Loading summary
Transcript1 lines
- [00:02]
A
People have been so busy with AI data centers and surging memory prices that they forgot about the EU Chips Act. And to be honest, I did too. Until I traveled to Antwerp for ITF World 2026 and someone at the media session mentioned that a Chips Act 2 is coming. But second breakfast so soon? In December 2025, the EU Court of Auditors released its special report on the first Chips Act. A quick read of this and other recent events says that Europe's current chip policy must change. It is not working. In today's video, A brief check in on the EU Chips Act in their December 2025 report, the Court of Auditors notes that the EU Chips act is a continuation of what is called the 2013 strategy. First proposed in May 2013. The strategy's longer name is the New European Industrial Strategy for Electronics. It was also branded as 10120. That stands for 10 billion euros of state funds to be mobilized for research and innovation, which is then to spur 100 billion spent by the industry into manufacturing. That 20 stands for the goal of the 2013 to double the economic value of EU semiconductor production from 10% to 20% by 2020. The effort was led by former European Commission Vice President Nayeli Kruse. She said at the time that she wanted Europe to produce more chips domestically than the US did an Airbus of chips. The EU left the implementation details down to a group of industry leaders, but there it failed. Several of the leaders privately said that this metric was not the right one to target, suggesting to use system cells rather than chip cells. The three European IDMs, STMicro, Infineon and NXP eventually turned against the concept too. 10 billion euro is a lot for a Deere, but insufficient to justify a semiconductor fab, let alone for multiple such fabs. The plan for going from 10 billion Euro of subsidies to 100 billion of overall investment was never figured out since the IDMs were the ones building and running the fabs. That sort of nixed the effort from the start. The industry simply passively made the motions until the issue fell off the radar. Then came Covid. When the world shut down, it triggered a global chip shortage that disrupted supply chains everywhere. Car factories reportedly shut down. Production reached 1975 levels go, causing used car prices to surge just like the United States. The European Commission suddenly got semiconductor pilled. And after the Americans proposed the US CHIPS act, the EU cobbled together EU Chips act under the feeling that it was an emergency. They had to do something fair. The problem was building the EU Chips act on top of the failed 2013 strategy, an effort that targeted a big, flashy and ambitious number like Europe's falling semiconductor market share, and then failed to achieve it. The Court of Auditors noted in their December report that perhaps in their haste, the EC did not do a full retrospective of the 2013 strategy's impact. They also did not do a full impact assessment or public consultation of the EU CHIPS act as it was being drafted. Had they done so, they might have better perceived that the EU CHIPS Act's tweaks and reorganizations did not properly address the issues surrounding the original strategy. The EU blames the COVID emergency for that, but this should have been done already. One thing I like about the EU CHIPS act is that it has this new pillar system with three nicely organized R&D, manufacturing and crisis response. Pillar one, the R& D pillar involves retaining Europe's lead in technological leadership and research. I've seen the phrase chips for Europe in 2013. The strategy focused on Europe's existing strengths. Pillar one, on the other hand, kind of looks ahead at new and developing technologies. Pillar two is the manufacturing pillar or security of supply. It's a framework for attracting new FAB investment to Europe. And finally there is Pillar three monitoring and crisis response. This is a new addition and sets up a coordination mechanism so so that the EU member countries can better foresee and handle chip shortages. Let us start with Pillar one. The Court of Auditors notes that it is in this pillar that they have seen the most progress. The EU CHIPS act set up what are called advanced pilot lines across Europe's research institutes, focusing on interesting technology regions. The Nano IC pilot line run by IMEC in Belgium for beyond 2nm stuff the Fames pilot line run by CA Ledi in France for RF, FDSOI embedded memories and so on. The Apex pilot line run by Franhofer in Germany for advanced packaging, the WBG pilot line for wide band gap stuff and PIX Europe for photonic integrated circuits. Some 20 institutions in 11 countries are involved in this one. IMEC's pilot line is called Nanoic and it focused on beyond 2 nanometer projects. It is noted to be the largest of the pilot lines and seems to have produced well on ITF World's second day, Imec researchers extensively discussed their Nano IC projects. There were several interesting technical presentations on Imec's recent Haina EUV machine as well as their capacitorless 3D DRAM project. The former is a nice step and the latter a possible breakthrough. Great advertisement for more chipsack funding, but research projects do not fruit within the next five or 10 years, if at all. This has no bearing on deterring another chip crisis. Moreover, while I appreciate that Europe creates and nurtures such ideas, I am also aware that many of those ideas and their entrepreneurs then go to the US or Asia to scale up. This is a flaw of Europe's venture capital and funding system and needs to be looked at. Regarding this, Perhaps the European Tech champions initiative, a 20 billion euro fund of funds idea, can help bring change. But let's see. One of the 2013 strategy's shortcomings had been insufficient attention paid to capacity. This tries to fill that gap. Now here we have a critical thing to mention. Articles in 2022 said that the EU CHIPS act will mobilize more than 43 billion euros of public and private funds, and I think that gave the impression that it was the same as the US Chips act, which actually had 50 billion or whatever of money disbursed from the federal government. But the EU CHIPS act did not pull all the 43 billion euros out of the EU budget. The amount of actual funds from Brussels was just about 3.3 billion euros and unfortunately this was not new euros but rather reallocated funds from pre existing pots. The reason for the 43 billion euro number is how the CHIPS act loosened state aid rules for semiconductor FAB funding. I didn't know this was a thing until recently, but the EU member states normally cannot subsidize private companies willy nilly. The EU CHIPS act thus allowed richer countries like Germany to hand over euros to companies like intel for their FAB in Magdeburg and TSMC's Fab in Dresden. Those were the flagship projects within this pillar. But the EU's other member states have funded a few other projects. Italy issued 292 million euro grant for an STMicro silicon carbide plant. That's pretty nice. And in 2023 the French issued 2.9 billion euros of grants for a global foundry's ST microfactory in Kroll. This one does fully depleted silicon on insulator technology. Italy proposed and had approved by the EU 1.3 billion euro state aid for advanced packaging and testing fabric to be built by the Singaporean startup Silicon Box. The third pillar is a mechanism for the various member states and the European Commission to monitor and respond to potential shortages. It is to create a toolbox of crisis measures that the EU can use to defuse issues. I think a few things mentioned have been common purchasing and priority orders as well as a supply chain monitoring scheme. To be honest, this one is so vague that I'm not even sure what counts as a failure. The Court of Auditors pointed out that there's no timetable, no deliverable the 2025 update on the EU Chips act says that there has been some information provided for some future analysis of Europe's position in the global supply chain. It doesn't sound like a lot has been done here. We talked earlier about how controversial the market share number target was. In 2022 the EU had less than 10%. More like 9% market share of the world's semiconductor market has measured by total revenue. The goal again was to double that to 20% by 2030. It looks like they're not going to make that goal. The EU Court of Auditors projects the 2030 market share number to be 11.7%. I think even that is charitable considering how badly digital logic is blowing up in the United States and Asia. A presentation at ITF World 2026 by Maria Marsed, who served as TSMC Europe's president for 16 years, had a slide showing that 9% falling to 6% for 2026. Now fabs take time to build TSMC Dresden coming online next year projected is going to generate some production, but I don't see how one phase of a relatively older FAB returns Europe to even 10%, let alone 20%. And sadly, intel cancelled their Magdeburg FAB plans as well as that for a packaging plant in Poland. I think this had a major effect on the whole pillar. In a few years Europe might start seeing the days of 10% global market share as the good old days. But you see, this is why the EU Chips act is such a failure. The 20% market share target made little sense in 2013 and it made little sense in 2023. Why were they targeting this again? It is a goal that both cannot be reached and discounts Europe's existing strengths like for instance analog and mature node chips. These sell well and are important as we will discuss later, but their prices do not hold a candle to a 2 nanometer wafer and Europe is still good in semiconductor manufacturing equipment. There is ASML of course, but also Bessi, ASM International, EV Group and aixtron. Such companies are benefiting from the AI boom too. An alternate measuring stick might be to look at EU27 semiconductor value chain revenues. One estimate by IDC in 2024 found that EU value chain revenue share surged to 11% in 2022 and is projected to stay at that proportion until 2030. The Court of Auditors report concluded that the 20% target by 2030 was too ambitious from the start and a reality check is badly needed. I dare to go a little further feeling that that target was not even the right metric to target. So three years down the drain Tilting at silicon windmills In a 2024 interview, Thierry Breton, who drove the CHIPS effort, gave a number about its impact. He says if we now have 67 semiconductor plant projects in Europe where before there were none, it's thanks to the efforts made with the European CHIPS Act. End quote. I was not able to track down the 67 fab projects under construction now and enabled by the EU Chips Act. The 2025 update notes seven state aid decisions and today there are about 11. Rather, 67 sounds a lot like the number of important projects of common European interest relating to microelectronics, which is about 68, but covers 5G, 6G, autonomous driving, AI and quantum computing research, not capacity or production. By the way, new FAB projects were getting approved before the EU CHIPS Act. Saying that there were none is wrong. Bosch and Infineon literally just opened new fabs in Austria and Germany in 2021. Generally speaking, the most positive thing oft said about the CHIPS act is that it helped bring semiconductors to the public forefront. Policymakers now know that chips are worth caring about and can use it to help set the agenda. I'm sure this is valuable, especially after the 2013 debacle, but after 3.3 billion Euro in four years of political capital, I feel like Europe should have gotten more awareness. Wasn't the deliverable back in 2022 when this was first proposed? And didn't all the fuss surrounding the first Covid chip shortage do enough of that? If this is what I'm hearing proponents say about the EU CHIPS act, then they are damning it with faint praise. By the way, I also want to point out that attention without education is pretty useless and likely self defeating. I know I am talking about the Internet and the Internet is all fake people, but I see more wrong semiconductor commentary online now than ever before. The recent Naxperia dumpster fire in the Netherlands has been the latest and most profound teaching example of the EU CHIPS Act's failure. Someone is going to cover this in way better detail than what I can do in a few minutes. But here is the fast rundown. Nexperia used to be part of Philips and today they make billions of boring mature node chips for mostly the car industry. These chips literally cost cents each, but because automotive companies are conservative, it can take years to redesign and re rate systems for New chips. Thus Nyxperia's chips are essentially irreplaceable despite running nodes that are decades old. In late 2017, before the chip crisis, Nyxperia was sold to a state owned Chinese company for $2.75 billion. Nobody cared at the time, including the US watchdog Cepheus. In my opinion, someone messed up there. Little changed for a few years. Then in 2019 control was handed over to another Chinese company called We Wingtech. Wingtech was run by a guy named zhang Shizang or Mr. Wing. In late 2025, the Dutch government then moved to take direct control of Nakesperia, citing concerns about IP theft and tech transfers to China. This was reported to be at the behest of the urgings of the American government during the Sino American trade war. And that is an angle. But I must also note that Mr. Wing has a troublesome past, having been convicted in China China, let me emphasize, for stealing trade secrets from ZTE. Mr. Wing also raised money to build fabs in China and saw Nysperia as a stepping stone to get there. There are also on the record concerns from Nexperia employees, ex employees about Mr. Wing's intentions. This isn't just smoke. Even so, the way the Dutch did it was unprecedented and it basically killed Wingtech's business. They and the mainland Chinese government responded ferociously. Naxperia relied on Chinese labor to package 70% of their chips and that turned out to be a significant weakness. In October 2025, the government banned entry of Nuxteria's staff as well as their Fab chips into China. This immediately cut off Nuxperia's customers from supplies of mature node chips, triggering another Covid like crisis for automotive customers. The US and Dutch scrambled and in the end announced an uneasy truce in where America suspends export restrictions for one year and China allows imports for packaging on a case by case basis. Nexperia is now looking for new packaging sources. Rumor is that they're expanding their secondary packaging fab in Malaysia. They also recently signed a manufacturing deal with an American semiconductor company called Polar to produce power MOSFETs. This event also severed Nyxperia into European and Chinese habs that are not getting along right now. Reminds me of the ARM China debacle. Wingtech sued Nuxt for over a billion dollars claiming that they're being bullied right now. This whole gong show is still ongoing, so more developments are to be expected as we wait. Please read this NRC piece written by friend of the channel Mark Hayink back When it happened in late 2025, the Naxperia thing hardly made the news in the United States. I think everyone was too busy rubbing themselves all over about the latest Claude four point whatever. Naxperia would have happened even if Europe had 20% global market share. And I think it demonstrates all all the things that the first EU Chips act got wrong regarding actual resiliency, like an excessive focus on leading edge nodes without investment into trailing edge or mature nodes. Naxperia had nothing to do with 2 nanometer or even 5 the shouting out of big beautiful fabs while ignoring the equally important role of packaging. The silicon box project aside and those guys are advanced packaging. Outside of packaging there remain other landmines like photoresist, EDA tools and chemicals. Another one not working closely with big customers to figure out their chip weaknesses and letting them sleepwalk into crisis. At some point these carmakers and others need either larger inventories or and I know this is easier said than done, they need to design for more chip interchangeability and the talk of chip sovereignty without doing or thinking too much about who actually makes the decisions. The Intel Magdeburg fab fiasco also applies here. The EU staked 1/2 of their chips bet on a company with a then uncertain future with only tentative ties to Europe. However, to quote not friend of the Channel Ray Dalio, pain plus reflection equals progress. The EEC should properly reflect on this next period mess when writing up EU Chips Act 2.0. So what should be done for 2.0? Well it sort of depends on how creative you want to get. Perhaps one interesting and risk tested concept is one modeled in what Japan did. The Japanese strategy had dual tracks. On one end there is the TSMC led joint venture and Kumamoto Kyushu. This is the big flashy joint venture that started out doing just image sensors but has now graduated to making 3 nanometer wafers. It is probably going to be very successful. And then on the other end of Japan Hokkaido we have the riskier more niche option Rapidis Japan's advanced node higher speed foundry. JSM and rapidus target different niches in the market and share similar investors. The book on Rapidus remains out but so far it looks ok. The EU already has half of that with esmc. Perhaps they can gather together and really create the Airbus of chips this time a smaller fast response and fully integrated foundry somewhere. Big question to figure out where with technology owned and licensed perhaps from imic. A key point is to ensure that there is equity and working participation from all of Europe's system companies from automotive to mobile to networking to whatever arm. 2 the big constraint is of course budget a leading edge FAB costs a lot. The Japanese government is estimated to have put billions into rapidus first two fabs. Exactly how much in what form is not easy to discern. And the EU budget is only so much. Perhaps the Dutch and Germans can fund it together and the FAB can be half in Germany and half in the Netherlands. So in conclusion, I don't know what the EU is going to do in the end. Let us see what they're going to propose, but what they did before is not going to work again. With regards to this, I think AI is going to be a distraction. Like I said, I was recently at ITF World in Antwerp, my first time back in two years, and I admit I was a bit struck by the change in mood from the last show back in 2024. The major topics discussed at the time involved efficiency and sustainability. I think there was also some discussion about supply chain resilience. This year nothing mattered more than AI. These presentations are sort of amusing in that they all start off by saying that AI is the new technology that changes everything, defines today's era, and so on. It was like a chant. There is a risk that state aid gravitates to the new and sexy. Like for instance the possibility. Like the French AI lab Mistral makes her own AI chips. They will probably want to throw money at that, but I think that will be a mistake. Let the hot money chase the AI. If the EU wants to avert more crises like Nyxperia, then it needs to get to shoring up the older boring stuff. Then they can think about moving on from there. Alright everyone, that's it for tonight. Thanks for watching. Subscribe to the Channel, Sign up for the Patreon and I'll see you guys next time.