What a Shooting War with China Would Look Like in 2030

The premise of this article is that the United States and the People’s Republic of China are at war in 2030. Not a trade war, not a technology contest, not a proxy conflict: a shooting war, with American and Chinese forces engaging each other directly in the western Pacific, and with nuclear weapons on both sides operational enough to shape the outcome. The premise is counterfactual, but it is not exotic. Every serious wargame of the last decade has run this scenario, and the analytic community has converged on expectations specific enough to summarize. This article assembles them: how such a war would begin, what the first weeks would look like, where it would break, and how it would end.

I. Why 2030 Is the Year Everyone Wargames

The scenario year is not arbitrary. Three trajectories converge on 2030. First, the nuclear balance: the Pentagon’s annual China Military Power Report estimates that China passed 600 operational warheads in 2024 and will exceed 1,000 by 2030, with a launch-on-warning posture arriving “this decade” alongside lower-yield options.¹² Second, the alliance balance: by 2030 Japan’s 2022 decision to spend 2 percent of GDP and field counterstrike missiles will be fully implemented, Australian nuclear submarines will be rotating through Western Australia, and the Philippine basing network will be complete.³⁴⁵ Third, the industrial balance: the American munitions surge that began in 2025, with Tomahawk contracts above a thousand a year, SM-6 output above five hundred, and several munitions families quadrupled, will have had four years to rebuild magazines that a 2026 war in the Middle East drew down.⁶⁷⁸

The result is a window in which China’s arsenal is large enough to complicate American deterrence without being large enough to threaten disarmament, and in which the United States has rebuilt some, but not all, of the capacity its own wargames say it needs. That is precisely the configuration strategists fear most: neither side clearly dominant, both sides believing the other might blink, and time running out.⁹

II. The Pathways In: How a Shooting War Starts

A war in 2030 does not begin with a declaration. It begins with a move below the threshold of invasion, most plausibly a blockade: the People’s Liberation Army Navy sealing the sea lanes around Taiwan, boarding or sinking neutral shipping, and establishing a quarantine that the United States either accepts, which ends the Taiwan relationship, or contests, which starts the war.

Fiona Cunningham’s Under the Nuclear Shadow shows why this path is the one to watch. China, unable to match the United States conventionally and confining its nuclear weapons to a high-threshold deterrent it treats as “blunt and ineffective instruments” of coercion, spent two decades substituting precision missiles, offensive cyber, and counterspace for escalation dominance.¹⁰ A blockade weaponizes exactly that substitution. It is deniable at first, and it is coercive only if the United States believes it cannot be broken cheaply.

Escalation theory supplies the mechanism by which the blockade turns into a shooting war. Thomas Schelling’s central insight was the “threat that leaves something to chance”: deterrence and compellence work because each side manipulates a shared risk of catastrophe that neither fully controls.¹¹ Glenn Snyder’s stability-instability paradox adds the darker corollary, that a stable nuclear balance may actually embolden conventional aggression, because each side assumes the other will not escalate.¹² Wargame participants consistently show this: in the CSIS/MIT nuclear-enabled games, the teams that escalated did so not from confidence but from desperation, and the teams that started wars did so believing they could be kept local.¹³

The historical precedent is the 1958 Quemoy crisis, when the United States privately threatened nuclear use over offshore islands and the shelling abated by October.¹⁴ That crisis is usually read as proof that nuclear deterrence works at the edge. The sobering alternate reading is that it worked because the credibility of the threat was never actually tested. In 2030, with Chinese conventional forces vastly stronger than in 1958 and a nuclear arsenal China did not possess then, a replay of that brinkmanship runs on compressed timelines.

III. The Conventional Opening: Geography, First Strikes, and the First Two Weeks

The single most important fact about a 2030 war is geography. RAND’s U.S.-China military scorecard established the baseline: China’s power atrophies quickly beyond a few hundred kilometers from its coast, and the United States retains the ability to decisively defeat Chinese action at distances beyond a thousand kilometers.¹⁵ But Taiwan sits roughly 160 kilometers off the Chinese coast, inside the envelope where the PLA is strongest, and the American bases that could intercede, Guam and the Japanese archipelago, sit inside the range of China’s DF-26 “Guam Express” ballistic missiles.¹⁶¹⁵ Okinawa alone hosts roughly two-thirds of the American personnel stationed in Japan, concentrated on a handful of bases that no dispersal plan can move.¹⁷

This asymmetry produced the most consistent finding in all the wargames. In CSIS’s 24-iteration “First Battle of the Next War,” the defense of Taiwan succeeded in most runs, but in every base-scenario iteration the United States lost two aircraft carriers and between 7 and 20 cruisers and destroyers, roughly 270 aircraft (168 to 372 across runs), and an average of about 7,000 personnel including 3,200 killed.⁹¹⁸ Japan lost another 112 to 122 aircraft. Up to 90 percent of the aircraft losses came on the ground, at bases the United States could not move and could not sufficiently defend.¹⁸ In the 18 pessimistic iterations, with a degraded American force and reduced allied participation, China still never achieved a clear victory; 14 runs ended in stalemate and 4 in Chinese defeat. Only in the extreme variants, Taiwan fighting alone or an all-out exchange from the start, did the invasion succeed.¹⁸

The 2024 nuclear-enabled follow-up, Confronting Armageddon, produced the same opening: by the end of week one, the United States and Japan had typically lost 270 aircraft and 20 ships including two carriers; at least one carrier was lost in week one in every one of the 15 runs, and a second in 9 of them.¹³ The lesson both games teach is that the first two weeks are survivable for the defense but catastrophic in consumption. The question is not whether the fleet survives week one. It is whether anything is left for week five.

IV. The Consumption Problem: Munitions and the Length of War

Here is the number that should govern all sober thinking about 2030: the entire global inventory of LRASM anti-ship missiles, roughly 450, was expended within the first week of every CSIS iteration, and the joint American stockpile of long-range precision munitions, about 5,000 JASSM and LRASM-family weapons, was exhausted within three to four weeks.¹⁸¹³ The campaigns these games describe are not decided by tactics. They are decided by magazine depth, and every wargame says the deeper magazine wins the fourth week.

The 2030 American position is better than the 2022 baseline because the surge happened. Scranton Army Ammunition Plant more than doubled its 155mm output, from 14,000 to 36,000 shells a month, on the way toward a 100,000-shell goal.¹⁹ Tomahawk production, which had fallen below a hundred a year in FY2018-19, is now contracted above a thousand a year, with deliveries tripled between the first halves of 2025 and 2026.⁷ SM-6 is above 500 a year, against a Navy plan for 1,800 in the fleet,²⁰ and AMRAAM is running at 1,900 a year;⁶ PAC-3 MSE is at 550 and rising.²¹ The FY2027 budget request moves total defense resources to roughly $1.5 trillion, a 28 percent discretionary increase, against a Chinese budget of about $277 billion that the Pentagon itself believes is understated by a third to two-thirds.²²²³

Yet the arithmetic of a Pacific war is unforgiving. China fields an estimated 3,500 to 3,600 conventional ballistic and cruise missiles across its rocket force, including 1,300 medium-range ballistic missiles and 500 intermediate-range missiles aimed at the American logistics chain, and its production lines run in a closed economy where the state sets the target.²⁴ The United States enters a Pacific war with bases thousands of miles from the fight, a sealift fleet that has atrophied, and an industrial base that, even at the new rates, takes years to replace a week of combat. The wargames’ consensus judgment is blunt: none ended with the United States running out of will, but several came close to running out of weapons, and the difference between winning and losing was measured by whether resupply reached the theater before the magazines ran dry.¹⁸¹³

V. The Economic Dimension: Taiwan, Chips, and the Global Nervous System

A shooting war in 2030 would detonate the global economy before the first missile exchange, because Taiwan is not a strategic island: it is a factory. Taiwan produces more than 90 percent of the world’s most advanced semiconductors and 92 percent of advanced logic capacity below 10 nanometers, and TSMC alone commands roughly 70 percent of global foundry revenue.²⁵ The 2025-2026 artificial intelligence buildout, the largest capital surge in the American economy, runs on exactly these chips. A blockade of Taiwan is simultaneously a siege of Hsinchu and an embargo on the world’s computing supply chain. Economic warfare would be an instrument of war from hour zero, not a follow-up.

China’s vulnerabilities are symmetric and deep. Roughly 80 percent of its imported crude oil transits the Strait of Malacca, a chokepoint in somebody else’s waters, and a third of global LNG trade passes through the South China Sea.²⁶²⁷ Beijing’s response has been strategic stockpiles and the Belt and Road alternative routes, but neither replaces the sea lanes within the window of a war. Against that, China holds the rare-earth counter: roughly 69 percent of global mine output and about 90 percent of refined processing, with American dependence at 80 to 100 percent for key magnets and metals, mitigated only partially by a January 2027 defense procurement ban and the slow reopening of Mountain Pass.²⁸

The net effect is mutual strangulation. Each side can cut the other’s critical inputs, and neither can strike without being struck in return. This is what great-power war looks like inside an integrated economy: a reciprocal seizure of the other’s Achilles’ heel, with the collateral damage landing not only on the belligerents but on the global system both sides built. The dollar remains the reserve currency and the payments system both sides use, which gives the United States a financial weapon and China a motive to make de-dollarization a war aim. Neither outcome is stabilizing.

VI. The Escalation Question: From Conventional to Nuclear

The most carefully quantified literature concerns the point at which a conventional war goes nuclear. Between the arsenal counts and the wargame runs, the expected shape is now fairly clear.

The arsenal context: the United States fields about 1,770 deployed warheads of roughly 5,000 total; China fields about 620 warheads total by the Federation of American Scientists’ count, of which only about 34 are deployed, most of the force held in central storage while new silo fields fill.²⁹ Even a 2030 arsenal of 840 to 1,000 warheads, the range the games model, remains roughly a fifth of the American total inventory.²¹³ That asymmetry is why escalation dynamics in the games run the way they do.

In the CSIS/MIT games, 7 of 8 first uses of nuclear weapons came from Chinese teams facing imminent conventional defeat in weeks three to five of the campaign, with the amphibious fleet attrited and logistics below sustainment: “gambling for resurrection” rather than striking for advantage.¹³ The typical Chinese first use was limited, 12 to 30 weapons, targeted at Taiwanese ground forces rather than at American cities or fleets, and the games’ modeling attributed 80,000 to 350,000 Taiwanese civilian deaths to such strikes even with military-only targeting.¹³ The one American first use, a panicked preemptive counterforce strike after week-one losses of 270 aircraft and 20 ships, failed to coerce and nearly guaranteed conflagration. Countervalue strikes, attacks on cities, produced conflagration in every run, with casualties in the hundreds of millions.¹³ Three of the 15 runs ended in full strategic nuclear exchange.¹³ The Center for a New American Security’s earlier game reached an equally stark endpoint: the Chinese team, having set its no-first-use doctrine aside in play, ended the exercise with a high-altitude nuclear detonation near Hawaii, and the authors described the whole dynamic as a “quick race to the bottom.”³⁰

Two structural changes make 2030 worse than the games’ baseline. The first is China’s stated move toward launch-on-warning with lower-yield options, which compresses decision time on both sides and reduces the window for the off-ramps the games found decisive.²¹³ The second is the American homeland’s thin missile defense: 44 ground-based interceptors at two sites, built for a limited attack, with no exo-atmospheric capability against hypersonic gliders, and that limitation matters precisely because systems like the DF-17 hypersonic glider are built to defeat it.³¹³² The replacement Sentinel ICBM program, meanwhile, slipped into the early 2030s after a Nunn-McCurdy critical breach; in 2030 neither side can disarm the other, and both can inflict catastrophic damage.³³ Herman Kahn’s escalation ladder, 44 rungs from subcrisis maneuvering to spasm war, remains the standard map of this territory, and the wargames’ consistent finding is that even well-intentioned players climb it far faster than they expect.³⁴ That is the classic stability-instability world, with one difference: the stability is much more fragile when one side’s arsenal is growing by a hundred weapons a year and its doctrine is shifting toward warning-based launch.

The honest summary from the report itself is that the frequency of nuclear use inside a game cannot be read as a probability of nuclear use in reality.¹³ What the games do establish is the condition: nuclear use becomes plausible exactly when a conventional conflict is being lost, on a timeline measured in weeks, by a leadership facing regime survival. In 2030, that condition is not exotic. It is the expected trajectory of a war that starts with a blockade and runs out of weapons.

VII. The Ends of the War: What Victory Means

The wargames are unified on the terminal condition: there is no total victory. In the CSIS/MIT runs, the outcomes distributed across five end states: Chinese withdrawal under phased attack, a Chinese enclave on Taiwan, full strategic conflagration, a return to the status quo ante, and inconclusive attrition.¹³ The CSIS “First Battle” study reached the same place, a defense that succeeded at costs that damaged the American position for years, and its conclusion became the field’s watchword: “Victory is not enough.”⁹¹⁸ A Taiwan that survives the war but loses its economy, its ports, and a generation of its young people is not a victory in any meaningful sense. An American alliance that holds Taiwan but loses two carriers and its credible capacity to deter the next crisis is not a victory either.

The regime dimension cuts both ways. A failed invasion and occupation would destabilize Communist Party rule in ways the Chinese system is not built to absorb; the American assessment is that Beijing may consider nuclear use precisely when a conventional defeat in Taiwan threatens regime survival.¹¹³ On the American side, a 2030 war fought over a chip foundry would produce a political reckoning about the exposure itself: why 90 percent of the world’s most advanced chips, and the world’s most dangerous maritime chokepoints, were left so concentrated for so long. The aftermath of the war, in every wargame, is a world in which the American position is diminished even when the military outcome is favorable, and the international order is permanently renegotiated.

VIII. What One Can Expect

Assembled, the expectations are these. A 2030 shooting war would begin below the threshold of invasion, most likely as a blockade, and would cross into open combat within days as the United States contested it. The first two weeks would be the costliest naval and air campaign in history, with a carrier or carriers lost, hundreds of aircraft destroyed largely on the ground, and the anti-ship missile inventory exhausted in the first week. The global economy would enter recession within days as the chip supply chain and the sea lanes seized simultaneously. The conventional war would likely be decided, if it is decided at all, in the fourth to sixth week, by magazine depth, industrial throughput, and the willingness of elected governments to absorb thousands of casualties a week for an outcome the strategists themselves describe as not enough. And if the war ran toward a decisive Chinese defeat, the nuclear question would move from academic to operational in the space of weeks, with the balance of evidence saying the first use would be limited, Chinese, and aimed at the battlefield, and the counter-response determining whether the war stopped at an exchange that still killed hundreds of thousands, or ran to a conflagration killing hundreds of millions.

None of this is prophecy. It is the compiled expectation of the people whose profession is thinking about exactly this. The premise that a war is possible in 2030 is the premise of every major wargame, every defense budget, and most of the last decade’s strategic writing. The most defensible expectation is the one the wargame authors reach when their runs are done: a shooting war with China in 2030 is a war neither side can win, in the only sense of winning that would justify it, and the rational work of the next four years is making the scenario one that never has to be run for real.

Notes

PRH | huffmanwrites.org | © Philip Huffman