STMicroelectronics and GlobalFoundries to build wafer fab in France While the US continues to debate how and whether to subsidize its semiconductor manufacturing industry, one of its leading lights – GlobalFoundries – has announced a joint venture to build a wafer fab with Swiss concern STMicroelectronics and "significant financial support from the State of France". The two companies today …
[France] has some of the highest tax rates in the world, a whopping 45% top marginal rate
If you have too much money, feel free to give me some :-P
Crolles might well be a viable investment. We’ve known that the New European Fab at Crolles has been a done deal for four or five years, and that it was going to be FDSOI. GloFo need to make 22FDX, and STMicro FDSOI22. Both are good, economic nodes for industrial and automotive. Fairly cost-insensitive sectors, which also leverage a lot of value. $50 of chips can make a $100k end-product rather easily.
Totally unlike Intel Magdeburg which is insanity: an attempt to sell leading-edge products that are relatively low-margin from a high-salary, high-input-energy-cost environment. And where the equipment capex is 50x that in Crolles (perhaps more), with almost all the output value being effectively the risk premium on owning that equipment due to the high possibility of it becoming stranded over the business cycle. Instead of just exporting the steppers, EUV sources etc direct from ASML Netherlands and Austria, to Taiwan, Singapore, USA and making an immediate large profit margin for EU companies. I’ve never once seen any of its advocates note that out of the $80bn intended to be spent on EU CHIPS, $70bn is a deliberate direct *subtraction* from EU exports! There aren’t going to be any more EUV sources in the world, they are being sold as fast as they can be made, just fewer of them will be exported.
The two investment business cases couldn’t be more different.
Depends on how big said chips are. You get around 600 10mm dies per 300mm wafer. A CPU might use multiple 14mm (200m^2) dies, so obviously you get a lot less. This fab is aimed at various less cutting-edge things so might get more. 400 million chips would probably be optimistic, 50 million might be a bit of an underestimate.
Zero consumer products. Not the right market. This is going to be mostly industrial and automotive parts.
If you want a “feel”: production is usually quoted in wafers per month, this fab is fairly average at 50k wpm. World production is about 8 million wpm. To reach its ambition of 20% world production, Europe would need around 30 such fabs.
An investment this size is suitable for something rather niche, like automotive. The real problem for the automotive industry is that in its own head its very important, but it’s tiny for semiconductor. Total annual EU vehicle production is 10 million vehicles, each of which might have 100+ chips. But, apart from anything self-driving, most of those are small microcontrollers, maybe 3mm2 each. Therefore, the whole of EU production for automotive can be satisfied by just 40k wafers. Less than one months production of a single fab. If you build a fab “because it’s strategic for the automotive industry”…..what’s the fab making the other 11 months of the year, to make it economic to run the fab at all?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
