This paper provides an overview of semiconductor front-end production capacity data, building on the OECD semiconductor taxonomy (OECD, 2024[1]) and the ongoing development of the OECD Semiconductor Production Database to help policymakers understand the complex geography of semiconductor front-end manufacturing and inform actions towards more resilient global semiconductor value chains. It contributes to broader efforts to map the global semiconductor value chain (OECD, 2025[2]).
The analysis highlights the importance of considering at least three key characteristics of fabs - node density, process technology and business models - to understand the geographic distribution of current and upcoming semiconductor manufacturing capacity and interdependencies between companies and economies.
Global wafer fabrication capacity is concentrated: five economies, namely China, Chinese Taipei, Korea, Japan, and the United States, together account for nearly 90% of global capacity. Within these economies, capacity is further concentrated in a small number of firms, with the ten largest manufacturers accounting for around half of total global production capacity. Upcoming investments are heavily focused on the same major producing economies, suggesting that existing geographic concentration may persist.
The distribution of capacity also differs significantly by process technology and chip type. For example, advanced logic and commodity memory are dominated by a limited number of highly specialised fabs, while analog, mature logic, and power semiconductors are typically produced in fabs with mixed manufacturing capabilities.
Data also suggest limited substitutability of production between fabs. For example, a fab optimised for analog or power chips cannot produce logic or memory chips. Even within the same process node, different process technologies are not interchangeable. This structural rigidity creates bottlenecks in specific parts of the value chain, which in turn can amplify the impact of supply disruptions and limit the scope for remedial actions. Enhancing the resilience of the global semiconductor value chain therefore requires a shared understanding of where such bottlenecks lie and how interdependencies can be managed.
However, the analysis also reveals that current commercial datasets are not sufficient for answering key policy questions. They do not adequately capture technological detail, they vary in completeness and consistency across regions and they are currently not designed for policy use. This currently limits the ability of policymakers to assess dependencies, evaluate diversification strategies, or monitor long-term trends. Improving the quality, granularity, and reliability of semiconductor manufacturing data should thus be in the collective interest of all economies.
Taken together, the findings suggest several implications for future policy work. Co-operation among like-minded economies will be increasingly important to identify and address structural vulnerabilities along the semiconductor value chain, particularly in areas where production is concentrated. Jointly developing more reliable and policy-relevant data – through shared taxonomies, better industry engagement, and complementary national efforts – would help governments base their decisions on better evidence. Finally, exploring opportunities for greater geographic diversification, could contribute to reducing systemic risks and strengthening the overall resilience of the global semiconductor value chain.
The OECD Secretariat will continue analytical work to support semiconductor policymaking, including, for example, on combining available production capacity data with information on chip demand. This work aims to assess whether risks of excess capacity and unnecessary redundancies can be identified, and to understand which chip types should be prioritised for investments to ensure a more resilient value chain.
Additional work will involve extending the data upstream and downstream from front-end manufacturing. The downstream extension, including back-end manufacturing (assembly, test and packaging), could provide information on how different types of chips depend on a combination of front-end and back-end manufacturing technologies. The upstream extension, including information about semiconductor manufacturing equipment and subsequently chemicals and other material inputs, would potentially provide insights into how different types of front-end manufacturing depend on specific types of manufacturing equipment.