Part 1 established the demand—why India needs data centers. Part 2 answers the supply question: where does the infrastructure buildout actually happen, and what separates the states capturing capital from those watching it flow elsewhere.
Microsoft: $17.5 bn. Amazon, 24 hours later: $35 bn. Google's Vizag AI hub: $15 bn. That's $67.5 bn committed in under 60 days—and these aren't distant promises. Microsoft's Hyderabad campus goes live mid-2026. Google's gigawatt-scale facility is already under construction.
India's data center capacity sits at 1.5 GW today. It needs to hit 14 GW (~10x) by 2035. Seven states are chasing this opportunity. But December 2025 revealed the pattern: capital concentrated in three states that built the infrastructure first.
What do Developers & Hyperscalers Actually Evaluate
In the site selection - Incentives matter, but they're tertiary. These four deal-breakers come first, and if a location fails any of them, no amount of tax holidays rescue the deal.
1. Power Infrastructure (The Non-Negotiable)
Can the state deliver 100 MW continuous to a single site? Not on paper—actually, operationally, today?
Hyperscale facilities need proximity to 220kV or 400kV substations. Beyond a certain distance, you're building your own substation: $15-25M in additional capex and 12-18 months added to timelines. On a project where speed to revenue determines IRR, that delay kills the economics even before you calculate the cost.
Tamil Nadu doesn't just offer concessional electricity tariffs—they provide dedicated power feeders. Maharashtra guarantees 24×7 supply through state DISCOMs with dual grid infrastructure. These aren't marketing claims in policy documents. They're operational realities determining whether a site works at all.
Subsidized electricity rates mean nothing if the grid can't handle the load. Developers filter locations first by power availability, then optimize for cost.
2. Approval Timelines (Speed as Infrastructure)
Maharashtra cut approval timelines from 18 months to six through actual single-window implementation. It's 12 months of carrying costs on land you've already acquired, 12 months closer to revenue, 12 months ahead of competitors.
On a 30 MW facility, those 12 months represent $3-5M in opportunity cost. On a 100 MW hyperscale project, you're talking $10-15M in saved carrying costs plus faster time to market.
Telangana's approach: 100% stamp duty reimbursement matters, but their Data Centre Economic Zone with pre-allocated land near existing power infrastructure matters more. You're no longer fighting land conversion battles or environmental clearances.
Single-window clearances only work if the window has authority. States where "single-window" means "one place to collect documents that still route through twelve departments" aren't competing.
3. Land Availability (Not Just Acreage—Proximity)
Hyperscalers need 50-100 contiguous acres. But not just any 50 acres—parcels near substations, fiber routes, and flat terrain with low flood risk. That's not general commercial inventory. That's industrial corridors, ex-manufacturing sites, or purpose-built DC parks.
States that land-banked these parcels three years ago are closing deals today. States announcing they'll "identify suitable land" are 36 months behind—by which time the capital has deployed elsewhere.
The land equation isn't about total available acreage in a state. It's about shovel-ready parcels that pass all the other filters simultaneously. How many sites in your state can deliver 100 MW power within 2 km, fiber within 5 km, flat terrain, and industrial zoning without conversion battles?
4. Fiber Connectivity (Cables Landed, Not Planned)
Mumbai commands 45% of upcoming capacity for one reason: multiple submarine cable landing stations within city limits. Latency to Singapore, Europe, North America—all measured in tens of milliseconds.
Chennai hosts cable landing stations connecting Southeast Asia.
States without cable landing stations aren't competing for international gateway traffic. They're fighting for edge computing and domestic enterprise colocation—different use cases, different economics, different anchor tenants.
You can't retrofit submarine cable infrastructure quickly. Landing stations require coastal access, regulatory approvals spanning multiple agencies, and coordination with international consortia.
Then Come the Incentives
After a site passes the infrastructure filter, incentives optimize economics:
- Electricity duty exemptions (Maharashtra: 60% for 15 years) save $2-4M annually on a 30 MW facility
- Stamp duty waivers (Telangana, UP: 100% reimbursement) cut $1-3M upfront
- Capital subsidies (UP, Karnataka: 10-25% of eligible costs) offset construction expenses
- Renewable energy waivers (Karnataka, Tamil Nadu) address ESG mandates and reduce cross-subsidy charges
The Three-Tier Reality: Where Capital Is Actually Going
Tier 1: The Infrastructure-First States
These states didn't announce policies and wait for developer interest. They built substations, allocated land, streamlined approvals, and created certainty. When hyperscalers arrived with $67.5 billion, the infrastructure was operational.
Maharashtra (Mumbai, Navi Mumbai, Pune)
Twelve submarine cable landing stations. Eighty-five operational data centers—more than Tamil Nadu and Karnataka combined. Maharashtra will capture 40-45% of India's upcoming capacity.
The competitive advantage isn't incentives—it's infrastructure layering: dual power grids with guaranteed 24×7 supply, approval timelines compressed from 18 months to six, industrial corridors with pre-zoned land near transmission infrastructure, and twelve submarine cable stations creating India's densest fiber convergence point.
Tamil Nadu (Chennai)
East coast gateway to Southeast Asia. Thirty-five operational facilities. Chennai complements Mumbai geographically—Mumbai handles west-bound traffic to Europe and Middle East, Chennai handles east-bound to Singapore, Thailand, Malaysia.
Tamil Nadu treats data centers as industrial infrastructure: dedicated power feeders with concessional tariffs, priority renewable allocation, and building codes streamlined at the bureaucratic level. Automotive and manufacturing digitization creates enterprise colocation demand beyond hyperscale—Tier 2 and Tier 3 suppliers need low-latency access to OEM systems, creating steady baseload underneath hyperscale deployments.
Telangana (Hyderabad)
First state in India to launch a dedicated data center policy—being first meant proving the model worked, not copying templates. One hundred percent stamp duty reimbursement, subsidized land in pre-approved zones, dual power grids engineered specifically for data center loads before hyperscalers arrived.
Telangana didn't wait for federal infrastructure funding. They invested in dual grids, allocated land near substations, and created Data Centre Economic Zones where approvals, power, and land converged.
Tier 2: The Execution Race
These states understand the infrastructure-first model. Whether they capture Tier 1 status depends on execution speed over the next 18 months.
Karnataka (Bangalore)
Thirty operational facilities, massive IT talent pool, enterprise demand from startups to MNCs. Competitive policy: waivers on cross-subsidy surcharges for renewable sourcing, land subsidies outside Bangalore Urban district, building code relaxations.
The missing piece: power infrastructure lagging policy ambition. Karnataka is investing in 400kV transmission corridors and substation expansions now. The question is whether grid upgrades complete before developers commit the next capital wave.
Uttar Pradesh (Noida, Greater Noida)
Twenty-five operational facilities. Proximity to Delhi creates enterprise demand—government agencies, financial services, corporate headquarters. UP's competitive move: capital subsidies specifically designed for hyperscale facilities that meaningfully offset construction costs on 50+ MW projects.
The challenge: power reliability. Delhi-NCR faces summer deficits and load-shedding. UP is addressing this with dedicated transmission corridors and fast-tracking captive power plant approvals. If they deliver in 18 months, they capture Delhi-NCR demand.
Pune (Maharashtra)
Benefiting from Maharashtra's statewide infrastructure while offering 30-40% lower land costs than Mumbai. As Mumbai saturates, Pune becomes the natural overflow—if it delivers the same power and fiber connectivity. Currently competing for enterprise colocation rather than hyperscale anchors. Pune's advantage: it's already within Maharashtra's infrastructure envelope. The substations exist. The approval process is identical.
Tier 3: Policy Without Prerequisites
These states announced policies without building infrastructure. The gap between announcement and operational deployment: 3-5 years minimum.
Gujarat
Policy allows captive power plants—which signals developers can't rely on the state grid. Gujarat has industrial land and coastal access. What it lacks: transmission infrastructure at hyperscale capacity and operational submarine cable landing stations. "Developing" and "operational" represent a 3-5 year gap.
West Bengal (Kolkata)
Launched policy in 2021. Operational deployments remain limited. West Bengal needs to demonstrate approvals move at promised pace and power infrastructure can support multiple hyperscale facilities simultaneously. Policy credibility requires proof points—those take 24-36 months.
Odisha
The wildcard. Planning a submarine cable landing station in Puri that could create a new east coast gateway. But "planning" and "operational" represent 3-5 years. During that window, Mumbai and Chennai are landing six new cables and locking in 10-15 year hyperscale commitments.
If Odisha lands a cable by 2027 with 200+ MW substation capacity nearby, they create a legitimate Chennai alternative. If timelines slip to 2029-2030, the window closes—capacity has already deployed elsewhere.
The Tier structure isn't static. Four forces could elevate Tier 2 states to Tier 1 status—or expose vulnerabilities in current leaders. These aren't distant trends; they're deployment decisions happening in 2025-2026 that determine the 2027-2030 landscape.
Two states have a shot at Tier 1 status if they execute in 18 months: Karnataka and Uttar Pradesh. Both are investing in grid capacity now. Whether they succeed depends on one question: can they complete transmission upgrades before the next capital wave allocates?
The infrastructure lead time is 18-24 months. States that started building in 2024 are capturing 2026 deployments. States announcing investments in 2025 are competing for 2027—if they execute flawlessly.
India will add 500+ MW annually through 2030—that's $7-10 billion in facility investment compounding for a decade. December's acceleration isn't temporary. It's the new baseline.
"The $100 billion is deploying now. Capital doesn't wait for infrastructure to catch up. It goes where infrastructure already exists."
.jpg)
