On September 18, 2026, Premier Cho Jung-tai inaugurated the Taiwan-Matsu No. 4 subsea cable. Focus Taiwan reported that the 294 km link, built by Chunghwa Telecom and laid by Dong Fang Offshore, connects New Taipei's Bali District with Dongyin, Nangan and Juguang townships in Lienchiang County. Cho called it part of the government's "preemptive preparations" for communications resilience. Trade press put combined Taiwan-Matsu capacity at about 1.9 Tbps, though that figure rests on secondary reporting and we could not confirm it independently.
The case for heavy state involvement
The strongest argument for treating Matsu connectivity as a state project is that the market will not build it. A few thousand residents on remote islands cannot generate the revenue to justify a 294 km seabed cable, and a purely commercial operator would rationally under-invest in a fourth route. Resilience is also a public good: the benefits of a cable that rarely carries traffic accrue to hospitals, banks and government offices that cannot easily price that insurance.
The record supports this. According to Focus Taiwan, a Chinese fishing boat and a cargo ship damaged cables No. 2 and No. 3 in February 2023, causing nearly two months of outages. In January 2025, natural deterioration struck both cables in two separate incidents. MODA's January 22, 2025 release said the No. 2 cable failed from aging while No. 3 was already damaged, leaving both main lines inoperable. The ministry told Chunghwa Telecom to activate microwave backup and prioritised government offices, banks and hospitals. It also said the cables had been designated critical infrastructure.
That is a serious failure mode, and the premier acknowledged as much: residents had to complain to the central government before service recovered.
Why this is the proportionate response
The response is notable for what it avoids. MODA did not impose redundancy obligations on private carriers or expand its powers over content or traffic. It funded physical infrastructure. According to MODA's pre-completion release, the project was paid for from the Forward-Looking Infrastructure Development Program budget. Once operating, it gives Taiwan three subsea cables to Matsu, and each of the four townships gets at least two microwave systems for outside connectivity.
This is the pro-innovation way to buy resilience. Capital is directed at a documented market failure, the asset sits with an existing licensed operator, and no new compliance burden lands on smaller providers or on users. It also avoids a trap common in infrastructure policy, where governments respond to outages with sweeping security mandates that raise costs for everyone and do little for remote islands.
MODA also announced a 5G smart-island rollout for Matsu with three telecom operators on the same day. Subsidising coverage in places where commercial returns are thin follows the same logic. The ministry has said it continues to subsidise Taiwan's three major operators to extend mobile coverage and redundancy to rural and outlying areas.
Where the risk remains
The weak point is the assumption that more routes mean more resilience. Cables are only independent if they do not share failure modes. Cables landing in the same stretch of seabed, or relying on the same repair vessels, can be cut by the same event. The 2023 incident hit two cables at once. Taiwan's policy should therefore be judged on route diversity and repair readiness, not on the cable count.
MODA's own release shows a second vulnerability: the project was slowed because the Middle East conflict delayed raw-material shipments. Repair capacity is also supply-chain dependent. A fourth cable is worth much less if a cut cannot be fixed within days. Taiwan should publish repair-time targets, maintain standby cable stock and keep a regional repair-ship arrangement ready, rather than only announcing capacity.
There is also a transparency gap. Neither the primary sources we could verify nor the press coverage explains how the new cable's route differs from the existing ones, or how quickly failover has been tested. The microwave and satellite layers the premier cited are real backups, but the premier said capacity would "far exceed demand" without publishing any test data. Policy that depends on backup should publish the results of failover drills, as is already normal for power grids.
What good looks like
First, MODA should report failover test results for Matsu at least annually, covering time to restore, traffic carried on microwave and satellite, and the services given priority. Second, it should disclose the physical separation of the four cables' routes so independent engineers can judge diversity. Third, any future subsidy should be tied to open, interoperable infrastructure so smaller carriers can lease capacity instead of being locked out.
The model Taiwan has chosen, state financing of physical redundancy delivered by licensed operators, is far better than mandating private resilience or imposing broad restrictions. It treats the problem as an engineering and capital issue, not a pretext for regulating speech or traffic. But a cable that nobody has tested under stress is a promise, not resilience. The test of this policy will come the next time a cable fails, and MODA should be measured by how quickly Matsu recovers.