Chip Geopolitics: When Manufacturing Became Statecraft

No industry has become an instrument of statecraft as quickly as semiconductors. This story covers the export control regime, the silicon shield debate, the equipment chokepoints, China’s response and what companies actually do about it — part of the Taiwan Company Stories hub.

How did chips become a security issue?

Through the convergence of military and commercial technology. Advanced processors underpin artificial intelligence, cryptography, sensors, communications and weapons systems, so leadership in chip manufacturing became a strategic capability rather than a commercial matter.

The pandemic shortage then demonstrated that dependence was not theoretical. Automotive plants stopped for want of inexpensive chips, and governments discovered that no domestic policy tool could produce them.

Those two realizations — strategic importance and physical dependence — produced the export control regime, the subsidy programmes and the intense diplomatic attention that semiconductors now receive.

What do the export controls actually restrict?

Advanced manufacturing equipment, particularly extreme ultraviolet and advanced deep ultraviolet lithography systems, high-performance computing chips, and the tools and services required to maintain advanced production, along with restrictions on personnel and technology transfer.

The design targets capability rather than volume: mature-node production continues largely unrestricted while advanced logic and memory manufacturing face escalating limits. The theory is that leading-edge capability confers military advantage while ordinary chips do not.

Enforcement extends extraterritorially through the foreign direct product rule, restricting products made anywhere using American technology, which brings Taiwanese, Korean, Japanese and European suppliers into the regime regardless of their own governments’ positions.

Why is equipment the chokepoint?

Because a handful of companies make the tools that advanced manufacturing requires, and no alternative exists. ASML in the Netherlands is the sole supplier of extreme ultraviolet lithography; Applied Materials, Lam Research, KLA and Tokyo Electron dominate other critical steps.

These machines require continuous service, spare parts and software updates from the manufacturer, so restricting support degrades existing capacity over time rather than only preventing new purchases — a far more powerful lever than a simple sales ban.

Developing indigenous alternatives is a decade-scale challenge involving optics, precision mechanics, materials and control systems at the limits of physical possibility. China is investing heavily and progressing, and the timeline remains long.

What is the case for the silicon shield?

That Taiwan’s indispensability raises the cost of any disruption to a level that deters it. A conflict affecting the island would halt advanced chip supply globally, damaging every major economy including those with interests on both sides.

The physical argument reinforces the economic one: fabs cannot be captured intact and operated by an occupier, since they depend on foreign equipment, foreign service engineers, foreign materials and specialized staff who would not remain. Seizing them would produce rubble rather than capability.

The deterrent therefore operates in both directions: those who might disrupt gain nothing, and those who depend on the output have strong reasons to prevent it.

What is the case against it?

That deterrence based on economic cost assumes decisions are made on economic grounds, which historical experience does not support. Political and security calculations have repeatedly overridden economic rationality.

Diversification also erodes the shield deliberately. Every subsidized fab elsewhere reduces the global cost of disruption to Taiwan, which is precisely the objective of the subsidy programmes and precisely why some Taiwanese observers view them with concern.

A third argument holds that extraordinary value attracts attention rather than deflecting it, and that being indispensable makes Taiwan more central to great-power competition than it would otherwise be.

How has China responded?

With enormous investment in domestic capability, focused on mature nodes where it can compete now and on the equipment and materials required to eventually reach the leading edge without foreign supply.

Mature-node capacity expansion is the most immediate consequence, creating price pressure across the segment that affects Taiwanese, Korean and Western producers — the dynamic described in the UMC story.

Progress at the leading edge has been more limited but not absent, with domestic manufacturers producing advanced chips using older equipment through multiple patterning techniques at high cost and limited yield. The long-run question is whether indigenous equipment development succeeds.

Export controls have also accelerated Chinese investment in the entire supply chain, which may produce a genuinely independent ecosystem over a longer horizon than the controls were designed to address.

What do companies actually do about this?

Comply, diversify geographically, segment their customer bases and avoid public positions. Manufacturers operate under overlapping and sometimes conflicting regulatory regimes, and compliance functions have grown accordingly.

Operationally, firms maintain separate supply chains for different markets, restrict technology transfer between operations, and structure investments to limit exposure to regulatory change. This fragmentation raises costs across the industry.

Strategically, most Taiwanese companies keep their highest-value activity at home, expand incrementally in permitted jurisdictions and avoid concentration in any single market — a defensive posture that reduces both risk and growth.

What is the realistic outlook?

Continued fragmentation of the semiconductor industry along political lines, higher costs throughout, Taiwan retaining leading-edge concentration for the foreseeable future, and Chinese capability advancing in mature nodes while remaining constrained at the frontier.

The industry that spent forty years globalizing is being partially re-nationalized, and the efficiency lost is substantial. Consumers and companies pay for it through higher prices and slower innovation than an integrated industry would deliver.

For Taiwan the outcome is uncomfortable: greater strategic attention, greater economic importance and greater exposure simultaneously, with limited ability to influence decisions made in Washington, Beijing or Brussels that determine its industry’s conditions.

What happened to Chinese memory and logic ambitions?

They advanced considerably before controls tightened and have since faced constraints on equipment and technology access that limit progress at the frontier while allowing continued expansion in mature and legacy nodes.

Domestic manufacturers have demonstrated advanced-node production using older equipment through techniques that raise cost and reduce yield substantially. Whether this is a stepping stone toward independent capability or a demonstration with limited commercial scalability is genuinely disputed among analysts.

Memory has followed a similar path, with domestic producers gaining share in less advanced products while facing barriers in the highest-value segments where the technology gap and equipment restrictions bite hardest.

How do allied governments coordinate controls?

Through negotiated arrangements rather than through a formal multilateral regime. The Netherlands and Japan, home to critical equipment suppliers, have aligned their export policies with American restrictions through bilateral understanding, and Korea and Taiwan participate through their companies’ compliance.

Coordination is imperfect and politically costly. Equipment suppliers in allied countries lose substantial revenue from restrictions, and their governments must balance security cooperation against domestic commercial interests, producing periodic friction over scope and timing.

The system’s effectiveness depends on this alignment holding. Unilateral controls would be substantially less effective given the distribution of critical technology across several jurisdictions, which gives allied governments genuine leverage over the regime’s design.

What are the second-order effects on the industry?

Duplicated capacity, higher costs, slower technology diffusion and fragmented standards. An industry that achieved extraordinary efficiency through global specialization is being partially reorganized along political lines, and the efficiency loss is real.

Research and development is affected too. Collaboration between institutions and companies across the divide has diminished, talent mobility has been restricted, and the informal knowledge exchange that accelerated the industry for decades operates more narrowly.

The cost falls on everyone eventually through more expensive electronics and slower improvement. Whether that cost is worth the security objectives it serves is a political judgement, but the cost itself is not seriously disputed by industry participants.

How do Taiwanese companies manage political risk operationally?

Through geographic distribution of production, separate systems and data for different jurisdictions, conservative public communication and legal structures that limit exposure to any single regulatory regime.

Business continuity planning has become substantially more serious, covering scenarios that were previously considered unthinkable, including extended supply interruption and loss of access to specific markets or facilities.

None of this is discussed publicly in detail, for obvious reasons. Companies avoid statements that could be read as political positions, and the practical measures are handled as ordinary risk management rather than as a response to any specific scenario.

What is the long-run equilibrium?

Probably a partially fragmented industry with parallel supply chains for politically sensitive applications and a continuing global market for everything else, with Taiwan retaining leading-edge concentration and China building substantial independent capability in mature and eventually advanced nodes.

That outcome is more expensive and less innovative than the integrated industry it replaces, and it is the direction current policy on all sides points toward. Whether it stabilizes there or fragments further depends on political developments no industry participant can forecast.

What should businesses outside the industry take from this?

That supply chain decisions now carry political risk that ordinary commercial analysis does not capture, and that any dependency on a single jurisdiction for a critical input deserves the kind of scrutiny previously reserved for credit or currency exposure.

The semiconductor case is the most visible instance of a general shift. Critical minerals, pharmaceutical ingredients, batteries and industrial equipment all show similar concentration patterns, and the policy attention now directed at chips will likely extend to them.

Why does mature-node policy matter as much as the leading edge?

Because most chips in cars, appliances, industrial equipment and infrastructure are made on mature processes, and a supply disruption there stops factories just as effectively as an advanced chip shortage would. The 2021 automotive crisis was a mature-node event.

Policy attention has been concentrated on leading-edge capability for security reasons, while the practical economic vulnerability sits lower down the technology ladder, where Chinese capacity expansion is largest and least restricted. That mismatch is increasingly recognized and increasingly difficult to address without disrupting supply that industry currently depends on.

How do talent restrictions affect the industry?

Significantly and quietly. Rules restricting nationals from working on certain advanced manufacturing abroad, alongside company policies and visa considerations, have constrained the movement of experienced engineers that the industry historically relied on.

Since semiconductor capability lives substantially in people rather than in documentation, limiting engineer mobility is among the most effective restrictions available, and it has slowed capability transfer more than equipment controls alone would have.

Frequently Asked Questions