Meta-Reliance 168 MW AI data center in Jamnagar

The headline number is 168 megawatts, but the more revealing fact is what the Jamnagar lease completes. Meta now holds three connected layers of infrastructure inside India: compute (the built-to-suit data center) ^[1], the pipes that move data in and out (Meta's Project Waterworth subsea cable system, reported at more than 50,000 km across 24 fiber pairs linking the US, India, Brazil, and South Africa) ^[2], and a software layer (a 2025 joint venture with Reliance selling Llama-based enterprise AI to Indian companies). A single data center is a real-estate decision; owning compute, transport, and applications at once is a country strategy.

Note the structure of the deal: Meta is not building or owning the building. Reliance builds, owns, and operates it, and Meta leases the full capacity and pays for all the power and water ^[1]. That keeps the multi-year capital and the local regulatory, land, and utility relationships on Reliance's books while Meta gets dedicated capacity it can switch on without managing Indian construction and grid interconnection itself. It is the same asset-light playbook hyperscalers use elsewhere, applied to a market where a local partner's permits and land access are the hard part.

Reliance's pitch is that it is a single-window provider: it brings the land, renewable power, connectivity through Jio's fiber network, proximity to India's western submarine-cable landing stations, and managed operations under one contract ^[3]. For a hyperscaler that wants capacity fast without assembling a dozen local vendors, that bundle is the product. It also explains why Meta's first India compute lands inside an existing Reliance industrial complex in Jamnagar rather than a greenfield site.

The cooling choice is not an environmental footnote; it is risk management. Hyperscale AI sites can consume millions of gallons of water a day, and more than half of India's data centers already sit in water-stressed regions ^[4]. By designing Jamnagar around desalinated seawater and renewable power from the start ^[1], the partners are trying to pre-empt the exact permitting and community fights that freshwater-cooled facilities are starting to trigger across India. Whether that engineering promise survives contact with operations is the open question local critics keep raising.

The Meta deal is one data point inside a much larger reallocation of capital. Global hyperscalers have directed roughly $400 billion into India's AI ecosystem over the past year, most of it into data centers and the energy infrastructure to power them ^[5]. India's own data center capacity has climbed from well under half a gigawatt in 2020 to around 1.5 GW in 2025, and Nomura projects it to reach roughly 7 GW by 2030 ^[6].

The imbalance driving that build-out is stark: India generates an outsized share of the world's data but holds only a low-single-digit percentage of global data center capacity ^[2]. Policy is leaning into the gap, with India offering tax exemptions for foreign cloud providers and treating data centers as strategic national infrastructure ^[6]. For Meta, getting in now means locking down scarce, cost-efficient capacity before the land-grab fully prices in.

The official framing is clean: renewable power, seawater cooling, Meta covering every rupee of energy and water. The reception on the ground is not. In Gujarat-local and Indian business communities online, the dominant reaction skews skeptical, and it clusters on two anxieties: that a water-scarce region is taking on a notoriously thirsty industry, and that the permanent local job count will be small relative to the resource cost. Several residents near the site describe land already being acquired and hiring underway, but for specialized AI-infrastructure roles largely filled from outside the state.

That skepticism is not purely vibes. Reporting on India's data center pipeline notes a recurring lack of clarity around how much water and energy approved facilities will actually use, even as more of them land in water-stressed areas ^[4]. The desalinated-seawater design is meant to answer exactly this critique, which is why how Jamnagar performs in practice matters beyond this one project: it will become the reference case other Indian states cite when the next hyperscaler comes knocking.