European 2nm chip fab a “futile endeavour,” says think tank
The white paper has been prepared, in part, in response to the 2030 Digital Compass decadal plan produced by the European Commission, which sets out the aim to establish leading-edge semiconductor manufacturing in the European Union (EU). The goal is to operate semiconductor fabrication plants (fabs) with 2nm process nodes within the EU by the end of this decade.
The think tank’s white paper (see The lack of semiconductor manufacturing in Europe) concludes that while building and equipping a wafer fab would be technically possible it would only be sustainable if supported by a robust business case. Otherwise it would represent a waste of tens of billions of euros that could be more usefully spent on other things. It would also run against the long-term trend of Europe’s reduced investment in chip manufacturing.
Total wafer capacity by region and technology node as of December 2020 (millions of wafer starts per month). Source: Stiftung Neue Verantwortung.
The figure shows Europe has barely invested in chip making over the last 10 years (capacity up 18 percent) while China’s capacity has tripled.
The author concludes that Europe creates almost no demand for leading-edge silicon and could not expect to sell chips to areas where the demand is high, essentially the United States. The US fabless chip companies will buy either from TSMC, Samsung or possibly Intel who all have plans for leading-edge fabs in the United States.
Next: What demand?
“It is true that NXP and ST do use 7nm and 5nm production, but just not enough,” Jan-Peter Kleinhans, author of the white paper, told a press conference.
Estimated share of 7nm and 5nm wafer shipments by customer in 2021. Source: Stiftung Neue Verantwortung.
The figure shows no named European customers with European demand just a part of the RoM zone.
Jan-Peter Kleinhans is director of the “Technology and Geopolitics” project. Currently his work focuses on the intersection of global semiconductor supply chains and geopolitics. He went on to say there are four potential routes for Europe towards 2nm, none of which look likely achieve the desired result.
1) Incentivize Samsung and or TSMC to invest in Europe.
2) Create an EU consortium to build and operate a fab using licensed process technology from Samsung or TSMC
3) Create an EU consortium to build and operate a fab using its own IP
4) Rely on Intel to create leading-edge foundry capability within Europe.
Kleinhans pointed that there is almost no incentive that Europe could provide to attract Samsung or TSMC to come to Europe because there is no significant customer base here. “The problem is a European foundry needs to attract US customers, which it will not,” he said. Added to this is the issue that the cost of operating is considerably higher than in Asia because of electricity costs.
This makes option two or three more likely to be implemented but no more likely to succeed.
Finally Intel’s offers to turn to foundry supply out of Ireland and Israel and potentially build another fab in Europe should be welcomed but are unlikely to get to the leading-edge. But Intel should be welcomed because serving Europe with trailing-edge silicon for automotive and industrial applications are areas where Europe does generate demand.
In response to questions about the European Union invoking extreme subsidy measures to pay a leading-edge foundry to manufacture in Europe Kleinhans said that this would likely be a contravention of World Trade Organization (WTO) rules.
There is the possibility that the European Union and nations around the world may abandon the WTO and re-introduce tariffs and other trade measures to achieve their goals of strategic independence in semiconductors. But this is not something that Kleimhans’ report considered.
The main conclusion of the white paper is that a focus on leading-edge wafer fabrication in pursuit of technological sovereignty in semiconductors is ill-advised. Without a business case it will likely waste billions of euros of public and private money, the report asserts and argues that the money would be better spent building up Europe’s chip design capability at the leading-edge so that European demand at the leading-edge can be of global significance. If that can be achieved within a decade or two, then Europe may have the leverage to attract manufacturing capacity.
This is something that the European Commission has recognized in the past but has been unable to do anything about. Over the period 2010 to 2020 Europe’s share of the global fabless semiconductor domain has fallen from 4 percent to 2 percent, Kleinhans observed. With the leading proponent, Dialog Semiconductor, set to be acquired by Renesas, the downward trend will likely continue.
Related links and articles: