Summary: Meta said on June 9, 2026 that it would lease 168 MW of the first phase of an AI-enabled Reliance data center in Jamnagar, its first AI data center in India. Reliance will handle design, construction, renewable power, connectivity, and operations, while Meta reimburses energy and water costs. The facility is expected within two years and will use renewable power and desalinated seawater cooling.
Meta said on June 9, 2026 that it has struck a deal with Reliance Industries to lease 168 megawatts — the first phase of an AI-enabled data center in Jamnagar, Gujarat — with room to expand afterward. It is Meta’s first AI data center in India, delivered as a built-to-suit arrangement: Reliance handles everything end to end, from design and construction through renewable power, connectivity, and day-to-day operations, while Meta reimburses the full cost of the energy and water it ends up consuming. Completion is expected within two years.
Strip away the real estate and what remains is an energy-engineering story. “We’re proud to be working with Reliance to build our first AI-enabled data center in India,” said Mark Zuckerberg. Mukesh Ambani, Reliance’s chairman, described the partnership as “a transformative moment for India’s digital infrastructure.”
Renting Capacity Instead of Building It
The structure is the part worth studying. Instead of acquiring land and building for itself, Meta is renting capacity from Reliance. That converts capital expenditure into operating expenditure and hands delivery risk to a partner on the ground; Reliance, for its part, gets to sell the power, water, and heavy-engineering muscle accumulated across its refining and telecom empire to buyers facing global AI demand. According to Meta, the facility will back its global infrastructure and AI computing needs and dovetails with the Project Waterworth subsea cable system — for the company, India works as a node in its worldwide network, not merely a market.
There is history here. Meta put $5.7 billion into Jio Platforms in 2020, and the pair formed a joint venture in 2025 to put Meta’s open-source AI models in the hands of Indian enterprises. The new agreement moves things beyond equity and licensing into physical infrastructure — and the campus taking shape at Jamnagar, which Reliance is developing, ranks among the world’s largest data center campuses.
Renewables for Power, Seawater for Cooling — and Meta Settles the Bill
The energy package is what makes this announcement stand out. Alongside the data center, Meta unveiled close to 1 gigawatt of fresh renewable contracts across India: 837 MW of solar and wind from CleanMax, sited in Rajasthan and Karnataka, which lifts the two companies’ cumulative collaboration past 900 MW, plus another 88 MW from Fourth Partner Energy spread across Tamil Nadu, Karnataka, Maharashtra, and Uttar Pradesh. The site will run on renewable power and shed its heat with desalinated seawater, and Meta picks up the entire energy and water bill.
Writing water into the contract is more than messaging. Where a data center gets its water is turning into a cost line that bites, and Jamnagar happens to sit at the heart of Reliance’s refining operations, where desalination and power infrastructure already stand. That existing base is exactly what allows Reliance to put a bundle with power and water included on the table.
India’s Moment, and What Brought It Forward
Money for AI infrastructure is converging on India. Microsoft has pledged $17.5 billion by 2029; Amazon tacked $35 billion onto plans that already totaled $75 billion; Google has a $15 billion AI infrastructure hub in the pipeline; OpenAI, working through Tata, has locked in 100 MW and is aiming for a full gigawatt. The country’s data center capacity climbed from roughly 375 MW in 2020 to about 1.5 GW in 2025, and KPMG expects it to clear 8 GW by 2030. Policy adds to the pull: foreign cloud providers serving overseas customers out of Indian data centers are exempt from tax through 2047.
Reliance shares rose more than 2% once the news broke.
The Ceiling on AI Growth Is Measured in Megawatts
Put 168 MW in perspective: major AI training campuses draw anywhere from several hundred megawatts into gigawatt territory, and how long it takes to secure electricity has become the first choke point of AI expansion. Meta’s answer was to put power ahead of everything else — renewable contracts signed in step with the data center, so energy risk is pinned down before construction begins, and seawater cooling chosen so freshwater disputes never arise. Set against labs forced to grab compute wherever it exists, what Meta has actually purchased here is a dependable energy supply.
What This Means for Infrastructure Strategy
Three readings. One: leasing is on the rise — AI demand is swelling faster than self-build cycles can run, so bundles with “power and operations included” will keep multiplying, and infrastructure capacity at scale is being packaged into a tradable service. Two: what emerging markets bring to the table is energy and land rather than chips — India is settling into the role of compute landlord, taking in global training and inference workloads. Three: a word for product and platform teams — latency-sensitive services will keep sitting near users, but the map for batch AI workloads is being redrawn, and how you are taxed, what electricity costs, and whether water is available will move your cost structure more than the model you pick.