TSMC looks to double down on U.S. chip factories as talks in Europe falter

"European chip and auto companies, for their part, are mostly lined up against the idea. They would prefer subsidies for the older-generation chips that are heavily used by car manufacturers and are in short supply.

Many of TSMC's most lucrative customers, such as Apple, are U.S.-based, while its European customer base is made up of mostly of automakers buying less-advanced chips."

Oh boy... This is exactly why EU will always stay third behind USA, China...

"We don't need e mobility, we have the best combustion engines!" Tesla owns VW now.

"We don't need Apple like chips"

This hurts. Apple and rest does many things differently and way better than EU. We should learn from them.

Or do I miss something?

Not to mention it takes years to build these fabs. Do they think the automotive industry will use the same chips forever?

These fabs are already there, it would mainly be increasing capacity.

As far as I understand, both Tesla and VW use the same older automotive chips (in the thousands). The self driving and infotainment chips are of course newer but very few. They send messages to the older more reliable chips.

Tesla Model S is a 2009 design if you think about it, though they did upgrade the touchscreen computer once in 2019

That sounds like a pretty low information hot take. While it's probably true that car manufacturers are calling for more manufacturing of the chips they use right now to deal with the shortages, that's a 2021 and 2022 issue. I don't see any plausible reason why Europe would not see the value in having advanced chip fabs over the next 15 years. It's mostly likely that behind the scenes TSMC demanded more in subsidies than Europe was willing to give and that the Americans' pocket books are wide open at the moment.

I think Europe is more likely to remain third due to bureaucracy, multilateralism and a regulatory and investment philosophy that favors European businesses and citizens over international businesses. The cost of being first is high anyway.

Eh, yes and no. New computer, new screen, new grill, new batteries, new engine, probably a bunch of other stuff I can't think of off the top.

Yep. This is sooo frustrating. Why does Europe, and Germany especially, just NOT grasp digitalization at all? I'm growing ever more angry that almost our politicians are 80 year old mega-boomers who are voted in by other boomers and couldn't care less about any interests other than their own. Young people and their climate & computors be damned.

Other people have called this the greatest strength of Nintendo: build stuff with slightly older parts that are cheap and proven to work, so that you don't have to sell at a loss

That aside, I'm in the process of designing an USB gadget and I find process shrinks quite pesky. Of course it'll reduce production costs for the manufacturer of the IC but it also reduces flanking times which causes more electromagnetic interference issues for me.

So insisting on using large ICs built using old process nodes is a very valid decision if you work in an area where reliability trumps a $0.001 per piece cost saving. Like in cars, robots, or industrial machines.

That said, I see it more and more that German companies outsource all of their production until they eventually have so little building skills left inside their building that innovation stops. It's kind of the same issues that drove Boing to recycle old designs, but in Europe it's happening with thousands of small companies, instead of one big one.

Tesla owns VW? Telsa is overvalued and still doesn't make profit from car manufacturing. And it's USPs get fewer and fewer and especially full driving autonomous won't happen in the near future despite Musk's claims

I feel the same way. Talk to your local IHK = chamber of commerce.

I did and mine will soon run an article series in all of northern Germany to educate business leaders about electronics, innovation and how that'll create jobs and lucrative patents.

It was some work for me to collect statistics and experts to cite, but I expect it'll help my business too by increasing demand.

I went with industrial cameras as the example product because they have a market yearning for innovation in factory robotics and optical AI.

It's a little bit more complicated. TSMC wants subvention for building the fabs, but the chips that will be produced will be sold elsewhere. So the EU gains nothing from it but costs. The EU needs a european chip manufacturer and developer not a subsidiary of Intel or TSMC

Tesla are good electric cars, the best in their categories, but Europe has no problem to produce good electric cars that are the best in their categories too.

The VW group is doing particularly well these days, with the ID4 or the Taycan for example.

Guess what by putting a factory doing the newest node you educate a whole new generation of engineers that can build and run those next fabs.

You've bought the CNBC fud. Tesla made 4x the profit from cars than bitcoin or credits. Valuation is based on the future, not on the past. Exponential growth matters - just look at the last ten years, then extrapolate a few years more. If you did not predict 500k Tesla cars delivered in 2020, back in 2015, then you can not credibily comment on current valuation ( which is based on the future).

We already have em, but they lag behind TSMC by decades

Valuation is based on who farted at wall street.

If you extrapolate, VW will overtake Tesla soon in electric cars.

The reality is more that untapped markets follow S curves. The electric car market will not be bigger than what the total car market has been but replace it. Electric cars are not there yet, but the market valuation is already far beyond that.

They do have a point, subsidies for something the EU doesn't really use are sort of wasted taxpayer money. But I think a leading edge chip fab in the EU would be strategically important.

Tesla made 438 millions profit, 100 million are from bitcoin. If we ignore the profit from carbon credits (they made more from credits than from bitcoin) they made 4x100 millions profit from cars by your math. So combined they made more profit than they did. What comes next? Are they building the perpetual motion machine?

Except Nintendo's strength and main selling ponit was never their hardware, but their vast entertainment IP roster(Super Mario, etc.) which they go to great lengths to enforce its protection. So it's not a good comparison to products who's performance is dependent on being on the latest tech nodes.

I was on the market for a new car this year in Europe and was happily surprised that every European brand now, not just VW, is offering a fully electrical model, at least, and several traditional models are offered as plugin-hybrids as well.

Audi has the e-tron[1] already, and plans to have every model it sells come with a fully electric version before 2030[2].

BMW has been selling the i3 fully electric model since 2013, now has several other electric cars[3] (including the iX, the first one it designed from scratch to be an electric car, and the ix3, an electric version of the popular X3) and most of its models come as plugin hybrid (I ended up buying an X1 plugin-hybrid, the electric models are still really expensive compared to hybrid, which are themselves really expensive compared to the old combustion engine models - but worth it because you save loads of money if the electric range is enough for your daily trips, like for me, and due to government bonuses and much lower taxation!).

Mercedez seems to have started a bit late, but now it's keeping up with BMW and also offers many plugin-hybrid models as well as the new EQ line[4] which is fully electric (I drove the EQC and it's amazing - the luxury of Mercedez with the modernity of a Tesla - but just a little too pricey for me).

Other brands are following the same pattern. For example, Seat has a new sub-brand called Cupra[5] which offers a more affordable fully electrical, but still pretty awesome, model called Formentor which seems to be getting pretty popular in my country. Volvo now offers the XC40 with a pure electric engine[6] or plugin-hybrid... and it has a subsidiary to challenge Tesla called Polestar[7] which makes highly performant electrical vehicles similar to Teslas.

Finally, Renault Zoe[8] is a fully electrical, small vehicle that is pretty popular in Europe (where small cars are more convenient to use in cities).

[1] https://www.audi.com/en/experience-audi/models-and-technolog...

[2] http://www.mobiletechdaily.com/audi-plans-to-electrify-every...

[3] https://www.bmw.com/en-au/discover/electromobility/electric-...

[4] https://www.mercedes-benz.com/en/eq/

[5] https://www.cupraofficial.com/

[6] https://www.volvocars.com/intl/why-volvo/human-innovation/el...

[7] https://en.wikipedia.org/wiki/Polestar

[8] https://en.wikipedia.org/wiki/Renault_Zoe

>Tesla owns VW now.

One is a company that make less money selling cars than trading bitcoins and emission certificates. The other is selling more cars than anyone else. It's less than 500.000 cars a year versus 2.8 million cars with just one brand from the VW group (out of 12).

Edit: According to VW's news site "Volkswagen delivered around 5.3 million vehicles in 2020". That does not include vans or trucks.

Tesla likely won't survive as a car company when other car makers stop needing their emissions certificates. Tesla would be deep in the red today if it didn't live off other income than cars. Look at their official numbers if in doubt. In 2025 Tesla will likely be an investment firm selling some cars.

See https://en.wikipedia.org/wiki/List_of_Intel_manufacturing_si... - Intel has a 14nm in the EU (Ireland) and has planned a 7nm fab also - so "the EU" also plays in this space. A TSMC fab also would be good too of course but things aren't that bad.

The EU wants a homegrown solution - as usual, they are providing frontloaded loans for this from the NextGenEU ~700B EUR fund.

https://digital-strategy.ec.europa.eu/en/news/member-states-...

I'd say car performance is pretty independent of the performance of the embedded CPU

Yeah, I made that comment more out of emotion and without doing any research, that's on me, sorry.

TSMC buys their next gen stuff from ASML (Netherlands), so the problem is not knowledge transfer, it's good old manufacturing and scaling up. (And that means labor, and traditionally labor laws in the Members States meant that offshoring large scale production was the smart choice ... strategic importance or not, if it doesn't make sense economically and business-wise, then that means it'll be instantly worthless strategically too.)

You missed that TSMC buys their next gen EUV lithograph machines from the EU (from ASML in the Netherlands).

What the EU needs is specialization (comparative advantage), not becoming a TSMC assembly outpost.

This happened to Kodak.

Unfortunately it is a truism that MBA led companies are incentivized for short term profits, which translates to selling future innovation by outsourcing now.

They aren’t thinking about what the next generation will be doing (I.e. they will be contracting for the companies still innovating.)

Did you publish your research?

>"We don't need e mobility, we have the best combustion engines!" Tesla owns VW now.

VW is one of the biggest EV manufacturers. Maybe you should pick a different example.

>The other is selling more cars than anyone else. It's less than 500.000 cars a year versus 2.8 million cars with just one brand from the VW group (out of 12).

Plus 231600 EVs sold in 2020 across all brands. It's no longer a battle between an EV company vs an ICE company. It's a battle between an EV company vs an ICE+EV company where the EV portion alone is half the size of the first EV company.

https://www.motorauthority.com/news/1130931_vw-group-sold-al...

I see ID.3s and ID.4s every day. Renault Zoe are quite common. Sometimes you get to see an e-up.

Teslas are exceedingly rare. I remember being excited about them in 2017 but nowadays I don't care anymore.

Of course because I am living in Germany it feels like half the cars on the road are VWs so VW dominance is not surprising, at least that is how I feel it subjectively.

That explains why global foundries stopped building leading edge fabs.

It only sounds good in the news, in practice it is a money pit, not a competitive advantage. You only do it because of national security.

The buy a tool from ASML. TSMC built up the knowledge to use that tool, to achieve these processes, themselves.

Knowledge transfer is key.

They will publish it in their magazine, but they'll remove company identifiers

I'd say the working relationship is the important thing. (yaay, synergy!)

Naturally TSMC knows their tools and processes, works with product companies (nvidia, AMD, Qualcomm) to actually implement their design on the target process node. And at the same time works with the various tool vendors to refine the process, the tools themselves. But as lately we saw it's increasingly impossible to do it "alone", both Intel and Samsung works with many-many vendors. (Intel - among others - funded the push for EUV by buying ASML equity in 2012.)