The EU Chips Act, a new-generation industrial policy tool for the European Union, is producing appreciable results, but not those expected by its early advocates, writes Mathieu Duchâtel.
Mathieu Duchâtel is Director of International Studies at the French think tank Institut Montaigne.
For two years, there were doubts as to whether TSMC, the Taiwanese semiconductor giant, intended to build a fab in Germany. Very high production costs, managerial cultures said to be incompatible, foreseeable difficulties in recruiting engineers, as elsewhere in Europe or the United States, were reasons behind the blur.
The announcement on Tuesday (8 August) of a €10 billion investment in a foundry in Dresden (for which TSMC’s board had approved a 3.5 billion euro investment), thought to be subsidized up to 50% by the German government in a joint venture with three major European players in the sector – Bosch, Infineon and NXP – has put an end to this speculation.
Not only is this excellent news for Saxony’s semiconductor cluster: TSMC’s decision will also benefit the European semiconductor ecosystem and the stability of Europe’s automotive supply chain, if it is – very likely – approved by the European Commission.
Overly ambitious objectives
The European Chips Act is producing results, except that they differ from the overly ambitious objectives that Thierry Breton had originally set for it: 20% of the world semiconductor market by 2030, compared with less than 10% today; production on European soil of 2-nanometer generation chips, whereas the most advanced foundry in Europe by 2023 is Intel’s Leixlip foundry in Ireland, producing 14-nanometer chips; and a quest for “technological sovereignty”.
The goal of doubling Europe’s global market share is unlikely to be achieved. The EU Chips Act authorises exemptions to the prohibition of state aid if companies and governments demonstrate that a fab project is “first-of-a-kind”, a loose concept that has considerably relaxed the EU’s notoriously strict competition law.
Since the beginning of the year, the Commission has approved a €292.5 million subsidy from the Italian government for the construction of an STMicroelectronics silicon carbide substrate manufacturing plant in Sicily, and €2.9 billion in state aid from the French government for STMicroelectronics and Global Foundries. The French plan will see the construction at Crolles of a new site specializing in FD-SOI, a power electronics technology essential for reducing the energy consumption of integrated circuits.
According to the French Economy Ministry, the STMicroelectronics/Global Foundries fab will increase overall European semiconductor production by 6%.
What’s next? In a good scenario, the Chips Act will enable three additional investments, all in Germany, where the federal government has earmarked a €20 billion budget for the semiconductor industry.
In addition to TSMC, Infineon and Intel could benefit from this in their expansion plans in the east of the country. This budget is far from negligible, but by comparison, TSMC plans to invest three times as much in 3- and 5-nanometer production in southern Taiwan alone.
Quick maths tell a blunt story: without the EU Chips Act, these investments would have been unlikely to take place in Europe, but they won’t be enough to increase Europe’s market share vis-à-vis Taiwan. Trends in semiconductor equipment spending tend to show that Taiwan’s and South Korea’s advance is in fact still increasing.
Revival of G7 countries’ industrial strategies
What’s more, TSMC is not setting up in Europe to produce the “2-nanometer generation technologies” that Thierry Breton called for, and which the Taiwanese company is the only one in the world to manufacture today.
In Dresden, TSMC is meeting the demand from the automotive sector for mature technologies, on a 28/22 nanometers process technology. The announcement of TSMC’s introduction in Germany of its 16/12 nm Finfet process technology also responds to the developing needs of “fast-growing automotive and industrial sectors”.
Between the political vision of 2 nanometers and the commercial viability of such an enterprise, which could, in theory, be reached through compensatory but risky state aid, there is the entire TSMC strategy, and in the background, that of the Taiwanese state.
TSMC is surfing on the revival of industrial policy in the G7 economies, simultaneously building fabs in the USA, Germany and Japan, where state aid packages reduce the production cost gaps with Taiwan.
One would almost forget that, until 2022, TSMC had never ventured outside the Chinese-speaking world. Its only factory outside Taiwan was in Nanjing, Mainland China.
This new strategic shift will not, however, change the fact that TSMC is deeply rooted in Taiwan’s unique industrial ecosystem. The company intends to keep the majority of its production volume in Taiwan.
This already was Taiwan’s choice in the early 2000s, to respond to Chinese attempts to absorb its semiconductor industry. When 4-nanometer production begins at the first TSMC plant in the USA in 2026, TSMC will be mass-producing 2-nanometers at the Hsinchu Science Park.
Aim for industrial ‘cluster effects’
Finally, the term “technological sovereignty” is misleading. “Economic security” would be more appropriate to describe the modest revival of industrial policy in Europe.
TSMC’s heavily subsidised fab investment in a sector particularly vulnerable to geopolitical risk and natural disasters will contribute to Europe’s economic security if it is approved by the European Commission.
Unlike in Arizona, TSMC will not be bringing its network of Taiwanese suppliers to Germany. The profit margin for mature nodes needed by the automotive industry today is insufficient for them to see Europe as an attractive investment proposition. TSMC will have to rebuild a network of European suppliers – hence the joint venture with key European companies.
As a result, the Chips Act contributes, albeit modestly, to a dynamic of Europeanisation of semiconductor supply chains that benefits Europe’s resilience. This will be all the more the case if Intel and Infineon’s plans are successful.
If they do go ahead, Germany will be the European country to benefit most from the Chips Act. In this case, the Commission will have authorised precisely what European competition law sought to prevent by prohibiting industrial subsidies: concentration in countries with sufficient budgetary leeway to support large-scale projects.
If the aim is to relaunch the semiconductor industry in Europe however, the cluster effect is far more realistic than the sprinkling of subsidies across the whole European Union.
The inevitable concentration in Germany must, however, be remedied by two corrections. Firstly, public policies need to cultivate the other European clusters around Eindhoven, Leuven and Grenoble. Secondly, the EU and member states need to respond as quickly as possible to the talent crisis, a threat to European industry that is more imminent than any other, which public policies have so far failed to address, and against which decisive action could benefit the whole of Europe.