The cloud touches the seabed
TeleGeography forecasts that more than 1,010,000 kilometres of new submarine cable will be deployed between 2026 and 2035, with average annual investment of approximately US$5 billion (Mauldin, 2026). This is a forecast for the cable market as a whole, not for artificial intelligence alone. Even so, the growth of data centres, cloud services, training workloads and inference is reshaping demand for long-distance connections.
The figure allows the debate to move elsewhere. Artificial intelligence is usually explained through models, chips and data centres, yet coordination between those facilities depends on a physical network crossing the oceans. Data is processed where energy, land, cooling, incentives and computing capacity are available. It must then travel between territories governed by different rules on sovereignty, security and data localisation.
The cloud is therefore not outside the material world. It is sustained by landing stations, specialist vessels, repeaters, data centres, capacity agreements and cables laid or buried on the seabed. The cloud touches the seabed.
Submarine cables turn the ocean into a material, political and territorial infrastructure. Starosielski (2015) follows these networks from the ocean floor to their landing stations, showing that their operation depends on coastlines, territories, companies, regulatory frameworks and environmental conditions. From the anthropology of infrastructure, Larkin (2013) approaches such networks as material forms that enable circulation across space while also organising promises, hierarchies and relations of power.
Submarine cable routes overlap with jurisdictions, fishing, navigation and other uses of maritime space. They are therefore matters of ocean governance and geopolitics, rather than engineering alone (Bueger & Liebetrau, 2021; Takei, 2012; Van Logchem, 2014). A cable does not simply join two points on a map. It connects seabeds, coastal landing stations, permits, companies, electricity systems and data centres. Its distribution is not neutral either: socioeconomic, political and geographical factors influence where access points and planned routes are concentrated (Franken et al., 2025).
Who owns the seas? Humboldt and Firmina
There is no single legal answer to the question of ocean ownership. In the territorial sea, the coastal state exercises sovereignty. In the Exclusive Economic Zone, it holds rights over resources and certain environmental responsibilities, while other states retain freedoms including navigation and cable-laying. On the continental shelf, coastal states hold rights over the seabed and subsoil, but no general power to prevent cables owned by others. No state may claim sovereignty over the high seas. The seabed beyond national jurisdiction, known legally as the Area, and its mineral resources are governed by the principle of the common heritage of humankind (Davenport, 2018; Takei, 2012; United Nations, 1982; Van Logchem, 2014).
These divisions do not by themselves explain who controls digital circulation. Maritime space may be governed by public and international regimes, while cable systems, fibre pairs and available capacity are owned or allocated to companies, consortia and public bodies through different arrangements.
The Humboldt project makes these layers visible. In June 2026, Chilean authorities approved the installation and operation of a system linking Santo Domingo in the Valparaíso Region with Sydney, alongside a second route to Panama. The latest official information describes more than 21,000 kilometres of submarine infrastructure, with installation expected to begin in the final quarter of 2026 and service planned for 2028 (Gobierno de Chile, 2026).
Humboldt is presented as a route capable of turning Chile into a digital node between South America and the Asia-Pacific region. Google and the state-owned company Desarrollo País created Humboldt Connect with equal stakes. That company is associated with the activation and commercialisation of selected dark-fibre pairs, however, so ownership of the entire system, project finance, reserved capacity and commercial rights should not be treated as the same thing. Infrastructure does not have one simple form of ownership (Desarrollo País, 2025; Gobierno de Chile, 2026).
The Firmina cable provides a contrast. Built by Google, it links the east coast of the United States with Las Toninas in Argentina, with additional landings at Praia Grande in Brazil and Punta del Este in Uruguay. Its design allows the system to be powered from one end if the power source at the other becomes temporarily unavailable. Telxius provides infrastructure and commercial access at several points, but that collaboration should not automatically be described as shared ownership of the cable (Koley, 2021).
Humboldt combines state and corporate participation in the name of regional integration. Firmina shows a hyperscale company building a route designed around its services and resilience requirements. Together, they reveal a paradox of digital sovereignty. A state may require local data centres or protect the domestic residence of information while remaining dependent on oceanic capacity controlled contractually and technically by transnational companies.
Since January 2026, this expansion has also taken place under the newly effective BBNJ Agreement on marine biodiversity beyond national jurisdiction. The treaty creates mechanisms concerning marine protected areas, impact assessment, marine genetic resources and technology transfer. It does not make every cable automatically subject to a full environmental assessment, nor does it determine cable ownership. It does introduce a more demanding framework for examining the cumulative effects of new activities on the high seas (United Nations, 2026).
The available research argues against environmental sensationalism. Telecommunications cables occupy a very small proportion of the seabed and their effects are generally localised, especially during installation or recovery. Disturbance may nevertheless matter more in coral systems, seagrass meadows and other sensitive habitats. The forecast expansion makes route selection, burial methods, maintenance and the fate of retired systems legitimate questions, without treating the seabed as an inert platform (Calhoun et al., 2025).
When connectivity is called development
The anthropology of development provides a way to examine the language used to legitimise these infrastructures. Connectivity, modernisation, digital inclusion, resilience and growth describe genuine needs, but they can also present political choices as inevitable consequences of technical progress.
Arturo Escobar showed how development operates as a regime of representation. It defines which territories appear to be behind, which forms of knowledge are recognised as authoritative and which interventions become reasonable (Escobar, 1995). Mark Hobart examined how expert knowledge can produce ignorance by relegating local practices and knowledge (Hobart, 1993). James Ferguson demonstrated how translating political conflicts into technical problems can expand administrative and corporate power while hiding its distribution (Ferguson, 1994).
Applying these perspectives does not mean denying the usefulness of cables. It means asking what connectivity does beyond carrying data.
A new route may reduce latency and expand capacity while concentrating ownership. It may turn a coastline into a regional node without ensuring that nearby communities take part in the decision. It may attract data centres while deepening dependence on foreign clouds, contracts and jurisdictions. It may be presented as Latin American integration even when much of the traffic and value flows towards platforms located outside the region.
It may also be understood as infrastructure that enables data extraction and valorisation, in dialogue with what Couldry and Mejias (2019) call data colonialism. Applying that framework to submarine cables is AIthropology Lab’s own analytical extension: the cable does not extract data from the seabed in the manner of a mine, but provides the material channel through which data travels towards systems capable of storing and processing it, and of capturing its value. The ocean floor is thereby incorporated into chains of value extraction that also depend on energy, labour, minerals, water and territory.
George Marcus offers a suitable method for studying this architecture. His multi-sited ethnography proposes following objects, connections and conflicts across different locations (Marcus, 1995). Here, following the cable means beginning at the data centre, reaching the landing station, tracing the maritime route, identifying permits and legal regimes, reconstructing ownership of individual fibre pairs, and asking who purchases capacity and for what purpose.
That journey moves attention away from the abstract promise of connection and towards its concrete conditions.
Who chooses the route?
Who receives public capacity and who depends on private agreements?
Which coastline becomes a node, and which remains peripheral?
What environmental knowledge informs the route?
Who is able to repair the connection when it fails?
Submarine expansion does not prove that artificial intelligence alone is conquering the oceans. It shows something more precise. The global reorganisation of computing requires the sea to be made into a coordinated architecture of circulation. States, corporations, ecosystems and communities meet there under unequal relations.
Artificial intelligence does not float above those conflicts. It travels through them.
References
Bueger, C., & Liebetrau, T. (2021). Protecting hidden infrastructure: The security politics of the global submarine data cable network. Contemporary Security Policy, 42(3), 391–413. https://doi.org/10.1080/13523260.2021.1907129
Calhoun, E., Fletcher, R., McOwen, C., Ondo, I., Kamath Cannanure, V., Wrottesley, J., Clare, M., Bressie, K., & Wopschall, R. (2025). Submarine cables and marine biodiversity. UNEP World Conservation Monitoring Centre & International Cable Protection Committee. https://doi.org/10.34892/V9NB-TR83
Couldry, N., & Mejias, U. A. (2019). Data colonialism: Rethinking big data’s relation to the contemporary subject. Television & New Media, 20(4), 336–349. https://doi.org/10.1177/1527476418796632
Davenport, T. (2018). The high seas freedom to lay submarine cables and the protection of the marine environment: Challenges in high seas governance. AJIL Unbound, 112, 139–143. https://doi.org/10.1017/aju.2018.48
Desarrollo País. (2025, 4 de junio). Desarrollo País y Google se unen para crear Humboldt Connect, empresa a cargo del primer cable submarino que cruzará el Pacífico Sur. https://desarrollopais.cl/desarrollo-pais-y-google-se-unen-para-crear-humboldt-connect-empresa-a-cargo-del-primer-cable-submarino-que-cruzara-el-pacifico-sur/
Escobar, A. (1995). Encountering development: The making and unmaking of the Third World. Princeton University Press.
Ferguson, J. (1994). The anti-politics machine: “Development,” depoliticization, and bureaucratic power in Lesotho. University of Minnesota Press.
Franken, J., Reinhold, T., Dörnfeld, T., & Reuter, C. (2025). Hidden structures of a global infrastructure: Expansion factors of the subsea data cable network. Technological Forecasting and Social Change, 215, 124068. https://doi.org/10.1016/j.techfore.2025.124068
Gobierno de Chile. (2026, 22 de junio). Cable Humboldt: Aprueban instalación del cable submarino de Google que unirá Chile con Asia-Pacífico. https://www.gob.cl/noticias/cable-humboldt-aprueban-instalacion-del-cable-submarino-de-google-que-unira-chile-con-asia-pacifico/
Hobart, M. (Ed.). (1993). An anthropological critique of development: The growth of ignorance. Routledge.
Koley, B. (2021, June 9). Hola, South America! Announcing the Firmina subsea cable. Google Cloud Blog. https://cloud.google.com/blog/products/infrastructure/announcing-the-firmina-subsea-cable
Larkin, B. (2013). The politics and poetics of infrastructure. Annual Review of Anthropology, 42, 327–343. https://doi.org/10.1146/annurev-anthro-092412-155522
Marcus, G. E. (1995). Ethnography in/of the world system: The emergence of multi-sited ethnography. Annual Review of Anthropology, 24, 95–117. https://doi.org/10.1146/annurev.an.24.100195.000523
Mauldin, A. (2026). Forecasting the next 10 years of submarine cable investment and kilometers. TeleGeography. https://resources.telegeography.com/forecasting-the-next-10-years-of-submarine-cable-investment-and-kilometers
Starosielski, N. (2015). The undersea network. Duke University Press. https://doi.org/10.1215/9780822376224
Takei, Y. (2012). Law and policy for international submarine cables: An Asia-Pacific perspective. Asian Journal of International Law, 2(2), 205–233. https://doi.org/10.1017/S204425131200001X
United Nations. (1982). United Nations Convention on the Law of the Sea. https://www.un.org/depts/los/convention_agreements/texts/unclos/unclos_e.pdf
United Nations. (2026). Agreement on Marine Biological Diversity of Areas beyond National Jurisdiction. https://www.un.org/bbnjagreement/en
Van Logchem, Y. (2014). Submarine telecommunication cables in disputed maritime areas. Ocean Development & International Law, 45(1), 107–122. https://doi.org/10.1080/00908320.2014.867195