How a single semiconductor bottleneck locks the US into impossible choices.

Taiwan has become the single point of failure in three separate US strategic systems. The Defense Department cannot sustain a military guarantee without it. The Commerce Department cannot enforce chip containment without it. The Treasury cannot maintain its budget authority without TSMC Arizona producing at scale.

In May 2026, all three requirements collided.

Taiwan's prosecutors began investigating Nvidia server smuggling into China. The US delayed arms shipments to Taiwan without explanation. The Treasury committed $20 billion to TSMC's Arizona expansion. The Commerce Department shifted chip controls from export licensing to server-level enforcement. China, reading these signals clearly, accepted the RTX 5090D domestic ban.

None of these events contradict the others. That contradiction is the mechanism.

The Three Locked Systems

The United States operates three overlapping strategies toward Taiwan and semiconductors. Each makes sense in isolation. Together, they are mutually indefeasible.

System 1: Military Commitment. The US has offered Taiwan implicit security guarantees for seventy-five years. The guarantee rests on superior naval capacity and overwhelming air power. Taiwan's survival requires continuous flow of munitions, integrated defense systems, and F-16 upgrade packages. These transfers cost $5-8 billion annually. They require Congressional advance approval. They require 18-24 month lead times for manufacturing and delivery. Taiwan needs them continuously. The US can stop them, but only by ending the guarantee.

System 2: Chip Containment. The US enacted CHIPS Act in 2022 to reshore semiconductor manufacturing. The goal: prevent China from accessing leading-edge chip technology through strict export controls. The mechanism: TSMC cannot sell advanced chips to China. The enforcement: licensing, inspection, sanctions on violators. The vulnerability: China smuggles chips through intermediaries, specifically Nvidia server components. The escalation: move enforcement from export licenses to server-level controls, meaning tracking individual server shipments to verify they are not entering China. This requires TSMC to cooperate.

System 3: Budget Authority. The US Treasury borrowed $20 billion from CHIPS Act authorization to invest in TSMC's Arizona fab capacity. The justification: onshore production reduces China exposure. The reality: TSMC receives the capital. TSMC's Q1 2026 earnings showed $514 million annual profit. The Arizona fab is losing money. TSMC can redirect the capital. The US cannot force TSMC to complete the Arizona build. The US can only make continued investment contingent on loyalty. But loyalty to what? To Taiwan, which the US is de-prioritizing. Or to Arizona production, which requires China market access to achieve scale.

The trap: All three systems assume Taiwan remains a functioning US-aligned entity. All three are real. All three cannot be simultaneously satisfied within 12-24 months. 

The May 2026 Signal

May 9—Taiwan prosecutors initiate investigation. Reuters reports that Taiwan's judicial authorities are investigating Nvidia H100 server smuggling into China. H100s are export-controlled. Taiwan has caught the smuggling chain. Taiwan's message: we found your problem. Now what do you do?

Same day—US arms sales suspended. Military transfer notifications, normally public announcements, stopped. Taiwan expected delivered systems. They stopped arriving. No timeline for resumption given. The message: arms sales are now conditional.

May 15-18—TSMC Arizona expansion announced. The Treasury announces $20 billion in additional CHIPS Act funding for TSMC Arizona. This is new capital. This is public commitment. The message: TSMC is now a US domestic company for government support purposes.

May 20 onwards.Chip enforcement escalates. The Commerce Department shifts from "you need a license to sell X to Y" to "you must certify every shipment of server hardware and prove it is not entering China." This requires real-time tracking. This requires TSMC cooperation. The message: we are enforcing supply chain by logistics.

Late May—China's response. China issues guidance that domestic AI companies should reduce reliance on Nvidia RTX series. The signal: China accepts that cutting-edge Nvidia access is permanent. China is developing Plan B. The message: your containment is working, but we are moving toward autonomy.

These four events are causally independent. Together, they reveal the three-way bind. 

Why All Three Cannot Coexist

The Defense System requires Taiwan to survive as a functioning state. Survival requires that Taiwan's government remains intact, its economy functioning, its population stabilized. Taiwan's economy depends entirely on semiconductor exports. If Taiwan loses the ability to export chips, the economy collapses within 18 months.

The Chip Containment System requires that China cannot access leading-edge semiconductors. The enforcement mechanism is TSMC-based. But Taiwan's economic survival requires TSMC to operate inside Taiwan and export globally, including to China for non-military applications. The restriction prevents military-grade chip access but not commodity chip access. The enforcement depends on TSMC accepting lower profitability and higher operational risk. TSMC accepts this in exchange for US security guarantee for Taiwan.

The Budget Authority System requires that the $20 billion Arizona investment produces return for TSMC. But Arizona production is expensive. Arizona labor costs are 10x Taiwan labor costs. Without access to high-volume commodity chip markets, Arizona production cannot achieve efficiency needed to be profitable. TSMC's profit margins come from customers willing to pay premium prices. Those often mean Chinese tech companies.

The arithmetic: Military guarantee → Taiwan survives → Taiwan exports globally → TSMC serves China market → profitable operations → Arizona investment viable.

Chip containment → China cannot access TSMC → TSMC cannot serve China profitably → Arizona loses economic basis → TSMC diverts capital → Taiwan loses investment protection.

Budget authority → TSMC Arizona profitable → TSMC must access profitable markets → China market is only option → containment fails.

Pick any two. The third fails within 12-24 months.

The Structural Response

The US has three theoretical responses.

Option 1: Abandon military commitment, focus on containment and budget. Taiwan falls. TSMC becomes Chinese asset. The trade: China obtains all Nvidia AI chips and can develop advanced capabilities freely.

Option 2: Abandon chip containment, focus on military and budget. Taiwan survives with Chinese chip access. TSMC Arizona becomes profitable because TSMC serves Chinese markets freely. The trade: the US surrenders the containment goal entirely.

Option 3: Abandon budget authority, focus on military and containment. Taiwan survives. TSMC remains profitable through premium pricing. Arizona fab becomes government-subsidized strategic asset. The trade: the $20 billion becomes permanent loss. Congress will not accept this indefinitely.

The US cannot execute any of these choices because each makes explicit what the current architecture obscures: that Taiwan cannot remain a functioning state while serving all three functions simultaneously.

The current architecture obscures this through deniability of choice. The Pentagon delays arms sales and claims it is budget-related. The Treasury invests in Arizona and claims it is about onshoring. The Commerce Department escalates enforcement and claims it is about compliance. None of them claim Taiwan is being deprioritized. This deniability has a time limit.

Within 12-24 months, one of three things happens: Taiwan's government formally requests that the US choose between military protection and chip export restrictions. TSMC's capital allocation reveals the priority by shifting production or exports. Or China makes the choice explicit by offering TSMC deal terms that require choosing between Chinese cooperation and US cooperation.

When the choice becomes undeniable, the current architecture fails.

The Strongest Objection

The strongest counterargument does not dispute the May 2026 events. It disputes whether they constitute a structural trap or merely a temporary coordination problem. The objection says the three systems are not actually incompatible over a 12-24 month horizon; they are incompatible only if the US fails to execute with precision and speed.

This objection is structurally serious. It correctly identifies that execution matters. But it confuses timing with capability. The reading offered here does not claim the contradiction is instant. It claims the contradiction becomes binding within 12-24 months because of structural features of semiconductor production and political feasibility.

Specifically: TSMC Arizona requires 18-24 months of continuous capital commitment to reach breakeven. If capital is cut mid-project, the fab becomes abandoned. Chip containment requires real-time enforcement, which means TSMC must actively cooperate with US Commerce. Cooperation requires TSMC to believe Taiwan will survive. If TSMC perceives Taiwan as expendable, cooperation ends. Military commitment requires that Taiwan's economy remains functional, which requires that TSMC remains profitable, which requires China market access.

The objection claims all three can coexist if execution is perfect. The reading offered here claims that perfect execution is logically incompatible, not merely operationally difficult. A temporary coordination problem resolves. A structural contradiction persists and worsens as each system tightens.

The observable May 2026 signals suggest the decision-making process has already begun inside TSMC and inside the Taiwanese government. They are not waiting for the 12-24 month deadline. They are reading the signal that the deadline is accelerating. 

What Remains

The US locked Taiwan into a three-way structural contradiction in May 2026. The lock was not deliberate. It was the cumulative result of treating three separate policy systems as independent when they share a single point of failure: Taiwan's economic and political survival as a US-aligned entity.

The contradiction will resolve in one of three ways: explicit choice, where the US chooses one system and abandons the other two; market choice, where TSMC reallocates capital and exports, forcing the government to follow; or China's choice, where China makes the three options sufficiently costly that the US choice is made for them.

In all three cases, the outcome is predetermined. One system survives. Two do not.

This is not a forecast. This is an accounting of architecture. The architecture accounts for every actor: the Pentagon has a strategy, the Treasury has a strategy, the Commerce Department has a strategy, Taiwan has a government, TSMC has a board, China has a strategic calculus.

One actor remains unaccounted for in every system: the observer who sees the structure clearly but cannot change it.

Jerry van der Laan writes The Manifest Archive, analyzing power structures in geopolitics, finance, and institutions.