- Why the 1-GW headline matters (and what it really means)
- Who’s building vs who’s buying compute
- City-by-city odds: where the next GW lands
- Risks: power, water, clearances, cable landings
The short
- Signal: India’s DC market is graduating to gigawatt-class campuses as AI/cloud demand explodes. Major logos now talk in 500 MW–1 GW campuses, not 10–20 MW halls.
- Momentum: New multi-hundred-MW commitments (Vizag, Chennai, NCR) + revived National Data Centre Policy promise single-window clearances and standardised incentives.
- Ceiling: Grid and water are the hard constraints; renewable PPAs + coastal siting near cable landings will decide speed.
Why “1-GW” is the new yardstick
1 GW ≈ a full campus lifecycle across multiple buildings over 5–8 years. It signals locked-in land, long-lead power, and anchor buyers who can pre-contract capacity. For publishers and investors, the GW bar helps separate PR from shovel-ready plans.
Who builds (developers/colos) vs who buys (workloads)
Builders (recent moves)
- AdaniConneX — financing lines, 1-GW platform by 2030; Chennai & Noida builds in flight.
- Equinix — Mumbai estate + new CN1 Chennai interconnection hub.
- Nxtra (Airtel) — rapid capacity growth; green-energy PPAs; new hyperscale in Chennai.
- Iron Mountain (Web Werks) — JV consolidation; Navi Mumbai and pan-India expansion.
- Digital Connexion (Digital Realty–Brookfield–RIL) — scaling India platform targeting hyperscale.
Buyers (compute demand)
- Hyperscale cloud — AWS/Azure/Google expanding regions and AI-ready zones.
- Frontier AI — model training/inference footprints (foundation models, RAG, agentic).
- Domestic demand — BFSI, telecom, public sector stack, video/commerce, SaaS.
Recent headline commitments (illustrative)
| Player | Headline | Location(s) | Read as |
|---|---|---|---|
| ~$6B plan incl. 1-GW DC + 24/7 power stack | Visakhapatnam, Andhra Pradesh | Coastal campus near cable landings; large renewables slice | |
| OpenAI | Scouting partners for ≥1-GW India DC | TBD (multi-state evaluations) | Frontier AI compute anchor; likely phased build |
| AdaniConneX | Platform targeting 1-GW by 2030 | Chennai, Noida, + | Grid + clean-energy integration; multi-city rollout |
| Equinix | CN1 (Chennai) + Mumbai estate expansion | Chennai, Mumbai | Interconnection hub; subsea proximity |
| Nxtra | Capacity to ~450 MW in ~3 years; new Chennai hyperscale | Chennai, Mumbai, NCR | Renewable PPAs, brownfield + greenfield mix |
Note: “1-GW” is a lifecycle target—delivered in phases; energisation depends on grid upgrades, PPAs, and clearances.
Where the next GW likely lands (near-term odds)
| Metro/Cluster | Why it wins | Watch-outs |
|---|---|---|
| Vizag (AP) | Coast + cable landings + proactive state; large land banks | Transmission upgrades; balanced green/thermal mix for 24/7 |
| Chennai (TN) | CLS hub; established DC parks; new interconnection campus | Monsoon hardening; water + urban planning constraints |
| Mumbai–Navi Mumbai (MH) | CLS density; rich ecosystem; developer pipeline | Power costs/availability; land scarcity |
| NCR (Noida) | Policy support; space for large campuses; captive renewables | Seasonal cooling loads; grid redundancy |
| Hyderabad (TS) | Existing hyperscale region; talent; policy comfort | Transmission ramp speed; water-wise design |
Risks & path dependencies
- Power: Transmission capacity, low-carbon PPAs, and “essential service” status for priority restoration.
- Water: Cooling choices (air/liquid), state-wise extraction charges, and WUE transparency.
- Permits: Single-window timelines vs 30+ approvals; coastal siting/CRZ and CLS permissions.
- Subsea: Cable landing station access/ILD licensing; ROW for terrestrial backhaul.
FAQ
- Is 1-GW “one building”? No—think multi-building campus phased over years.
- Do all gigawatts go to AI? Not just AI; mix of cloud, database, analytics, content. AI raises density/power needs.
- What decides siting? Power first; then subsea proximity, land, policy, climate risk, and skilled ops.