Intel is still talking up the potential benefits of an EU partnership as it lobbies European nations to support a new leading-edge foundry facility. The company has pledged to spend $20B on a facility or facilities, with a lifetime investment of up to $100 billion in the project over several decades. This is not necessarily unusual — foundries are expensive to build and most companies upgrade them multiple times — but it speaks to Intel’s long-term willingness to invest in a plant.
According to the Financial Times, Intel executives are willing to build multiple facilities in several member states to secure funding and support. “We could put manufacturing on one site and packaging on another,” Greg Slater, Intel vice-president of global regulatory affairs, part of the team exploring possibilities for expansion in Europe, told FT. “We are well placed to make this an ecosystem-wide project, not just a couple of isolated paths in one member state,” he said. “We do believe that this is a project that will benefit Europe at large.”
Intel has explored the idea of building a fab in Germany, the Netherlands, Belgium, and Italy thus far, on a site of roughly 1,000 acres. The company would initially build a pair of fabs on the same site at roughly $20 billion, or $10 billion per fab. The estimated facility lifetime spending is around $100 billion.
There are two significant unanswered questions regarding such a facility: Would it be a leading-edge fab and to what degree would it be subsidized? According to French officials, there’s a 30-40 percent cost gap between building a fab in the US versus in Asia, and much of that difference is due to government support. According to both Intel and the French government, the question of which customers would be likely to use the foundry and where those customers are located geographically would also play a part in the fab’s final location. Semiconductors circle the globe multiple times on their manufacturing journey, but Intel and the EU may be looking to reduce global dependencies and the impact these can have on semiconductor manufacturing.
Most of the semiconductor customers in Europe are companies that are off the leading edge. Intel sounds at least theoretically open to the idea of building a plant to focus on legacy technology or nodes rather than insisting on a leading-edge facility as the only kind of factory worth building. Building foundries focused on the needs of non-leading-edge foundries wouldn’t be as sexy as a leading-edge deployment, but it will probably be easier to bring up the facility and start cranking out wafers. Intel’s current plan for a $20B investment in Europe is separate from its $7B expansion of the Leixlip, Ireland facility, which will eventually transition to 7nm and build leading-edge chips whether Intel pursues larger fabs in mainland Europe or not.
A new facility in Europe with strong European support would help anchor Intel’s Foundry 2.0 plan and position the company as an alternative to TSMC. The Taiwanese foundry currently handles most of the world’s contract manufacturing, but Intel wants to change that in the future. Of the top three manufacturers shown in the diagram above, Intel and Samsung both produce large amounts of their own equipment. TSMC is the only pure-play foundry among the three, and it dominates chip production for every company that doesn’t own its own fabs. The two firms have already tangled with each other over the need for a European foundry in the first place, with TSMC downplaying any such effort and Intel offering the idea of new semiconductor facilities — with some hefty subsidies attached. What sort of subsidies will undoubtedly depend at least in part on whether the final plan calls for a leading-edge EUV semiconductor plant or a trailing-edge facility designed for older nodes and/or planar silicon.