Purpose-built subsea routes that land close to major AI compute clusters shorten network hops and cut cross-border transfer costs, which matters for distributed training, dataset replication, and hybrid-cloud inference. Practitioners estimating downstream egress costs and replication windows in the India-Southeast Asia corridor should note the capacity, latency, and landing geography this build adds.
What happened
A consortium including Microsoft, Tata Communications, Singtel, ASEAN Cableship and NEC Corporation will build the I-2SEA submarine cable connecting India with Malaysia and Singapore, Reuters and the Economic Times reported. The system will span 3,600 km, include landing stations at Machilipatnam (Andhra Pradesh) and a new landing near South Chennai, and is expected to be service-ready in Q4 2029. A joint statement carried by the Economic Times says the system is open for capacity commitments and will provide dual Indian landings, with Machilipatnam offering the shortest subsea path to Hyderabad. Reuters additionally reports Lightstorm expects the build to raise the number of AI and cloud zones it connects across India to 29 from 19, and that the company has discussed a potential India listing in mid-2027.
Technical context
Purpose-built subsea systems like I-2SEA typically target three goals relevant to AI workloads: higher aggregate capacity for hyperscale transfers, lower latency between data-center clusters, and route diversity for resilience. The joint statement emphasizes integration with Lightstorm's terrestrial network to extend that reach into regional data-center-interconnect fabrics, which is the pattern hyperscalers and GPU infrastructure providers commonly require for low-latency links and peering on-ramps.
For practitioners
A new landing near Hyderabad and Chennai shortens subsea distance to major Indian AI clusters, which can lower round-trip time for multi-region synchronous training and reduce egress costs for cross-border dataset staging. Dual landings and route diversity are the kind of measures that typically raise uptime and reduce maintenance-related outages, which matters for production ML pipelines and large checkpoint transfers.
What to watch
Watch for published route maps, per-fiber-pair capacity, lit-versus-dark-fiber commercial terms, and latency figures once carriers publish SLAs and capacity products, plus regulatory approvals and landing-station construction timelines that could shift the stated Q4 2029 in-service target.
Key Points
- 1A Microsoft-Lightstorm-led consortium will build the 3,600 km I-2SEA cable linking India, Singapore and Malaysia by Q4 2029.
- 2Dual Indian landings near Hyderabad and Chennai shorten subsea distance to AI clusters, cutting latency for cross-border training and inference traffic.
- 3Lightstorm expects the build to expand its connected India AI/cloud zones from 19 to 29, with capacity open for commercial commitments.
Scoring Rationale
A concrete, multi-party hyperscaler-backed subsea build with real landing geography, capacity targets, and a firm in-service date materially affects AI network topology in a major growth corridor; kept at the upper end of notable since it is an incremental (not paradigm-shifting) infrastructure build.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
