Executive View
The subsea cable industry enters 2026 with a rare combination of strong demand, strategic urgency, and operational scarcity. The internet's international backbone still depends overwhelmingly on fiber systems lying on the seabed. That infrastructure is now being pulled in three directions at once: AI is raising bandwidth needs, hyperscalers are taking more direct control of routes, and governments are treating cable resilience as a national-security issue.
The result is not a simple build boom. It is a build boom colliding with tight manufacturing slots, an ageing cable-ship fleet, skilled-crew shortages, crowded landing corridors, and more intense scrutiny of who owns, lands, repairs, and supplies critical infrastructure.
A few headline numbers frame the picture. Used international bandwidth passed 6.4 Pbps in 2024 after tripling from 2020. More than US$13 billion of new subsea systems are expected to enter service across 2025-2027. The industry sees a typical 150 to 200 cable faults worldwide each year, with most caused by anchors and fishing activity. And an estimated US$3 billion in fleet investment is needed to keep repair capability from falling behind.
The bottom line is that customers are no longer only buying capacity. They are buying deliverability. The strongest commercial position in 2026 rests on a credible schedule, secured ship and factory access, realistic repair commitments, and a resilience story that can survive procurement, insurance, and regulatory scrutiny.
Why 2026 Is an Inflection Point
Subsea cables have always been exposed assets, but the industry has historically absorbed that risk through redundancy, cooperative maintenance zones, and a small but specialized repair fleet. What changes in 2026 is the simultaneity of the pressures. Demand is rising, the build pipeline is full, geopolitical incidents receive board-level attention, and the fleet that keeps the system repairable is ageing faster than it is being replaced.
For two decades, cable conversations often started with capacity and route economics. The new conversation starts earlier in the supply chain: who can manufacture the wet plant, which vessel can lay it, which jurisdiction will permit the landing, where spares are stored, and how quickly a fault can be repaired when several customers in the same zone need attention at once.
Three shifts define the new landscape. Demand is bigger and less forgiving: AI training, cloud replication, streaming, gaming, and enterprise workloads all require higher-capacity routes, and the absolute amount of bandwidth added each year is now large even when growth rates moderate. Ownership is changing: Google, Meta, Microsoft, and Amazon are not just buying capacity, they are shaping route design around their own data-center maps and latency priorities. And repair capacity is the weak link, because cables can be financed faster than new specialist vessels and crews can be created, which makes installation and maintenance access the practical constraint.
Demand: AI Turns Bandwidth Into Strategic Infrastructure
International bandwidth demand has continued to compound from a much larger base. Used bandwidth tripled between 2020 and 2024 and moved beyond 6.4 Pbps. The percentage growth rate has eased from the pandemic surge, but the volume added each year remains enormous. That distinction matters: a lower growth rate on a much bigger base still creates a larger physical build requirement.
AI is the new force multiplier. Training workloads need high-capacity links between campuses. Inference pushes traffic closer to users, which increases the value of low-latency regional routes and diverse intercontinental paths. Cloud providers also need more private, predictable inter-data-center fabric, which is why subsea capacity is becoming part of AI infrastructure rather than a telecom afterthought.
How to read market sizing
Subsea market estimates vary because analysts draw the boundary differently. Some count only telecom systems and wet plant. Others include power cables, installation services, offshore energy links, or wider marine infrastructure. For planning purposes, investment flow and project queue are more useful than a single total-addressable-market number.
The differences matter operationally. Looking at international bandwidth, traffic continues to compound from a larger base, with content and cloud providers driving a large share of demand, so route planning should assume sustained pressure on transoceanic and intra-Asia capacity. Looking at new cable investment, more than US$13 billion of systems are expected to enter service across 2025-2027, which makes factory slots, vessels, marine survey capacity, and landing readiness the scarce inputs. And looking at broader subsea infrastructure, market forecasts can look much larger when power, offshore energy, and installation services are included, which is useful for understanding competition for ships, yards, and skilled labor rather than for sizing a single market.
The Build Pipeline: Hyperscalers Take the Wheel
The biggest structural change is not only how much cable is being built; it is who is controlling the build. Traditional carrier consortia still matter, especially in Asia, but hyperscalers increasingly design routes around their own cloud regions, AI clusters, and data-center energy strategy. This creates more private capacity and can tighten the wholesale capacity available to carriers and enterprise customers.
Meta's Waterworth is a planned global multi-continent system of roughly 50,000 km, showing how far hyperscalers are willing to go to own strategic connectivity. Google's Sol is a trans-Atlantic route linking the United States, Bermuda, the Azores, and Spain, adding path diversity and private cloud capacity on one of the world's busiest corridors. Google's TalayLink is an Australia-Thailand route designed to add a diverse Indian Ocean path, signaling the value placed on bypassing congested or politically sensitive corridors. Meta's Candle reinforces the strategic role of Southeast Asia in hyperscaler network design.
Carrier and consortium models remain commercially relevant in Asia alongside private hyperscaler systems. The I-AM system, backed by NTT DATA, Sumitomo, and JA Mitsui Leasing with ASN and OMS in supplier roles, connects Japan, Malaysia, and Singapore with additional regional connectivity, spanning about 8,100 km with an initial design capacity of 320 Tbps.
Singapore and the Asia-Pacific lens
Singapore remains one of the most important landing and interconnection hubs in the world. Recent research points to 28 cables already connecting to Singapore and at least 13 more in development or scheduled over the next few years. Bifrost, Echo, Nongsa-Changi, SEA-H2X, and SJC2 are central to the 2025-2026 regional picture, while I-AM extends the Singapore-Japan-Malaysia axis into the late-decade pipeline.
Singapore's role is not just geographic. It combines landing density, data-center demand, regulatory predictability, and a growing repair ecosystem. Keppel's acquisition of Global Marine Group in 2025 brought a six-ship cable-lay and repair fleet under Singaporean ownership, while regional cable ships based in Singapore add to the country's strategic relevance. That places customers, systems, landing stations, and repair capability in the busiest part of the map.
Resilience: Faults Are Routine, but the Pattern Is More Serious
Subsea cables break every year. The industry is built around that fact. What has changed is the public and political interpretation of damage in sensitive waters. Cable faults in the Baltic, around Taiwan, and in the Red Sea have pushed subsea resilience into national-security discussions. Even when incidents are accidental, the economic consequence can look strategic.
The September 2025 Red Sea disruptions were a good example. Multiple cables were damaged in a route corridor that carries important Asia-Europe and Middle East connectivity. The likely cause was commercial shipping activity rather than a confirmed act of sabotage, but traffic still had to be rerouted and customers still felt degraded performance. For buyers, the practical lesson was not attribution. It was corridor concentration.
The disciplined reading is not to sell fear but to focus on resilience. Elevated route risk requires route diversity, faster permitting, better spares planning, clearer incident communication, and maintenance capacity that is backed by real vessel access.
Chokepoints matter more than headlines
Chokepoints create outsized impact. The Red Sea, the Mediterranean approaches, the South China Sea, the Taiwan Strait environment, and the dense waters around Singapore all concentrate risk in different ways. A single cut is manageable when spare capacity and alternate routes are abundant. A cluster of faults in a narrow corridor, during a period of ship scarcity, is a different operating problem.
This is why new routes increasingly market themselves on diversity, not only capacity. The customer question is changing from "how many terabits?" to "what happens when the usual corridor is unavailable?"
The Repair and Supply Bottleneck
If demand were the only issue, the answer would be to build more cable. The harder truth is that the supply side is constrained. The wet-plant manufacturing base is concentrated, lead times are longer, specialist cable ships are ageing, and the same shipyards and crews are needed for both new builds and maintenance cover.
Ships are the binding constraint
The maintenance model depends on enough cable ships being available, modern enough, and close enough to reach faults quickly. That assumption is under pressure. Industry fleet analysis indicates that a large share of dedicated maintenance vessels will pass normal service life before 2040 unless replacement orders accelerate. The investment required to sustain repair service levels is estimated at roughly US$3 billion, while cable kilometers in the water continue to rise.
For customers, the effect will show up in waiting time, repair uncertainty, and higher prices for guaranteed cover. Maintenance is no longer a low-visibility aftercare line item. It is becoming a premium resilience product.
Factories are tight too
Manufacturing lead times that once could be discussed in roughly one-year terms have stretched toward multi-year planning windows on major systems. The end-to-end supplier base remains narrow, with SubCom, Alcatel Submarine Networks, NEC, and HMN Tech among the few firms able to supply major telecom systems at global scale. Capital is moving into the sector, but factories, marine assets, and trained crews do not appear overnight.
The constraints compound one another. Longer manufacturing lead times mean customers may underestimate how early they need to commit. An ageing repair fleet makes repair-time promises harder to honor during busy periods. Limited skilled crews mean even available ships can be delayed by specialist labor constraints. Crowded landing corridors create permitting, marine-coordination, and route conflicts that can delay activation. And rising geopolitical screening means ownership, vendor choice, and repair-vessel nationality all receive more scrutiny.
Policy Response: Repair Permission Becomes Part of the Product
Governments are moving subsea cables from a specialist telecom topic into critical-infrastructure policy. The ITU and ICPC launched an international advisory body on submarine cable resilience, followed by working groups focused on timely deployment and repair, risk monitoring and mitigation, and geographic diversity. The EU's 2025 cable-security action plan frames the problem across prevention, detection, response and repair, and deterrence. In the United States, cable-landing-license rules and proposed legislation reflect greater concern over supply-chain, ownership, and foreign-adversary risk.
For operators, the policy direction is mixed. Faster repair permitting and harmonized standards are helpful. More ownership scrutiny, vendor screening, reporting obligations, and security documentation add work. In practical terms, the ability to get permission to repair is becoming almost as important as the ability to repair technically.
Three policy fronts stand out. On permitting, repair plans, territorial-water access, cabotage rules, and vessel clearance need early coordination. On security screening, vendor choice, ownership, repair vessels, and data-routing implications will attract more questions. And on regional harmonization, ASEAN guidance and ITU work may reduce friction over time, but near-term processes still differ by jurisdiction.
Outlook: 2026-2030 Scenarios
Forecasting the subsea market precisely is less useful than tracking which constraints ease and which intensify. Three paths are plausible.
In the base case of managed strain, AI and cloud demand remain strong, lead times stay elevated but manageable, and repair times drift upward in busy corridors, while operators with secured access can charge a resilience premium. In the upside case, new vessel orders, factory investment, and better permitting reduce the gap by the late 2020s, which is good for global resilience but compresses the scarcity premium. In the downside case, a major chokepoint incident occurs during a period of vessel scarcity, repair delays become public, insurance scrutiny rises, and aggressive service-level agreements turn into reputational risk.
Signals to monitor
Several indicators will reveal which path the industry is on. Cable-ship orders, meaning net new vessels and replacement orders, are the clearest indicator of whether repair capacity is catching up. Factory lead times matter too: a sustained move back below about 24 months would signal easing pressure in the wet-plant supply chain. Fault duration in the Red Sea, the Baltic, Taiwan-adjacent waters, and Southeast Asia matters more than headline fault count. Policy execution is worth watching to see whether ITU, EU, ASEAN, and national initiatives actually shorten repair approval windows. And the balance between private and consortium systems, particularly the hyperscaler share of new systems, will shape the future customer base and the availability of wholesale capacity.
Final Takeaway
The subsea cable industry is not short of demand. It is short of slack. In 2026, the strongest players will be those that can translate scarce physical capability into credible commercial promises: build slots that are real, repair cover that is defensible, permits that are planned, and resilience that is designed before the fault happens.
That makes the work more technical and more strategic. The winning position is not simply more bandwidth. It is the ability to get a system live, keep it live, and prove how it will recover when the seabed does what the seabed does.
