In a fresh note, Morgan Stanley analysts, including Mayank Maheshwari, say recent comments from RIL’s chairman and policymakers “point to a runway of NAV creation as large as 15% of RIL’s NAV” from the AI infrastructure opportunity alone.
The brokerage noted that “AI economics point to ~12–14% ROCE” as Reliance leverages deep partnerships with Meta, Google, and Microsoft, while executing a multi-gigawatt data centre build-out anchored by low-cost renewable power and in-house battery solutions.
“Reliance has reinvented itself every decade, and AI is set to reshape its equity story,” the report said, adding that generative AI deployment “enables large-scale capital deployment while unlocking value creation with synergies across energy, digital, consumer, and media verticals.”
The brokerage retained Reliance Industries as a top pick, maintaining an Overweight rating and an unchanged price target of Rs 1,803, implying a 24% upside from the February 6 close of Rs 1,451.
Central to the thesis is a capex-heavy push into hyperscale computing. Morgan Stanley estimates that Reliance will invest $12–15 billion in AI infrastructure to build a 1GW data centre, assuming $0.8 million per MW for the powered shell and related infrastructure, with the remainder spent on chips. Of this, the brokerage expects RIL to underwrite around 25% of the compute capacity, translating to roughly $7 billion in infrastructure and $5 billion for 250MW-equivalent AI chips to be deployed directly by Reliance.
The remaining 75% of capacity could be monetised as Data Centre as a Service for hyperscalers and large model providers, with indicative annual revenues of $1.5–1.6 million per MW, the note says. For a 100MW block, Morgan Stanley’s model estimates revenues of $822 million, EBITDA of $743 million, a 31% net margin, 12.2% ROCE, and 18.3% ROE, underscoring what it describes as attractive “AI data centre infrastructure and inference economics,” even before accounting for tax incentives.The brokerage adds that the macro and policy environment is unusually supportive. “This, along with the 20-year tax holiday for hyperscalers in India’s recent Budget and GPU sourcing from the US under the India–US trade framework, supports a significant acceleration of the ‘Powering AI’ story,” it said.
Putting the scale in context, Morgan Stanley estimated that India would need ~9GW of AI DC power by 2030, compared to China and the US at ~48GW and ~102GW,” with Malaysia – which has rolled out similar tax incentives – already having “~8GW of capacity for datacenters in the works.
On the demand side, Morgan Stanley expects AI compute in Asia ex‑China to be “powered by gas and batteries,” with about 50GW of new coal and gas‑based generation through 2030, backed by renewables as gas markets stay oversupplied. China’s compute build‑out, by contrast, is projected to be met “largely by renewables and batteries as unit economics improve,” while affordability remains “top of mind for policymakers.”
Reliance’s integrated energy and new‑energy businesses are positioned as key enablers of this AI build‑out. The report flags that data centres are “unlocking new demand in Energy Storage Systems (ESS)” as gas turbine shortages and grid constraints force operators toward “alternative, fast deployable power solutions.” It points to 2026 as a potential “inflection point where ESS becomes a viable mainstream component of datacenter power architecture,” and notes that RIL “in the recent call upped its LFP cell pack/battery production capacity to 100GWh vs. 30GWh previously.”
Given that data centres are “significant energy consumers,” Morgan Stanley argued that the Reliance can “underwrite more than 20GW of internal power demand, supporting 100GW of its solar panel capacity and 30-40GWh of its own battery capacity.” In solar, it highlights that while “there is oversupply in India on solar panel manufacturing capacity, there are limited integrated manufacturers like RIL,” which is planning an integrated polysilicon‑to‑module chain scaling to 20,000 tonnes of polysilicon and 20GW each of ingots, wafers, cells, and modules by FY28.
The digital partnership layer is the other critical pillar. “Google Cloud is partnering with RIL to offer generic AI solutions to enterprises, while RIL also plans to offer its own AI agents built on Google Enterprise,” the report noted.
On the consumer side, the Google‑RIL tie‑up will see Jio subscribers offered “access to Google’s Gemini Pro AI model, 2TB of cloud storage and other functionalities,” with 5G AI add‑on plans starting at “Rs51/month” for existing 4G customers. Morgan Stanley also points to RIL’s strategic tie‑ups “with Meta, Google and Microsoft” and its plan to provide “AI as a service to subscribers” while building “multi‑gigawatt datacenters,” arguing that these partnerships de‑risk both technology and utilisation risk.