You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
I am currently asking myself how the European Chips Act will affect the wafer size trend within Europe. As we all know, Europe is especially strong in power devices and sensors, both applications mainly run on 8" wafers. Increasing wafer size of course would increase the UPH of such chips but the investment costs coming with this is tremendous. What is your opinion on this? Will we see a shift in wafer size trends in Europe or will 8" still remain the size for the future? Do you know any articles dealing with this topic? Especially with the background of the "new" materials like GaN or SiC which are not even available on a big scale for 8".......
If you look at current announcements you can see that investments are planned at both sizes depending on economy of scale. Bosch has done a 300mm fab in Dresden for old BiCMOS technology for capacity, ST/GF will invest together in 300mm in Crolles for FDSOI 22nm and below, ST/Intel (ex-TowerJazz) in 300mm in Agrate for a large product mix. At the same time you have ST investments in GaN in 200mm and Infineon having power products in Villach in 300mm. I expect 200mm fabs by everybody in EU to receive investments to increase capacity as demand looks to be solid for a while.
Consider that, even for "older" or "non-CMOS" technologies, moving to 300mm makes sense when the market pass a certain point. The advantages in unit cost and overall yield justify the investment provided you are ready to access capital and persuade shareholders to reduce your EBIT for a while. It is the (un)willingness of investing ahead of time, betting on being capable of capturing market share, that makes the biggest difference between Asia, US and EU.
The costs quoted for 300mm fabs are usually for bleeding edge and high capacity. If your products can satisfy your market with 30-50Kwafer/month and you do not need immersion and EUV then the costs are much more manageable. And, for running costs and identical volume of chips, a 300mm is much cheaper to run than a 200mm as not only you get an increase in area, but tools are more productive and consume less resources. One element is also the size of chips: some are so small that you can already get thousands to tens of thousands chips per 200mm wafer so the volumes for the switch to be doable are much higher. Also consider that for IDMs the overall cost for their product has to include the packaging. The business case is then different than foundries as a reduction of chip fabrication impact only part of their cost. Sometimes the packaging is more costly than the silicon.
In conclusion, provided that the EU Chip Acts in its final form will really generate an improvement in conditions of investments in Europe, it will be across the board.
Then is still to be seen if and how many of the non-european companies, will open fabs locally as they have business models that are not restricted to current european strenghts.
The old role of government in industry was to financially support research which trickled down, indirectly and sporatically, to industry. This was smart and successful, not because governments could pick the right things to support (they couldn’t) but because it was a ”pick them all” approach where 1 or 2 things out of the group of hundreds resulted in a big impact.
This was also smart because it was forward looking beyond the horizon of private capital, even speculative venture capital, which permits the government to “stay in its lane” and not be a nuisance or competitor to private capital.
So if you have the principles of long horizon, forward looking and pick them all, this is very different from the US or Europe Chips acts. The most-bad approach would be direct subsidies to NXP, STM, Infineon, Bosch and ABB that replaces private capital, invests in the past tech and looks to fix the short-term supply chain capa issues. And yet this probably exactly what these companies are begging for.
The most-good approach would be support physics and chemistry (and materials science, biology, etc) science and engineering research which would create trained technologists, patents, and eventually some of it would result In a big winner. You could use patents as a metric for progress. This is the best way forward.
Government role is only one element. TSMC and the Korean memory manufacturers would not be where they are without the state and bank aid they got between 2000 and 2010, when they were building capacity and pushing competitors and IDM out of the maket thanks to access to capital and tax conditions that nobody was having in the rest of the world and to investors not asking for real dividends. They executed perfectly and now they can further grow with their own revenues. In other places state aid has been wasted on badly thought out and executed schemes or the industry (eg TI, ST, Motorola/Freescale, ...) preferred to rely on slow growth but safe "niches" not requiring investments and maximising ROI and cash flow keeping shareholders happy. TSMC dividend has been zero until 2002 and only 4 times has been over 3NT$/share (~0.1$/share) since then.... At the same time the share went from about 10 to 80-90NT$. Tell me wherever else the shareholders would have kept investing between 20% to 60% of revenue each year for those financial results if they were not having external help or guidance (those with a bit of time can search on the public conditions of land lease to TSMC and on the largest shareholders).
Again, without perfect execution aid is just a waste but is fundamental to change the market situation in face of incumbent position.
I found this interesting: https://digital-strategy.ec.europa.eu/en/library/european-chips-survey
The survey says, for 2021-2025
Aggregate demand grows 4x between 2019 and 2025
Discrete devices are the largest group today and in the future
90-65nm is the largest node today and in the future
The survey represent the view of the current demand side in Europe. As there is no more consumer digital the view is is relatively narrow and is one of the sources of lack of understanding among the actors. At macroeconomic view, EU consumes as much advanced digital as the US, it is just that it is in systems that are not produced in EU hence, on the Commission side, the lack of domestic production is seen as a weakness akin to the one that emerged with covid and the initial lack of vaccine due to lack of chemical precursors. On the other hand, the buyres of chips look only at current demand and given their main product cycles length (car, industrial tools and aircraft evolve much more slowly than mobile phones and computing) still do not see the impact of advanced nodes and see only mature needs. The semiconductor producers are in the "we will invest after we get orders" and "we will wait for customers to tell us what to do" mode and so are just putting capacity in now. What all of them are missing, in my opinion, is that no EU or US market exists in reality, there is only the World market and so investment should be done looking to service beyond the region. As it is exactly was is also already happening: 70% of the components bought in EU comes from outside EU and 70% of the components produced in EU are sold outside the EU.
Basing decision on EU (or US) market is a nonsense. Producing advanced nodes in EU make sense in order to sell them worldwide as there is a demand, and not to serve domestic needs. Domestic needs already do not justify having the installed capacity for automotive ......
I am currently asking myself how the European Chips Act will affect the wafer size trend within Europe. As we all know, Europe is especially strong in power devices and sensors, both applications mainly run on 8" wafers. Increasing wafer size of course would increase the UPH of such chips but the investment costs coming with this is tremendous. What is your opinion on this? Will we see a shift in wafer size trends in Europe or will 8" still remain the size for the future? Do you know any articles dealing with this topic? Especially with the background of the "new" materials like GaN or SiC which are not even available on a big scale for 8".......
