Data Centres in India: The Business Model Behind Power-Hungry Digital Infrastructure
A data centre is often marketed as digital real estate. Economically, it is a power-constrained infrastructure business with expensive fit-out, demanding uptime requirements and long customer sales cycles. Land matters, but reliable power, network connectivity, cooling efficiency and contracted utilisation usually matter more.
Finin2min Summary
- Announced megawatts are not the same as commissioned, contracted or revenue-generating capacity.
- Revenue can be based on colocation space, power committed, managed services and connectivity.
- Utilisation and power-usage efficiency have a direct effect on return on invested capital.
- Long contracts can improve visibility but create concentration and customer-credit risk.
- Water, grid access, renewable claims, cyber security and physical resilience are material diligence areas.
India's growing digital demand, cloud adoption and AI workloads support the infrastructure case, but a strong sector story does not guarantee strong project economics. Data centres require heavy upfront investment before full occupancy. Delays in power connection, customer ramp or equipment procurement can materially extend the payback period.
Read capacity in four stages
Separate planned, under-construction, commissioned and contracted capacity. A press release may aggregate all four. Revenue begins only when the facility is ready and customers deploy equipment or consume contracted service. Investors should ask for live IT load, contracted load, billing commencement and pipeline rather than relying on total project megawatts.
Understand the revenue contract
Retail colocation sells smaller footprints and services to many customers, while hyperscale arrangements may commit large capacity to a few cloud or technology clients. Contracts may include power pass-through, minimum commitments, escalation, service-level credits and fit-out obligations. Gross revenue can therefore overstate economic value when electricity is largely passed through.
Power and cooling drive operating economics
Power is both the core input and a capacity constraint. Power Usage Effectiveness compares total facility energy with IT equipment energy; lower is generally better, but climate, utilisation and measurement boundaries matter. Cooling design, redundancy and renewable sourcing can change cost and capital needs. Efficiency should be assessed alongside uptime and service quality, not in isolation.
Capital structure must match ramp-up
A facility may have construction debt before contracted cash flow stabilises. Interest during construction, equipment replacement and customer-specific fit-out can increase funding needs. Project models should stress delays, lower occupancy, power-cost changes, customer concentration and refinancing. The relevant return is cash yield on total deployed capital after maintenance—not an EBITDA multiple alone.
What the Viral Version Usually Misses
Viral maps showing 'data-centre capacity' can mix operational sites with announcements and can confuse building power with IT load. They may also treat AI demand as unlimited. AI workloads can require higher rack density and different cooling, so existing capacity is not automatically suitable. The sector opportunity must be translated into site-level contracts and cash flow.
Worked Scenario: A 30 MW project with slow occupancy
A project commissions 10 MW in phase one at a total deployed cost of ₹750 crore. It signs 6 MW, but only 3 MW is billing in the first year because customer deployment is staggered. The model that assumes 60% utilisation from day one materially overstates cash flow. A better model separates commissioned, contracted and billing load and includes fit-out timing, electricity pass-through and service-level credits. The same announced capacity can produce very different returns depending on ramp.
Practical Decision Checklist
- Reconcile announced, commissioned, contracted and billing MW.
- Separate electricity pass-through from value-added revenue.
- Review grid connection, redundancy and renewable sourcing evidence.
- Measure customer concentration and contract renewal exposure.
- Stress occupancy ramp, capex overruns and refinancing.
- Assess water, cyber, physical-security and community constraints.
Article-Specific Q&A
Is a data centre a real-estate investment?
It has real-estate elements, but power, equipment, connectivity, operations and customer contracts are central. Valuing it like an ordinary building can miss the main risks.
What does MW capacity mean?
Definitions vary. It may refer to IT load, facility load or planned power. Diligence must confirm the measurement and operating stage.
Why is utilisation so important?
Most capital is committed before the facility is full. Higher billing utilisation spreads fixed costs and improves return, while slow ramp leaves expensive infrastructure underused.
Does renewable power make a facility green?
It is one factor. The claim should specify sourcing, matching period, additionality and treatment of backup generation, water and equipment lifecycle.
Are hyperscale contracts always safer?
They can provide scale and visibility but may concentrate revenue in a few customers with negotiating power and specific technical requirements.
What is the key AI-related risk?
Higher-density AI racks may require different power and cooling. A facility designed for traditional loads may need additional capital before it can serve those workloads.
Sources and Verification Trail
- Ministry of Electronics and Information Technology: Official digital infrastructure and cloud-policy context. — https://www.meity.gov.in/
- Central Electricity Authority: Primary electricity-system data and planning source. — https://cea.nic.in/
- Bureau of Energy Efficiency: Official energy-efficiency programmes and standards context. — https://beeindia.gov.in/
- IndiaAI: Official AI compute and infrastructure context. — https://indiaai.gov.in/