What CSIS found about Chinese research ships near undersea cables

A Center for Strategic and International Studies (CSIS) analysis reports a statistically significant association between Chinese-flagged research vessels and a larger estimated share of observed time near subsea infrastructure on the high seas than for most other national fleets. The study measured automatic identification system (AIS) vessel-tracking data from January 1, 2020, through December 9, 2025. Its measure is proximity to mapped infrastructure, including fiber-optic cables and gas pipelines; the authors’ discussion of possible military use is an interpretation of those patterns. Read the CSIS analysis.

What the CSIS study measured

CSIS says its dataset covered 223 research vessels, comparing Chinese-flagged vessels with fleets from 47 other nations. The researchers estimated each vessel’s annual share of observed time on the high seas near subsea infrastructure. They used AIS tracks alongside geographic data for cables, pipelines and other areas; cable and pipeline map layers were given a 1-kilometer buffer to account for location imprecision.

In the baseline statistical model, the association between a Chinese flag and the measured share was positive, with a coefficient of 0.0315 and p < 0.001. That result describes an association in this dataset. It does not assign a purpose to any particular voyage. CSIS says most vessels in the dataset spent less than 1% of their observed time near the infrastructure in a given year.

How the fleet comparisons differ

CSIS also compared how often each fleet ranked above the 80th percentile in five annual comparisons of high-seas time near infrastructure. The measure is relative to the study’s comparison set.

Seven Chinese research vessels ranked above the 80th percentile in 2024. The report describes the Russian flag’s association in its baseline model as negative, but not statistically significant; it does not give a cause for that result.

The two vessels in the case studies

CSIS’s case studies examine Xiang Yang Hong 06 and Ke Xue. The researchers describe the vessels’ AIS tracks as survey-like patterns near named cable systems. Those track interpretations supply context for the fleet-level result, while the report treats any mission purpose as an inference.

CSIS describes Xiang Yang Hong 06 trips near Guam from February through May 2024, close to the Apricot and Asia-America Gateway (AAG) systems. Its July 2024 tracks were near ACC-1, AAG and SEA-US. The report also describes an expanded pattern from August 14 to October 27, 2024.

For Ke Xue, CSIS describes tracks near TGN-IA, Apricot and JUPITER from March through June 2024, followed by a northern mission near multiple cable systems from July 15 to August 28. The report’s broader discussion of possible seabed analysis relevant to submarine operations or infrastructure sabotage is the authors’ interpretation of the recorded patterns. The study’s reported findings concern vessel time near infrastructure, rather than a documented cable cut, tapping or packet interception.

What the statistical limits mean

CSIS reports that its baseline result remains positive under several robustness checks, though statistical precision weakens under some error structures. The report also gives an adjusted R² of about 1.3% in one passage. Adjusted R² estimates how much variation in the measured outcome a model explains; here, the authors describe that share as limited. The association is a result from the analysis, while the model’s explanatory power and the interpretation of individual tracks constrain what can be concluded about specific missions.

General Atomics supported the CSIS report, and General Atomics Intelligence’s Optix platform supplied the AIS data used in the analysis. CSIS recommends greater maritime information-sharing, coordinated patrols near infrastructure and regional cable-repair capacity.