Reliance Jamnagar AI data center
Last reviewed
Sources
18 citations
Review status
Source-backed
Revision
v1 ยท 2,222 words
Improve this article
Add missing citations, update stale details, or suggest a clearer explanation.
Suggest edit
CtrlK
Last reviewed
Sources
18 citations
Review status
Source-backed
Revision
v1 ยท 2,222 words
Add missing citations, update stale details, or suggest a clearer explanation.
The Reliance Jamnagar AI data center is a planned gigawatt-scale artificial intelligence computing campus that Reliance Industries, India's largest company by market value, is building at Jamnagar in the western state of Gujarat. The project pairs a data center buildout with NVIDIA AI chips and Reliance's own renewable-energy manufacturing, and it sits at the center of an AI strategy that chairman Mukesh Ambani has framed as a national effort. The plan was first announced through an NVIDIA partnership in October 2024, formalized under a new subsidiary called Reliance Intelligence in August 2025, and folded into a roughly 10 trillion rupee (about $110 billion) seven-year AI investment program unveiled in February 2026. As of mid-2026 the campus is under construction. Reliance says a first phase of more than 120 megawatts (MW) is targeted to come online in the second half of 2026, with the full campus described in reports as eventually reaching about 3 gigawatts (GW). Most of the scale, spending, and chip figures attached to the project are announced or planned rather than operational.
Jamnagar is already the site of Reliance's refining hub, widely cited as the world's largest single-site oil refinery, with a crude-processing capacity of roughly 1.4 million barrels per day and a 7,500-acre footprint. Reliance is layering an AI data center campus onto this industrial base alongside the adjacent Dhirubhai Ambani Green Energy Giga Complex, a 5,000-acre renewable-manufacturing project. The pitch is that existing land, power, cooling, and connectivity, combined with captive green electricity, give Reliance a cost and speed advantage over hyperscalers that must build such infrastructure from scratch.
The numbers below mix confirmed announcements with figures attributed to specific reports. Capacity, GPU, and dollar amounts should be read as planned or announced unless noted otherwise.
| Attribute | Detail (as planned or announced) |
|---|---|
| Location | Jamnagar, Gujarat, India (adjacent to Reliance's refinery and green-energy complex) |
| Developer | Reliance Industries, via its Reliance Intelligence subsidiary |
| First announcement | NVIDIA partnership, October 2024 (NVIDIA AI Summit, Mumbai) |
| Subsidiary launch | Reliance Intelligence, August 29, 2025 (RIL 48th AGM) |
| Investment program | About 10 trillion rupees (~$110 billion) over seven years, to 2033 (announced February 2026) |
| Initial data center | 1 GW running NVIDIA Blackwell GPUs (announced October 2024) |
| Reported full scale | About 3 GW campus (Bloomberg report, January 2025) |
| First phase | More than 120 MW targeted for second half of 2026 |
| Estimated data center cost | $20 billion to $30 billion for the ~3 GW build (Bloomberg report) |
| Power | Reliance green energy; about 10 GW of renewables (solar in Gujarat and Andhra Pradesh) |
| Google role | Dedicated "Jamnagar Cloud region" for Reliance, using Google Cloud |
| Meta joint venture | About 8.55 billion rupees (~$100 million), 70% Reliance / 30% Meta; Llama-based enterprise AI |
| Status (June 2026) | Under construction; mostly planned and phased |
At Reliance Industries' 48th annual general meeting on August 29, 2025, Mukesh Ambani announced Reliance Intelligence, a new subsidiary dedicated to AI. He gave it four stated missions: building gigawatt-scale, AI-ready data centers powered by green energy and engineered for training and inference at national scale; bringing together leading technology companies and open-source communities; delivering AI services for consumers, small businesses, enterprises, and sectors of national importance; and creating a home for researchers, engineers, and product builders. Ambani compared the ambition to the disruption caused by Jio, Reliance's telecom arm, saying the unit "promises to deliver AI everywhere for every Indian."
Two partnerships were unveiled at the same meeting. Google and Reliance said they would build a dedicated "Jamnagar Cloud region," described by Alphabet chief executive Sundar Pichai as bringing AI and compute from Google Cloud, "powered by clean energy from Reliance, and connected by Jio's advanced network." Separately, Reliance and Meta agreed to form a joint venture capitalized at about 8.55 billion rupees (roughly $100 million), owned 70% by Reliance and 30% by Meta, to offer an enterprise AI platform built on Meta's Llama open models, sold as a service for use cases such as sales, marketing, customer service, IT, and finance. Reports at the time said the venture was expected to close in the fourth quarter of 2025. (The $100 million joint-venture figure is far smaller than the headline data-center numbers and should not be confused with them.)
In February 2026, Ambani put a price tag on the broader effort: a program of about 10 trillion rupees, near $110 billion, to be spent over roughly seven years through 2033 on gigawatt-scale data centers, a nationwide edge-computing network, and AI services tied to Jio. He described the spending as long-term "nation-building capital" and argued that the main bottleneck for Indian AI is infrastructure, saying "the biggest constraint in AI today is not talent or imagination. It is the scarcity and high cost of compute." He also said India "cannot afford to rent intelligence" and pledged to drive down AI service costs the way Jio drove down mobile-data prices.
The Jamnagar project has been described at several capacity levels, and the figures should be kept distinct. The initial commitment, announced with NVIDIA in October 2024, was a 1 GW data center running NVIDIA's Blackwell-generation GPUs. In January 2025, Bloomberg reported that Reliance was planning a far larger facility of about 3 GW, which would make it among the world's largest by power draw. For comparison, that report noted the facility would "dwarf" what was then cited as the largest operating data center, a roughly 600 MW Microsoft site in Virginia. Estimated cost for the larger build was put at $20 billion to $30 billion. By the February 2026 plan, Reliance described "multi-gigawatt" data centers under construction at Jamnagar, with more than 120 MW expected to come online in the second half of 2026 as the first phase.
The distinguishing feature is power sourcing. Reliance intends to run the campus largely on its own renewable electricity rather than the grid, drawing on a roughly 10 GW renewable buildout anchored by large solar projects in Gujarat (including Kutch) and Andhra Pradesh. That electricity is meant to come from the Dhirubhai Ambani Green Energy Giga Complex, a 5,000-acre, roughly $10 billion project housing five "gigafactories" for solar photovoltaic modules, batteries, electrolyzers, fuel cells, and power electronics. Reliance has begun commissioning solar manufacturing lines toward a target of about 10 gigawatt-peak per year, with plans to expand further, and has said a battery gigafactory with an initial 40 gigawatt-hour annual capacity would start up in 2026. Powering AI compute with captive solar, wind, and storage is central to the project's economics and to Ambani's framing of it as green AI, though the renewable factories are themselves still ramping.
The anchor technology partner is NVIDIA. At the NVIDIA AI Summit in Mumbai in October 2024, Reliance and NVIDIA announced a collaboration to build AI infrastructure in India, with NVIDIA committing to supply its Blackwell AI processors for a 1 GW Reliance data center. NVIDIA chief executive Jensen Huang used the event to argue that India should build its own AI rather than export data and import intelligence, saying "you should not export data to import intelligence" and predicting that India would become a country that exports AI. Specific chip counts and compute figures for the facility were not disclosed.
The software and cloud layer brings in several global firms. Google's role centers on the Jamnagar Cloud region for Reliance workloads, and in February 2026 Jio also struck a consumer deal to give millions of Indian users free access to Google's Gemini AI Pro tier. Meta's role is the Llama-based enterprise joint venture. OpenAI entered the picture in February 2026, when it and Reliance announced a partnership to add AI-powered conversational search to the JioHotstar streaming service, built using OpenAI's API and supporting text and voice queries in multiple Indian languages. Some 2025 reporting also described earlier OpenAI-Reliance discussions around distributing ChatGPT through Jio. Brokerage Morgan Stanley has called the AI push an underappreciated growth driver for Reliance, estimating that the infrastructure opportunity could add on the order of 15% to the company's net asset value over time and that a data-center-as-a-service model could eventually generate several billion dollars in annual revenue. Those are analyst projections, not company guidance.
The Jamnagar project is the largest single piece of a wider Indian effort to build domestic, or "sovereign," AI capacity. India's government has run the IndiaAI Mission, which by early 2026 had provisioned tens of thousands of subsidized GPUs for startups and researchers, with more being added. At the India AI Impact Summit 2026 in New Delhi in February 2026, Indian conglomerates and global investors collectively pledged on the order of $240 billion toward AI infrastructure in the country. Reliance's roughly $110 billion seven-year commitment was the headline, but it was not alone: the Adani Group outlined plans to expand its data center platform from about 2 GW toward 5 GW, and Tata Group and Tata Consultancy Services described AI data centers starting near 100 MW and scaling toward 1 GW, with OpenAI named as an anchor tenant.
For Ambani and Reliance, AI is a natural extension of the Jio playbook of using massive capital spending to build infrastructure, undercut incumbents on price, and capture a large user base. Jio serves on the order of 450 million telecom subscribers, JioHotstar reports hundreds of millions of streaming users, and a consumer JioAICloud service had tens of millions of users by 2025. The strategy is to combine that distribution with Jamnagar compute so that Reliance both sells AI infrastructure and embeds AI across its energy, retail, telecom, and financial-services businesses.
The clearest caveat is that very little of the project is operational as of mid-2026. Reliance has begun construction and is targeting a first phase of more than 120 MW in the second half of 2026. The headline figures, a roughly 3 GW campus, $20 billion to $30 billion in data-center capital, and the wider $110 billion seven-year program, describe plans that will play out over many years and remain subject to execution.
Several risks attach to that timeline. A 3 GW AI campus would be unprecedented in scale; as of early 2025, the largest operating data centers worldwide drew well under 1 GW, so Reliance would be building at a size with few precedents. Power is the recurring industry constraint: AI campuses are straining grids worldwide, and "time-to-power," meaning how fast firm electricity can actually be delivered, has become a gating factor for data-center timelines. Reliance's answer is its own renewable supply, but those solar, battery, and electrolyzer factories are themselves still being commissioned. The capital outlay is also enormous relative to historical data-center spending, and some of the marquee partnerships, including the Meta joint venture and various Google and OpenAI tie-ups, are early-stage or consumer-facing rather than guarantees of demand for gigawatts of training compute. The "world's largest" framing comes largely from a January 2025 Bloomberg report rather than from confirmed, built capacity, and the competitive landscape has shifted quickly as multi-gigawatt campuses are announced globally. In short, the Reliance Jamnagar AI data center is a credible, well-capitalized, and partly under-construction project, but its most striking numbers remain announced ambitions rather than delivered facts.