Asia’s AI and cloud expansion is putting pressure on a less visible part of the digital infrastructure stack: the undersea cables that connect data centers to global networks. Kapil Kumar Jain, vice president of global network services at Tata Communications, says that if cable and end-to-end network investment fails to keep pace with computing capacity, businesses risk losing returns on AI spending through outages, higher latency and rising operating costs.
Cable failures have disrupted banks and cloud services
Undersea cables are the physical foundation of cross-border digital activity, not distant telecommunications infrastructure. More than 99% of international data travels through undersea cables rather than satellites, according to the International Telecommunication Union. Banks, fintech platforms, cloud providers and supply chains all rely on these links to move data continuously.
Recent outages have shown how widely disruption can spread. Cable damage in 2024 interrupted internet services across parts of Africa, affecting banks, fintech platforms and cloud operations. In 2025, damage to cables in the Red Sea severely disrupted connectivity between Asia and the Middle East. Traffic had to take longer routes, increasing latency for cloud services and costs for businesses that depend on real-time digital operations.
AI workloads are driving network upgrades
AI workloads and the growth of hyperscale data centers are pushing up bandwidth requirements. The article identifies 400 Gbps as the current benchmark, with 800 Gbps connections beginning to emerge. As governments and companies race to expand computing capacity and data centers, whether networks can scale alongside them has become a practical constraint on AI infrastructure in Asia.
Jain cited Tata Communications research finding that 77% of business leaders rank AI investment as a strategic priority at board level, while 65% still rely on legacy infrastructure or facilities that remain under development. As AI traffic becomes harder to predict, businesses need data-center interconnects that can scale quickly with demand to maintain stable operations and manage costs.
Without sufficient capacity and alternative routes between undersea cables, terrestrial networks and data centers, congestion or outages can constrain data flows even when computing resources are in place. Network reliability therefore affects more than IT operations: it can determine whether AI projects deliver as intended.
Singapore and India serve as regional traffic hubs
Singapore and India are key data transit hubs in Asia. Singapore connects Southeast Asia and hosts one of the world’s densest data-center clusters. India links Asia with the Middle East and Europe; Mumbai and Chennai are important cable landing locations, while the country’s data-center capacity is expanding. Their strategic positions and infrastructure support cross-border data flows, making route diversity and resilience increasingly important.
Debate over Singapore’s AI ambitions often focuses on computing power, talent and regulation. But the speed, reliability and security of data transmission also affect how effectively those facilities can operate. Bottlenecks at coastal landing stations could erode the speed advantage of transoceanic cables, so network planning needs to consider subsea cables, terrestrial fiber and data-center connections as one system.
ASEAN guidelines recommend a 10-day repair approval window
Resilience depends on more than laying additional cables. It also requires capacity planning, diverse routes, protection and repair readiness, and ongoing lifecycle management. Monitoring systems, spare parts, links between subsea and terrestrial networks, and clear long-term governance all shape whether damage to one cable leads to wider service disruption.
Public policy is also addressing repair response times. The recently adopted ASEAN Enhanced Guidelines for Strengthening Undersea Cable Resilience and Repair recommend keeping repair approval processing within 10 days. The recommendation links repair readiness to regional economic stability, though resilience in practice will depend on coordination among network operators, repair resources and alternative routes.
For businesses, the capacity and returns of computing facilities and data centers are relatively easy to measure. Cables, landing stations and backup routes are less visible and harder to justify as standalone investments. Yet insufficient connectivity can leave costly AI infrastructure exposed to outages, congestion and higher operating expenses. As AI use expands, the capacity, route redundancy and repair readiness of Asia’s undersea networks will remain central to the continuity of digital businesses.