Meta plants AI infrastructure in India

Meta and Reliance Industries have announced an agreement for an AI-enabled data center in Jamnagar, in the Indian state of Gujarat. Reliance will build the facility, which has an initial capacity of 168 MW, and Meta will lease it with options to scale. The central point is straightforward: Meta is placing part of its AI compute capacity closer to one of its largest markets, rather than relying only on already dense infrastructure regions in the United States or Europe.

An AI-enabled data center, in this context, means infrastructure designed to support compute-intensive workloads such as training, inference, and services powered by AI models. Meta says the site will support its products and AI capabilities in a country where its social apps, messaging services, and business tools reach a very large audience. The 168 MW figure does not directly measure model performance. It describes the electrical capacity available to run servers, networking, cooling systems, and related equipment.

The announcement also has an energy layer. Meta says it has separately contracted nearly 1 GW of new clean and renewable energy in India through agreements with CleanMax and Fourth Partner Energy. CleanMax accounts for 837 MW of new solar and wind projects, while Fourth Partner Energy adds 88 MW across several Indian states. Reliance is also expected to support the Jamnagar data center with renewable energy, and Meta says it will cover the full cost of the energy and water used by the facility, which will be cooled with desalinated seawater. Those details do not settle the environmental debate around AI, but they show that energy sourcing is now part of the infrastructure story.

For the AI ecosystem, the move illustrates a practical shift: competition is increasingly fought through cables, megawatts, land, cooling, and local partnerships, not only through model releases. In India, Meta is combining an existing relationship with Reliance, dating back to its 2020 investment in Jio Platforms, with a new physical layer for compute. This is not a consumer product launch or a new model announcement. It is an industrial signal. Bringing AI capacity closer to the markets that use it will require infrastructure agreements that matter as much as software progress.