India's data centre sector is rapidly emerging as one of the country's fastest-growing commercial real estate asset classes as artificial intelligence (AI) fuels demand for built-to-suit facilities and hyperscale campuses. Recent developments, including the Reliance-Meta partnership and AirTrunk's USD 30 billion investment commitment, have reinforced the sector's momentum, while CBRE estimates cumulative investment commitments could exceed USD 180 billion in 2026.
As AI adoption gathers pace, the conversation is gradually shifting beyond investment announcements. Power availability, specialised infrastructure and project economics are increasingly influencing where and how data centres are developed, making them a distinct segment within India's commercial real estate market.
Why Built-To-Suit Wins
Global hyperscalers account for nearly 50-55 per cent of India's data centre market activity, while cloud service providers secured more than 300 MW of capacity during the first quarter alone. The growing share of hyperscalers is accelerating demand for built-to-suit (BTS) campuses designed to support AI workloads and specialised computing infrastructure.
Anshuman Magazine, Chairman & CEO- India, South-East Asia, Middle East & Africa at CBRE, said the structure of the market itself limits speculative development because built-to-suit facilities are committed before construction begins. He added that hyperscalers prefer customised campuses as they provide greater flexibility in securing power access and meeting specialised AI infrastructure requirements.
Further, Magazine said speculative AI-focused construction remains limited because the capital expenditure involved is difficult to justify without an anchor tenant. He added that India's data centre pipeline continues to remain demand-led, reflecting genuine long-term expansion plans rather than capacity being built ahead of proven demand.
Is Land Still King?
Traditional enterprise data centres are designed to support 5-10 kilowatts (kW) per rack, while AI-ready facilities require 40-100+ kW per rack, significantly increasing demand for power infrastructure, advanced cooling systems and high-density computing environments over conventional real estate specifications.
Against this backdrop, Amit Sarin, Managing Director of Anant Raj, said land accounts for only a small proportion of the overall cost of developing a data centre, while access to reliable power infrastructure remains a far more critical consideration. He added that AI-ready facilities can accommodate significantly higher computing capacity within a relatively smaller land footprint.
Further, Sarin said the company has not encountered power allocation or grid connectivity issues in Haryana. He added that capacity creation is being undertaken in phases, allowing operators to align expansion with customer demand while maintaining commercially viable utilisation levels.
Can Returns Keep Pace?
The growing interest in data centres is also reshaping how institutional capital evaluates commercial real estate. Unlike traditional asset classes, investment returns are increasingly linked to infrastructure readiness, long-term lease visibility and capital intensity rather than location alone.
Echoing the focus on project viability, Binitha Dalal, Founder and Managing Director of Mt. K Kapital, said data centres are no longer competing only within their own sector for investment. She added that investors increasingly compare them with residential, commercial, retail and hospitality projects, where conventional real estate assets often offer stronger margins.
Further, Dalal said rising land values are only one part of the equation, as power infrastructure, cooling systems and specialised construction account for a substantial share of project costs. She added that the commercial viability of AI-ready facilities will ultimately depend on disciplined capital deployment and efficient execution.
Will AI Demand Sustain Growth?
As AI investment gathers pace, an important question is whether India's data centre market can sustain its current growth trajectory beyond a handful of global hyperscalers. With AI-native cloud providers, global capability centres (GCCs) and enterprise users emerging as new occupiers, the breadth of future demand is increasingly coming into focus.
Building on this shift, Magazine said concentration among a handful of global hyperscalers is a legitimate consideration, but long-term contracts substantially mitigate the risk. He added that hyperscalers typically sign agreements spanning 10-15 years, providing developers and lenders with strong revenue visibility while the demand base continues to diversify.
Offering an investment perspective, Dalal said sustained AI adoption is strengthening investor interest in data centre real estate despite the sector's high capital requirements. She added that project selection and execution discipline will remain key differentiators as investment activity gathers pace.
Can Expansion Stay Disciplined?
India currently has 164 operational data centres, with more than 700 MW of capacity under construction. Installed capacity is projected to reach 1.7-2.0 GW by the end of 2026, underscoring the need for disciplined supply addition as AI-driven demand continues to accelerate.
Reiterating the need for disciplined growth, Sarin said capacity addition is being planned in phases rather than through large speculative expansions. He added that the company does not see a significant risk of capacity running ahead of demand, as expansion continues to be aligned with customer requirements.
On India's competitive positioning, Sarin said the country remains at an early stage of its data centre growth journey and continues to offer relatively lower capital and operating costs than many global markets. He added that these advantages strengthen India's competitiveness as AI and cloud infrastructure deployment gathers pace. |
