Mumbai Strengthens Its Position As India’s Data Centre Capital: Report

Mumbai has reinforced its position as India’s data centre capital, accounting for 40% of national capacity and 44% of live IT capacity, according to Knight Frank’s Asia-Pacific Data Centres 2025 report.

The city’s capacity rose 14.3% in the first half of 2025, surpassing the 4GW milestone, with 591MW operational, 185MW under construction, and 3.2GW in the pipeline. This growth builds on India crossing 10GW of total data centre capacity in H2 2024, supported by 1.4GW live and 400MW under construction. Rapid cloud adoption, data localisation mandates, and the expansion of local fintech and BFSI firms are driving demand.

Knight Frank’s analysis reveals that Mumbai recorded 97.6MW of take-up in H1 2025, translating into a tight vacancy rate of just 5.4%, compared with India’s overall colocation vacancy rate of 12.3%. Absorption remains broadly aligned with the multi-fold growth in supply over recent years. Moreover, two-thirds of the city’s capacity currently under construction is already pre-leased, reflecting strong demand commitments.

Despite the robust growth, the city faces a short-term supply gap for large-scale deployments. Only three live sites currently support hyperscale deployments of over 2.5MW, and just one site offers more than 10MW of available capacity. The distribution of live capacity skews heavily toward smaller deployments: ten sites provide less than 1MW, five sites fall in the 1–2MW range, while three sites offer over 3MW. Such fragmentation creates opportunities for well-capitalised global operators and joint ventures to deliver high-capacity facilities. The 500MW NAV2 campus announced by NTT and Blackstone-Panchshil Realty’s 500MW AI-focused facility exemplify this trend.

As larger operators seek hyperscale deployments, secondary markets are gaining attention. Hyderabad, with over 500MW of new capacity in the pipeline, is positioning itself as a hyperscale-first market. STT GDC India has signed an MoU with the Telangana government to develop a 100MW campus, while NTT has committed INR 10,500 crore (approximately USD 1.25 billion) to establish a 400MW AI-focused data centre campus. Hyderabad is India’s second-largest data centre market at 2.1GW, followed by Chennai (1.6GW), New Delhi (712MW), and Bengaluru (307MW).

Shishir Baijal, Chairman and Managing Director, Knight Frank India, says, “Mumbai has firmly established itself as the epicentre of India’s digital infrastructure growth. With over 3GW in the pipeline and strong policy support for green data centre parks, the city is attracting sustained global investment. Its subsea cable connectivity, scalable power infrastructure, proximity to enterprise hubs, and progressive policies consolidate Mumbai’s position as India’s data centre capital.”

The Knight Frank report highlights that APAC is witnessing unprecedented expansion. In H1 2025, the region saw nearly 13GW of new project announcements, a 160% increase over the same period last year, requiring funding exceeding USD 180 billion. Technology giants are leading the investment surge. Amazon is projected to exceed USD 100 billion in capital expenditure for 2025, up from USD 82 billion in 2024, while Microsoft invested USD 55 billion in 2024 and has committed over USD 33 billion this year. Combined, Microsoft, AWS, Google, and Meta have committed more than USD 160 billion in 2025, reflecting the scale of infrastructure development.

Fred Fitzalan, head of data centres Asia-Pacific at Knight Frank, observes, “The sheer volume of new projects in APAC underscores the region’s strategic importance in the global digital infrastructure landscape. Coordinating rapid growth is complex. Operators must match technology advances with energy requirements while delivering facilities in line with evolving demand.”

Alongside hyperscalers, GPU-as-a-Service providers are expanding rapidly, seeking multi-megawatt capacity, and introducing greater diversity into leasing dynamics. Creditworthiness and accelerated deployment timelines remain challenges, but innovative guarantee structures are enabling operators to compete effectively. “Operators must now design facilities with flexible capacity for either cloud or AI workloads,” Fitzalan adds. “Sites that combine proximity to parent facilities with adequate power allocations are preferred, although national grid constraints and permitting delays pose obstacles.”

Other APAC markets are also evolving rapidly. Johor has become Southeast Asia’s fastest-growing hub, with supply nearly doubling to 5.8GW by Q2 2025. Take-up in H1 2025 reached 260MW, dominated by social media and AI workloads. Tokyo retains its key hub status with aggregate capacity exceeding 4.2GW, although leasing slowed to 41.1MW in H1 2025 due to limited supply. Melbourne’s supply nearly tripled to 4.7GW, attracting all four major US cloud providers, with AI driving 95% of colocation take-up. Seoul continues to attract strong investment with LG U+ expanding an 87.2MW footprint and Macquarie Asset Management acquiring the 40MW Hanam Data Centre.

Investment strategies are also evolving. Infrastructure and private equity capital increasingly partner with operators on developer-led powered shells, achieving faster time-to-power deployment. Grid capacity and energy availability remain key constraints, while geopolitical considerations influence timelines. In North Asia, US and Chinese cloud providers compete for the same capacity, pushing rental values higher. The AI ecosystem continues to expand beyond hyperscale deployments, requiring digital infrastructure that is flexible, scalable, and future-ready.

Knight Frank’s Asia-Pacific Data Centres 2025 report underscores that Mumbai, with its robust infrastructure, policy support, and pipeline, will continue to lead India’s data centre growth while APAC markets chart unprecedented expansion. The report highlights the importance of synchronising technology evolution, energy demand, and global investment flows to ensure sustainable, high-capacity digital infrastructure that meets growing cloud and AI requirements.