27-Dec-2024
Privatize Imec
Preamble The following is a translation of an article that I originally published in Dutch, on September 6, 2021. Since its publication the situation has only gotten worse. Here's the thing: the European politicians have discovered that Europe is lagging far behind in chip production. Now, if there is one publicly funded institute that has contributed to this, it is imec. But what is our politicians' solution? Even more subsidies to imec! And not just a few percent, but an order of magnitude more. One has to wonder what these masterminds are thinking.
When I graduated in 1985 as an electronics engineer, it was obvious: our generation would put Leuven on the map as the new Silicon Valley. The catalyst would be a recently established “superlab,” the Interuniversity Microelectronics Center (imec). The goal of imec was “to strengthen the Flemish microelectronics industry.”1 This would require start-up subsidies, but these would be phased out within a reasonable period of time.
Fast forward to today. What was actually strengthened during this time was imec itself. The annual subsidies were not phased out at all: they have instead increased sharply. In 2020, imec received 109.6 million euros of “structural funding” from the Flemish government.2 Things did not stop there. Another 44 million euros came from the EU. And then there were the project subsidies granted by a number of Flemish and European bodies in the framework of cooperation agreements. Sometimes it seems as if noone at imec moves until a subsidy tap is opened somewhere.
And what does imec actually do with all these subsidies? For the most part they are used to co-finance the R&D of a number of the largest chip chip manufacturers in the world: Intel, Samsung and TSMC. These are all non-European companies without factories in Europe, let alone Flanders. Their shareholders should be grateful to the Flemish taxpayer for his generosity. They get the added value and profits, we get some technological prestige.
I look at it with vicarious shame, because if there is one sector that does not need subsidies it is the electronics sector. Consumers are fond of the end products it enables. Consider the smartphone with its handy apps with which you always stay connected, the Internet where you all your information, the laptop as a virtual meeting platform, the modern car full of safety features. Or think of the smart devices that provide sporting and medical guidance, or the miracle of cochlear implants for the hearing impaired. It is all possible thanks to chips. The massive demand and wide applications are resulting in an impressive mobilization of private investment and entrepreneurial spirit. Government involvement is unnecessary.
As usual with subsidies, the direct cost is only one aspect. In addition, there are the hidden costs due to unfavorable side effects. As mentioned, a lot of the developed know-how goes to large foreign companies. The Flemish taxpayer thus sponsors the competitors of Flemish and European companies.3 Moreover, the brilliant engineers who work in imec's comfortable environment are not available for more productive tasks, such as helping to build a Flemish Silicon Valley, where companies run the show instead of a government institution.
There is also a problem in the area of chip design. Besides a well-known name as Melexis, there are a number of Flemish chip design companies, including spin-offs from KULeuven and imec itself. Since imec's inception, these companies have faced unfair competition from imec's subsidized design services.4 Attempts to do something about this have been systematically ignored. In a tumultuous conversation about this, it was once made very clear to me made clear that spin-off companies serve to promote imec's growth rather than the other way around.
And why does imec need to be a Design Center for TSMC in Europe?5 That is a purely commercial activity. The justification that imec sometimes used is that they only engage in activities that the market does not offer. That isn't so difficult, of course, if you shut the market out first.
Admittedly, imec plays the game quite well. After all, what is imec really? Sometimes they behave as company, for example in protecting intellectual property and developing commercial activities. Sometimes they behave as research institution, for example in winning project grants. (In contrast to real companies, imec is subsidized for the full 100% of the proposed cost.) The pleasant consequence of this ambiguous structure is that imec can pick the role that suits them best in any situation.
Imec's model is regularly defended with the argument that R&D for chip manufacturing is becoming so expensive that individual companies can no longer pay for it. Admittedly, it is quite plausible that R&D collaboration is appropriate. However, there is no reason why that would require taxpayer money. In fact, the successful model where the interested companies pay for their R&D themselves is proven and available from imec's closest partner: the Dutch technology company ASML.6
In summary, ASML develops the machines to manufacture chips, and imec the process technology implemented with those machines. To the market, they appear to be technological twins. Consequently, they have the same customers, such as the well-known names Intel, Samsung and TSMC. But that is where the similarity ends. ASML is a commercial company. To finance the expensive R&D, ASML's customers are asked to contribute, via participation in capital raises and direct cash injections. The program is called the Customer Co-Investment Program.7 There is no reason why one could not use a similar scheme for imec's technology.
Therefore, I would like to conclude with a suggestion to politicians. It would be in the public interest to get rid of the ambiguous structure of imec. Consider privatizing their commercial activities. ASML can serve as an example. Since the technology developed is state-of-the-art you will undoubtedly get a good price. Moreover, from then on you will avoid the annual subsidy expense. You can then spend that money on core tasks, such as infrastructure for mobility, water management and basic education.
And what about the fundamental research done at imec? Bring that back to where it always belonged: the university.
Source
-
About imec's original purpose: https://nl.wikipedia.org/wiki/IMEC ↩
-
Imec's 2020 numbers: https://www.imec.be/nl/press/imec-presenteert-succesvolle-jaarresultaten ↩
-
Roland Duchâtelet (Melexis, X-Fab) on grants to imec: https://www.tijd.be/politiek-economie/belgie/federaal/duchatelet-voorstel-meerwaardebelasting-is-gedrocht/9820449.html ↩
-
imec competes with Flemish chip design companies: https://www.imec-int.com/en/what-we-offer/ic-link/asic-design ↩
-
imec as Design Center for TSMC: https://www.tsmc.com/english/dedicatedFoundry/oip/value-chain-aggregator/imec ↩
-
On the need for cooperation on R&D and partner ASML: https://www.tijd.be/ondernemen/technologie/in-de-space-race-om-chips-wil-iedereen-een-stukje-van-leuven/10321488.html ↩
-
ASML's Customer Co-Investment Program: https://www.asml.com/en/news/press-releases/2012/asml-announces-customer-co-investment-program-aimed-at-accelerating-innovation ↩