Although the light of the European semiconductor industry does not seem to be as dazzling as in the past, the recent succession of measures has revealed the ambition of the EUROPEAN Union to "revitalize".
On February 8, the European Commission passed the €45 billion European Chip Act, which aims to ensure the resilience of the chip supply chain and reduce international dependence, and enhance the EU's influence in the global semiconductor chip supply chain and global value chain.
It is worth noting that the EU not only plans to increase its share of global chip production from the current 10% to 20% by 2030, but also clearly lists the development plan of its advanced process, including the construction of 10nm and below node FD-SOI test line, 2nm below the process node FinFET/GAA test line, 3D heterogeneous advanced packaging test line, etc.
At the level of advanced technology, the EU, which is basically "speechless", is eager to "recover the lost land", and the realization of these lofty goals must "rely on" the support of foundry giants.
However, in addition to Intel's high-profile platform guided by IDM2.0, which announced that it plans to invest 80 billion euros in Europe in the next decade to build or expand production capacity, and establish research and development and design centers in Germany, Ireland, France and Italy, TSMC's factory construction trend since last year's contact with the European Union is still confusing, and Samsung also seems to be "intentional or unintentional" as a wall view, and the dream of advanced technology in Europe still needs to be marked with a big question mark.
Advanced processes in Europe's pain
In the past few decades, the EU has made achievements in various segments of semiconductors, such as Arm in the IP field; ASML is the dominant lithography machine; IDM has NXP, ST, Infineon and other support facades, and has also laid a deep "home foundation" in the automotive and industrial fields.
In addition, Europe also has a complete silicon carbide substrate, epitaxial, device and application industry chain, and is in a leading position in the global wide bandgap power electronics market. In the application field, Continental and Bosch are the largest semiconductor buyers in Europe; Ericsson and Nokia are the "gold owners" of European cable and wireless chip demand; Siemens, Philips, Signify and ABB are active in industrial electronics. In the context of the global semiconductor division of labor, Europe undoubtedly has the right to speak in some key areas of the semiconductor industry chain.
But even if the EU holds the advantages of traditional fields such as power devices, MCUs, and sensors, the "stubborn disease" behind the seeming brilliance needs to be faced more directly.
According to the Boston Consulting Group, the proportion of EU semiconductor manufacturing capacity in the world has dropped from 44% in 1990 to about 10% at present, and local companies are mainly longer than industrial and automotive semiconductors, and their competitiveness in the field of high-end logic and memory is insufficient, and the process nodes are still hovering in mature processes; the world's top ten Fabless lists, in recent years, rare European companies have been shortlisted. Undoubtedly, the lack of advanced manufacturing is already the largest and most obvious "short board" in Europe.
From the historical traceability point of view, on the one hand, the development of European semiconductors is concentrated in the automotive business, and the strength in industrial, medical, defense and other fields is also very strong, but these areas do not involve advanced processes. Over the past 20 years, it can be argued that Europe has effectively stopped investing in advanced manufacturing capabilities and instead weighed the "cost-effective" route of outsourcing its advanced chip production. On the other hand, Europe has helplessly or been forced to miss many outlets such as smart phones, cloud computing, big data, 5G, and automatic driving, and even the "change" of supporting soldiers in the field of high-end chips and advanced processes has also been buried.
Whether it is the lack of production capacity and supply chain shortages that have spread in recent years, or the strong return of manufacturing under the game of great powers, almost making the auto industry in Europe, an old automobile industry, "hurt the bones", the eu's semiconductor industry crisis has become increasingly "obvious".
Naturally, the EU is not willing to be constrained everywhere. On the one hand, expanding production capacity has become an inevitable lesson to be made up. On the other hand, making up for the shortcomings in advanced manufacturing is already a "hill" that the EU's semiconductor attack must cross.
This is not only about the status quo, but also about "resupplying" the future. Europe's high-impact semiconductor demand for automotive, industrial electronics, and wired and wireless infrastructure is increasing demand for mature and advanced chip portfolios. It is expected that in 2024 and the following period, the 5nm process is expected to gradually become popular in the automotive semiconductor market, and the penetration rate of the 7nm and 16nm processes is also expected to increase. It can be seen that the use of advanced process chips in the automotive field will become an inevitable trend. According to an ASML report, Europe's own demand for advanced semiconductors is expected to grow at a rate of five times the demand for other, more mature semiconductors over the next decade. Although automotive electronics has always been a proud field of European semiconductors, if it does not fill the "gap" in the field of advanced manufacturing, Europe or automotive semiconductors, power semiconductors and other fields will lose their inherent advantages.
In the face of this crisis, the chip bill, which has a lot of "ink" on this, in addition to the top-level design from the policy, capital and other levels to "strive for excellence", support NXP, Bosch, ST, Infineon and other local enterprises to accelerate development, promote the cooperation of giants such as TSMC, Samsung and Intel, or can quickly achieve results and the only feasible way.
Foundry giants have their own concerns
For investing in Europe, in addition to intel's high profile and aggressive performance, TSMC and Samsung do not seem to be keen.
Intel said it will bring advanced technology to Europe to build the more balanced and resilient semiconductor supply chain the world needs. As the first phase of the investment plan, Intel will invest 17 billion euros in Magdeburg, Germany, to build two advanced technology plants, planning to start construction in the first half of 2023. If subsidies and necessary funds are in place, official production is expected in 2027.
In addition, Intel's investment plan involves investing 12 billion euros in existing plants in Ireland to expand production capacity, which will bring Intel's total investment in Ireland to more than 30 billion euros. Intel also said it was in talks with Italy on a €4.5 billion investment plan to produce state-of-the-art chips. Its investment plans also include the establishment of an R&D and design center in France, the expansion of laboratories in Poland, and the strengthening of computer laboratory cooperation with local researchers in Spain.
Such a comprehensive full-value chain output and active expansion plan, as Isaiah Research pointed out, Intel's investment plan in Europe will help the EU to reach 20% of the global goal of advanced process production by 2030, plus there should be relevant subsidies between the EU and Intel, so Intel will have such a positive expansion plan. Intel's expected construction of a 20A plant in Germany will also help europe to have advanced process capacity and avoid excessive concentration of advanced process capacity in Asia and North America.
In terms of TSMC, since August 2021 announced that it accepted Germany's invitation to build a factory in Europe, it has not followed, and Samsung seems to have little "interest" in setting up factories in Europe.
Behind the foundry double ambition, pointing out that Europe is not a hard wound suitable for betting on the "treasure land" of advanced technology. Some analysts believe that investing in Advanced European processes is riskier than investing in mature processes, because the advanced manufacturing ecosystem must be built from scratch and take longer to generate a return on investment. In addition, taking into account the political environment, industry support, talent and living expenses of European countries, the comprehensive cost of building factories in Europe is 20% or more higher than in Asia.
In fact, the United States faces the same challenge. As an insider who knows the inside story pointed out, TSMC's Wafer Tek plant in the United States has been established for more than 24 years, but has it made a profit? Differences in culture and nature doom some industries to be unable to be transferred. For Intel's investment, he said bluntly, some investments are doomed to drift, but now it is the moment of Truth, which is very important for political correctness.
Looking at the current influence of geopolitics and the great power game, the order of the semiconductor industry chain shaped over the years is undergoing unpredictable deconstruction.
As jiwei network previously analyzed, the "two-way choice" between Intel and the EU is quite intriguing. Among the three manufacturers who master advanced process technology, TSMC reported last year that negotiations with the EU side had been "not happy and scattered", Samsung has not yet sent out the news of the layout of Europe, Intel's pan-European layout and the EU's generous support to Intel, vaguely outlining more mutual trust between the "transatlantic partners" in the EU semiconductor industry planning, and the East Asian semiconductor industry is intentionally or unintentionally ranked in the alternate position for cooperation.
Returning to the EU's "calculations", it seems that in order to successfully attract top giants and new industry players to the EU, the EU needs to improve its attractiveness, and European policymakers need to provide sufficient incentives and effective implementation.
But I am afraid that what hurts the EU the most is the huge amount of capital needed to achieve the goal. Some analysts say that assuming that the market size of the semiconductor industry will grow to $1 trillion / year by 2030, the associated capital expenditure required to achieve this growth is about $825 billion. In order for Europe to grow its share to 20%, the total investment in Europe must be about $264 billion.
For the EU, perhaps the challenges of time and technology are just beginning.
Investing in Europe "money" is uncertain
If you want to wear a crown, you will bear its weight.
Intel "poured out" to "preemptively" land in Europe, and accordingly had to undertake a series of challenges that followed.
According to the analysis of Tao Zhi (pseudonym), a veteran of the industry, for Intel, although there are government subsidies for capital, but to set up factories in various places, how should the company's account be distributed? Are there so many people who can cope? There are no answers yet.
"If Intel's new factories are mainly based on advanced processes, there will be only a few chip suppliers who can use the first process, and they will still be dominated by Intel, Qualcomm, NVIDIA, AMD, MediaTek or FPGA manufacturers." Under the strong partnership, the first decimal program will still be in the hands of a few semiconductor companies, but Intel is currently outsourcing advanced processes to TSMC, whether the newly established technology can have the strength to undertake large orders from European customers in the future, or it is still Up in the Air (hanging in the air). Tao Zhi said cautiously.
Judging from the response of the related parties of the industrial chain, that is, the relevant materials and equipment manufacturers in the United States, it may also be mixed. Tao Zhi revealed that on the one hand, they are optimistic, because this will directly increase revenue, and believe that the probability of success is higher. On the other hand, it involves risks such as transportation and whether talent training is in place, and the corresponding cost will increase relatively, which is quite distressing.
What deserves more long-term attention is that these factories are expected to be built in three or four years, but will they face the risk of overcapacity after completion? Chen Xiang, senior analyst of Jiwei Consulting, pointed out that Intel's foundry expansion in Europe will increase its share in the European chip market, and the world's chip production capacity demand will tend to be stable, and it will not fluctuate as much as the outbreak of the epidemic in the past two years, and the growth rate of capacity demand will gradually decrease, but the growth of market share will make Intel stand firm in the generation industry.
However, Isaiah Research believes that Intel's expected plants in Europe are mainly advanced processes, such as the expansion of 4nm processes in Ireland and the construction of 20A plants in Germany. Considering the limited customer base of the advanced process at present, if there is no additional customer or a large number of product applications, the advanced process (especially 4/5nm) may indeed face the risk of overcapacity in the future, but the fab will also think about the production line, machine allocation and other measures to deal with it at the same time.
"For Intel, TSMC and Samsung have continuous expansion plans and long-term stable cooperation customers in advanced processes, whether they can grab large customers in the future, especially the orders of American manufacturers such as AMD, Nvidia, and Qualcomm are the focus of market attention, and now there are indeed some customers who negotiate early cooperation with Intel, but they are very early, and there are still uncertainties." Isaiah's research further analyzed.
It will be difficult to change the pattern within three or five years
From the perspective of the industry, behind Intel's enthusiastic "signing" and large bets with the EU, it is not only to reshape the glory through IDM2.0, or it will have an impact on the current foundry pattern.
Chen Xiang believes that Intel has carried out a wide range of foundry expansion in Europe, one is to enhance its own foundry strength and enhance its competitiveness in the foundry industry, and the other is to expand the manufacturing capacity of advanced processes in Europe and fill the gap in the European generation industry. Intel is also planning to establish an OEM R&D center in Europe while expanding its foundry, which will accelerate Intel's R&D speed in advanced processes and help Intel catch up with TSMC and Samsung on the process node. Therefore, Intel's expansion may impact the current foundry pattern and return to the forefront of the generation industry.
Isaiah research pointed out that Intel has recently actively invested in Europe, competing with TSMC and Samsung, and is looking forward to returning to its leading position in semiconductors. In the short term, Intel's expansion in Europe has limited impact on the foundry pattern, because Intel's new plant in Germany has a mass production plan at least 2027, and after mass production, there is indeed an opportunity to compete with large manufacturers such as TSMC and Samsung, but it is still necessary to evaluate the status of Intel's process research and development and mass production yield at that time.
Despite Intel's ambitions, whether it is the radical iterative route of advanced technology or the large-scale expansion in the United States and Europe, the road to surpassing TSMC is probably still very far away.
In this regard, Isaiah research pointed out that TSMC is currently the leader of advanced processes, and its advanced processes below 7nm are iterated every two years, so Intel's goal of upgrading the generation every year can be described as quite positive. In the current situation, Intel's current advanced process technology falls at about 7nm, lagging behind TSMC's two-generation process node, if Intel successfully steps into the 4nm process in the second half of this year, it is still behind TSMC. As for Intel's expected mass production of 20A process in 2024, it will compete head-on with TSMC's 2nm process, and it is believed that Intel's investment and technical talents may have the opportunity to achieve the target, but yield and production capacity are still problems that Intel continues to face.
It is worth being alarmed that in the face of foreign foundry giants often investing hundreds of billions of dollars, carefully carrying out the arms race of technology and production capacity, coupled with the discourse advantages in the supply chain, after the process upgrade and production capacity are cashed, the follow-up development of the mainland generation industry may produce a bottom-up control or squeeze, we need to think in the long run, and we need to make a hole.
(Proofreader/Li Yan)