Taiwan and the Silicon Chokepoint
Why one island shapes the military, trade, and technology policy of every major power. The single most concentrated point of geopolitical risk in the modern economy.
(by a single company, TSMC)
(on an island smaller than Switzerland, which holds 9 million)
(a little below Switzerland or Poland; it grew 8.8% that year on AI-driven chip demand, its fastest in fifteen years, and is forecast to grow 11% in 2026)
A note on framing. Taiwan is a topic where panic and dismissal are both common. Neither helps. The actual situation is unusually concrete - one island, one company, a measurable strait, two governments with public commitments - and the data is unusually clear about what is and is not at stake. The page below tries to walk through the situation as a careful adult reader would want it explained, without either war drums or false reassurance.
Why Taiwan matters
Taiwan is a small island a hundred miles off the coast of mainland China. It has 23 million people, a democracy with regular peaceful changes of government, and an economy that ranks around 20th in the world. None of that, by itself, would make it the center of strategic concern for the United States, Japan, Europe, China, or anyone else. What makes Taiwan central is a single industry: the manufacturing of advanced semiconductors, also known as computer chips.
The chips that run modern smartphones, data centers, cars, missiles, satellites, medical devices, and almost every other piece of complex electronics come, in disproportionate share, from a single company. It is Taiwan Semiconductor Manufacturing Company - TSMC for short. TSMC is the world's largest "foundry" - a company that does not design chips of its own but makes them for other companies. Apple, Nvidia, Qualcomm, AMD, and a long list of the world's largest technology companies design chips and then send the designs to TSMC to be physically manufactured.
On the most advanced chips - the ones with the smallest features, the highest performance, the highest profit margins, and the most strategic importance - TSMC has a share approaching 90% of the world's production. Samsung in South Korea and Intel in the United States can also make chips at a comparable level, but neither comes close in volume. In the second quarter of 2026 TSMC took 72.5% of global foundry revenue and Samsung 5.9%, by TrendForce's count. The most advanced chips themselves depend on machines made by a single Dutch company called ASML. The dependency chain narrows as you go up the technology ladder, and the very top of the ladder runs through a single island off the coast of mainland China.
That concentration is what makes Taiwan a chokepoint rather than a country among many. A serious disruption of Taiwanese chip production - whether from an invasion, a blockade, a natural disaster, or a major industrial accident - would produce shortages and price increases across most of the modern economy within months. Estimates of the global cost of a complete loss of Taiwanese chip production range from several trillion dollars in the first year to substantially more if the disruption persisted. There is no scenario in which the world economy absorbs a Taiwan crisis without significant pain.
How TSMC got there
The story of TSMC's rise is worth knowing because it explains why the concentration is hard to undo. TSMC was founded in 1987 by Morris Chang, a Chinese-born, American-trained engineer whom Taiwan's government had recruited to build a chip industry. With state backing, he pursued a then-unusual idea: a chip company that did not design any chips of its own, only manufactured them for other companies. At the time, every major semiconductor company designed and made its own chips. The "fabless" design model that exists today (where companies like Apple, Nvidia, and Qualcomm design chips and outsource the manufacturing) only became possible because TSMC made the foundry model viable.
From the late 1990s onward, TSMC invested heavily in a continuous cycle of building newer, more advanced manufacturing facilities. Each generation cost more than the last. By the 2010s, the cost of building a state-of-the-art chip factory had risen to about $20 billion, and only three companies in the world were even attempting it: TSMC, Samsung, and Intel. By the 2020s, Intel had fallen behind technically, and Samsung was struggling to match TSMC at the leading edge. TSMC kept investing through every economic cycle, kept hiring the best engineers, and kept its factories running with high precision. The result was a slow accumulation of advantages that compounded into near-monopoly at the cutting edge.
The other piece of the story is geographic. About 90% of TSMC's manufacturing capacity is located on the island of Taiwan. The company is building factories in Arizona, Japan, and Germany - a deliberate response to the concentration risk - but they are smaller and a generation or more behind. Arizona, where TSMC now plans to invest $265 billion, has made 4-nanometer chips in volume since late 2024 and targets 2 nanometers by the end of the decade; Taiwan has been mass-producing 2-nanometer chips since late 2025. The overseas plants will not catch up to the Taiwan facilities for at least a decade. The same is true for Samsung's overseas expansions and Intel's foundry effort. As of 2026, anyone who needs the very best chips needs Taiwan.
Where the world's chips come from
The semiconductor industry is more spread out than the headline suggests, but the leadership at the top is concentrated. The numbers below give a rough picture of where chip manufacturing happens, who designs the chips, and who makes the equipment used to build them. Capacity shares are 2025 estimates by SEMI, the industry association.
The picture is one of multiple chokepoints rather than a single one. Taiwan is the most concentrated and most exposed, but a serious disruption at any of the major nodes - the Dutch lithography machines, Japanese materials, Korean memory, or US-designed software - would also cause significant pain. The system that produces the world's chips is a deeply interdependent network with few backup options. Building new ones takes a decade and tens of billions of dollars per node.
The China question
The government of the People's Republic of China (PRC) holds that Taiwan is part of China and must eventually be "reunified" with it. It has held this position since 1949, when the losing side of the Chinese civil war, the Republic of China (ROC), retreated to Taiwan. China's leader, Xi Jinping, says Beijing will "strive for peaceful reunification" but "will never promise to renounce the use of force." In Taipei, President Lai Ching-te's Democratic Progressive Party (DPP) government says the ROC and the PRC "are not subordinate to each other" and that only Taiwan's people can decide its future. Beijing will not deal with that government, whose leaders it calls separatists. The opposition Kuomintang (KMT) opposes independence, holds that both sides belong to one China while each interprets that differently, and favors engagement with Beijing. Washington "acknowledges" Beijing's position but takes no stand on Taiwan's sovereignty. It sells Taiwan arms for self-defense under the 1979 Taiwan Relations Act and "opposes unilateral changes to the status quo," while leaving open whether it would fight.
What has changed is the underlying military balance. In the 1990s and 2000s, the United States Navy could have intervened in a Taiwan crisis with high confidence of success. By the 2020s, that confidence had eroded. China has built more naval ships, more anti-ship missiles, more air defense, and more missile-based systems specifically designed to make a US intervention costly. War games no longer assume a clean American victory. In the most-cited public series, run 24 times by the Center for Strategic and International Studies in 2023, an invasion usually failed, but the United States and its allies lost dozens of ships and hundreds of aircraft. The strategic geography of the Western Pacific has shifted, and Taiwan sits at the center of it.
What has not changed is the underlying difficulty of an actual military operation. Taiwan is a hundred miles off the Chinese coast, separated by a strait that is rough most of the year. The mountainous east coast of the island is hard to land on. The west coast (where landing is feasible) is heavily fortified with prepared positions, missile sites, and the ability to flood beaches. Modern amphibious invasions are extraordinarily difficult under the best of circumstances, and a Taiwan invasion would be one of the largest amphibious operations in history. China's military has no recent experience of amphibious combat; its largest opposed landing, on Hainan Island, took place in 1950. The military effort would be massive, and the political and economic costs would be enormous regardless of whether the invasion succeeded.
What people on the island want, and why the polls disagree
Everything above treats Taiwan the way most coverage does: as a chokepoint in a supply chain and a point of contention between Washington and Beijing. That framing leaves out the twenty-three million people who live there and who get a vote, in the literal sense, on most of what happens next. Their preferences do not map cleanly onto either of the outside framings, and the surveys that measure those preferences produce headlines that appear to contradict each other. Both of those facts are worth understanding before reading any scenario.
Start with identity, which is the least ambiguous of the measures. National Chengchi University's Election Study Center has asked the same question since 1992: do you consider yourself Taiwanese, Chinese, or both? In 1992 more respondents chose Chinese than Taiwanese, 25.5% against 17.6%. In the first half of 2026, 2.5% chose Chinese only, close to the record low of 2.4% in 2023 and 2024. About 31% chose both, and 63% chose Taiwanese only. This is a generational inversion rather than a swing, and it has continued under governments of both major parties. Whatever else is contested, the population's sense of who it is has moved a long way and shows no sign of moving back.
Preference about the island's political future is where the surveys diverge, and the divergence is an artifact of how the question is built rather than a mystery about the public. The Election Study Center offers six graded options, from immediate unification to immediate independence with four gradations of status quo between them. The large majority chooses one of the middle options rather than either end. In the first half of 2026, keeping the status quo indefinitely was the largest single answer, at 34.4%, up from 24.0% in 2018. Deciding later drew 25.5%, the status quo now with independence later 21.7%, and with unification later 6.3%; immediate independence drew 4.5% and immediate unification 1.3%.
The Taiwan Public Opinion Foundation asks something different and gets a different answer. Its question offers independence or unification and does not put the status quo on the list, leaving respondents to volunteer it. On that wording it found 51.8% for independence, 24.2% volunteering the status quo anyway, and 13.3% for unification in January 2025. The same survey then asked those who had chosen the status quo what they would pick if it could no longer be maintained. Of them, 39.5% switched to independence, 21.9% to unification, and 38.6% still insisted on keeping it forever.
These are the same population answering differently shaped questions. Each number is accurate for the question it answers. Read together, they say three things. A large majority prefers the present arrangement to either alternative. Support for unification is small and has been shrinking for two decades. And a conditional majority would choose independence if forced to choose, which is why the wording of any specific poll matters as much as its result. That also explains a pattern outsiders find puzzling: Taiwanese voters can be unwilling to be governed from Beijing and unwilling to provoke a confrontation over saying so, at the same time.
The domestic politics that follow from this are not a footnote to the security question; they are the mechanism through which it operates. The 2024 elections produced a divided government. Lai Ching-te of the Democratic Progressive Party won the presidency with about 40% of the vote in a three-way race, while the Kuomintang won a plurality in the legislature and, with the smaller Taiwan People's Party, control of it. In 2025 civil-society groups launched recall campaigns against 31 Kuomintang legislators, with President Lai's public backing. Every one of them failed, across two rounds in July and August, and the president's approval fell sharply afterward. An opposition attempt to impeach Lai in May 2026 fell well short of the two-thirds required. The opposition holds the legislature until the January 2028 elections; the next test is the November 28 local elections, which include an opposition-backed referendum on scrapping the government's "nuclear-free homeland" energy policy.
The clearest consequence runs through the defense budget, which is where Taiwanese domestic argument and the deterrence question covered above become the same subject. Lai proposed a special defense package of about 1.25 trillion New Taiwan dollars (NT$), roughly US$40 billion, in late 2025. The legislature passed a reduced version of NT$780 billion, about US$25 billion, in May 2026. That cut 38% and funded only purchases of American weapons, leaving out domestic programs such as drones and the Chiang Kung missile-defense system. The cabinet is trying to restore the rest through a supplementary budget for 2026 and its 2027 budget, which puts total defense spending, on a NATO-style count, at a record NT$1.12 trillion, 3.01% of GDP. The US State Department called further delays "a concession to the Chinese Communist Party." The opposition says it supports defense spending but will not approve open-ended funds for vague proposals, citing corruption risks, and that domestic programs belong in annual budgets, which face more oversight. In August it passed a six-year drone plan of its own, larger on paper than the cabinet's. How much of the gap is budget discipline and how much is partisan obstruction is exactly the sort of question that outside observers answer according to what they already believe about the parties.
What this changes about the scenarios below is the location of the decisive variable. Most external analysis treats the outcome as a function of what Beijing decides and what Washington would do in response. A substantial part of it is a function of what Taipei decides: how much the island spends on its own defense, what it buys, how fast reforms to conscription and reserves move, and whether a divided legislature can agree on any of it. That is a democratic argument being had in public, by people with the most at stake, and it is the part of the picture that the chokepoint framing tends to leave out.
The paths from here
Dated September 2026, days after Xi Jinping's state visit to Washington, with a $14 billion US arms package for Taiwan still under review and Taiwan's local elections due on November 28. None of the realistic paths is comfortable. The base case is "more of the same, with slowly rising tension," but several other paths have non-trivial probability.
Status quo persists, with managed tension
Taiwan continues to operate as a self-governing democracy, China continues to claim it without trying to take it by force, and a careful set of unwritten understandings keeps the peace. Both sides issue strong rhetoric publicly while carefully avoiding the actions that would force a confrontation.
Will it happen? This is the base case and has been the actual situation for more than seventy-five years. It is also the path most fragile to a single misstep. The probability over the next decade is high but not certain - somewhere around 60 to 75% depending on which careful analyst you ask.
A slow squeeze that wears Taiwan down
China increases pressure on Taiwan through economic, diplomatic, and gray-zone military moves without crossing the threshold of open war. Over years, the political will of Taiwan to resist erodes, the willingness of allies to defend it weakens, and at some point Taipei accepts a "framework agreement" that effectively ends its self-rule.
Will it happen? Many observers judge that Beijing would prefer to gain control this way rather than risk a war. So far the pressure has not moved Taiwanese opinion its way: on the Election Study Center's measure, under 8% favor unification now or later. In Washington, President Donald Trump has called arms sales to Taiwan "a very good negotiating chip" and held a $14 billion package in abeyance, while his administration says its Taiwan policy has not changed. A slow squeeze that succeeds is plausible over a thirty-year horizon, hard to imagine in ten.
Blockade or partial quarantine
China announces some form of inspection regime, exclusion zone, or trade restriction on Taiwan, forcing a binary choice on the United States and its allies. The blockade may or may not include a small kinetic component (a few ships fired on, a few aircraft turned back).
Will it happen? This is the scenario several careful military analysts now consider most likely if China decides to actively pressure the situation. China's largest drills in three years, Justice Mission 2025 in December 2025, rehearsed blocking Taiwan's main air and sea routes. It uses China's strongest military advantages (sea and missile power close to its coast) and exploits the West's weakest political dimension (the difficulty of choosing which crisis to escalate). Probability is harder to estimate but likely higher than full invasion - perhaps 5 to 15% over a decade.
Full invasion attempt
A direct amphibious operation aimed at seizing the main island. The US and Japan likely intervene. The conflict spreads. Casualties on every side are large; the global economic disruption is severe; the political consequences reshape Asian alliances for decades.
Will it happen? Most careful analysts put the probability in the low single digits over the next decade. Not zero. Beijing's public commitment to unification, what many analysts see as a slow erosion of US deterrence, and the possibility of misperception in either capital make it real enough to plan against. Observers debate whether China's military will meet Xi's reported goal of being ready to invade by 2027 - a capability target, not a decision - after purges left its top command with only Xi and one vice-chair. The high consequences of the scenario justify a much larger share of strategic attention than the probability alone would suggest.
Negotiated framework arrangement
Some new political settlement is reached - perhaps an updated version of the 1992 Consensus, perhaps something new. Taiwan retains substantial self-government in exchange for clear commitments not to declare formal independence; China accepts the de facto status quo in exchange for clear guarantees against further Western military entrenchment.
Will it happen? This is the path many professional diplomats in both Washington and Beijing privately say they would prefer. The trouble is domestic politics on every side: no Chinese leader can be seen accepting a permanently separate Taiwan, no Taiwanese government can be seen accepting rule from Beijing, and no US administration can be seen abandoning a democratic partner. KMT chair Cheng Li-wun met Xi in Beijing in April 2026, but Beijing will not deal with the DPP government, which holds the presidency until 2028. The conditions for a real framework remain difficult.
The base case is continued status quo with rising tension. The middle scenarios - slow squeeze, blockade - are the ones with real probability over a decade. Full invasion is unlikely, but not so unlikely that it can be set aside. A negotiated framework would be the least violent outcome and is the hardest to build. The world economy and the global chip industry depend on choices that have not yet been made in Taipei, Beijing, and Washington.
Where serious analysts disagree
Taiwan is one of the topics where careful disagreement is much more illuminating than confident prediction. Each of the readings below is held by named analysts whose work is worth reading directly, and the data does not cleanly resolve any of them.
A Chinese move on Taiwan is more likely this decade than commonly assumed
The window in which China can act decisively while still being able to absorb the cost may be closing. Chinese demographics are about to bite hard. The economic slowdown will get worse before it gets better. Xi Jinping's domestic legacy depends on visible progress on Taiwan. The temptation to act before time runs out grows each year.
Held by: Hal Brands and Michael Beckley (the "Danger Zone" thesis), parts of the US Department of Defense, and a number of Asian strategists. Their data on Chinese capability and motivation is solid; the open question is how much weight to give to the political-cost side of Beijing's calculus.
Blockade is more plausible than invasion
An invasion is enormously difficult, requires perfect execution, and would visibly fail or succeed within weeks. A blockade leverages China's strongest assets, can be calibrated for political effect, and forces the United States to choose between accepting it and starting a war. The Western strategic establishment has spent the last decade preparing for invasion and may be under-prepared for the more likely scenario.
Held by: Lonnie Henley (former US intelligence official), Bonnie Glaser (German Marshall Fund), and a number of naval analysts. Their argument has gained traction in recent Pentagon assessments and is increasingly mainstream within the strategic-planning community.
Even a successful invasion would not give China the chips
TSMC's most advanced fabrication facilities depend on hundreds of suppliers, thousands of trained engineers, and constant technical support from Western partners. A military takeover would almost certainly result in destruction of the most advanced equipment, evacuation of key staff, and a global cut-off of supplies. The chips matter, but the chips are not actually obtainable through force.
Held by: Chris Miller (author of "Chip War"), and most semiconductor industry analysts. The strategic implication is that the chip dependency is more of a constraint on Chinese options than a prize Beijing could realistically capture.
The bigger risk is miscalculation, not strategic decision
Both Beijing and Washington have publicly committed to positions they cannot easily walk back. Both have domestic political pressures that push them to project firmness. The most likely path to crisis is not a deliberate decision by either side to start a war, but a misread incident that escalates faster than either capital can contain it. The strategic conditions resemble pre-1914 Europe more than they resemble the Cold War.
Held by: Graham Allison (Harvard) and the broader "Thucydides Trap" school, plus historians of the 1914 crisis. Their argument is that risk management has to focus on the conditions that allow miscalculation, not just on each government's stated intentions. Allison's specific framework - that rising-versus-ruling-power transitions historically end in war about 75% of the time - has been heavily contested on case-selection grounds (Richard Hanania, Arthur Waldron, and others argue the dataset is constructed to support the conclusion). The underlying point about miscalculation risk does not depend on Allison's specific statistical claim being correct.
Taiwan can hold longer than commonly thought
Difficult terrain, prepared defenders, and modern anti-ship and anti-air weapons all favor the defense. Add the speed at which the United States and Japan could intervene, and even a serious Chinese assault is far from guaranteed to succeed quickly. The "Ukraine model" - where Western support enabled a smaller country to inflict large costs on a larger invader - is more applicable to Taiwan than is sometimes assumed.
Held by: Advocates of the "porcupine" defense that successive US administrations have urged on Taiwan, parts of the US Marine Corps, and "deterrence by denial" strategists. The implication is that the right Western strategy is to make a Chinese assault clearly unprofitable, rather than to commit to direct intervention in advance.
What the available evidence actually supports is this. Taiwan is at higher risk than at any point in the last forty years, and the risk is more about pressure short of war than about war itself. The consequences of any major event would be globally severe. The room for skillful diplomacy and credible deterrence to prevent a crisis is still substantial - but is shrinking, not growing.
What this means for you
For most readers, the Taiwan question is not a daily concern. It does, however, sit underneath the value of investments, the price and availability of nearly every electronic device, the security of supply chains many businesses depend on, and the stability of the part of the world economy that touches the chip industry. Some practical observations:
If you depend on chips at work
This includes anyone in technology, automotive, medical devices, defense, aerospace, advanced manufacturing, and most physical-product businesses. The risk of a chip supply disruption is now a real input to capital planning rather than a low-probability tail. Companies that have done multi-source planning, kept some inventory buffer, and avoided single-foundry exposure on critical products are the ones positioned to ride through any disruption rather than be blindsided by it.
If you invest
A single-event risk of this scale is hard to hedge directly. Diversified global portfolios already partially absorb it. Concentrated bets on Taiwanese companies, Chinese stocks tied to mainland-Taiwan tension, or single-name semiconductor exposures carry more risk than the headline suggests. Companies building chip-fabrication capacity outside Taiwan (TSMC's Arizona facilities, Intel's foundry effort, Korean and Japanese expansions, European projects) have policy tailwinds for the next decade. None of this is investment advice. It is observing where the policy gradient and the security calculation point.
If you live in the region
Japan, South Korea, the Philippines, Australia, and a number of Southeast Asian countries are now visibly preparing for a more contested Asia-Pacific. Defense spending is rising. Alliance structures are tightening. Visa and immigration policies are shifting around technology workers. If you live in one of these countries, your government is making decisions now whose effects will be felt over decades; voting on the merits and watching what is actually committed to is more useful than reacting to the news cycle.
If you live anywhere else
The economic shock from a serious Taiwan crisis would be felt globally - in chip prices, in technology product availability, in defense spending, in stock-market volatility, and possibly in oil prices and shipping. There is no part of the world economy that fully escapes a Taiwan event. The good news is that this is one of the things diversified investing and a steady cash buffer were designed for. The bad news is that Western politics has under-invested in preparing the public for the possibility that a major shock could come from this direction.
If you're thinking about supply-chain risk
Taiwan is the headline chokepoint, but the broader lesson is that the modern economy contains a number of less-visible single points of failure - the Dutch lithography company, Korean memory makers, Strait of Hormuz oil shipments (restricted since March 2026), key undersea cables, a handful of pharmaceutical-ingredient suppliers, and a long list of others. Mapping where your own business or country's economy has hidden weak points is one of the most useful exercises a careful planner can do once a year. The lesson of Taiwan is not to obsess about Taiwan. It is to take chokepoint risk seriously generally.
The mechanics behind this
The Taiwan question sits on top of three deeper mechanisms that show up across this site. If the analysis above depends on ideas you want to understand first, these fundamentals make the conversation more legible:
