Summary
This scenario looks at how China could take Taiwan by fait accompli. The PLA turns a scheduled autumn exercise around Taiwan into a real invasion. Before the first wave lands it fires conventional missiles at US air bases on Okinawa and at Andersen Air Force Base on Guam. The trigger is political and gradual: a run of US arms packages, a Taiwanese leadership Beijing calls "separatist", and a Japan–China quarrel. Beijing reads these as closing its window before its 2027 readiness goal [3][5][6].
China moves first. The missile strikes in Japan frighten the Japanese government rather than unite it. Tokyo condemns the attack, but under the 1960 prior-consultation arrangement it delays approval for US combat sorties from its territory. It also keeps the Self-Defense Forces to defending Japan itself [1][4]. Washington needs about two weeks to decide, mobilise and reposition. When US airpower arrives in strength, the PLA already holds the ports and airfields of southern Taiwan.
This combines the conditions that CSIS identified in 2023 as most favourable to China. Japan withholds full access, US combat starts late, the JASSM-ER missile proves ineffective against ships at sea, and Taiwan's response is slow at first [1]. Even so, CSIS found that most pessimistic runs ended in stalemate, not Chinese victory [1]. This scenario therefore assumes one further step: a ceasefire.
Five to six weeks in, the PLA holds the south and the west-coast cities and is pressing on Taipei. A US surface fleet has been badly hurt trying to break through. In the light of those losses, the US and Japan accept a ceasefire brokered at the UN and by third parties. Taipei's government, cut off, without power and with half its army lost, accepts terms. The island is not conquered in a final battle; it is conceded.
Estimated losses, drawn from the pessimistic and "Ragnarok" iterations: US 300-500 aircraft, 2-4 carriers and 15-40 major surface combatants and submarines, and 5,000-10,000 casualties. China: 100-130 ships, 250-350 aircraft and 30,000-60,000 casualties. Taiwan: its whole navy and air force, and 40,000-85,000 army casualties [1].
At a glance
| Trigger | First mover | Duration | US losses | Chinese losses | Taiwan/ally losses | Form of Chinese victory |
|---|---|---|---|---|---|---|
| Political crisis (arms sales, Japan–China rift); invasion launched from a declared exercise | China (missile strikes on Okinawa and Guam, then amphibious and airborne assault) | About 6 weeks of high-intensity war, then ceasefire | 300-500 combat aircraft (mostly on the ground); 2-4 carriers; 15-40 surface ships and submarines; about 5,000-10,000 casualties | 100-130 ships (mostly amphibious); 250-350 aircraft; 30,000-60,000 casualties | Taiwan: entire navy, most of its air force, 40,000-85,000 army casualties. Japan: about 160 aircraft and about 14 ships if it is struck and fights; fewer if it stays out | Fait accompli on the ground, ratified by a ceasefire and Taipei's acceptance of terms |
Background and assumptions
What research says. In 2023 CSIS ran 24 iterations of a wargame of a 2026 Chinese invasion [1]. In most runs the invasion failed, provided four conditions held:
- Taiwan resists.
- The US joins "within days" with its full capabilities.
- The US can use its bases in Japan.
- The US has enough long-range anti-ship missiles.
In the authors' words, "Without them, the United States cannot use its numerous fighter/attack aircraft", referring to the Japanese bases [1]. Of the 18 pessimistic iterations, none produced a clear Chinese success. Four were decisive Chinese defeats and 14 were stalemates; only three of the stalemates were "trending toward China" [1]. China won outright in only two cases:
- "Taiwan stands alone": the US stays out.
- "Ragnarok": Japan is neutral and US bombers are neutralised. CSIS says plainly that this was "an unlikely scenario" [1].
This scenario is therefore pessimistic by design. It sits between CSIS's worst stalemates and Ragnarok, and it needs all of the following to be true:
- Short warning used poorly. CSIS's base case gives unambiguous warning at D-14, with the build-up masked by an exercise [1]. Here the warning arrives, but Washington and Tokyo hesitate to act on it. CSIS warns that hesitation is likely even after strategic warning [1].
- Japan restricts access. The US must consult Japan before using its bases there for combat that is not in Japan's own defence [4]. The scenario assumes Tokyo delays approval even after US bases on its soil are struck. CSIS shows that "Japan neutral" and "JSDF remains defensive" are among the variables that help China most [1]. Japan's 2025 record cuts both ways. Prime Minister Takaichi called a Taiwan contingency a possible "survival-threatening situation", then partly walked the remarks back as consistent with the existing government line [5]. Beijing's response, a travel warning and a renewed ban on Japanese seafood, showed it is willing to coerce Tokyo [10].
- US combat effectively starts around D+14. CSIS's "U.S. combat starts D plus 14" excursion let China put more forces ashore before its amphibious fleet was worn down [1].
- Missiles fall short. The scenario uses CSIS's "no maritime strike JASSM" assumption. It also assumes the US runs out of LRASM missiles within a week, as it did in every CSIS iteration [1][2].
- Taiwan is slow at first and then bends. The scenario applies CSIS's "Taiwanese forces paralyzed to D plus 4" excursion. It also assumes that after weeks without power and with losses mounting, Taipei accepts terms. CSIS notes that "the Taiwanese government might capitulate before the bitter end" [1].
- A capable PLA. The US Defense Department's 2025 report says Beijing expects to be able to "fight and win" over Taiwan by the end of 2027. It also says China in 2024 tested essential components of its Taiwan options, including strikes on sea and land targets and blocking access to key ports [3].
Likelihood. The published research makes this a low-probability path. Each assumption on its own is plausible; together they are unlikely. The main weakness is Japan. In every pessimistic iteration China struck bases in Japan [1], and an attack on Japanese soil is the event most likely to bring Tokyo into the war. The scenario only works if Beijing persuades Tokyo that the strikes are limited to US assets and that staying out will spare Japan.
Timeline
| Day | Event |
|---|---|
| D-365 | The PLA continues large exercises around Taiwan, building on Strait Thunder-2025A (April 2025) and Justice Mission 2025 (December 2025), the latter rehearsing a blockade of ports including Keelung and Kaohsiung [6][8]. Washington and Taipei treat them as coercion, not preparation for invasion. |
| D-300 | Beijing turns its political pressure on Japan into economic retaliation, extending the travel warning and seafood ban it imposed after the 2025 Takaichi dispute [10]. Tokyo plays down Taiwan contingencies in public. |
| D-180 | The PLA announces an annual autumn "joint exercise" cycle. Taiwan's intelligence notes an unusual number of civilian roll-on/roll-off ferries in maintenance. Washington asks for more surveillance but no posture change. |
| D-90 | A new US arms package is announced. China's Foreign Ministry promises "resolute countermeasures". The US Navy keeps one carrier group in Japan. |
| D-45 | The PLA Rocket Force disperses launchers under exercise cover. The US intelligence community reports ambiguous indicators; CSIS assumes these would appear from about D-30 [1]. |
| D-30 | China requisitions civilian shipping "for exercise support". The US and Taiwan see it but debate intent [1]. Taipei raises its readiness level but does not mobilise reserves, to avoid "provoking" Beijing. |
| D-21 | China declares closure areas east and west of Taiwan for "Justice Mission-2026". Japan protests the zones near Yonaguni. |
| D-14 | US intelligence judges an invasion likely. The President orders a show of force: a second carrier group toward the Ryukyus and bombers to Guam. CSIS's base case assumes this move, which puts those forces inside China's missile range [1]. |
| D-10 | Washington asks Tokyo to start prior consultation on combat use of its bases [4]. Japan's cabinet is split and the coalition partner resists. Beijing privately warns Tokyo that bases "used against China" will be treated as hostile. |
| D-7 | Taiwan calls up some reserves. China says this is an "escalation" and pushes its exercise ships to within 12 nautical miles. |
| D-3 | China evacuates some diplomats' families from Japan and Guam. US officials leak invasion warnings to deter Beijing; China calls them "disinformation". |
| D-1 | The embarked PLA amphibious groups turn from exercise tracks toward landing areas in southern Taiwan. The US and Taiwan detect the movement within hours. |
| D-day, 0300 | China's opening strikes hit Taiwan's air bases, radars, naval ports and command nodes. Ballistic and cruise missiles also strike Kadena and Andersen. CSIS found about 90% of allied aircraft losses were on the ground [1]. Taiwan disperses what it can. |
| D-day, 0600 | Missiles hit US surface ships east of Taiwan. CSIS found that "any surface ship that approached Taiwan was destroyed" [1]. The US President declares a national emergency. |
| D-day, 1200 | First landings near Tainan and Kaohsiung, with airborne assaults on airfields. Taiwan's command is disrupted until about D+4 [1]. |
| D-day, 1800 | Japan's Prime Minister condemns the strikes on Okinawa. Because no JSDF facility was hit, Tokyo does not declare a "survival-threatening situation" and postpones a decision on US combat sorties. |
| D+1 | US bombers from Guam and the continental US launch standoff missiles at the amphibious fleet. LRASM stocks are used up within days [1][2]. China condemns "US aggression". |
| D+2 | Taiwanese shore-based anti-ship missiles sink amphibious ships and escorts. In CSIS's "stands alone" run they sank 17 of each in 10 days [1]. The PLA reinforces the beachhead. |
| D+3 | Strikes on Andersen leave Guam largely unusable for fighters and tankers. The US moves fighters to Alaska and Hawaii, too far to generate sorties over Taiwan [1]. |
| D+4 | Taiwan's army counterattacks at the beachhead but is too late to throw it back. China takes the port of Kaohsiung, damaged, and an airfield near Tainan. |
| D+5 | Tokyo authorises US logistics and ISR support from Japan but not combat sorties. The JSDF stays defensive. China says it "takes note". |
| D+7 | The PLA has roughly 40-60 battalions ashore, in line with CSIS's pessimistic average [1]. Washington debates strikes on the mainland. Fear of escalation produces a ban on strikes against the mainland, an assumption CSIS tested [1]. |
| D+9 | US submarines sink ships in the Strait but take losses; in Ragnarok, 10 SSNs were lost before the climactic naval battle [1]. China shifts escorts to anti-submarine screens. |
| D+11 | Taiwan's power grid fails under strikes and fuel shortages, as rehearsed in Strait Thunder-2025A's simulated strikes on energy facilities [8][7]. Taipei appeals for international convoys. |
| D+14 | Washington gives its main forces orders to engage in full. This is the "combat starts D plus 14" excursion [1]. China has already restored Kaohsiung and the southern airfields and is landing armour and engineers. |
| D+17 | Tokyo grants limited combat access from bases in mainland Japan, not Okinawa. Sorties rise but losses on the ground rise too [1]. China fires missiles at Iwakuni and Misawa, keeping its attacks to US-occupied facilities. |
| D+21 | The PLA holds the southern third of the island, CSIS's "stalemate, trending toward China" pattern [1]. The US is running short of long-range munitions [2]. |
| D+24 | A US task group of cruisers, destroyers, two carriers and submarines approaches Taiwan to break the siege. Chinese submarines, aircraft and missiles badly damage it, as in Ragnarok [1]. Beijing offers "talks without preconditions". |
| D+28 | The PLA takes Taichung and opens a second front at Hualien, the sequence in CSIS's "stands alone" run [1]. Taiwanese forces fall back from river line to river line. |
| D+30 | The UN Security Council is deadlocked by Chinese and Russian vetoes. The General Assembly passes a call for a ceasefire. India, Indonesia and Brazil offer mediation. |
| D+33 | Markets in Japan and Korea fall sharply. With shipping insurance suspended across the region, Japan's economy ministry warns of energy rationing. Tokyo urges Washington privately to accept a ceasefire. |
| D+35 | The PLA reaches Taoyuan and the southern edge of the Taipei basin. Taiwan's casualties approach the scale of CSIS's "stands alone" run [1]. China proposes a ceasefire in place, followed by "consultations on reunification". |
| D+38 | The US President, facing thousands dead, two or more carriers lost and empty magazines, approves a ceasefire through intermediaries. Congress is divided. |
| D+40 | Taipei's cabinet, without power and with no relief in sight, accepts the ceasefire. Fighting stops along the line of contact. |
| D+60 | A framework agreement is reached. Taiwan's armed forces stand down. A "special administrative" arrangement is announced, and PLA garrisons occupy ports and airfields. The US withdraws to its treaty commitments and starts a long rebuild. |
World politics and allies
Japan decides the outcome. CSIS calls it "the linchpin" [1]. The 1960 prior-consultation commitment gives Tokyo a legal veto over US combat sorties from its territory when Japan itself is not under attack [4]. Hornung warns that the process has not been agreed in practice, and that a mismatch between Japan's political timeline and US operational needs could leave the US unable to engage in time [4]. In this scenario China exploits that gap by striking only US facilities and threatening Japanese cities. Japan's 2025 experience shows both its willingness to name Taiwan as a security concern and its readiness to back off under pressure [5].
Australia gives intelligence and some naval and air forces, but at too great a distance to change the result in time. The Philippines grants limited access but bars strike operations from its territory. CSIS rated "Philippines allows basing" as helpful but not decisive [1]. South Korea stays out to guard against North Korea. NATO and the EU condemn the invasion and threaten sanctions, but their first priority is Europe. India calls for restraint and mediates.
Russia backs China diplomatically and blocks action at the UN. In December 2025 Lavrov said Russia would support China if tensions over Taiwan escalated [9]. North Korea carries out missile tests, which keep US and South Korean forces in place. Iran and others exploit the US distraction. Much of the Global South treats the war as a civil conflict and calls only for a ceasefire. The UN Security Council is paralysed; the General Assembly's ceasefire resolution gives the diplomatic cover that ends the war.
How China wins
China wins by being in possession when the world tires of the war. It does not need to destroy the US military. The end state is a ceasefire in place after Taiwan's government accepts terms: demobilisation, PLA garrisons at ports and airfields, and a promise of "consultations" on Taiwan's status.
The US accepts this for five reasons:
- It arrived too late. Without timely use of Japan, US fighter aircraft could not shape the battle. Bombers alone, short of anti-ship missiles, could not stop the build-up [1].
- Its attempt to relieve Taiwan failed. It lost one or more carrier groups trying to break through [1].
- Its magazines were empty. CSIS warned that stocks of long-range precision munitions would run out in under a week and that the industrial base cannot be revived quickly [2].
- Its allies wanted a ceasefire. Japan wanted to end the threat to its cities; Europe and Asia wanted shipping back.
- Taipei consented. Fighting on would mean a war of liberation against a garrisoned island. The CSIS authors noted the same thing after a successful defence: the hard part is "finding an acceptable off-ramp" [1].
The shift in world standing is the real victory for China. US alliances in Asia lose credibility, Japan and Korea debate nuclear options, and China becomes the leading power in the Western Pacific.
Losses
| Side | Personnel | Ships | Aircraft | Other | Basis |
|---|---|---|---|---|---|
| United States | 5,000-10,000 casualties, perhaps 2,500-4,500 killed | 2-4 carriers; 15-40 cruisers and destroyers; up to 15 SSNs | 300-500 combat aircraft, mostly destroyed on the ground | Kadena and Andersen badly damaged; LRASM inventory exhausted | CSIS base case about 6,960 casualties and about 3,200 killed in 3 weeks; pessimistic average 484 aircraft and 14 ships; Ragnarok 4 carriers, 43 cruisers and destroyers, 15 SSNs [1] |
| China | 30,000-60,000 casualties (ground combat and at sea) | 100-130 ships, most of them amphibious | 250-350 aircraft | Amphibious lift largely spent but replaced by captured ports | Pessimistic averages 113-120 ships, 327 aircraft, 17,000 ground casualties; "stands alone" 70,000 ground casualties [1] |
| Taiwan | 40,000-85,000 army casualties, of which 15,000-28,000 killed; heavy civilian losses | Entire navy (26 major ships) | Most of the air force | Power grid and ports destroyed | Pessimistic average 22,000; "stands alone" 85,000 casualties, about 28,000 killed [1] |
| Japan | Hundreds to low thousands, mainly at US-shared bases | 0-14 | 0-160 | Okinawa bases damaged | Pessimistic average 161 aircraft and 14 ships if Japan fights [1]; lower here because the JSDF stays defensive |
All figures are ranges from wargame iterations. They are not forecasts.
What would change the outcome
- Early Japanese access. If Tokyo approves US combat operations at the outset, CSIS's results move strongly back toward Chinese defeat [1][4]. Agreeing the prior-consultation process in advance is the single most important fix.
- Using the warning. Mobilising and dispersing forces on D-day, rather than D+14, and keeping carriers outside China's missile range rather than as a forward show of force [1].
- Missile stocks. More LRASM-class missiles, and proven anti-ship capability for JASSM-ER, would allow the amphibious fleet to be destroyed at standoff range [1][2].
- Hardened and dispersed bases. Aircraft shelters and access to civilian airfields in Japan would cut ground losses [1].
- Taiwan's resilience. Rapid reserve mobilisation, protected energy supplies and a political commitment not to capitulate deny China the fait accompli. CSIS's "stands alone" run shows Taiwan can buy months [1][7].
References
- Mark F. Cancian, Matthew Cancian, Eric Heginbotham, The First Battle of the Next War: Wargaming a Chinese Invasion of Taiwan, CSIS, 9 January 2023. https://www.csis.org/analysis/first-battle-next-war-wargaming-chinese-invasion-taiwan (PDF: https://csis-website-prod.s3.amazonaws.com/s3fs-public/publication/230109_Cancian_FirstBattle_NextWar.pdf). Figures are for a notional 2026 war.
- Seth G. Jones, Empty Bins in a Wartime Environment: The Challenge to the U.S. Defense Industrial Base, CSIS, 23 January 2023. https://www.csis.org/analysis/empty-bins-wartime-environment-challenge-us-defense-industrial-base
- US Department of War (Defense), Military and Security Developments Involving the People's Republic of China 2025, released 23 December 2025 (USNI copy: https://news.usni.org/2025/12/24/pentagon-annual-report-on-chinese-military-and-security-developments-2); quoted in Lee Ferran, "China military buildup leaves US 'increasingly vulnerable': Pentagon report", Breaking Defense, 24 December 2025. https://breakingdefense.com/2025/12/china-military-buildup-leaves-us-increasingly-vulnerable-pentagon-report/
- Jeffrey W. Hornung (RAND), "Taiwan and Six Potential New Year's Resolutions for the U.S.-Japanese Alliance", War on the Rocks, 5 January 2022. https://warontherocks.com/2022/01/taiwan-and-six-potential-new-years-resolutions-for-the-u-s-japanese-alliance/
- Associated Press, "Japan and China trade barbs after PM Takaichi says an attack on Taiwan would be an existential threat", PBS NewsHour, 10 November 2025. https://www.pbs.org/newshour/world/japan-and-china-trade-barbs-after-pm-takaichi-says-an-attack-on-taiwan-would-be-an-existential-threat
- AFP report on the "Justice Mission 2025" drills of 29-30 December 2025, The Defense Post, 30 December 2025. https://thedefensepost.com/2025/12/30/china-rockets-drills-taiwan/
- Mark F. Cancian, Matthew F. Cancian, Eric Heginbotham, Lights Out? Wargaming a Chinese Blockade of Taiwan, CSIS, 31 July 2025. https://www.csis.org/analysis/lights-out-wargaming-chinese-blockade-taiwan
- AFP, "China Says Ends Two-Day Taiwan Drills With Simulated Strikes on Key Ports, Energy Sites" (Strait Thunder-2025A, 1-2 April 2025), The Defense Post, 3 April 2025. https://thedefensepost.com/2025/04/03/china-taiwan-drills-ports-energy/
- "Russia to support China in case tensions escalate in Taiwan Strait — top diplomat", TASS, 28 December 2025. https://tass.com/politics/2065867
- "China Halts Japanese Seafood Imports Amid Diplomatic Row", Caixin Global, 20 November 2025. https://www.caixinglobal.com/2025-11-20/china-halts-japanese-seafood-imports-amid-diplomatic-row-102384739.html
← The Quiet Strangulation: A Coast-Guard Quarantine of Taiwan The Ayungin Withdrawal: a Second Thomas Shoal clash that the United States chose not to widen →