In 2025, India’s data centre expansion has moved well beyond the stage where scale alone commands attention. The defining question is no longer how quickly capacity can be added, but how efficiently that capacity can be powered, cooled and operated in a market characterised by rising rack densities, prolonged heat stress and tightening grid constraints.

Electricity now accounts for more than 50 per cent of operating costs at a typical colocation facility, turning energy performance into a direct determinant of long-term competitiveness. The impact is already being felt beyond individual campuses. States hosting dense data centre clusters are reworking transmission planning assumptions. Karnataka’s decision to upgrade select infrastructure to 765 kV to support data centre loads underscores a broader reality: digital infrastructure has become a material and permanent category within state power systems.

Measuring footprint

New Indian campuses are being designed with sub-1.4 power usage effectiveness (PUE) targets, while recognising that realised performance will ultimately depend on utilisation patterns and climatic extremes. The challenge is intensifying, as artificial intelligence (AI)-ready deployments push rack densities up. Without adaptive power and cooling architectures, increased density can rapidly erode nominal efficiency gains.

Water usage has become an equally important consideration. Facilities reliant on evaporative cooling face heightened scrutiny in water-stressed regions, prompting a shift towards air-cooled and hybrid systems despite higher upfront capital costs. Water usage effectiveness is no longer treated as a sustainability footnote; it is increasingly viewed as a risk variable that can shape regulatory engagement and the social licence to operate.

Carbon intensity has emerged as a strategic metric. With grid electricity still dominated by fossil fuels in many states, operators are under pressure to decouple compute growth from emissions growth. This has forced efficiency optimisation and renewable procurement to advance in tandem, rather than as sequential phases.

Engineering efficiency

Cooling evolution

Cooling remains the single largest non-information technology (IT) energy consumer within data centres. Operators are moving away from static cooling designs towards adaptive thermal management. High-performance air-handling units, hot- and cold-aisle containment and variable-speed drives are now baseline requirements rather than differentiators. Control systems increasingly adjust temperature bands dynamically in response to real-time IT loads, humidity levels and external conditions.

Crucially, most large new builds are now liquid-cooling-ready, even where initial deployments remain air-cooled. This is less a marketing choice than an economic hedge. Retrofitting liquid cooling into live facilities is costly and disruptive; designing for it upfront preserves optionality as AI workloads scale.

Power chain optimisation

Losses across the electrical chain remain a persistent but often under-appreciated efficiency drain. Operators are paying closer attention to uninterruptible power supply (UPS) efficiency at partial loads, transformer sizing discipline and harmonics management. High-efficiency UPS systems operating in eco-modes are being deployed alongside architectures that preserve redundancy without locking facilities into inefficient operating envelopes.

The underlying philosophy is straightforward but uncompromising: reliability remains non-negotiable, but inefficient over-provisioning is no longer acceptable in a margin-sensitive market.

IT load discipline

On the IT side, efficiency gains are increasingly driven by utilisation rather than wholesale hardware replacement. Workload consolidation, server right-sizing and intelligent scheduling enable operators and tenants to reduce energy consumed per unit of compute. Several colocation providers now actively advise customers on IT-side efficiency, recognising that facility-level optimisation alone cannot deliver the next phase of performance gains.

Smart energy management

Facilities are being instrumentalised with dense sensor networks tracking temperature, airflow, power quality and equipment health. These data streams feed into integrated building and energy management platforms capable of automated, real-time intervention.

This capability becomes critical as renewable sourcing scales. Open-access power introduces variability that must be managed operationally. Facilities are investing in advanced scheduling tools, grid-event monitoring and tighter coordination between primary supply, backup systems and load management. While low-carbon backup fuels are under evaluation globally, adoption in India remains cautious, reflecting cost sensitivities and regulatory uncertainty.

Digital intelligence

Data centre infrastructure management platforms have evolved into active optimisation engines. AI-driven analytics are increasingly used to predict thermal anomalies, forecast load behaviour and recommend control actions that reduce energy waste without compromising uptime.

Operators deploy these systems for early fault detection, predictive maintenance and dynamic cooling optimisation. Granular energy accounting has also become central as enterprise customers demand transparent, auditable sustainability data, aligned with global disclosure regimes.

For global players operating in India, expectations around water and energy reporting continue to tighten. Companies such as NTT have outlined initiatives in 2025 to enhance water-efficiency measurement across their data centre portfolios, reflecting rising standards around both performance and transparency.

Renewables’ integration

Renewables’ adoption in India’s data centre sector has entered a phase defined less by ambition than by execution complexity. Open-access procurement, hybrid solar-wind power purchase agreements, interstate transmission connectivity and banking mechanisms now determine the viability of green-power strategies.

Hyperscale and AI-driven deployments are increasingly paired with dedicated renewable capacity. Google’s India expansion is being supported by long-term clean-energy sourcing, including a 150-MW solar project in Rajasthan being developed with ReNew and targeted for commissioning in 2026. In parallel, the Adani Group and Google have announced plans for an AI data centre campus in Visakhapatnam, explicitly linked to new green-energy infrastructure.

For colocation providers, renewable-backed power has shifted from differentiation to expectation. Enterprise requests for proposals increasingly treat access to clean power and verifiable reporting as baseline requirements rather than value-adds.

Operator initiatives

Nxtra

Nxtra by Airtel has articulated one of the clearest sustainability roadmaps in the Indian market. The company has committed to reducing average PUE by around 10 per cent by 2025, measured against a 2021 baseline, driven by cooling optimisation, power-chain efficiency and advanced monitoring. On the energy side, Nxtra has outlined a target to source more than 70 per cent of its electricity from renewable sources by 2030, supported by on-site solar installations and large-scale open-access procurement. Notably, its strategy is structured at a portfolio level, allowing renewable sourcing to scale alongside capacity growth rather than lag it.

STT GDC India

STT GDC India’s sustainability execution is closely anchored in engineering discipline rather than headline commitments. New facilities emphasise high-efficiency cooling architectures, optimised airflow design and modular electrical layouts designed to maintain performance across variable utilisation levels. At the group level, ST Telemedia Global Data Centres has reported over 50 per cent renewable energy usage across its global portfolio, with measurable reductions in emissions intensity. Indian operations are increasingly aligned with these benchmarks, reflecting an emphasis on repeatable design templates rather than one-off flagship projects.

Yotta

Yotta’s approach reflects the inherent tension between sustainability and AI-led scale. Its Navi Mumbai campus is IGBC Platinum-certified, incorporating energy-efficient building design, optimised cooling and power-management systems. At the same time, Yotta’s 2025 growth narrative is closely tied to high-density AI infrastructure and sovereign cloud requirements. The company has signalled a focus on improving operational PUE even as power densities rise, positioning sustainability as an ongoing operating challenge rather than a static certification outcome.

AdaniConneX

AdaniConneX’s sustainability narrative is inseparable from its access to large-scale renewable energy. As a joint venture combining data centre development with upstream energy capabilities, the company is positioned to integrate green power directly into campus planning. The proposed Visakhapatnam AI data centre campus, announced in partnership with Google, exemplifies this model by linking hyperscale infrastructure with dedicated renewable generation. The approach reflects a broader shift towards energy-integrated data centre development in India.

Others

CtrlS has outlined net-zero timelines while highlighting energy-efficient design innovations, including integrated solar features and high-efficiency cooling, alongside plans for large renewable-powered campuses. Sify continues to emphasise renewable sourcing and sustainability reporting across its portfolio. Iron Mountain has focused on renewable energy matching, enabling enterprise customers to attribute clean-power consumption within their own disclosures. NTT continues to refine global sustainability standards, shaping expectations across its Indian facilities.

Key takeaway from 2025

India’s data centre market in 2025 demonstrates that sustainability outcomes are being shaped less by intent statements and more by engineering choices, contracting capability and operational intelligence. The operators set to define the next phase will be for those able to deliver capacity growth alongside measurable efficiency gains, scalable renewable integration and software-driven optimisation, while maintaining resilience in an increasingly power-constrained environment.