When Taiwan begins Han Kuang 42 on August 5, the most revealing exercise may not involve a missile, ship or drone. During the wider civil-defense drills, mobile internet will be deliberately slowed across 14 cities and counties in northern and central Taiwan, affecting areas that contain roughly 70 percent of the population. High-bandwidth services will become largely unavailable for 30-minute periods while voice and text remain accessible.

It is a useful test, but it also exposes a weakness in the way resilience is commonly purchased.

Governments are good at buying visible assets. They can count vehicles, radios, sensors, servers and drones. They are less disciplined about buying the capability that connects those assets into a decision when information is incomplete, networks are degraded and several agencies are working from different versions of reality.

That distinction matters because Taiwan’s defense and resilience debate is entering a capital-intensive phase. The government has set a target of raising defense spending to 5 percent of GDP by 2030. The Legislature has spent months debating the scale and composition of a special defense budget, while the administration is also advancing major plans for drones and domestic defense production. The political argument is understandably focused on how much Taiwan should spend and which platforms it should acquire. The harder question is how procurement will determine whether those platforms can still produce coordinated action under pressure.

Taiwan’s hardware strength can become a systems-integration weakness if procurement stops at delivery. The island’s industrial model is exceptionally good at manufacturing reliable components, scaling production and improving physical systems. Crisis response creates a different source of value. Sensors owned by the military, coast guard, police, emergency services, utilities and local governments must exchange information, reconcile contradictions and support a common course of action. Foreign Minister Lin Chia-lung recently identified hardware-software integration as an area where Taiwan has historically been weaker. National resilience is where that gap becomes operational rather than theoretical.

The answer is not another centralized dashboard. A screen that collects more feeds can increase cognitive load without improving judgment. Nor is the answer to place an opaque AI model in the command chain and hope it produces the correct recommendation. Taiwan should define a decision layer as a procurement category with measurable requirements, just as it defines requirements for range, endurance, payload and cybersecurity.

First, systems in the civil-defense decision path should be required to operate when disconnected from the public cloud. A blockade, cyberattack or cable failure will not arrive in a convenient 30-minute interval. In 2023, two undersea cables serving the Matsu Islands were severed, forcing residents and institutions onto limited backup capacity. The cables had reportedly been cut 27 times in five years. Taiwan has already experienced the difference between normal connectivity and constrained connectivity. Procurement should assume that constraint is the operating environment, not an edge case.

Second, interoperability should be tested, not merely promised. Vendors should have to demonstrate that their systems can exchange data through documented interfaces, preserve provenance and continue functioning when one node or supplier is removed. A resilience architecture built around proprietary lock-in is brittle by design. Open interfaces do not require open access to sensitive data. They require the government to control the standards by which authorized systems communicate.

Third, every significant recommendation should be traceable. An operator should be able to see which sources were used, which assumptions were made, how confident the system is and what changed between two versions of a recommendation. That is not a cosmetic explainability feature. In a democracy, officials may later need to defend evacuation orders, resource allocation or emergency restrictions before elected representatives, auditors or courts. A decision that cannot be reconstructed is a governance failure even when it happens to be correct.

Fourth, human authority must be explicit. Decision-support software should narrow options, identify conflicts and accelerate analysis, but it should not blur accountability. Contracts should specify which actions require human approval, how an operator can modify or reject a recommendation, and how that intervention is recorded. The objective is not autonomous government. It is faster human judgment with a defensible record.

Fifth, data residency and supply-chain provenance should be treated as core resilience requirements. Taiwan should know where sensitive data is stored, whether a system can operate without phoning home, who maintains the underlying models and which jurisdictions can compel access to the technology. This standard should apply to domestic and foreign suppliers alike. Sovereignty is not achieved by placing a Taiwan label on a system whose critical dependencies remain outside Taiwan’s control.

These requirements will remain abstract unless Taiwan changes how contracts are accepted and paid. Han Kuang and the urban resilience exercises should become operational acceptance environments for digital systems, not only demonstrations for people and equipment. A contract should measure how quickly agencies restore a common operating picture after bandwidth loss, how much data remains attributable to its source, whether manual override works, whether local governments can continue independently, and how rapidly the system recovers after a node is removed. Vendors should be paid for verified mission continuity, not for installing software that performs well in a conference room.

This would also improve Taiwan’s industrial policy. A government-owned reference architecture, a conformance-testing program and controlled interoperability sandboxes would allow smaller Taiwanese companies to compete at defined layers instead of being excluded by large, closed platforms. Prime contractors could still integrate complex systems, but they would have to expose the interfaces needed for modules to be replaced, upgraded or locally maintained. Foreign suppliers would remain welcome, but they would contribute to a sovereign architecture rather than becoming the architecture.

The economic opportunity is larger than domestic procurement. Democracies across the Indo-Pacific and Europe are preparing for cyberattacks, natural disasters, infrastructure sabotage and gray-zone coercion. Taiwan already produces many of the physical building blocks they need: semiconductors, rugged computing, communications equipment, drones and sensors. If Taiwan also develops trusted standards for disconnected decision-making, it can export integrated resilience systems rather than capturing value only at the component level.

That is the strategic prize. The same discipline that made Taiwan indispensable in advanced hardware can be applied to the layer that turns hardware into coordinated action. Doing so would strengthen deterrence, deepen the domestic software and systems-integration sector, and give Taiwan another category of technology that partners have a direct interest in adopting and sustaining.

My company develops decision-support and data-fusion software, including systems intended for disconnected and high-accountability environments. That gives me a commercial interest in this subject, and readers should weigh it. It does not change the procurement principle. No vendor, including mine, should be trusted on claims alone. The requirements should be public, the interfaces should be testable and performance should be proven under the same degraded conditions Taiwan expects its people to face.

The August drills will ask whether Taiwan can keep communicating when bandwidth collapses. Procurement should ask the next question: can institutions still reach a fast, consistent and accountable decision? If the contract measures only the number of sensors delivered, it is measuring the input. Resilience depends on the decision that follows.

(This piece reflects the author's opinion, and does not represent the opinion of CommonWealth Magazine.)

CommonWealth Magazine welcomes op-ed submissions. Please send your article proposals to [email protected]

About the author:

Al Vigier is the founder and CEO of Caseway, a sovereign Canadian decision-support software company, and a Canadian Army veteran.