Subsea Cables and the AI Race: How the South China Sea Became Asia’s Digital Chokepoint

Executive Summary

The first article in this series examined the chips. The second examined the data centers. This one examines the wires.

Beneath the surface of the South China Sea runs one of the world’s most consequential concentrations of digital infrastructure — a dense web of subsea fiber-optic cables carrying the overwhelming majority of data traffic between Asia, the Americas, and the rest of the world. These cables are not passive conduits. They are strategic assets, and their security, ownership, and routing have become active fronts in the US-China technology competition.

The connection to AI is direct. As Southeast Asia’s data center buildout accelerates — driven by the dynamics examined in the previous article — the volume of data traversing regional cable networks is growing at a pace that was unimaginable five years ago. AI training workloads, inference traffic, cloud synchronization between hyperscale facilities, and real-time data flows between compute clusters and end users all depend on the same physical infrastructure lying on the ocean floor.

That infrastructure is now contested terrain. China has accumulated significant leverage over cable landing points, cable construction, and cable repair capacity across the region. The United States, Japan, Australia, and India — the four nations of the Quad — have moved to counter that leverage through coordinated investment, regulatory pressure, and an explicit framework for cable connectivity and resilience in the Indo-Pacific. The outcome of that contest will shape not only who controls the region’s digital highways, but who can effectively weaponize — or defend — them in a crisis.

Key Development

Subsea cables carry the weight of the global digital economy. More than 99 percent of transnational data communications worldwide travel on submarine cables, along with an estimated $10 trillion in daily financial transactions. The internet, for all its apparent diffusion, is physically anchored to a relatively small number of cable systems lying on the ocean floor. Disruption to those systems — whether through accident, sabotage, or deliberate interdiction — does not merely inconvenience users. It severs the arteries of modern commerce, communication, and increasingly, AI computation.

The South China Sea is the critical chokepoint. The South China Sea sits at the intersection of cable routes connecting East Asia to Southeast Asia, South Asia, the Middle East, Europe, and the Americas. Nearly every major cable system linking the data centers of Singapore, Malaysia, Indonesia, Vietnam, and the Philippines to global networks passes through or near waters that China claims under its expansive maritime territorial assertions. This geographic reality transforms a commercial infrastructure question into a geopolitical one.

China has built structural leverage across the cable ecosystem. China’s state-backed cable operator HMN Technologies — formerly Huawei Marine Networks — has constructed or supplied components for a significant portion of the cable systems serving Southeast Asia. The South-East Asia–Hainan–Hong Kong Express cable system connects China, Malaysia, Singapore, Thailand, and the Philippines, routing traffic through Hainan Island. Researchers have identified this concentration as an expression of what political scientists call “weaponized interdependence” — the capacity of a state that sits at the center of a shared network to gather intelligence through the panopticon effect, or to disrupt access through the chokepoint effect, at a moment of its choosing.

Western hyperscalers have begun routing around Chinese infrastructure. Google and Meta’s Apricot cable lands in Guam, Indonesia, Japan, the Philippines, Singapore, and Taiwan — conspicuously bypassing mainland China and Hong Kong. Meta and Amazon’s Bifrost cable follows a similar routing logic, connecting Guam, Indonesia, the Philippines, Singapore, and the US West Coast. The US Clean Network initiative, which effectively bans new cables directly connecting the United States to mainland China or Hong Kong, has formalized this divergence at the policy level. Washington has also provided training grants to foreign telecoms firms to steer them away from Chinese cable consortia — most visibly in the SEA-ME-WE 6 project, where the American company SubCom was selected over a Chinese consortium.

Sabotage has moved from theoretical to documented risk. In December 2024, subsea cables in the Baltic Sea were severed, triggering NATO to establish a Critical Undersea Infrastructure Network for threat intelligence sharing. In February 2025, Taiwanese coast guard personnel boarded a vessel suspected of damaging a cable connecting Taiwan Island and Penghu Island. The pattern of incidents — combined with the deployment of older, foreign-flagged vessels by actors seeking to obscure attribution — has elevated cable vulnerability from a contingency planning footnote to an active security concern for governments across the Indo-Pacific.

The Quad has responded with a formal and escalating framework. The Quad Partnership for Cable Connectivity and Resilience, launched in 2023, has continued to deepen through ministerial coordination and technical forums, including the Wavelength Forum hosted in New Delhi in July 2025. At the 11th Quad Foreign Ministers’ Meeting in New Delhi on May 26, 2026, the four nations confirmed they have provided tangible support to connect all Pacific Islands Forum countries via secure undersea cable networks by 2026, and reaffirmed the need to protect cable networks from threats and sabotage through closer coordination and information sharing. The US Congress has simultaneously advanced the Strategic Subsea Cables Act of 2026, which formalizes American engagement across international cable protection bodies and creates sanctions mechanisms for critical undersea infrastructure sabotage.

Strategic Analysis

The cable layer is where AI infrastructure meets geopolitical risk.

The first two articles in this series identified two physical dependencies of the AI economy: the chips that power computation, and the data centers that house it. Subsea cables represent the third dependency — and in some respects the most exposed one. Chips are manufactured in controlled facilities. Data centers are built on sovereign territory with physical security. Cables lie unguarded on the ocean floor across thousands of kilometers, passing through jurisdictional grey zones and repair windows measured in weeks.

For AI specifically, this exposure is growing rather than shrinking. As Southeast Asian data centers come online at scale — processing next-generation AI workloads across Johor, Jakarta, and Ho Chi Minh City — the volume and strategic value of data traversing regional cable systems increases proportionally. An AI training run distributed across data centers in Singapore, Malaysia, and Japan depends on consistent, low-latency connectivity between those facilities. A disruption to the cable layer does not merely slow traffic — it can interrupt computation in ways that are difficult to predict and expensive to recover from.

The bifurcation of cable infrastructure mirrors the bifurcation of the broader technology stack.

The routing decisions made by Google, Meta, and Amazon in their most recent cable deployments are not primarily commercial decisions. They are architectural choices about which jurisdictions’ digital infrastructure should be considered trustworthy for long-term investment. By building cable routes that bypass mainland China and Hong Kong, Western hyperscalers are physically instantiating the same logic that has driven the separation of semiconductor supply chains, cloud infrastructure standards, and AI development ecosystems.

This bifurcation creates a structural problem for Southeast Asian nations pursuing technological non-alignment — the strategy examined in the previous article. A country hosting both Western and Chinese data center infrastructure on its sovereign territory may find that strategy increasingly difficult to sustain if the underlying cable architecture forces a choice between connectivity ecosystems. The cable layer, in this sense, is where the abstract geopolitics of technology competition becomes a concrete infrastructure decision with long-term consequences.

China’s position is structural, not merely tactical.

China’s leverage over regional cable infrastructure is not primarily the product of deliberate sabotage planning. It is the accumulated result of years of commercial cable construction through HMN Technologies, strategic positioning at cable landing stations, and the geographic reality that the South China Sea lies between China and every major Southeast Asian market. That structural position gives Beijing options that do not require active disruption to be strategically valuable — the knowledge that a chokepoint exists, and that it could be activated under certain conditions, shapes the behavior of other actors even when no action is taken.

The repair vessel dimension reinforces this dynamic. Cable repair is a specialized business dominated by a small number of operators. China has developed its own cable repair vessel capabilities, reducing its dependence on the international repair ecosystem while simultaneously creating tools that carry dual-use potential. Allied concern about this has been explicit in both policy documents and legislative action.

The Quad’s response is real but incomplete.

The Quad’s cable connectivity framework represents genuine strategic intent translated into operational commitments. The confirmed delivery of cable connectivity to all Pacific Islands Forum countries addresses a specific vulnerability that China had actively exploited — small island states with limited connectivity options are disproportionately susceptible to dependence on whichever infrastructure provider arrives first. The Wavelength Forum and regulatory harmonization efforts among Quad partners are building the coordination architecture needed for a sustained allied response.

The gaps, however, remain significant. The Quad framework does not yet include a shared maritime working group with enforcement authority to respond to cable incidents in Southeast Asian waters. Permitting and repair coordination across ASEAN jurisdictions remains fragmented, creating delays that complicate rapid response. And the fundamental geographic asymmetry persists: China’s position in the South China Sea cannot be altered by investment or policy coordination alone. As the Stimson Center has noted, the Quad has consistently been more effective at announcing initiatives than delivering measurable outcomes — a structural accountability gap that will matter more as cable security challenges intensify.

Investor Takeaway

Cable routing decisions are a leading indicator of geopolitical alignment. Track which cable systems are being built, by whom, and where they land. The divergence between Western-backed routes — bypassing China and Hong Kong — and Chinese-affiliated systems routing through Hainan and other strategic nodes is accelerating. Countries and operators anchoring their connectivity to one system or the other are making durable architectural choices with long-term implications for data sovereignty and geopolitical risk exposure.

The cable security industry is an emerging investment category. Historically, subsea cable infrastructure was treated as a background utility. That framing is changing. The combination of documented sabotage incidents, legislative action through the Strategic Subsea Cables Act, and Quad coordination frameworks is creating a policy environment in which cable monitoring, protection, and repair capabilities will attract growing public and private capital. Companies with capabilities in underwater surveillance, cable laying and repair, and maritime domain awareness sit at an intersection of defense technology and critical infrastructure that is drawing increasing institutional attention.

Data center valuations in Southeast Asia carry embedded cable risk. The investment case for regional AI infrastructure assumes reliable, low-latency connectivity to global networks. Facilities with diversified cable connectivity — access to multiple systems with different routing paths and ownership structures — carry structurally lower disruption risk than those dependent on a single cable system. This factor is underweighted in current infrastructure valuation frameworks and will become more visible as geopolitical tensions in the South China Sea develop.

The Quad’s cable commitments signal durable allied infrastructure investment across the Indo-Pacific. The pattern visible in semiconductor policy, data center incentives, and now cable security — coordinated allied investment designed to build trusted alternatives to Chinese-affiliated infrastructure — represents a structural trend rather than a cyclical policy response. The geographic arc of that investment, from India through Southeast Asia to the Pacific Islands, defines the terrain on which the long-term technology competition will be contested.