- Meta signed its first AI data center deal in India: a 168-megawatt facility in Jamnagar, Gujarat, built with Reliance Industries and ready within two years.
- The site runs on renewable energy and is cooled with desalinated seawater, with Meta covering the full cost of energy and water for its operations.
- India’s data center capacity has quadrupled from roughly 375 MW in 2020 to about 1.5 GW in 2025, and is projected to top 8 GW by 2030.
- Microsoft, Amazon, Google, OpenAI, AirTrunk, and Adani have all announced major India infrastructure plays, lured partly by tax exemptions through 2047.
Here is a number worth sitting with: India’s installed data center capacity has jumped from about 375 megawatts in 2020 to roughly 1.5 gigawatts in 2025 — a fourfold increase in five years — and industry estimates say it could blow past 8 gigawatts by the end of the decade. This week, Meta decided it wants a piece of that curve. The company announced its first AI data center deal in India, partnering with Reliance Industries on a 168-megawatt AI-enabled facility in Jamnagar, Gujarat. The location, the cooling system, and the financial structure of the deal all tell a bigger story about where the global AI infrastructure race is heading next.
The Jamnagar Deal: What Meta and Reliance Actually Signed
A 168 MW Facility, Cooled by the Sea
Under the agreement announced Wednesday, Meta will lease capacity at Reliance’s new AI-enabled data center in Jamnagar, Gujarat. The 168-megawatt facility will be powered by renewable energy and cooled using desalinated seawater — an unusual engineering choice that sidesteps one of the most contentious issues in data center construction: freshwater consumption. Meta has committed to covering the entire cost of the energy and water required to support its operations at the site.
Reliance says the facility will be ready within two years and can be expanded over time. Critically, the data center will support Meta’s global infrastructure and AI computing requirements, not just local Indian workloads — plugging India directly into Meta’s worldwide network of AI facilities. Reliance will provide end-to-end services spanning design, construction, renewable power, connectivity, and ongoing operations.
Six Years in the Making
This deal did not come out of nowhere. Meta invested $5.7 billion in Reliance’s Jio Platforms back in 2020, and the two companies launched a $100 million joint venture last year to develop enterprise AI solutions for customers in India and overseas. Separately, Meta said it has contracted nearly 1 gigawatt of new renewable energy capacity in India through agreements with CleanMax and Fourth Partner Energy. The companies did not disclose the value of the new data center agreement or the specific AI workloads that will run from the facility.
Trend Insight — The structure here matters more than the megawatts. Meta is not building and owning the facility; it is leasing capacity while Reliance handles everything from construction to operations. That asset-light model lets hyperscalers expand into new geographies faster while shifting execution risk to local partners — expect it to become the template for AI infrastructure expansion in emerging markets.
Why Everyone Is Suddenly Racing to India
A Crowded Field of Giants
Meta is late to a party that is already crowded. Microsoft has committed $17.5 billion to India by 2029, Amazon has raised its total planned spending to $75 billion by 2030, Google is investing $15 billion in an Indian AI infrastructure hub, and OpenAI tapped Tata for 100 MW of data center capacity with ambitions of reaching 1 gigawatt. Just last week, Blackstone-backed AirTrunk announced a $30 billion plan to build 5 gigawatts of capacity in India by 2030, while Adani has pledged $100 billion for AI data centers and Tata Consultancy Services secured TPG funding for half of a $2 billion data center project.
The Policy Magnet: Zero Taxes Through 2047
New Delhi has engineered this rush deliberately. Among the incentives: foreign cloud providers get tax exemptions through 2047 on services sold overseas, as long as those workloads run from Indian data centers. Combine that with comparatively low construction costs, abundant renewable energy potential, and a massive domestic market for local data processing, and India has positioned itself as the natural overflow valve for a world that cannot build compute fast enough.
Trend Insight — The 2047 tax exemption is a 21-year policy commitment — an eternity in tech. India is effectively offering hyperscalers regulatory certainty that the US and EU cannot match right now, and the “run global workloads from India” clause means the country is exporting compute the way it once exported IT services. Watch for other emerging economies to copy this playbook.
The Bigger Question: Can the Spending Outrun the Skeptics?
The Revenue Math Critics Keep Pointing To
Not everyone believes the infrastructure boom is rational. A widely shared analysis circulating this week on GeekNews argues that generative AI infrastructure would need to generate over $2 trillion in annual AI compute revenue by 2030 to justify the data center investments and compute commitments already planned — against roughly 190 GW of planned capacity globally. Whether or not you accept that math, it frames the bet Meta and its peers are making: that AI demand will grow into the capacity being built, not the other way around.
Why India Hedges That Bet
Seen through that lens, the India strategy looks like risk management as much as expansion. Lower build costs, long-dated tax relief, renewable power, and a lease-not-own structure all reduce the downside if AI demand grows slower than projected. And if demand does materialize, Meta has a foothold in the one major market where capacity is projected to grow more than fivefold by 2030. Either way, the Jamnagar facility is a signal: the next phase of the AI infrastructure race will be fought as much over geography, energy, and policy as over chips.
Trend Insight — For business leaders, the takeaway is that compute pricing power is shifting. As capacity comes online in lower-cost geographies like India, inference costs should keep falling — good news for companies building on AI APIs, and another reason the “cheaper models” trend has structural tailwinds behind it.
Related
- Why Wall Street Just Handed Bezos Another $12 Billion
- Amazon Just Quietly Came for Redbubble’s Whole Business
- Claude Fable 5 and Mythos 5: Anthropic’s Fifth-Generation Frontier Models
- Decart’s Oasis 3 World Model Simulates Hours of Photorealistic Driving
Sources
- TechCrunch — Meta signs first AI data center deal in India with Reliance (Jagmeet Singh, June 10, 2026)
- TechCrunch AI — Artificial Intelligence news category
- GeekNews — AI infrastructure and developer news digest
AI Biz Insider · AI Trends EN · aibizinsider.com
댓글 남기기