Li Qize's Perspective: Who Bears the Risk of Retreating to the Second Island Chain?

Jennifer Cavana, a senior researcher and military analysis director at the American think tank 'Defense Priorities,' released the 'Asian Second Island Chain Strategy' on September 10, proposing to shift the focus of the U.S. military from the first island chain to Guam, the Northern Mariana Islands, Palau, the Federated States of Micronesia, and the Marshall Islands. Simultaneously, it advocates for gradually ending or significantly reducing joint defense commitments with Japan, South Korea, the Philippines, and Australia, and shifting from strategic ambiguity to 'clear non-intervention' on Taiwan.

This report has sparked intense debate within the Asian security community. Some Chinese commentators have interpreted it as 'a proposal for peace being attacked by the West,' but in reality, the main criticisms come from scholars, analysts, and media commentators, with no evidence that governments of Japan, South Korea, and Australia have 'attacked' in unison.

Cavana's report forces us to confront a fundamental question: For which Asian interests is the U.S. willing to bear the risk of war with a nuclear power? However, when evaluating this proposal, we should also ask: Does it reduce the risk of war in Asia, or does it reduce the risk of the U.S. being drawn into an Asian war?

The report argues that the retreat is not a complete withdrawal but a proposal to shift the focus of the U.S. military to the second island chain. Cavana points out that China's long-range missiles, unmanned systems, and intelligence, surveillance, and reconnaissance capabilities are continuously improving, making large fixed bases like Okinawa increasingly vulnerable. She advocates for increasing submarines and unmanned systems, dispersing and hardening bases, reducing the high-value target characteristics of frontline troops, and requiring wealthy allies to bear more self-defense responsibilities, all of which have sound military reasons.

However, the report ties together three policies of different natures: military redeployment, which is about troop survival and operational efficiency; ending alliances, which is about rewriting political commitments and regional order; and announcing non-intervention in Taiwan, which directly changes Beijing and Taipei's cost calculations for future crises. The first has military rationality, but it does not automatically prove that the latter two are equally reasonable.

The U.S.-Japan, U.S.-South Korea, U.S.-Philippines security treaties, and the U.S.-Australia-New Zealand Security Treaty (ANZUS) all require the contracting parties to take action in response to armed attacks within the agreed scope, but they do not predetermine the form of action and cannot bypass the constitutional authority of the U.S. president and Congress. Taiwan is not directly protected territory under these treaties, and a conflict in the Taiwan Strait itself will not automatically trigger the four treaties. The real risk of entanglement comes from the use of allied bases, attacks on U.S. or allied forces, and the gradual escalation of the crisis.

The Taiwan Relations Act is not a mutual defense treaty. It lists the use of non-peaceful means to determine Taiwan's future, including boycotts and embargoes, as a threat to peace and security in the Western Pacific and a serious concern for the U.S., and stipulates that the U.S. should provide Taiwan with defensive military equipment and services necessary for self-defense, and maintain the ability to resist the use of force or other coercion, but it does not promise automatic dispatch of troops.

So-called strategic ambiguity is to maintain decision-making space between 'reserving the ability to intervene' and 'not promising certain intervention.' Cavana's proposed 'clear non-intervention' is not just making ambiguity clear but preemptively closing a choice that the law and policy originally reserved.

The second island chain can indeed increase early warning and response depth, but distance is also a cost. Guam is about 2,700 to 2,800 kilometers away from Taiwan, which will significantly increase the burden of air refueling, ammunition replenishment, maintenance, sea escort, and sortie cycles for most tactical aviation missions. If the first island chain's sensing, airport, and port nodes are lost in the early stages of a crisis, the U.S. military may still have the ability to intervene but may lose the speed of timely intervention.

The second island chain is not a safe area outside of China's firepower. The Pentagon's 2025 report on China's military power lists the range of the Dongfeng-26 missile as 3,000 to 4,000 kilometers. Combined with the geographical distance, Guam is within its possible strike range. Longer-range missiles, bombers, submarines, and cyberattacks also make the second island chain not an absolute rear area.

The U.S. Government Accountability Office's 2025 report points out that Guam's integrated air defense and anti-missile system planning is distributed in 16 locations, with phased deployment scheduled from 2027 to 2032, but operational and maintenance responsibilities, cross-service support, personnel needs, and deployment schedules are not yet fully determined.

Cavana is not unaware of these problems. She advocates for dispersing forces to Guam, Tinian Island, Palau, the Federated States of Micronesia, and the Marshall Islands, rather than concentrating them all in Guam. The real question is: Can these islands obtain sufficient fuel, ammunition, maintenance, runways, air defense, and political access? If execution still heavily relies on a few mature hubs like Guam and Australia, frontline vulnerability may turn into concentrated risk in the rear.

The report acknowledges that some host countries may refuse access and that the existing forces in the second island chain and Australia are not sufficient to respond to major threats to U.S. trade on their own. This illustrates the dilemma of offshore balancing: it seeks to reduce commitments but still needs the trust, bases, and intelligence networks generated by commitments.

The report believes that China has the ability to interfere with Asian shipping lanes but has never completely cut off U.S. trade. China relies on foreign trade, and its ability and incentive to conquer Japan, South Korea, or the Philippines are very limited. Therefore, the U.S. does not need to maintain the current alliance for the worst-case scenario.

It is indeed an important correction to the cheap 'domino theory' that Beijing is not likely to conquer the entire Asia. However, the fact that Beijing has not yet completely blocked the shipping lanes is happening under the condition that the U.S. military is still present in the front line and the alliance is still functioning. This observation cannot directly prove that Beijing will make the same choice after the U.S. military withdraws and the alliance collapses.

Of course, we cannot reverse this and say that Beijing's restraint is completely caused by U.S. military deterrence. The problem is that the current experience is not sufficient to infer the results after the removal of deterrence. Cavana's report carefully observes 'things that have not happened under the current situation' but does not sufficiently address the counterfactual gap of 'what may happen after changing the situation.'

This is especially true for the Taiwan Strait. Clear non-intervention may reduce Beijing's anxiety about the sudden intervention of the U.S. military and may also reduce the moral risk of Taiwan's misjudgment that the U.S. will definitely rescue. However, it may also reduce Beijing's expected cost of using coast guard inspections, isolation, local blockades, cyberattacks, legal warfare, and economic coercion. These measures may not immediately trigger a large-scale war but may gradually change the status quo.

Cavana believes that after the U.S. withdraws, Taiwan may be more willing to reach a political compromise with Beijing, and this is a result that is 'manageable' for U.S. interests. This statement honestly presents the calculation unit of the report: it concerns