AI Data Boom Supercharges Global Subsea Cable Buildout and Investment Cycle

Artificial intelligence is turning subsea cables from background plumbing into a strategic choke point for the global digital economy, forcing a step change in how undersea infrastructure is financed, built and operated.

Why AI is driving new cable demand

Training and running large AI models requires moving vast datasets between geographically dispersed data centres, pushing demand for high‑capacity, low‑latency international connectivity far beyond what legacy systems were designed to handle. Subsea systems already carry more than 95% of global international data and voice traffic, and operators warn that the AI boom is making the current infrastructure model “unsustainable” without a new wave of investment.

Hyperscalers—large cloud and internet companies—now account for the majority of used international bandwidth and are driving most of the new demand. As their AI workloads scale, many have concluded that leasing capacity on existing consortia cables is no longer sufficient and are co‑financing or directly owning new routes to guarantee performance, security and price visibility.

Hyperscalers reshape the subsea investment model

Historically, telecom carriers dominated subsea systems, sharing costs through consortia and recovering investment via wholesale capacity sales. Today, cloud providers such as Google, Meta, Amazon and Microsoft are co‑owners of multiple systems and are expected to drive investments in new subsea initiatives to roughly 13 billion dollars between 2025 and 2027, almost double the previous three‑year period, according to TeleGeography estimates cited by industry reports.

This shift is altering both financing and design. New projects are increasingly structured around hyperscaler capacity‑planning cycles, with flexible architectures that allow capacity to be lit in phases, matching the ramp‑up of AI workloads and reducing upfront capital exposure. For investors, that means a deeper pipeline of long‑term, infrastructure‑like projects, but with cash flows tied to the technology strategies and credit quality of a small group of very large buyers.

Technology upgrades: from terabits to AI‑aware networks

AI is not only increasing traffic volumes; it is changing how cables are engineered and managed. Next‑generation systems are moving into the above‑60‑terabit‑per‑second class and using advanced wavelength‑division multiplexing and coherent optics to squeeze more capacity out of each fibre pair, while AI‑assisted management platforms monitor performance and optimise routing in real time.

Operators and vendors are developing “self‑aware” transport networks that can reallocate bandwidth automatically when AI jobs spike or when faults occur, which is critical for distributed training clusters that span continents. Automation and autonomous underwater vehicles are also making route planning, seabed surveying and cable inspection more efficient, potentially lowering operating risk but increasing upfront software and systems integration costs.

Geopolitics, resilience and route diversity

The same geopolitical tensions that shape AI regulation are also reconfiguring subsea routes. Governments and regulators in regions such as the EU are publishing resilience and security assessments of submarine infrastructure and considering funding for strategic cable hubs, reflecting concern over concentrated capacity and exposure to sabotage or espionage.

For hyperscalers, AI makes downtime more costly, so route diversity and jurisdictional diversification have become central to cable strategy. New systems are designed to bypass geopolitical bottlenecks, connect emerging AI and data‑centre hubs, and provide alternative paths if a key corridor is disrupted, which in turn spreads investment to new coastal markets and landing stations.

What this means for investors

For equity and credit investors, the AI build‑out is turning subsea cables into a higher‑growth, more strategically sensitive segment of digital infrastructure. Telecom operators remain significant owners, but a growing share of future demand and capital is concentrated in a handful of hyperscalers whose long‑term contracts can underpin project financing while also raising questions about bargaining power and counterparty risk.

Key variables to watch include the pace of AI data‑centre expansion, regulatory moves on critical infrastructure and data sovereignty, and technological changes that could further increase per‑cable capacity or alter economics. Rather than a one‑off boom, current forecasts and investment pipelines point to a multi‑year cycle in which subsea cables become an increasingly indispensable, AI‑optimised backbone of global connectivity, with pricing and risk shaped as much by regulatory and geopolitical choices as by raw traffic growth.