Tata Communications to Spend ₹4,350 Crore on a 10,000-km 'AI Super Highway' Linking India's Data-Centre Hubs

India's AI build-out has so far been told mostly through data centres: megawatts of capacity, GPU counts and liquid-cooling halls. On Monday, 26 October 2026, Tata Communications put the spotlight on the part of the stack that ties those buildings together. The company said it will invest about ₹4,350 crore, or roughly $400–450 million, in what it calls an "AI Super Highway", a network of about 10,000 km of AI-ready fibre connecting India's major data-centre hubs, cloud platforms and enterprises.

What Tata Communications announced

According to the company's announcement and its stock-exchange filing, as reported by The Tribune, Zee Business, Elets CIO and others:

• Scale: roughly 10,000 km of new, AI-ready fibre routes.
• Capacity: the build is expected to add about 6 petabits per second (Pbps) of capacity between FY2030 and FY2036. Current national capacity on these routes is put at about 1.25 Pbps, running at an estimated 70–80% utilisation.
• Phasing: the rollout comes in three phases. The first covers the Mumbai–Pune–Hyderabad–Chennai corridor. Later phases extend to East Coast and Western clusters, including Visakhapatnam, Bhubaneswar, Jamnagar and GIFT City.
• Global links: the network will connect these clusters to international cable landing stations and to Tata Communications' global backbone.
• Funding: the capital spending will be met from the company's operating cash flows.

Coverage quoted Rupesh Chokshi, the company's Chief Product and Technology Officer, on the plan. The company also said India's data-centre-to-data-centre connectivity market could exceed $1 billion over the medium term. One report, by NewsBytes, said the build would take three to five years and that Tata Communications is aiming for a market share of 30–40% in data-centre interconnect; other outlets did not carry those targets.

Why AI needs a different kind of network

Traditional enterprise and consumer networks are built around many small flows: web pages, video streams, app traffic. AI workloads behave differently. Training a large model, replicating data sets between sites, or serving inference from several data centres at once produces very large, sustained flows between a small number of facilities. Those flows need high capacity, low and predictable latency and resilient routes, so that a fibre cut does not stall a training run or take an AI service offline.

That is the gap Tata Communications is targeting. When existing long-haul routes are already running at 70–80% utilisation, a wave of new AI campuses would quickly congest them. Adding about 6 Pbps on purpose-built routes is meant to let data-centre operators and cloud providers spread workloads across cities, back up data to distant sites and connect Indian capacity to customers abroad.

The data-centre boom it is chasing

The announcement cited Ministry of Electronics and Information Technology (MeitY) figures showing India's installed data-centre capacity rising from 375 MW in 2020 to about 1,575 MW in 2026, more than a fourfold increase in six years.

The pace has been especially visible this month. In Chennai alone, Vultr and AMD have outlined a 20 MW AMD Instinct supercluster and CapitaLand has unveiled a liquid-cooled data centre built for AI workloads. Neysa has chosen STT GDC's Noida facility for an AI cloud build-out, and new entrants have announced gigawatt-scale ambitions. The first phase of the AI Super Highway, along the Mumbai–Pune–Hyderabad–Chennai corridor, runs through exactly the cities where that capacity is clustering.

The later phases are just as telling. Visakhapatnam and Bhubaneswar sit on the East Coast, where new subsea cable landings and large AI campuses are being planned. Jamnagar is a major industrial hub in Gujarat, and GIFT City is India's international financial services centre, where low-latency, high-reliability links matter to banks and exchanges.

What it means for India's AI ambitions

India's AI policy has focused on compute access, through the IndiaAI Mission's subsidised GPU capacity, and on home-grown models. Both depend on networks that can move data at scale between where it is produced, where it is stored and where it is processed. A dedicated, high-capacity backbone lowers a practical barrier for start-ups and enterprises that want to use GPU capacity in another city, and for global cloud and AI companies that want to serve users from Indian sites.

There is also a resilience and sovereignty angle. As more public services, financial systems and enterprise data move onto AI platforms hosted in India, the routes that connect those platforms become critical infrastructure. Multiple, diverse high-capacity paths reduce the risk that a single failure disrupts essential services.

For Tata Communications, which already operates one of the world's larger subsea and terrestrial networks, the plan is a bet that data-centre interconnect will become a major business line in India rather than a niche. Funding it from operating cash flows, rather than new debt or equity, suggests the company sees steady demand rather than a speculative build.

What to watch

The capacity numbers in the filing run to FY2036, so this is a long project. The key milestones will be the commissioning of the first corridor, the customers who sign up for capacity on it, and how quickly the East Coast and Western phases follow. Competition will matter too: telecom operators and data-centre companies are also building their own interconnect, and pricing for bulk capacity will shape how widely AI workloads can be spread across India.

If the build keeps pace with the data-centre boom, it could help ensure that India's AI capacity works as a connected national system rather than a set of isolated islands.

Sources

• https://www.tribuneindia.com/news/ai-infrastructure/tata-communications-to-invest-rs-4350-crore-to-build-indias-ai-super-highway-connect-major-data-centre-hubs
• https://www.lightreading.com/ai-machine-learning/tata-communications-to-build-india-s-ai-super-highway
• https://qz.com/tata-communications-ai-fiber-network-india-102626
• https://cio.eletsonline.com/technology/tata-communications-to-invest-450-million-in-indias-ai-super-highway-linking-ai-data-centre-corridors/76404/
• https://www.zeebiz.com/companies/news-tata-communications-to-invest-450-million-in-ai-superhighway-plans-10000-km-fibre-networkv-404554
• https://www.newsbytesapp.com/news/science/tata-communications-to-invest-450m-in-india-s-ai-super-highway/story
• https://hdfcsky.com/news/tata-communications-plans-450-million-ai-super-highway-shares-rise-1-73-percent