India’s data-centre sector is poised to draw investments of nearly $90 billion by FY35 across construction, power, cooling, networking and security infrastructure, according to a report released by KPMG in India this month, as the country’s installed capacity is projected to expand roughly tenfold over the next decade on the back of artificial intelligence-driven demand.
The report estimates that the construction-linked value chain opportunity alone will touch $30 billion by FY30 before scaling up to $90 billion by FY35, as hyperscalers, global operators and Indian conglomerates race to build capacity suited for compute-intensive workloads.
A market moving from the margins to the mainstream
India currently accounts for only 2-3% of global data-centre capacity despite generating close to a fifth of the world’s data, the report noted, pointing to what it calls considerable headroom for domestic expansion. That gap is beginning to close. The country’s installed capacity has more than tripled since FY19 to reach approximately 1.9 GW in FY26, with annual capacity additions rising from around 778 MW in FY23 to nearly 1,900 MW in FY26. According to the report, this is due to the long-term commitments from hyperscale cloud providers rather than the domestic enterprise and telecom demand that has anchored the sector.
Roughly 90% of this capacity remains concentrated in Bengaluru, Chennai, Delhi NCR and Mumbai, though the report also noted that Hyderabad, Pune and emerging Tier II locations such as Visakhapatnam are gaining traction as land availability and state incentives push new projects beyond the traditional metro clusters.
Globally, the data-centre market is expected to grow from about $160 billion in FY26 to $300 billion by FY30, with India’s share climbing from 2-3% to nearly 5% over the same period, which means a jump in market size from roughly $1.7 billion to $6.8 billion, the report stated.
AI is reshaping what a data centre needs to be
A central theme of the report is that artificial intelligence is not simply adding volume to existing demand but altering the physical and technical specifications data centres must meet. AI and high-performance computing workloads are five to ten times more power-intensive than conventional cloud applications, the report said, pushing rack densities from the 8-12 kW range typical of legacy facilities toward 50-60 kW and beyond, and making liquid cooling, which was once optional, a baseline requirement rather than an add-on.
By FY30, AI workloads are expected to make up around 55% of total domestic data centre capacity, rising to roughly 65% by FY35, per the report’s projections. It adds that only 25-30% of India’s existing data-centre stock is realistically capable of being retrofitted to meet these AI workload standards, meaning much of the coming capacity will need to be built new rather than upgraded.
The report tallied more than $120 billion in announced commitments from hyperscalers, global data-centre operators and large Indian conglomerates as of March 2026, with hyperscalers alone accounting for $50-55 billion earmarked for India, largely directed at GPU-dense, AI-first infrastructure.
Where the supplier opportunity lies
Within the broader value chain, the report breaks down where the money is expected to flow: design and construction account for the largest share at roughly $27 billion, followed by cooling infrastructure at $24 billion, power infrastructure inside data centres at $18 billion, and external power infrastructure like substations, transmission lines and grid augmentation at around $12 billion. Networking and security are estimated at $5 billion each.
The report also noted that cooling as a segment is worth watching, describing it as one of the most concentrated globally and in India, with liquid cooling technology almost entirely import-dependent at present. It draws similar conclusions for battery energy storage systems and advanced grid technology, where localisation remains limited despite India’s relatively strong position in transformers, switchgear and fibre optic cable manufacturing.
KPMG also pointed to India’s underlying cost advantages as a draw for both operators and suppliers: Mumbai ranks as the world’s second most cost-effective location for data centre construction, according to the report. As per ISI markets, the average residential power tariff lies in the range of $0.07 to $0.09 per kWh, while commercial and industrial rates typically range from $0.10 to $0.15 USD per kWh.
Policy support building momentum
The report credited a series of regulatory measures with accelerating the sector, chief among them the infrastructure status granted to data centres with more than 5 MW of IT load in the 2022-23 union budget, which it says opened access to longer-tenured debt at more competitive rates. It also highlighted IndiaAI Mission’s allocation of more than 34,000 GPUs, a 2025 draft national data centre policy proposing a 20-year tax holiday, and Budget 2026-27’s tax exemption running until 2047 for qualifying foreign cloud service providers operating through Indian entities.
The Digital Personal Data Protection Act, as per the report, is separately expected to add 1,800-2,000 MW of demand by 2027 as companies across banking, e-commerce, healthcare and government platforms are pushed toward domestic data storage and audit-ready infrastructure.
The report argues that gaps still remain in the segment, like the absence of a dedicated data centre R&D fund, unclear capital repatriation timelines for foreign investors, and financing costs for domestic developers that remain well above what global hyperscalers can access through offshore capital. According to KPMG, unless these gaps are addressed, much of this investment could end up benefiting global players more than Indian ones.