Tata Communications To Pour Up To $450 Mn To Build AI Connectivity Infrastructure

Tata Communications plans to build a new 10,000-kilometre fibre network across India. It will connect major data centres and AI hubs, then link with the company’s existing global network. The goal is to support rising demand for data-centre connectivity and computing infrastructure.

“AI-ready” means the network is designed for the demands of AI workloads. The article highlights high capacity, reliability and low latency. These features help transfer large amounts of data between computing infrastructure and data centres with fewer delays and dependable performance.

The first phase will cover the Mumbai-Pune-Hyderabad-Chennai corridor. Later phases will reach emerging data-centre clusters elsewhere in India. The programme is scheduled between FY30 and FY36 and will add around 6 petabits per second of capacity, financed through operating cash flows.

Tata Communications plans to invest $400-450 million, or around ₹4,350 crore, in the expansion. The money will build a new high-speed fibre network across 10,000 kilometres of India. The investment is intended to capture growing demand for data-centre connectivity and AI infrastructure.

The programme will add around 6 petabits per second, or pbps, of network capacity. Once complete, Tata Communications expects total capacity to reach approximately 7.25 pbps. This capacity is the network’s overall ability to carry data across its connections.

Construction is scheduled between FY30 and FY36. The company plans to finance the programme through operating cash flows. Its existing network is already operating at around 70%-80% utilisation, making additional capacity important as hyperscalers, cloud providers and AI companies expand their requirements.

The first phase will connect Mumbai, Pune, Hyderabad and Chennai. Tata Communications chose this route because it is one of India’s key growth corridors for AI and hyperscale infrastructure. The company will use it as the starting point for the wider national network.

The corridor will connect important computing and data-centre locations through new high-speed fibre. This creates a direct route for moving large data volumes between facilities. The network is designed to provide the capacity, reliability and low latency needed by AI and hyperscale users.

Later phases will extend connectivity to emerging data-centre clusters in other parts of India. The broader plan also includes links to international cable landing stations. That approach is intended to connect the country’s growing domestic AI infrastructure with global AI hubs and international data flows.

The new network will give AI and digital infrastructure users a faster, more dependable way to exchange data. Tata Communications says it is designed for the high capacity, reliability and low latency needed between computing infrastructure and data centres.

For example, an AI platform could use the fibre route to move large datasets between a data centre and separate computing facilities. Cloud providers and hyperscalers could also use the added capacity to connect infrastructure across major hubs. Faster transfer and fewer delays support the movement of data required by these operations.

The beneficiaries include AI platforms, cloud providers, global capability centres, enterprises and digital-native businesses. The network will first serve the Mumbai-Pune-Hyderabad-Chennai corridor, then expand to other clusters. International links will also support reliable data exchange with global AI hubs.

AI workloads depend on both computing power and data movement. GPUs and other systems perform the calculations, but data must reach those systems from storage, data centres or other computing facilities. As AI demand grows, the network carrying that data also becomes important.

A high-capacity network can carry larger data volumes at once. Low latency reduces the delay in transferring information between connected facilities. Tata Communications says its planned network is designed for these requirements, with the reliability needed by AI platforms, cloud providers and hyperscalers.

The article points to demand from foundation model builders, AI-native startups, enterprises, research institutions and public-sector programmes. E2E Networks is also expanding GPU capacity for training, inference and multimodal applications. Together, these investments show AI growth driving demand for both compute infrastructure and specialised connectivity.

Tata Communications’ domestic fibre routes will connect India’s main AI and data-centre hubs. The expansion will also connect those routes to international cable landing stations. These stations provide the connection point between domestic network infrastructure and international cables.

In practice, data can travel from a computing facility through the new national fibre network, reach a cable landing station, and then continue toward international infrastructure. The article says this will strengthen connectivity between India’s domestic infrastructure and global AI hubs. It is intended to support large volumes of data moving quickly and reliably.

The first domestic route is the Mumbai-Pune-Hyderabad-Chennai corridor. Later phases will reach other Indian data-centre clusters. Once connected with the company’s existing global network, these routes can support the needs of hyperscalers, cloud providers, AI platforms and enterprises operating across countries.

Bandwidth describes the amount of data a connection can carry over a given period. Latency is the delay before data reaches its destination. Network capacity is the total amount of traffic the wider network can handle. Higher bandwidth and capacity support larger data flows, while lower latency makes exchanges quicker.

For AI systems, these measures affect how efficiently data moves between computing infrastructure and data centres. A high-capacity fibre route can carry more traffic from many users. Low latency helps reduce delays between connected facilities. Reliability matters too, because users need consistent connectivity for data-intensive workloads.

Tata Communications plans to add around 6 pbps of capacity, taking its total to approximately 7.25 pbps. Its existing network is already 70%-80% utilised. The planned 10,000-kilometre network is therefore intended to provide more room for AI platforms, cloud providers and enterprises.