Meta has recently leased an AI-enabled data centre in Jamnagar with 168 megawatts of capacity from Reliance Industries. Reliance will build the data centre for Meta and deliver the facility within two years, with an option to scale. In an exclusive conversation, Ashish Banerjee, Sr Principal Analyst, Gartner breaks down the dynamics of the partnership to Rajneesh De, Group Editor, APAC Media & CXO Media.
What are the core strategic imperatives behind the Meta Reliance partnership?
This is not a routine data-centre lease; it is a structural race to anchor and isolate critical AI capacity. Meta gets large-scale compute closer to one of its biggest user markets, while Reliance converts its strengths in land, energy, fibre, utilities and execution into an AI-infrastructure platform. The strategic logic is clear: AI leadership will be shaped as much by access to power and cooling as by models and software. Jamnagar gives both companies a highly credible position in that new infrastructure stack.
How does it help Meta in terms of operational costs and market penetration?
For Meta, the benefit is less about cheap real estate and more about reducing execution friction. A build-to-suit lease lets Meta scale AI capacity without owning every layer of land acquisition, utilities, construction and local operations. India-based capacity can also improve latency, user experience and localisation options across Meta’s apps and AI services. But this should not be read as a cloud-market entry strategy. It is primarily about securing AI compute capacity, not selling enterprise cloud like AWS, Azure or Google Cloud.
Given Reliance’s captive power units in Jamnagar, how much does this help in energy efficiency?
Captive power helps most with energy assurance, not automatically with energy efficiency. For AI data centres, reliable megawatts are often harder to secure than land. Jamnagar’s advantage is the integrated energy, water and infrastructure ecosystem Reliance can bring. But efficiency will depend on PUE, server utilisation, cooling design and 24×7 clean-energy matching – not just captive generation. Renewable power and desalinated seawater cooling are positive signals, but the real test will be transparent operating metrics after commissioning.
This partnership adds 30% to India’s entire hyperscale capacity. How will this impact the overall Indians total hyperscale cloud capacity across all providers including AWS, Azure,Google Cloud and smaller domestic players?
I would be careful with the “30%” claim unless the denominator is clearly defined. Against India’s total operational data-centre stock of roughly 1.5 GW IT, 168 MW is closer to an 11% addition; against pure hyperscale-only capacity, it may look much larger. Either way, the impact is significant but not evenly available to the market: this is largely dedicated Meta capacity, not open cloud capacity for enterprises. It raises India’s AI infrastructure credibility more than it immediately expands multi-tenant cloud supply.
