समाचार Samāchār · Tech & AI · 26 October 2026
Tata Communications commits ₹4,350 crore to a 10,000 km fibre spine for India’s AI
The AI Super Highway will link data centre hubs from Mumbai to Chennai first, then the east and west coasts. It adds about 6 petabits per second between FY2030 and FY2036.
India’s AI build-out has a new layer. Tata Communications said on Monday it will invest ₹4,350 crore, about $450 million, in what it calls India’s AI Super Highway. The plan is roughly 10,000 km of new long-haul fibre. It links data centres, cloud platforms and enterprises at high speed and low latency. The company disclosed it in a stock exchange filing. It will fund the build from operating cash flows.
The numbers explain the urgency. The company’s network carries 1.25 petabits per second today and runs at 70–80% utilisation. The new routes add about 6 Pbps in three phases between FY2030 and FY2036. Phase one is the Mumbai–Pune–Hyderabad–Chennai corridor. Later phases reach Visakhapatnam, Bhubaneswar, Jamnagar and GIFT City, and tie the clusters to international cable landing stations. MD and CEO Ganesh Lakshminarayanan sees India’s data-centre-to-data-centre connectivity market passing $1 billion over the medium term. Rupesh Chokshi, who runs network services, said the AI economy needs more than compute.
The demand is on the record. Microsoft has committed $17.5 billion to cloud and AI in India. Google has committed $15 billion to an AI hub in Visakhapatnam. TCS’s HyperVault is building a campus of up to 1 GW in Hyderabad. MeitY puts installed data centre capacity at about 1,575 MW in 2026. It was 375 MW in 2020.
Compute gets the headlines. Power and pipes decide whether it works. Sovereign models and AI at population scale need data centres that talk to each other fast and stay up. This is private capital building that spine from its own cash. The next task is power. Every new campus needs firm electricity, and the grid must keep pace with the fibre. The timeline matters too. 2030 to 2036 is long for an AI cycle that moves in quarters. Faster rights of way and single-window clearances would bring it forward.
What a founder takes from it: India’s AI race will be won on power and connectivity as much as on chips. Founders building AI products should plan around the corridors where fibre, power and data centres meet. Mumbai, Pune, Hyderabad and Chennai come first.