TSMC Tells Europe: Subsidies Get You a Fab, Orders Keep It Alive

A senior TSMC executive warned on September 1 that European semiconductor subsidies, while effective at reducing upfront investment barriers, cannot replace the long-term customer demand that chipmakers ultimately need to keep factories running.

Speaking at the 2026 Invest in EU Forum held at Taipei's Nangang Exhibition Center — co-organized with SEMICON Taiwan 2026 for the first time — TSMC Senior Vice President and Deputy Co-Chief Operating Officer Hou Yung-ching said supply chain partners' primary concern is not incentive packages but whether a market can generate durable, multi-year orders.

"It's not just about building the facility," Hou said. "You have to ask whether the operation can survive, whether it can run sustainably. Subsidies can lower the initial hurdles for construction and investment — but once the facility is built, if there is no demand, they cannot help it sustain in the long run."

His remarks carry particular weight given the scale of semiconductor investment. A single advanced wafer fab can cost tens of billions of dollars, and any manufacturer must be confident that demand will persist not for one quarter but for years. Government money can catalyze a decision to build, but it cannot substitute for customer commitments, capacity utilization, or orders.

The forum panel — moderated by Industrial Technology Research Institute (ITRI) President Chang Pei-fen and featuring representatives from TSMC, Hon Hai Precision (Foxconn), Germany's Infineon Technologies, and European public- and research-sector stakeholders — focused on how Taiwan and Europe can deepen semiconductor cooperation through supply chain resilience.

TSMC's most concrete European commitment is the European Semiconductor Manufacturing Company (ESMC), a joint venture with Bosch, Infineon, and NXP Semiconductors currently under construction in Dresden, Germany. The fab is designed to produce chips on 28/22-nanometer and 16/12-nanometer processes, targeting automotive and industrial applications.

Hou confirmed that the Dresden plant is progressing on schedule and that European automakers have already committed demand. "We are building our first 28/22-nanometer wafer fab in Dresden, and it is on track," he said. "The primary purpose is to support automotive customers. Europe has very strong advantages in the automotive market, and we believe local demand is sufficient to support that fab."

The Dresden investment reflects TSMC's broader logic for overseas expansion: subsidies matter less than a verifiable, sector-specific end market — and Europe's deep-rooted automotive industry provides exactly that foundation.

Hou outlined three hurdles making overseas expansion especially difficult for smaller members of TSMC's supply chain. The first is basic infrastructure: land, electricity, water, and skilled workers. "Not every supply chain partner has our scale, where we can handle everything ourselves," he said, noting that utilities and recruiting often extend project timelines significantly.

The second challenge is regulatory complexity. Taiwanese companies operating abroad tend to follow the same procedures they use at home, but Europe confronts them with overlapping national, regional, and local rules. Many suppliers, Hou said, do not even know which agency to approach first, and end up bouncing between organizations to find the right contact.

Toon van Dijk, a board member of the European Semiconductor Regional Alliance, acknowledged the problem from the European side. "Europe still has different rules across different countries," he said, adding that grid capacity — not electricity generation — is the key bottleneck in the Netherlands, where planning for new grid infrastructure alone can take three to five years before construction begins.

Hou suggested that Taiwan's science park model could offer a useful template — not for its physical infrastructure, but for its administrative design. "Many people think replicating a science park is about copying the physical facilities. That is not it," he said. "The real advantage is that there is a government office inside the park — a single window." In Taiwan's Hsinchu Science Park and similar zones, companies with regulatory questions can reach a single point of contact who coordinates across agencies, rather than navigating multiple bureaucracies on their own.

On the question of whether Europe could attract cutting-edge fabs below seven nanometers — beyond ESMC's current scope — Hou was measured. "We have to go back to the fundamental market question: Where is the demand? Is the demand real? Are customers willing to commit — and not just for one year, but for the long term?"

Forum materials projected that European demand for sub-seven-nanometer wafers could reach 300,000 wafers per month by 2040, equivalent to roughly six fabs. But Hou emphasized that advanced manufacturing depends heavily on economies of scale; a single isolated fab would not achieve optimal cost or efficiency on its own.

To reach that threshold, Europe would need to cultivate local demand from data centers, electric vehicles, and industrial AI applications, while also attracting commitments from non-European customers willing to have their most advanced chips produced on the continent.

Hou called on European policymakers to distinguish between two categories of supply chain investment: factories built primarily for resilience, which need long-term demand to survive, and uniquely European capabilities in equipment, materials, or technology that no other region can easily replicate. The two categories, he argued, warrant different subsidy and investment frameworks.

For TSMC, the Dresden factory has automotive demand as its foundation. Attracting advanced-process investment will require Europe to demonstrate something that government grants alone cannot manufacture: a chip market that is large, real, and built to last. （Related： TSMC Charts Three-Pronged Course to Build Europe's Chip Ecosystem ｜ Latest ）

Original Article In Chinese