AI
India’s 8.33 GW Data Centre Pipeline Reshapes the Three-City Race
India’s data centre pipeline has hit 8.33 GW, Knight Frank says, with Mumbai, Hyderabad and Chennai taking most of the planned build. Vizag’s $15B Google hub is the sleeper.
India’s data centre pipeline has hit 8.33 GW, Knight Frank India said in its June 2026 update. That’s more than five times the country’s 1.6 GW of live operational capacity, with Mumbai, Hyderabad and Chennai taking most of the planned build. The race between the three cities is the visible story. The sleeper is Vizag, where Google’s $15 billion AI hub with AdaniConneX is delivering gigawatt-scale single-tenant capacity that wasn’t on the map a year ago.
An 8.33 GW Pipeline Built on 1.6 GW of Live Capacity
India’s data centre industry isn’t short on ambition. The pipeline Knight Frank India mapped has grown to 8.33 GW of planned capacity, of which just 1.6 GW is live and operational today. Spread across 271 facilities covering roughly 23 million square metres of land, the 5x gap between what’s running and what’s planned is the gap the rest of this story is about.
The consultant breaks the future build into three buckets. About 0.32 GW is under construction right now. Another 2.92 GW sits at the “committed” stage, where a hyperscaler or operator has signed. The remaining 5.41 GW is in early-stage planning, with land secured and feasibility work started but no construction yet. The committed slice alone has roughly doubled inside two years, Knight Frank India said.
What changed is the demand mix. Cloud computing, AI workloads and data localisation requirements under India’s Digital Personal Data Protection Act (DPDP Act, 2023) now push hard on hyperscale campuses rather than smaller enterprise rooms. “Hyperscaler and colocation facilities are expected to dominate India’s data centre market over the next decade,” Knight Frank India told the Council on Energy, Environment and Water (CEEW) in stakeholder consultations. IndiaAI Mission demand, IDC-tracked cloud growth and DPDP-driven localisation are pulling the same direction.
| City | Pipeline (GW) | Subsea Cable Landings | Primary Demand Driver |
|---|---|---|---|
| Mumbai | 3.75 | Multiple, western coast | BFSI and IT services exports |
| Hyderabad | 1.93 | None (inland) | GCCs and AI startups |
| Chennai | 1.36 | Multiple, eastern coast | Manufacturing and hyperscaler cloud |

Why Mumbai Still Leads the Three-City Race
Mumbai takes 45% of the future build, the largest single share of any Indian market. The city’s live capacity has run from 96.8 MW in 2016 to 766.6 MW in 2025, nearly eightfold growth in less than a decade. That base, plus the concentration of international subsea cable landings at Versova and along the western coast and a fibre density built around the BFSI and IT services industries, keeps Mumbai at the front of the queue for new hyperscaler deals. Replicating that cable footprint in another city would take years of permitting and trenching.
Three numbers inside Mumbai’s pipeline tell the same story. About 0.17 GW is under construction, 1.54 GW sits at the committed stage, and 2.21 GW is still in early-stage planning. Even the early-stage slice alone exceeds Chennai’s entire forward build, and that’s why the BFSI GCCs anchoring India’s AI finance demand cluster around the western city rather than spread evenly across the country.
Hyderabad and Chennai Are Climbing Different Stairs
Hyderabad’s 1.93 GW pipeline has come up faster than anyone expected. The Telangana government has run an active data centre policy since 2020, offering capital subsidies, land at subsidised rates in tier-II corridors outside the main city, and dedicated single-window clearances. Global Capability Centres (GCCs) for US banks and tech firms have piled in, and Microsoft and Google have already committed hyperscale campuses. The city sits close to Bengaluru’s chip-design corridor, which keeps latency low for AI workloads bouncing between training and inference.
Chennai is climbing a different staircase. Its 1.36 GW pipeline is smaller, but Chennai’s western edge already hosts one of the densest subsea cable clusters in Asia, with fresh landings from the Bay of Bengal and Africa routes arriving regularly. The combination of cable diversity, a manufacturing belt that consumes its own private cloud, and Tamil Nadu’s separate data centre policy makes Chennai the most carrier-grade of the three.
The operators building this capacity aren’t a who’s who of familiar names yet, but the field is consolidating fast:
- AdaniConneX, the Adani Group and EdgeConneX joint venture, targets 1 GW of hyperscale capacity by 2030 across Mumbai, Chennai, Hyderabad and Visakhapatnam.
- Yotta Infrastructure (Hiranandani Group) operates a 50,000-GPU-ready NM1 campus in Navi Mumbai and is expanding to Chennai and Bengaluru.
- NTT GDC India runs 18 data centres across Mumbai, Bengaluru, Delhi NCR and Chennai.
- Sify Technologies operates the Tidel Park campus in Chennai at 3.6 MW IT power and is adding capacity in Mumbai and Bengaluru.
- Reliance Jio has acquired a 40-acre plot and is planning a multi-hundred-megawatt campus.
- STT GDC India runs the Global Cloud Data Centre in Chennai and continues to expand in Mumbai.
Vizag’s $15 Billion Single-Tenant Bet
None of the three front-runners, though, matches the size of the single deal now under construction in Vizag. On October 14, 2025, Adani Enterprises (through AdaniConneX) and Google announced a $15 billion AI hub in Visakhapatnam, Andhra Pradesh, the largest investment either company has committed to India. The plan covers five years from 2026 through 2030 and combines gigawatt-scale data centre operations, a new international subsea cable gateway on India’s eastern coast, and co-investment in transmission lines and clean energy generation inside Andhra Pradesh.
The compute footprint splits across three campuses at Tarluvada, Adavivaram and a third location, with combined capacity approaching 1 GW. By comparison, that single gigawatt roughly matches Chennai’s entire 1.36 GW forward pipeline. Airtel has signed on as a connectivity partner, and the project is structured to land multiple international subsea cables in Vizag, complementing the existing landings in Mumbai and Chennai. Eastern coast landings also reduce the latency for traffic from Singapore and Southeast Asia, a route that currently has to backhaul through Mumbai or Chennai.
Read Google’s $15 billion AI hub announcement from October 2025 or the Adani press release on the partnership and the structure is clear: one anchor tenant, its own campus, its own power purchase agreement, its own subsea landing.
“Visakhapatnam is now set to become a global destination for technology, and we are thrilled to be the architects of this monumental journey,” Gautam Adani, Chairman of the Adani Group, said in the October announcement. Thomas Kurian, CEO of Google Cloud, framed it as the foundation for India to compete in the AI age: “Working with Adani, we will bring our cutting-edge resources closer to communities and customers alike.” That single-tenant structure is what makes Vizag different from any deal currently on the three-city boards, and it is exactly why India’s GCC pilot-to-scale problem now has a real anchor to grow around.
The Power and Water Constraint Behind the Headlines
What could break all of this is power and water, not land. Data centres accounted for roughly 0.5% of India’s electricity consumption in 2024, per Cornell and IEA figures cited by CEEW, and that share is projected to more than double by 2030. Total committed investments across the sector have reached roughly $95 billion through 2025 and are expected to cross $100 billion by 2027, per CBRE data in the same study. Land is being allocated. Capital is showing up. The grid is the bottleneck.
Water tells a parallel story. Indian data centres used roughly 150 billion litres of water in 2024, mostly for cooling, and that figure is also projected to more than double by 2030. Smaller operators under 5 MW still draw mainly on municipal supply. Fifteen Indian states have notified dedicated data centre policies, but only five of those policies embed explicit sustainability provisions, CEEW found. Diesel generator sets remain indispensable for uptime, which means each new megawatt also requires a fossil-fuel backup. Liquid cooling, which uses far less water than air-based systems, is gaining traction but still accounts for a minority of new builds, and only the largest operators have the capital to retrofit.
CEEW’s data centre infrastructure and power/water study is blunt about what comes next.
Stakeholders strongly endorse the need for an integrated land-water-energy nexus framework to guide siting and improve long-term resilience of data centres.
CEEW stakeholder consultation, 2026.
What the Three Cities Are Competing For Next
India’s data centre capacity has nearly tripled from about 520 MW in 2020 to roughly 1.5 GW by mid-2025, per JM Financial and CBRE, and is projected to reach 4.5 to 6.5 GW by 2030, per Colliers and S&P Global. The 8.33 GW pipeline Knight Frank mapped is the upper bound of that curve. The next round of hyperscaler deals, the ones that decide whether Hyderabad closes its gap with Mumbai and whether Chennai keeps its carrier-grade edge, will be settled by who locks in power purchase agreements and water reuse plans first.
Vizag is no longer competing with the three-city group from behind. It is competing with them from the front, on a single-tenant gigawatt-scale deal the others cannot match without their own anchor tenant. The biggest variable in India’s 8.33 GW pipeline isn’t land or capital. It’s whether the country can power and cool the compute it has already ordered, before the next hyperscaler signs another campus.
-
NEWS1 month agoGoogle Search Profiles Build a Follow Graph Inside Discover
-
CRYPTO1 month agoXPL Rallies 30% Ahead of Plasma One Card Tier Launch
-
GAMING1 month agoMicrosoft Xbox Layoffs Start in July as Sharma Slams 3% Margin
-
AI3 weeks agoOracle Cuts 21,000 Jobs in a Year, Cites AI in 10-K Filing
-
AI1 month agoSpaceX’s Google Deal Turns a Rocket Company Into a Cloud Landlord
-
AI1 month agoMoonshot AI Targets $30 Billion in China’s Fastest AI Funding Sprint
-
AI6 days agoWhatsApp Meta Business Agent Reaches India, With a New Pricing Meter
-
NEWS1 month agoOppo’s ColorOS 17 Eligibility List Leaves A-Series Buyers Behind
