Underneath the western Pacific, between Taiwan, Japan and the Philippines, run the glass threads that carry most of the data between the United States and Asia. The water there is shallow, the shipping lanes are crowded, and China — which claims sovereignty over Taiwan and has a well-documented habit of sending vessels to look at things — is right there. It is a bad place to put critical infrastructure. It is also where the critical infrastructure is.
A new risk index, reported by Ars Technica, says exactly this: the danger of sabotage to subsea cables is highest around Taiwan, Japan and the Philippines, partly because of Beijing’s claim to Taiwan and partly because the cables connecting the region to the US have little choice but to pass through those shallow, busy waters. Geography, it turns out, has opinions about routing.
Among the 10 trans-Pacific cables the index flags as most at risk, several are still under development — and several belong, in whole or in part, to American tech giants. There is the TPU project, owned by Google (yes, also what Google calls its AI chips; the company likes its TPUs everywhere). There is the ORCA cable, owned by Meta. And there is E2A, co-owned by Verizon alongside firms from Japan, South Korea and Taiwan. All three sit in the top five risk scores. Build your cable in the water everyone is watching and, well, everyone is watching your cable.
Why keep building there at all? Because the internet is, at its core, a physical object, and physics charges by the mile. Data wants the shortest path; every detour costs time, and time costs money — a few extra milliseconds to a hedge-fund trade, a sluggish response to a cloud customer. The straight line between California and Taipei happens to run through contested water, and the straight line is where the value is. The risk index is, among other things, the price quote for that straight line.
The report’s authors walk through the scenarios. Cut one cable and you “risk overwhelming current available capacity on direct routings between the United States, Taiwan, Japan and the Philippines,” with traffic shoved “onto slower, less direct routes.” Cut several — say, during a geopolitical crisis over Taiwan — and you could “significantly reduce headroom and even overwhelm trans-Pacific capacity as a whole,” forcing US–Indo-Pacific communications to detour around the rest of the globe, with heavy economic fallout for internet traffic, cloud services and financial trading.
Who would do the cutting? Everyone is quietly acquiring the means. The major powers keep expanding their fleets of military submarines and undersea drones, and all of them are developing what the trade calls dual-use technology — undersea cable cutters that work equally well for repair and for sabotage. A cable cutter, it turns out, is a cable cutter. The tool does not know which press release it belongs to.
Meanwhile, a separate October 2026 report from the Center for Strategic and International Studies, a Washington think tank, found that China has deployed research vessels equipped with seabed-mapping technology near trans-Pacific cables. Tracking the movements of more than 200 research ships from multiple nations between 2020 and 2025, the CSIS analysts concluded that China-flagged ships stood out as outliers in the sheer amount of time spent loitering around subsea infrastructure. If you were ever going to cut a cable — purely hypothetically — step one is knowing precisely where it lies and what the seabed around it looks like. Step one appears to be well underway.
The tech companies publish lovely maps of their cable networks as monuments to connection and engineering prowess. The index suggests a second reading of those maps: as a targeting list. Anyway, the cables will keep going in, because the data has to flow and there is too much money in flowing it. Somebody just has to hope everyone with cutters stays in a repair frame of mind.