Securing an advanced chip supply chain comes with additional costs depending on the regulatory regime in place. Since January 2025, the BIS has tightened certain controls and expanded the Entity List, contributing to the fragmentation of the global semiconductor market. The surcharges associated with this fragmentation may reflect a reorganization of technological access.
The essentials
- Advanced semiconductor supply chains can face surcharges depending on their access to export control regimes.
- Controls rest on regulatory categories and end-user authorizations, creating access differentials depending on applicable regimes.
- Additions and updates consolidate an evolving control regime; their permanence is not established.
- This fragmentation may favor certain established players at the expense of open competition in advanced computing markets.
- By 2030, the choice of technological cooperation regime will determine whether current surcharges become a permanent feature in the price of all digital infrastructure.
The surcharge is not a market accident
For thirty years, the global semiconductor market operated on a simple principle: maximum specialization. TSMC manufactured in Taiwan, ASML built its lithography machines in the Netherlands, American designers drew chips from Silicon Valley. Each link in the chain optimized itself independently, and costs fell.
This model produced remarkable efficiency. It also produced concentrated dependence: In September 2022, the White House cited a BCG/SIA estimate stating that Taiwan concentrated 92% of global manufacturing capacity for chips under 10 nm, on a contested island. When Washington decided to treat this dependence as a strategic risk, it sought to strengthen domestic production and reduce dependencies, and someone had to pay the bill.
BIS obligations depend on the item, destination, end user, and transaction in question. Available analyses focus on the cost of parallel self-sufficient chains compared to an integrated global chain. These surcharges may be associated with geographic fragmentation of access regimes rather than a particular commercial situation.
Two speeds: formal allies and the rest of the world
Documented access differentials fall under control rules, country groups, and authorizations, while certain commercial agreements are only being considered or negotiated. Certain exceptions and lists of approved companies exist in the EAR, but they do not constitute proof of preferential bilateral regimes reducing surcharges. It remains significant, but it is manageable for industries that can pass part of the gap on to their margins or their customers.
For everyone else—Indian equipment manufacturers, Vietnamese assemblers, European integrators not covered by sectoral agreements—surcharges remain significant with each AI project, each defense system, each industrial robot that cannot source freely on the global market. And as automation markets in Asia accelerate, demand for advanced chips does not slow.
The BIS architecture, the American Entity List subject to export control, plays the role of an invisible customs tariff. This list has regularly expanded since January 2025. Each addition reduces the space of accessible suppliers without a license, lengthens qualification delays for alternatives, and increases the costs of parallel chains.
The distinction Zingales poses, and facts validate
Luigi Zingales, economist at the Booth School of Business, has long defended a distinction that semiconductor geopolitics suddenly makes very concrete: the difference between being pro-market and being pro-business. A pro-market regime protects competition, rules of the game, and consumers. A pro-business regime protects established companies, their rents, and their privileged access to public decision-makers.
Export controls on advanced chips are fundamentally a restrictive national security regime, with conditional authorizations for certain actors. TSMC is a company appearing on BIS regulatory lists and a beneficiary of American industrial support, reflecting its strategic importance. Samsung negotiated an agreement with the United States concerning its investments. Intel receives CHIPS Act subsidies to rebuild domestic capacity that the market alone would never have justified at this cost.
In The State of Capitalism (2025), Zingales argues that this shift, when states favor their national champions at the expense of open competition, erodes confidence in capitalism itself. Rules of access vary depending on regulatory categories and applicable authorizations. Companies that do not have a government allied with Washington pay the price for others.
Access to advanced technologies varies depending on regulatory regimes and applicable authorizations, without this ruling out the role of productivity, industrial capacity, or other factors.
The security argument deserves to be taken seriously, and examined
The competing reading is solid, and it deserves to be presented honestly. Economists close to Dani Rodrik or Mariana Mazzucato, who questions the boundaries between common good and public intervention, argue that certain goods cannot be managed as pure commodities. Advanced semiconductors, tools of military artificial intelligence, state cryptography, and guidance systems, may belong to this category.
This argument has a logic. A chip manufactured in a potentially hostile country can contain hardware vulnerabilities. The traceability chain of advanced components poses verification challenges. If national security involves controlling flows of dual-use technologies, then additional surcharges could be considered the price of legitimate insurance.
The security argument hits limits that facts bring to light. Entities added to the BIS list are not all military actors: some are civil equipment manufacturers, universities, or research centers. A successively expanding list can become an instrument of industrial policy rather than a targeted security instrument. This distinction determines who bears the surcharge and on what grounds.
Where the cost spreads
The advanced semiconductor is computing infrastructure, and computing is now the infrastructure of everything else. Generative AI runs on 3-nanometer GPUs. Next-generation radar detection systems do too. Warehouse robots that transform Asian logistics carry advanced image-processing chips.
Smart electrical grids use increasingly powerful power semiconductors.
Surcharges at this level of the chain can spread downstream, affecting the cost of data centers and AI services. It increases the cost of defense systems, therefore equipment budgets. It increases the cost of automation, which slows productivity gains in sectors that need them most. And as AI is already redrawing labor markets in Asia-Pacific, a slowdown in computing infrastructure investments has real social consequences, not just industrial ones.
Europe is in a particularly uncomfortable position. It has no top-tier player in advanced chip manufacturing; STMicroelectronics and Infineon work on finer nodes. The 2024 EU Chips Act aims to raise Europe’s share of global production to 20% by 2030, from about 9% today. But ongoing investments depend on access to technological equipment subject to export control regimes. Europe is reducing dependence on one chain by building dependence on another.
What regime minimizes surcharges without abandoning collective security
The question facing us by 2030-2035 is not whether fragmentation will disappear. Current trends indicate probable fragmentation. Export controls are now bipartisan in the United States, supported by the administration and Congress. China is building its own capacities, SMIC is progressing on mature nodes, even if access to the most advanced nodes remains limited. The movement is underway and investments are committed.
A cooperation architecture remains to be defined to contain surcharges while preserving justified security objectives. Three trajectories are emerging.
The first is hard fragmentation: each technology bloc builds its independent chain, surcharges stabilize significantly for the world outside formal agreements, and countries without solid technological alliances structurally pay more to access computing infrastructure. This is the default scenario if no multilateral framework is put in place.
The second is the expansion of bilateral agreements: Washington expands its network of preferential agreements to more partners, gradually reducing surcharges for signatories. This is the pro-business scenario in its least unfavorable version; it reduces costs for allies without resolving the underlying question of open competition. The European Union, India, and certain Southeast Asian countries have an interest in negotiating in this direction.
The third, more ambitious, would be a multilateral verification regime on dual-use technologies, inspired by IAEA mechanisms for civil nuclear. The idea: establish technical standards for traceability and auditing of advanced chips, allowing more actors to prove the security of their chain without going through a bilateral agreement with Washington. The American CHIPS Act and the EU Chips Act both contain information-sharing clauses that could serve as a foundation. Analyses from the Center for Strategic and International Studies (CSIS) have explored this path, without it yet being politically backed.
Signals to watch: the rate of adding entities to the BIS list, SMIC’s progress on advanced nodes, and Intel’s decision on its European sites, indicators of investor confidence in rule stability.
The price of confidence
What is at stake goes beyond the arithmetic of surcharges. Semiconductor markets need stable rules for long-term investments to be possible. Building an advanced chip factory costs between 10 and 20 billion euros, equivalent to one year of the education budget of a medium-sized country, and requires a multi-year timeline. Predictable stability of access rules remains necessary for multi-year investments in advanced semiconductor infrastructure.
Zingales is right on one point: when rules of the game become a diplomatic variable rather than a stable framework, confidence in the market erodes. And this erosion has its own cost, distinct from the surcharge of parallel chains, a cost of uncertainty that BCG models do not yet capture.
Technological sovereignty is a legitimate objective. A regime that imposes significant surcharges on all actors without a seat at bilateral negotiations amounts to club policy, not market policy. The CHIPS Act and EU Chips Act have not yet opened the path to an open, verifiable, and predictable regime.
Sources
- SupplyICs / MIT Industrial Performance Center, 2026 Semiconductor Supply Chain & Export Controls, March 2026, https://supplyics.com/insights/supply-chain/2026-semiconductor-supply-chain-export-controls/
- Boston Consulting Group, Semiconductor Supply Chain Analysis, March 2026 (cited via SupplyICs/MIT Industrial Performance Center)
- Luigi Zingales, The State of Capitalism with Luigi Zingales (2025), Andersen Institute, https://anderseninstitute.org/the-state-of-capitalism-with-luigi-zingales/
- EU Chips Act, Regulation (EU) 2023/1781 of the European Parliament and of the Council, entered into force in 2024
- CHIPS and Science Act (United States, 2022), Bureau of Industry and Security (BIS), Department of Commerce
- Center for Strategic and International Studies (CSIS), analyses on technological dependencies and semiconductor export controls
