Microsoft has announced it is joining a consortium led by Singapore-based Lightstorm to build a next-generation subsea cable, I-2SEA, connecting India with Singapore and Malaysia. The 3,600-kilometer infrastructure addresses the surge in traffic driven by AI workloads, cloud applications, and hyperscale architectures, in a region where bandwidth demand is growing at breakneck speed and the data market has become strategic.

The news comes as India solidifies its position as a crucial node for global cloud providers. Microsoft has repeatedly strengthened its presence in the region with Azure data centers, and this cable is part of a web of connections aimed at reducing latency and increasing resilience between the subcontinent and the financial and tech hubs of Southeast Asia. The move follows well-established patterns in the submarine infrastructure sector: hyperscale players are no longer content to lease capacity on existing cables but instead join laying consortia to gain granular control over routes and available bandwidth.

For those evaluating enterprise AI deployment, the reinforcement of submarine backbone networks adds an important piece to the on-premises versus cloud puzzle. Cables like I-2SEA enable more efficient data flows to regional data centers, but the decision on where to run an LLM—especially for sensitive workloads—involves variables that go well beyond simple bandwidth availability. Data sovereignty, latency requirements for real-time inference, total cost of ownership (TCO), and regulatory compliance are pushing many Indian organizations to explore local, self-hosted stacks, often in hybrid setups.

The Indian landscape, in particular, is shaped by a shifting regulatory environment around data residency and growing enterprise focus on protecting personal and industrial data. In this context, more robust connectivity does not automatically imply a wholesale move to the cloud; rather, it can enable hybrid architectures where LLMs run on-prem for low-latency inference, leaning on the cloud only for periodic training or orchestration services. The trade-offs remain complex and are measured in operational costs, in-house skills, and the ability to manage data pipelines across distributed physical nodes.

The investment in I-2SEA can also be read as a sign of AI supply chain consolidation in the Asia-Pacific region. Singapore and Malaysia are becoming hubs for data centers hosting training and inference hardware, while India represents an enormous market for data consumption and production. For industry observers, the vertical integration of physical networks by cloud giants raises open questions about dependence on a few controlled infrastructures, a factor that strengthens the case for on-premise and air-gapped solutions in regulated sectors or where business continuity is critical.