MIST

Based on 47 RIPE Atlas measurements from GeoCables monitoring infrastructure, March–April 2026.

When you read the technical sheet for the MIST cable — Malaysia-India-Singapore-Thailand, brought into service in 2024 by Orient Link — you'll find an unambiguous number: 8,100 kilometres of fibre, 12 pairs, five landing stations across four countries. It's the kind of specification that invites an obvious expectation: packets travelling across MIST should spend a certain minimum amount of time in the glass, and that minimum should track the cable's length.

Our probes don't quite agree.

Across 47 measurements from Tuas, Singapore to Mumbai, India — the one direction our probe geography currently exercises on MIST — the fastest observed round-trip is 42.88 ms. At the speed of light in silica fibre (about 200,000 km/s), that works out to roughly 4,290 km of optical path end-to-end. Less than 54% of what the cable's headline length suggests.

This is the kind of reading that, a year ago, would have sent us hunting for a measurement bug. Today, it's the most interesting thing about the cable.

The two lengths

The discrepancy isn't an error — it's geometry. When a consortium announces an 8,100 km cable, that number totals every kilometre of fibre that the construction ship laid down, including branches. MIST's five landings aren't in a straight line. Chennai and Mumbai sit on opposite coasts of India, separated by 1,300 km of peninsular terrain. Satun and Morib anchor the Southeast Asian stretch. Tuas hangs off the southern tip of Singapore. These aren't all reachable through a single spine of fibre.

A cable connecting those five points inevitably has branches. The construction cost of 8,100 km is real. What isn't real is the idea that any one packet traverses all of it. Packets pick the shortest segment that connects their endpoints — and between Tuas and Mumbai, that segment is the arc across the northern Bay of Bengal and the Arabian Sea, skipping the Indian peninsular detour and the Chennai landing entirely.

Orient Link, the company that owns MIST, is a single-purpose operator built around this specific cable. Its commercial model depends on the fact that a cable's value to customers comes from the number of regional points of presence it offers, not from the straight-line distance between any two of them. Two Indian landings let the operator sell connectivity on both coasts of the subcontinent without forcing customers to build their own backhaul across peninsular India. Satun and Morib add optionality for Thai and Malaysian traffic. Five landings are five markets, and it's five markets of demand that justified the capital to lay 8,100 km.

Great-circle distance between Tuas and Mumbai is about 3,900 km. Adding the roughly 10% overhead that real cable routes incur over the geodesic — cables hug continental shelves, avoid seismic zones, curve around territorial waters — gets us to 4,290 km. That's the 42.88 ms reading, almost exactly.

Why this isn't a problem

It's tempting to read a "below floor" measurement as a sign that something is wrong with either the calculation or the cable. Neither is true. MIST is working as designed. Orient Link built a cable with five landings because they wanted five points of presence, not because they wanted every landing to be in the path of every packet. A trunk-and-branch topology — a main lane of fibre with spurs leading to secondary landings — is the standard way to offer regional coverage without forcing every user onto a single linear highway.

This is the same architecture that shapes several of the cables we've written about. Apricot is explicitly hub-and-spoke: a Japan-Indonesia spine with branches to Guam, the Philippines, Taiwan, and Singapore. Bifrost uses a trunk with branches for its North American–Southeast Asian reach. TAM-1, the Trans Americas pure-play cable, presents as a hub whose sub-arcs consistently measure below any length-derived floor. All three, like MIST, accept that their headline kilometre count is a statement about the cable's footprint, not the distance any single packet travels through it.

What the 47 measurements actually show

The floor, at 42.88 ms, is stable: the lowest ten readings cluster between 42.88 and 46.1 ms. Above that floor, variance is significant. Average RTT is 101.93 ms, standard deviation 44.17, maximum 290.77. A 2.4× spread between minimum and average is the signature of a path where the baseline is fast and well-defined, but where individual measurements occasionally take a longer detour — likely hitting congestion at one of the transit points in either Singapore or Mumbai, or being briefly routed around a busy node by one of the operators running traffic over the cable.

None of this suggests a fault. A 2024 cable, not yet two years into service, is still in the period where transit providers are tuning their BGP preferences and discovering what traffic patterns actually work. Variance like this flattens over the following year or two as routing settles.

The India-Singapore corridor

MIST joins a cluster of recent cables addressing the India–Southeast Asia corridor: India-Europe Xpress, AAE-1, SeaMeWe-6. These cables compete for the same broad purpose — delivering low-latency transit between the Indian subcontinent and the Southeast Asian hub cities — but each takes a different geometric bet. AAE-1 and SeaMeWe-6 route westward through the Suez corridor and back out toward Europe. MIST stays in the Indian Ocean, serving India-to-Singapore traffic directly.

The reason this corridor has attracted so many new cables in the last five years is a structural shift in regional traffic. Indian data centres — in Mumbai, Chennai, Bangalore — have grown by order of magnitude over that period, and a large share of their traffic terminates in Singapore, where AWS, Google Cloud, and Microsoft Azure run their primary South and Southeast Asian regions. MIST is the infrastructure response: the shortest feasible path between two poles of that corridor, without intermediate geography and without competing for capacity with Europe-bound flows.

For operators who want the shortest path between Mumbai and Singapore — specifically the path that doesn't hairpin through an intermediate geography — MIST is the answer. And our 42.88 ms floor is the evidence that, for at least one direction of traffic on at least one transit provider, the cable delivers exactly the path its geometry suggests it should.

What we're watching next

The one missing element is the reverse direction. Our current probe footprint gives us Tuas→Mumbai but not Mumbai→Tuas. Once a Mumbai-based probe joins the measurement rotation, we'll know whether the path is symmetric (the floor stays at 42.88 ms either way) or whether there's a systematic asymmetry from different transit provider preferences on the two ends. That's the kind of cross-check that turns a single-direction reading into a complete picture of how an 8,100 km cable chooses to behave in practice.