Executive Summary
The global semiconductor industry has become the arena for the most consequential industrial policy competition since the Cold War. Since 2022, governments worldwide have committed over $380 billion in direct subsidies, tax incentives, and infrastructure investments to attract and retain chip manufacturing capacity. The United States' CHIPS and Science Act allocates $52.7 billion; the EU Chips Act mobilises €43 billion; Japan's semiconductor strategy commits ¥3.9 trillion ($26 billion); South Korea's K-Chips Act provides ₩340 trillion ($260 billion) in tax incentives over ten years; and China's Big Fund Phase III adds $47 billion on top of existing programmes.
This paper analyses the semiconductor subsidy race through the lens of strategic trade theory and subsidy competition games. We demonstrate that the competitive dynamics exhibit characteristics of a multi-player Prisoner's Dilemma: each nation's subsidy programme is individually rational given competitors' actions, but the collective outcome — massive fiscal expenditure, redundant capacity investment, and distorted global allocation — is Pareto-dominated by a cooperative equilibrium. The Semiconductor Industry Association estimates that planned fabrication capacity, if fully realised, would exceed projected global demand by 30–40% by 2030, implying that a significant portion of subsidised investment will yield below-market returns — a classic outcome of subsidy races.
The Strategic Landscape: Why Semiconductors Are Different
Semiconductors occupy a unique position in the global economy that justifies heightened policy attention, if not necessarily the specific subsidy mechanisms currently deployed. The industry exhibits several characteristics that distinguish it from ordinary manufactured goods:
Extreme learning curves and scale economies. Semiconductor fabrication is the most capital-intensive manufacturing process in history. A leading-edge fabrication facility (fab) costs $20–28 billion and takes 3–5 years to construct. Once operational, unit costs decline steeply with cumulative production volume — the learning curve effect — creating powerful first-mover advantages and natural barriers to entry. The Boston Consulting Group (BCG) estimates that production costs for a mature fab are 40–60% lower than for a newly commissioned one, implying that early entrants in each technology generation enjoy lasting cost advantages.
Extreme geographic concentration. Despite the industry's global importance — semiconductors are embedded in virtually every electronic product, from smartphones to medical devices to military systems — fabrication is extraordinarily concentrated. Taiwan's TSMC alone manufactures approximately 90% of the world's most advanced chips (sub-7nm). South Korea's Samsung accounts for much of the remainder. This concentration creates what the US Department of Commerce has termed a "single point of failure" vulnerability: a conflict in the Taiwan Strait, a major earthquake, or a severe drought affecting Taiwan's water-intensive fabs could disrupt global electronics supply chains for years.
Dual-use characteristics. Advanced semiconductors are critical components of military systems, intelligence capabilities, and strategic technologies including AI. This dual-use nature transforms semiconductor policy from an economic issue into a national security imperative, justifying the involvement of defence and intelligence establishments in what would otherwise be commercial policy.
The Subsidy Race as a Multi-Player Prisoner's Dilemma
The Brander-Spencer model of strategic trade policy provides the theoretical foundation for understanding semiconductor subsidies. In their seminal 1985 paper, Brander and Spencer demonstrated that in industries with significant economic rents (profits above the competitive level), a government subsidy can shift those rents from foreign to domestic firms — a "profit-shifting" motive for industrial policy.
However, the Brander-Spencer model's policy implications depend critically on the assumption that only one government subsidises. When multiple governments simultaneously subsidise — as in the current semiconductor landscape — the game transforms into a Prisoner's Dilemma. Each government's subsidy erodes the rent-shifting gains of others, while the collective fiscal cost escalates.
Consider a simplified three-player game between the US, EU, and China. Each chooses between Subsidise (S) and No Subsidy (N). If only one player subsidises, that player captures rent-shifting gains at low cost. If all subsidise, the gains are largely offset by competitors' subsidies, but withdrawal is costly because it cedes strategic advantage. The dominant strategy for each player is Subsidise — regardless of others' choices — producing the Nash equilibrium {S, S, S}, which is Pareto-dominated by {N, N, N} (or, more realistically, by a coordinated agreement with modest, targeted subsidies).
The empirical data confirms this dynamic. Semiconductor subsidies have escalated in clear reactive patterns: the US CHIPS Act (2022) prompted the EU Chips Act (2023), which accelerated Japan's and South Korea's programmes, which reinforced China's Big Fund III. Each round of subsidies triggers competitive responses, ratcheting total fiscal commitments upward in a process economists term "subsidy escalation."
The Overcapacity Problem: Lessons from Solar and Steel
History offers cautionary precedents for subsidy-driven capacity expansion. The global solar panel industry experienced a subsidy-driven boom in the 2000s and 2010s, with governments offering generous production subsidies and guaranteed purchase prices. The result was massive overcapacity: by 2012, global solar panel production capacity was approximately double demand, triggering a price collapse that bankrupted many subsidised manufacturers and wiped out much of the public investment.
The steel industry presents an even more direct parallel. Chinese government subsidies drove a tripling of Chinese steel production capacity between 2000 and 2015, creating global overcapacity that depressed prices, triggered trade conflicts, and resulted in massive losses for unsubsidised producers worldwide. The OECD estimates that global steel overcapacity persisted at approximately 500 million tonnes (roughly 25% of capacity) through the late 2010s.
The semiconductor industry faces similar risks. The SIA's analysis of announced fab construction projects reveals that if all planned facilities are completed on schedule — a significant assumption given construction delays and labour shortages — global wafer fabrication capacity will exceed projected demand by approximately 30–40% for mature-node chips (28nm and above) by 2030. Leading-edge capacity is less at risk of oversupply due to concentrated demand from AI and high-performance computing, but even here, the combined capacity of TSMC's Arizona fab, Intel's Ohio and Germany fabs, Samsung's Texas expansion, and multiple Chinese facilities may strain demand absorption.
The Distributional Question: Who Benefits from Semiconductor Subsidies?
Public discourse frames semiconductor subsidies as investments in national competitiveness and security. A more granular analysis reveals a complex distributional picture. The primary beneficiaries of semiconductor subsidies are: (a) the small number of firms capable of building and operating advanced fabs — principally TSMC, Samsung, and Intel; (b) the equipment suppliers (ASML, Applied Materials, Tokyo Electron) whose order books swell with subsidised construction; and (c) the skilled workforce in semiconductor engineering, whose wages have risen by 25–40% since 2022 due to demand competition.
The costs are borne by taxpayers and, through opportunity cost, by the public services that foregone fiscal revenue would have funded. The IMF's Fiscal Monitor 2025 notes that semiconductor subsidies represent a significant fiscal commitment for several countries: the US CHIPS Act's $52.7 billion allocation is roughly equivalent to the entire annual budget of the National Science Foundation. The EU's €43 billion mobilisation, while partly private co-investment, diverts public resources from other priorities including digital skills, broadband infrastructure, and basic research.
The geographic distributional effects are equally significant. Subsidies compete for a fixed pool of fab construction projects. When the US offers $39 billion in manufacturing incentives, it attracts fab investments that would otherwise have located in Asia. When the EU offers comparable incentives, it competes with both the US and Asian alternatives. The net effect is a redistribution of manufacturing investment based on subsidy generosity rather than underlying economic efficiency — precisely the distortion that the WTO's Agreement on Subsidies and Countervailing Measures was designed to prevent.
Alternative Approaches: From Subsidy Race to Coordinated Strategy
The game-theoretic analysis suggests that the current subsidy race is a stable but suboptimal equilibrium. Shifting to a more efficient outcome requires either cooperative agreements or mechanism design innovations that alter the incentive structure.
Multilateral subsidy disciplines. Extending WTO subsidy disciplines to cover semiconductor industrial policy could constrain the escalation dynamic. However, WTO enforcement is slow and the political will for new subsidy commitments is limited, particularly given US scepticism toward WTO dispute resolution. A more feasible approach might be a plurilateral "semiconductor subsidy code" among major producing nations, establishing transparency requirements, subsidy caps, and dispute resolution mechanisms outside the WTO framework.
Coordinated capacity planning. Rather than each nation independently subsidising capacity to achieve self-sufficiency, a coordinated approach would allocate capacity investments based on comparative advantage and risk diversification. This approach — analogous to NATO's defence burden-sharing arrangements — would achieve the security objective of geographic diversification at lower total fiscal cost. The political obstacles are formidable, as each nation prefers to host the most advanced facilities, but the potential fiscal savings (estimated at $80–120 billion relative to current announced programmes) create a substantial cooperation dividend.
Demand-side rather than supply-side intervention. Current programmes overwhelmingly subsidise fab construction (supply-side). An alternative approach would subsidise chip procurement for strategic applications (defence, critical infrastructure, AI research) while allowing market forces to determine where production locates. This approach would achieve security objectives — ensuring access to critical chips — without distorting the global allocation of manufacturing capacity.
Implications for GDEF's Finance & Economy Working Group
The semiconductor subsidy race represents a microcosm of the broader challenge facing the international economic order: how to manage strategic competition in critical technologies without descending into a collectively destructive subsidy war. The game-theoretic analysis presented here demonstrates that current trajectories are fiscally unsustainable and economically inefficient. GDEF's Finance & Economy Working Group will advance proposals for semiconductor subsidy coordination in its programme on International Economic Governance Reform, drawing on the analytical framework presented in this paper.
References & Sources
- Semiconductor Industry Association, 2025 State of the US Semiconductor Industry. semiconductors.org/resources
- Boston Consulting Group, Strengthening the Global Semiconductor Supply Chain in an Uncertain Era, 2025 Update. bcg.com/publications
- OECD, Industrial Policy for the Semiconductor Sector: International Approaches. oecd.org/sti
- IMF, Fiscal Monitor: Industrial Policy and Public Finances, October 2025. imf.org/fiscal-monitor
- Brander, J.A. and Spencer, B.J. (1985). "Export Subsidies and International Market Share Rivalry." Journal of International Economics, 18(1–2), 83–100. doi.org/10.1016/0022-1996(85)90006-6
- US Department of Commerce, CHIPS for America: Implementation Strategy. nist.gov/chips
- European Commission, European Chips Act: Building Europe's Semiconductor Ecosystem. ec.europa.eu/chips-act
- OECD, Measuring Distortions in International Markets: The Semiconductor Value Chain. oecd.org/trade
- Irwin, D.A. and Pavcnik, N. (2004). "Airbus versus Boeing Revisited: International Competition in the Aircraft Market." Journal of International Economics, 64(2), 223–245. doi.org/10.1016/j.jinteco.2003.08.006
- Miller, C. (2022). Chip War: The Fight for the World's Most Critical Technology. New York: Scribner. simonandschuster.com