Taiwan’s Security Beyond the Cross-Strait Force Balance

Opening

Welcome to today's China report.

This week, rather than skip publication, we widened our usual window to reports published within the past sixty days. Only two qualified, which is fewer than usual, and that quiet patch in the literature is worth saying out loud. Both reports sit just outside a conventional cross-strait force-balance audit. One is about Taiwan’s institutional and social resilience. The other is about the legal, economic, and regional machinery around deterrence. They earn their place because ships and missiles matter, but so do the systems that have to keep working after the first shock.

GMF on Taiwan’s resilience

The German Marshall Fund’s Next-Generation Perspectives on Taiwan is an eleven-essay collection by early-career American and European researchers after a study week in Taiwan. Its big claim is that Taiwan’s security can’t be read off an order of battle. The People’s Liberation Army, or PLA, can attack confidence, communications, supply chains, and political cohesion before it fires a shot. So Taiwan needs armed forces that can act when headquarters goes quiet, civilians who know their jobs, digital rules people trust, and partners that can keep drones, chips, finance, and diplomacy moving.

The collection gets there by stacking mechanisms. Cognitive operations exploit rigid command culture and personal vulnerabilities inside the military. Cyberattacks prepare the ground for physical strikes. A state that leaves civic groups outside official planning wastes local skill. Drone partnerships spread production and complicate a blockade. And European coordination raises the economic cost Beijing would expect to pay. Put those pieces together, the authors say, and resilience denies the PLA a quick political collapse even if Taiwan starts with fewer platforms.

What does that rest on? It’s a collage, not one research design. One essay cites Taiwan’s National Security Bureau counting more than nineteen thousand influence messages from seven hundred and ninety-nine abnormal accounts during a January drill. Another uses the bureau’s figure of roughly two point six million attempted cyber intrusions a day in twenty twenty-five. Those are useful measures of activity. They aren’t measures of persuasion, access, or military failure. The cognitive-warfare chapter also draws on one joint interview with three researchers at Taiwan’s government-linked defense institute. Its strongest causal support comes from a separate experiment with roughly thirty thousand participants showing that advance warnings can reduce belief in false claims.

The administrative evidence is sharper. By March twenty twenty-six, more than one hundred thousand people had received resilience training against a target of four hundred thousand. But the collection doesn’t show what those trainees could do under pressure. And its drone chapters supply an awkward scale check: Taiwan aims for annual capacity of one hundred and eighty thousand drones by twenty twenty-eight, while Ukraine was using about two hundred and seventy thousand a month in early twenty twenty-six. The same chapters say Taiwanese producers still depend on Chinese batteries, magnets, and other upstream inputs.

I checked the twenty twenty-three China Security Report from Japan’s defense-ministry National Institute for Defense Studies. It also documents PLA influence activity, but it explicitly can’t tell whether a large fall in Taiwanese confidence in American intervention was caused by Chinese messaging. That’s the missing variable here: effect.

The report does give us clocks. Taiwan’s Artificial Intelligence Basic Act has a two-year implementation window from January twenty twenty-six, and the drone plan points to twenty twenty-eight. No enforceable privacy and risk rules by then, or production that misses the target and remains tied to Chinese inputs, would cut against the collection’s policy confidence.

Here’s the internal challenge. One essay calls Taiwan’s whole-of-society resilience committee perfectly positioned to test economic scenarios. Another says that committee had met only seven times and still lacked permanent operating authority. Both can’t carry the same reassuring implication. The collection is right that resilience has to be built before a crisis. But until exercises publish restoration times, degraded-command performance, and verified civilian tasks, it demonstrates an agenda, not resilience achieved.

NBR on middle-power statecraft

The National Bureau of Asian Research’s Statecraft at the Crossroads of the Indo-Pacific is another roundtable, developed from a conference co-sponsored by Taiwan’s foreign ministry and Canadian and Taiwanese institutions. It isn’t a force ledger either. Its claim is that framing Taiwan only as a contest between Washington and Beijing hides the agency of Taiwan, Japan, Southeast Asian states, and other middle powers.

Its chapters range widely. One maps the motives behind past American interventions. One goes back to the original United Nations debate to argue that Resolution twenty-seven fifty-eight settled who represented China, not sovereignty over Taiwan. Others look at damaged undersea cables, Japanese security policy, small-group defense arrangements, semiconductor decisions, and a proposed economic version of the North Atlantic Treaty Organization that could answer coercion quickly without becoming a formal alliance.

The logic is modular deterrence. Spread useful capacity across many governments and firms. Agree in advance how to attribute cable sabotage, repair damage, share intelligence, redirect trade, and impose costs. Then Beijing can’t disable the whole response by pressuring one capital. That’s plausible. But it requires those actors to see the same incident the same way and to accept costs at the same time. The report mostly shows that coordination forums exist, not that this political conversion has happened.

The evidence varies a lot. The legal chapter’s close reading of United Nations transcripts and the two votes in nineteen seventy-one is strong evidence for what the resolution did and didn’t say. The semiconductor chapter is the rare original dataset here: sixteen survey rounds from twenty eighteen through twenty twenty-five, usually forty to sixty managers drawn from about two hundred listed Taiwanese firms, with stratification and weighting. In twenty twenty-five, about eighty percent considered Taiwan for new investment, while interest in China rebounded to twenty-five percent and the United States reached roughly fifteen percent. That supports selective diversification, not a clean economic split.

The cable chapter uses a real stress test. An earthquake in two thousand six cut eight cables, slowed some regional connections by more than ninety percent, and left repairs running for forty-nine days. That proves vulnerability. It doesn’t prove deliberate cable cutting would deter or unite a coalition. More seriously, the chapter says a simultaneous blockade and cable attack could immediately erase two to five percent of global gross domestic product. Its note points instead to an estimate of roughly ten percent for a full Taiwan war. The number printed in the claim isn’t established by the source beneath it.

The report’s timed forecasts come mainly from the business survey. Roughly a third of semiconductor respondents late in twenty twenty-five expected American protectionism to remain structural beyond twenty twenty-eight. A TrendForce estimate cited by the report puts China near half of mature-node chip capacity by twenty thirty. Later survey rounds and actual capacity can catch those out. The roundtable’s bigger forecast, that networked middle powers will strengthen deterrence, has no deadline or agreed outcome measure. As written, it can’t really fail.

And regional agency may not mean coalition-building. I checked the twenty twenty-six State of Southeast Asia survey from Singapore’s Institute of Southeast Asian Studies, now named the ISEAS–Yusof Ishak Institute, whose main general funding comes from Singapore’s education ministry. Among two thousand and eight regional opinion leaders, just eighteen point two percent chose more minilateral groupings as the best way to strengthen the Association of Southeast Asian Nations. More preferred fixing internal divisions or expanding trade within the region. It’s a purposive online survey, not a public-opinion poll, but it directly complicates the assumption that giving middle powers agency points them toward Taiwan-linked deterrence.

That’s the report’s real weakness. It’s very good at identifying tools. It’s much thinner on coalition consent and behavioral effect. Evidence that would settle this is concrete: standing cable-repair commitments, pre-agreed attribution rules, funded joint logistics, and a measurable change in PLA gray-zone behavior after those arrangements begin. Until then, “collective deterrence” names the hoped-for result.

What to watch

The nearer test is implementation, not another headline weapons count: Taiwan’s two-year digital rulemaking, published results from resilience exercises, and funded drone output with its upstream dependencies exposed. After that, watch whether regional governments turn cable security and economic coordination into standing commitments; if they keep choosing flexibility over alignment, the second report’s central mechanism stays on paper.