Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/not-enough-demand-in-europe-for-a-dedicated-foundry.14049/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Not Enough Demand in Europe for a Dedicated Foundry?

Daniel Nenni

Admin
Staff member

The lack of semiconductor manufacturing in Europe Policy Brief​

As part of the 2030 Digital Compass decadal plan, the European Commission aims to establish cutting-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. This would require tens of billions of Euros in public and private investment. To make this investment strategically sound in the long-term, such an “EU foundry” must have a solid business case based on substantial demand in the market, especially in the highly competitive market of cutting-edge chip manufacturing which has almost insurmountable barriers to entry. Unfortunately, chasing the 2nm fab is a futile endeavor with a very real risk of wasting billions of Euros in public and private money. This idea lacks a business
case due to the following factors.

First, an EU foundry would predominantly serve European customers, but there are very few semiconductor companies in the EU designing chips on 7nm or 5nm nodes today. Most types of chips that Europe’s leading semiconductor companies produce do not benefit from cutting-edge manufacturing. Thus, companies did not invest in cutting-edge fabs for almost two decades. This lack of cutting-edge chip designs in the EU directly translates into miniscule demand for cutting-edge contract chip manufacturing. Therefore, before investing in supply (creating cutting-edge fabs), Europe needs to create demand by investing substantially in its own chip design capabilities.

Second, it is at best overly optimistic and at worst naïve to hope an EU foundry would attract orders from US chip design companies. It is highly likely that the two companies at the forefront of cutting-edge chip manufacturing, TSMC in Taiwan and Samsung in South Korea, will establish advanced fabs in the United States. Not least because the United States has the largest chip design industry by far. Thus, to alleviate national security and business continuity risks, within the next three to four years US fabless companies will be able to order advanced chips from TSMC’s and Samsung’s US foundries. This outlook further deludes the unique selling point of an EU foundry. What is the rationale for a US fabless company to order chips from TSMC EU rather than from TSMC Taiwan or TSMC US?

Third, it is not by chance that the market for cutting-edge chips manufacturing consolidated substantially over the last 20 years: Skyrocketing investment costs for fabs and the need for extensive R&D collaborations across the entire supply chain to advance the cutting-edge, all while maintaining high utilization rates to amortize equipment costs within a few years, led to most companies dropping out of “More Moore Scaling.” Only two companies operate cutting-edge fabs: Samsung in South Korea and TSMC in Taiwan.

Even if Europe were able to set up an advanced fab through a concerted effort and tens of billions of Euros of private and public money, this money and resources would be better spent in other areas where Europe is even more dependent on foreign technology providers: designing cutting-edge logic chips (such as processors for data centers, high-performance computing, artificial intelligence and mobile applications). Europe’s Achilles heel is the lack of fabless companies that design chips. Once Europe’s chip design prowess is rejuvenated, the region will be in a much stronger position to think about how best to invest in its manufacturing capabilities.

 
Dan, what leads you to believe that "Europe’s chip design prowess will be rejuvenated" ? I've seen Europe's share of worldwide chip design and manufacturing monotonically shrink over the last 30 years. Mobile chips went to US/Korea/China. Wired telecom industry pretty much in ruins - some of it sold out to Huawei - compared to 30 years ago (may well be similar in NA/Canada). Automotive may be at risk from upheaval of EV transition.

Some of this probably goes with a move from regional standards - remember how we used to have different mobile standards, phones and chips for Japan, US and Europe - to global ones.

You're an EDA veteran - you'll know the worldwide EDA sales region breakdown. This is public information:

"Geographically, Synopsys’ revenues in North America (47% of total) were $405.8 million, while that in Europe (10%) was $89.3 million. Revenues from Japan (9%), Korea (12%) and the Asia Pacific (21%) were $80.5 million, $105.3 million and $180.4 million, respectively."

Now, it's not always quite that simple - perhaps licences might get booked through one [HQ] region and used in another in some cases. But this tells the story. That used to be about 1/3 US, 1/3 Japan, 1/3 Europe 30 years ago.

We may or may not like the way things are going here. But something pretty major would have to change to reverse this trend. I'm not seeing what that is right now.
 
According to the latest quarterly ESDA numbers:

EMEA: $457.2 million
Americas: $1,304 million
Japan: $228.2 million
Asia Pacific: $1,042 million

So Europe does have a significant EDA base but certainly not enough to justify many billions of dollars worth of leading edge fabs. So I agree with the German thinktank (above). Focus on fabless companies then fabs. Do not build fabs and expect fabless business to come like the Field of Dreams movie because that is just a movie.



Dan, what leads you to believe that "Europe’s chip design prowess will be rejuvenated" ? I've seen Europe's share of worldwide chip design and manufacturing monotonically shrink over the last 30 years. Mobile chips went to US/Korea/China. Wired telecom industry pretty much in ruins - some of it sold out to Huawei - compared to 30 years ago (may well be similar in NA/Canada). Automotive may be at risk from upheaval of EV transition.

Some of this probably goes with a move from regional standards - remember how we used to have different mobile standards, phones and chips for Japan, US and Europe - to global ones.

You're an EDA veteran - you'll know the worldwide EDA sales region breakdown. This is public information:

"Geographically, Synopsys’ revenues in North America (47% of total) were $405.8 million, while that in Europe (10%) was $89.3 million. Revenues from Japan (9%), Korea (12%) and the Asia Pacific (21%) were $80.5 million, $105.3 million and $180.4 million, respectively."

Now, it's not always quite that simple - perhaps licences might get booked through one [HQ] region and used in another in some cases. But this tells the story. That used to be about 1/3 US, 1/3 Japan, 1/3 Europe 30 years ago.

We may or may not like the way things are going here. But something pretty major would have to change to reverse this trend. I'm not seeing what that is right now.
 
Dan, I'm not disagreeing about the fabs. Simply that critical mass in fabless to ever think about fabs (which is what your original comment implied for me) is a long, long road from here.

As the numbers in your more recent "China has only 5% of global market share" posting shows. Even if the Europe fabless share number of 1% is artificially suppressed as much of the European fabless design work is for US - and to a much lesser extent - Japanese and Korean companies, the effective share of worldwide fabless design is very, very low. IP is a slightly different story.
 
Thanks for sharing my analysis here, @Daniel Nenni -- great to see that I'm not the only one who's skeptical about the EU's plans.

Also, thanks for the idea to look at EDA revenue regions, probably a decent proxy for chip design efforts per region. :)

Cheers.
JP
 
Dan, I'm not disagreeing about the fabs. Simply that critical mass in fabless to ever think about fabs (which is what your original comment implied for me) is a long, long road from here.

As the numbers in your more recent "China has only 5% of global market share" posting shows. Even if the Europe fabless share number of 1% is artificially suppressed as much of the European fabless design work is for US - and to a much lesser extent - Japanese and Korean companies, the effective share of worldwide fabless design is very, very low. IP is a slightly different story.

China has only %5 of the semi market share, but is the 50%+ of the market itself. That's a way bigger issue if one thinks a bit.

The moment a domestic contender gets in the niche, expect foreign semi to get a boot.

This is already happening with Chinese automotive use chips under the guise of new car safety regulations.
 
@Paul2 , "This is already happening with Chinese automotive use chips under the guise of new car safety regulations." -- any sources / links for that? Sounds interesting.
 
As this "techno-nationalism" trend continues, my view is that market access, and therefore a region's Consumption TAM, becomes a more important lever. So the discussion will move away from the optimization of the global integrated supply chain, or whether a geo has enough demand from design company HQs in their region to justify fabs in a given region. Instead, watch for the (re)emergence of tariff and non-tariff trade barriers. Countries or regional blocks will say - if you want access to our TAM, more of the value-add (design and manufacturing) needs to be done in our country/region. Those firms in the semi supply chain that move value-add into the region will benefit from exemptions to the trade policies. These policies might well be deployed in concert with the more classical financial incentives seen today in China (Big Fund et al), US (CHIPS Act) and corollary legislation in the EU and India. For the economic carrot/stick to work, it needs to be more expensive for high volume producers to import than to design and/or make in the region.

This model will only work for China, US, EU and perhaps Japan (or a South East Asia regional block) and in time, India. Is this a desired or optimal model? No. But countries and regions may not be able to resist the temptation as a lever to "incent" a morphing of today's highly concentrated supply chain into a more globally diversified model, one that increases the perception of supply chain security for national security and critical infrastructure applications. And they will use the national security exemptions to WTO and international trade law to justify the approach. Solely deploying financial incentives will likely be insufficient to meet the goals of these countries/regions.

Readers of this blog well understand the tremendous inefficiencies inherent in diversifying supply geographically, and in the risks associated with global overcapacity by redundant, regional investments. Thus for the EU, the Quad (US, Japan, Australia, India), South Korea, Taiwan and Singapore, there should be a well constructed alliance strategy to optimize the use of public funds while improving supply chain security objectives, including that of diversifying the existing model. China's own actions with respect to threats to the sovereignty of nations in the South and East China Seas, human rights, debt diplomacy, compliance with obligations taken under international agreements, Intellectual Property and other issues will determine whether it drives a further de-coupling from the global supply chain, or provides reasons for re-integration.
 
I believe a Quad+1 of semiconductor industry alliance is forming:

( US + Japan + Taiwan + South Korea) + Netherlands.

The Quad of Semiconductor share many common national interests and security concerns.
 
Back
Top