The pivot accelerated last week when Reliance incorporated Reliance Enterprise Intelligence Limited (REIL), a joint venture giving Meta a 30% stake alongside Reliance Intelligence‘s 70% holding. The partners have jointly committed an initial ₹855 crore to develop, market and distribute enterprise AI services — the clearest signal yet that Ambani’s AI ambitions have moved from boardroom presentations to balance sheet reality.
Morgan Stanley analysts now estimate Reliance will deploy $12-15 billion on AI infrastructure to build a 1-gigawatt datacenter, underwriting roughly 25% of capacity itself. “We estimate an ROCE of approximately 11% on these initial investments, based on recent token prices from LLM providers,” said Mayank Maheshwari at Morgan Stanley, adding that the firm expects remaining capacity could be leased as “Datacenter as a Service” to hyperscalers and large language model providers.
The valuation thesis is straightforward: “We believe it could be valued at a minimum of 2x P/B, if not higher, considering the valuations of global peers such as GDS in Asia (US players like CoreWeave and Equinix trading at even richer multiples),” Morgan Stanley noted. At twice the $15 billion capital invested by 2027, that implies a $30 billion valuation for the AI vertical alone.
Also Read | Reliance Intelligence is the new Jio for Mukesh Ambani: Can AI leap help RIL become deep-tech giant?
From Inference to Infrastructure
The strategy operates on two tracks. Reliance will utilize its initial 100-megawatt Gen AI datacenter capacity — scaling over two years — to address enterprise inference demand through its joint venture with Meta on small language models, alongside partnerships with Google and Azure. Morgan Stanley estimates annual revenues of around $1.5-1.6 million per megawatt for the “Datacenter as a Service” model.But the real scale play involves Reliance underwriting more than 20 gigawatts of internal power demand to support 100 gigawatts of solar panel capacity and 30-40 gigawatt-hours of battery capacity — turning datacenters’ voracious energy appetite into a captive market for Reliance’s clean energy vertical.”We estimate that RIL will spend approximately $12-15bn on AI infrastructure to develop a 1GW datacenter, underwriting about 25% of the capacity itself (roughly $7 billion for datacenter infrastructure and $5 billion for the 250MW of chips RIL will deploy directly),” Maheshwari said.
Reliance’s Dual Game
The corporate structure reveals Ambani’s audacious dual-track approach to AI dominance — simultaneously partnering with and competing against Silicon Valley’s biggest names. At Reliance’s recent Q2 earnings call, Anshuman Thakur, Head of Strategy at Reliance Jio Infocomm, drew a sharp distinction between the telecom arm and the AI venture.
“Think of Reliance Intelligence as AI companies which are coming out with products and solutions. Reliance Intelligence is not looking to do its own LLMs at this point in time but the products and solutions, AI-based product and solution companies,” Thakur explained. “Jio is the user of those capabilities which are going to be built. Jio would also work with Meta and would work with OpenAI and Google and Microsoft to use their products and services as well.”
Then came the telling line: “Reliance Intelligence is competing with the Metas and the Googles for the AI products that are coming into the market.”
While Jio positions itself as a customer deploying AI services from global tech giants, Reliance Intelligence aims to build competing AI products and solutions — creating an unusual dynamic where Reliance both partners with Meta through a joint venture while simultaneously competing with Meta and Google in AI products. It’s a strategy that requires delicate balancing: cooperating on infrastructure and foundational models while competing on enterprise applications and AI services.
Also Read | Two billion-dollar disruptions, one Reliance: Can Mukesh Ambani deliver?
The Jamnagar Gambit
The physical manifestation of this pivot is taking shape in Jamnagar, where Reliance Intelligence has begun constructing gigawatt-scale, AI-ready datacenters powered by green energy. The company announced a partnership with Google Cloud to establish an AI-focused cloud region at the site, leveraging the same location that houses Reliance’s massive oil refining complex.
At August’s annual general meeting, where the word “AI” appeared 80 times, Ambani framed the opportunity in characteristically expansive terms: “A decade ago, digital services became a new growth engine for Reliance. Now, the opportunity before us with AI is just as large, if not larger. Jio promised and delivered digital everywhere and for every Indian. Similarly, Reliance Intelligence promises to deliver AI everywhere for every Indian.”
The Meta partnership, announced at AGM as a 70:30 joint venture with an initial $100 million commitment, aims to deliver open-source AI models to Indian businesses.
The Valuation Question
Whether the market assigns Reliance’s AI vertical a $30 billion valuation depends on execution, competitive positioning, and investor appetite for infrastructure plays in an increasingly crowded AI landscape. Morgan Stanley’s 2x price-to-book assumption reflects valuations of Asian peer GDS, while US-based CoreWeave and Equinix trade at even richer multiples.
The bet is fundamentally this: that Reliance can leverage its capital access, energy infrastructure, and telecom distribution to capture a meaningful share of India’s AI infrastructure buildout — and that the market will value this franchise at hyperscaler multiples rather than traditional infrastructure returns.
For Ambani, who built fortunes first in petrochemicals, then in retail, and most recently in telecom through Jio’s disruptive entry, the AI pivot represents a fourth act. The question is whether artificial intelligence will prove as transformative for Reliance as the black gold that built the empire — or whether this time, the new oil stays liquid only on balance sheets.
(Disclaimer: The recommendations, suggestions, views, and opinions given by the experts are their own. These do not represent the views of The Economic Times.)