 20 Percent of Semiconductor Manufacturing by 2030
The European Union is considering plans that would put it on course to manufacture up to 20 percent of all leading-edge semiconductors in the world by 2030. There’ve been reported conversations between EU officials, Samsung, and TSMC, with an eye towards increasing the number of semiconductors manufactured in Europe. Now, officials at the EU are reportedly prepping a draft proposal on the topic for consideration by various member states and the European Parliament.
Currently, about 10 percent of the world’s semiconductor manufacturing facilities are located in Europe, but this figure makes the situation look better than it is. The vast majority of European foundries are off the leading edge. Intel has a facility in Ireland that’s currently building 14nm and has been cleared for expansion to 7nm, but there’s not much else. GlobalFoundries is spending $1.4B this year to expand capacity at its Dresden facility, but that foundry builds chips on 22FDX (FD-SOI), 28SLP, and >40nm nodes. GlobalFoundries hasn’t canceled its 12FDX follow-up, but the company is slow-walking the node and does not expect to have it in production in New York before 2023 or 2024.
The EU’s Goal Is to Reduce ‘High-Risk Dependencies’
According to the draft proposal, one of the goals of this project is to reduce high-risk dependencies on technology companies in the United States and Asia. These goals make sense, given what’s been happening across the larger semiconductor industry these past few years. Intel, in the United States, has had problems with its 10nm and 7nm node transitions. The entire semiconductor industry is currently bottlenecked and 300mm chip production is highly concentrated in Taiwan, Korea, and Japan:
Some of the rumors around the EU’s plans in this area have involved the formation of a new foundry company, while others suggest a partnership with TSMC or Samsung. Both the pandemic-created semiconductor shortage and the US-China trade war have disrupted supply chains, and it may make good geographic and geopolitical sense for the EU to collectively improve its semiconductor manufacturing capabilities.
If the EU is serious about reducing “high-risk dependencies,” it’ll have to do more than just build a new foundry. The semiconductor industry is heavily optimized for efficiency. Resiliency and efficiency are not always opposed to one another, but foundries face huge fixed costs. Build a new foundry, and the owners of said foundry will seek to run it as close to full as possible. Furthermore, the current semiconductor shortage isn’t just being driven by a lack of wafer starts. A shortage of ABF substrate has also caused problems.
The semiconductor industry is characterized by high barriers to entry and long development cycles. Companies such as AMD, Nvidia, and Qualcomm cannot switch their manufacturing partner from TSMC to Samsung or vice-versa without a significant investment of time and effort. Building resiliency into the system means creating an ecosystem that favors companies that maintain additional, deliberately untapped manufacturing and production capacity at every relevant stage of the supply chain. Building additional foundry capacity in Europe may not protect the bloc from the types of high-risk dependencies currently giving everyone a headache.