Semiconductor manufacturers acquire expertise through overseas suppliers, engineers and investment relationships, according to October research that questions whether building domestic factories alone can secure technological leadership.
The working paper, by Pinelopi Koujianou Goldberg, Réka Juhász, Giulia Lo Forte and Jeff Thurk, combines industry history, a manufacturing model and evidence on international knowledge flows. It argues that access to foreign expertise often depends on deliberate commercial relationships.
For businesses planning new chip factories, the analysis puts supplier cooperation, staff mobility and access to established manufacturing networks alongside buildings and machinery. A production site can be located within a country’s borders while the knowledge needed to operate it efficiently remains distributed internationally.
The paper was also released by the Centre for Economic Policy Research on October 2. It is research circulated for discussion, not a peer-reviewed verdict on a particular country’s semiconductor policy.
Manufacturing experience travels through business relationships
Chip fabrication involves repeatedly depositing, removing and patterning materials on semiconductor wafers. A foundry is a manufacturer that produces chips for other companies, including businesses that design chips without owning fabrication plants.
Owning the machinery does not settle how efficiently a factory will produce working chips. Engineers must diagnose defects, adjust processes and match equipment to the conditions on a particular production line.
One measure of that performance is yield, the share of chips produced on a wafer that work as intended. Better yields spread manufacturing costs across more saleable products.
The researchers describe equipment suppliers and manufacturers as holding complementary information. A supplier may understand a machine’s controls, while its customer understands problems emerging inside the factory. Working together can produce improvements that neither business could make as readily alone.
Suppliers can then carry some of what they learn into later customer relationships. Engineers moving between international operations and foreign investment can transmit expertise through other routes. These are organized exchanges involving companies and people, rather than knowledge becoming freely available simply because production has begun nearby.
The evidence comes from historical production records
The model uses 17,089 quarterly survey responses collected by the Global Semiconductor Alliance between 2004 and 2019. These are responses about wafer purchases, not a count of factories or individual companies.
The records include prices, quantities, manufacturing locations and wafer characteristics. They cover selected integrated circuits and exclude memory chips. The authors examine experience accumulated in a particular technology, experience with other technologies and experience generated by producers elsewhere.
Their estimates assign a substantial contribution to learning across borders, particularly for countries entering a technology after others have already accumulated production experience.
Historical records cannot directly measure the performance of a factory opening in 2026. The statistical model also requires assumptions about costs, prices and learning. Its estimates should not become a guaranteed return from hiring foreign engineers or signing a supplier agreement.
Factory investment and access to knowledge can pull in different directions
The authors argue that policies encouraging domestic production are more likely to develop local capabilities when companies retain access to international sources of advanced knowledge.
They also explain why restrictions on those relationships can impede technological catch-up. Governments pursuing national security or supply-chain resilience may accept that cost. The research examines the trade-off without deciding that commercial efficiency should override those objectives.
For a manufacturer, this affects the assessment of a new location. Equipment installation, technical assistance, training arrangements and the ability to work with customers and suppliers can influence the time needed to reach dependable production.
Our earlier coverage of the costs of relocating electronics production examined the expense of recreating supplier networks. The new paper adds a related constraint: those networks can carry production knowledge as well as physical components.
A domestic factory can reduce reliance on imported finished chips while continuing to depend on foreign equipment, expertise and cooperation. The location of the building and the location of the knowledge remain separate investment questions.
