On 27 October 2026, former Singaporean politician Calvin Cheng posted an infographic titled "Indonesia's Internet Dependence on Singapore". It claimed about 90% of Indonesia's international internet traffic passes through Singapore, and it asked one word: "Cut?" Communication and Digital Minister Meutya Hafid answered on Instagram the same day. She wrote that cables landing in Singapore mostly run through Indonesian waters, and that if they were cut, 80% of Singapore's connectivity capacity would also be threatened (Bisnis, 28 Oct 2026). Her message was that digital infrastructure should be protected jointly, and she added that "we need not be furious".
Neither side's post is official policy, and Cheng is not in the Singapore government. But the exchange is useful because both numbers describe a real structural fact, and the sensible response is neither retaliation nor complacency.
The dependence is real, and Komdigi has said so itself
It would be easy to dismiss Cheng's figure as provocation. It is not far from what Indonesian officials have said. Katadata reports that Denny Setiawan, Komdigi's director for digital infrastructure strategy and policy, said at the Selular Award on 8 June 2026 that "right now 90% of our traffic depends on Singapore" (Katadata, 27 Oct 2026). The wording differs slightly: Cheng's graphic says international traffic, while the official's remark carries no qualifier. The order of magnitude, however, is shared.
The strongest case for treating this as a national-security problem is straightforward. A country of Indonesia's size, with the Bisnis report citing 267 million internet users, routes most of its external traffic through a single foreign hub. Any political dispute, outage or accident at that hub becomes Indonesia's problem. Indonesian planners are right to worry about that concentration, and a minister who ignored it would be negligent.
Why the threat of cutting cables is the wrong frame
Meutya's reply points at the other side of the ledger. Cables that land in Singapore mostly cross Indonesian waters, so the dependence is mutual. Her 80% figure is an assertion in a social-media post, not a published engineering analysis, and I could not find a primary source for it. It should be treated as a political estimate.
The legal frame matters more than the rhetoric. Under the UN Convention on the Law of the Sea, Article 113 requires every state to make the wilful or negligent breaking of a submarine cable beneath the high seas a punishable offence (UNCLOS Part VII). That article covers the high seas, not Indonesia's archipelagic waters. For those waters, Article 51(2) says an archipelagic state "shall respect existing submarine cables laid by other States and passing through its waters without making a landfall" and must permit their maintenance and replacement (UNCLOS Part IV). Not every cable serving Singapore fits that description, because many land in Indonesia too. But for transit cables, a government that cut or obstructed them would be breaching an obligation Indonesia has accepted.
There is also a practical reason to avoid escalation. A country that wants foreign cable operators and data-centre investors to commit capital for decades cannot also be seen to treat landing rights and transit routes as bargaining chips. The cost of that reputational damage would fall on Indonesian businesses and users first.
The better answer is already on the table
On 20 October, a week before the exchange, Setiawan told the Bisnis Indonesia Forum that Komdigi is drafting a roadmap for submarine cables and data centres. The stated aim is "multi-gateway" connectivity, so that internet access from eastern Indonesia does not have to pass through Singapore. According to Bisnis, the plan would develop cable landing stations on each large island, with at least two landing points per island. It also favours "open" landing stations that several operators can share, and it ties data-centre growth to power supply coordinated with the state utility PLN and the energy ministry (Bisnis, 20 Oct 2026). The report mentions a route via Manado but gives no capacity numbers or deadlines.
This is the proportionate response. It reduces concentration risk by adding routes and exchange points rather than by restricting anyone's traffic, and it is consistent with an open-internet approach: more paths mean more competition, lower latency for eastern provinces and less exposure to any single failure.
What the roadmap should get right
The public record on the roadmap is thin, so the following are tests rather than criticisms of a finished plan.
- Open access that is real. Shared landing stations only reduce costs if tariffs and access terms are transparent and non-discriminatory. A mandated-use model, where traffic must be routed domestically regardless of price, would raise costs and could cut performance.
- Demand, not just pipes. Eastern cables will be underused unless content, cloud and peering move east. Data centres need reliable power, which the Bisnis report flags as a coordination task with PLN and the energy ministry.
- Permitting speed. Landing licences, seabed permissions and repair-ship access across several agencies are where cable projects typically stall. Clear timelines would matter more than new targets.
- Targets with dates. The reported plan has none yet. A published roadmap should state the capacity expected through non-Singapore gateways and the year it is expected.
- No data-localisation creep. Diversifying physical routes is different from forcing data to stay onshore. The roadmap should not become a vehicle for the latter.
Bottom line
Cheng's "Cut?" was careless, and Meutya's counter-claim is an estimate, not evidence. Yet together they illustrate a sound policy conclusion: a hub concentrating 90% of a large country's traffic is a vulnerability for that country and for the hub. The way out is more landing points, shared infrastructure and eastern demand, delivered on a published schedule, not threats against existing cables that Indonesia is bound by treaty to respect. The roadmap is the right instrument. It now needs numbers and dates.