Talk of building leading-edge foundry capacity has accelerated in Europe, but Taiwan’s Economy Minister thinks the business case for such efforts is weak. “TSMC has said repeatedly that the most advanced technology will definitely be (produced) mainly in Taiwan,” Minister Wang said. “As for how Taiwan and the EU can cooperate, companies have their arrangements and considerations, and it can be further discussed.”
TMSC’s CEO said this week that the company “currently have no further fab expansion plan in other areas such as Europe. But we did not rule out any possibility.” That could be interpreted as a bit of a rebuff to EU officials, who were supposed to sit down to talk about fab expansion plans with both Intel and TSMC later this week.
Does a Leading-Edge Foundry in Europe Make Any Sense?
Europe once held more leading-edge manufacturing than it does today. There’s no single reason for this. GlobalFoundries has several manufacturing plants in Dresden, but these are older factories producing on legacy nodes. Companies such as STMicroelectronics, Infineon, and NXP are still major industry players, but they fab on older nodes or outsource to TSMC for leading-edge capability.
According to a recent report from a German think tank Stiftung Neue Verantwortung (New Responsibility Foundation), the EU’s efforts to build a 2nm leading-edge fab is likely to fail for several reasons. First, electricity prices are typically much higher in Europe than in Asia. Second, there isn’t enough EU-based demand to justify the investment. SNV thinks there’s little chance of US companies signing up to be EU customers given existing relationships with both Samsung and TSMC, as well as Intel’s efforts to ramp up its foundry business.
Image by Stiftung Neue Verantwortung via EENewsEurope. There are no European companies known to be building at 7nm or 5nm. The chart shows the estimated share of 5nm and 7nm wafer starts by customer.
The fact that Europe has just two major fabless chip manufacturers left, with one of them, Dialog, about to be sold to Renesas, undercuts the idea of an independent European market. SNV argues that working with Intel is to be preferred, not because Intel will automatically deliver leading-edge manufacturing, but because the capacity Intel does build will be useful for automotive and industrial applications.
Intel is already installing 7nm equipment at Leixlip, Ireland. That node won’t be leading edge by the time it comes online, but just as Intel’s 10nm is considered a match for TSMC’s 7nm, Intel’s 7nm should be comparable to TSMC 5nm. Intel claims it wants to return to the leading edge and to build a robust foundry business as part of its IDM 2.0 strategy. Europe claims it wants to account for 20 percent of the high-end chip manufacturing business by 2030. Taiwan and TSMC both claim Europe’s plan is a bad idea, and Samsung’s opinion on all of these events is unknown.
It’s easy to see the European Union working with Intel in some capacity to expand x86 manufacturing at its existing plant. What would be more radical would be a deal between Intel and the EU to build new client foundry facilities somewhere in Europe. Intel would be betting that the EU was serious enough about boosting its semiconductor capacity to invest in it, and the EU would be betting Intel could deliver enough customers to fill the foundry.
Intel’s CEO, Pat Gelsinger, has called for the United States and Europe to redevelop their semiconductor supply chains and reduce our dependence on China for chip manufacturing. A major European deal would be a way for Intel to signal how serious it is about IDM 2.0, regaining the leading edge, or both depending on the particulars of the hypothetical agreement.
Now Read:
- TSMC Commits $100 Billion to 3-Year Manufacturing Expansion
- Intel’s CEO Travels to Europe to Talk Foundries as Silicon Nationalism Rises
- Intel in Talks to Start Producing Automotive Chips in Just 6-9 Months
No comments:
Post a Comment