“India’s AI data centre race will not be won by the fastest builder — it will be won by whoever controls the power, the land, and the fibre underneath it all.” — The Mobile Times
India’s AI data centre infrastructure is already being carved up, and most Indian operators are watching from the sidelines. Foreign hyperscalers are committing billions while domestic players scramble for policy clarity, and the window to own this infrastructure — rather than rent it — is closing faster than the sector acknowledges. The AI data centre race is real, and India’s strategic position in it is far from guaranteed.
The TMT Position
- Microsoft, Google, and Amazon have collectively pledged over $30 billion in India cloud and AI infrastructure through 2026, dwarfing domestic commitments
- Reliance Jio and Adani are the only Indian conglomerates with the land, fibre, and capital to genuinely compete at hyperscale — everyone else is a tier below
- India’s power grid deficit, particularly reliable 24×7 green power, remains the single biggest structural constraint no policy announcement has solved
- Most people wrongly frame this as a tech story — it is fundamentally an infrastructure ownership story with 30-year revenue consequences
In This Article
Why India’s AI Data Centre Race Determines Who Profits From the Next Decade
The AI data centre race is not a future event. It is happening now, and the investment numbers make the stakes impossible to ignore. India’s data centre capacity is projected to exceed 2,700 MW by 2026, up from under 900 MW in recent years, according to JLL and CBRE tracking reports. That tripling of capacity sounds impressive until you measure it against what Southeast Asian rivals like Singapore and Malaysia have already locked in. The gap between ambition and deployed infrastructure in India remains embarrassingly wide.
Reliance Industries committed $25 billion toward new energy and digital infrastructure through 2026, with AI compute specifically called out in its investor communications. Adani Group’s data centre arm, AdaniConneX, has announced over 1 GW of capacity pipeline across Indian metros. These are not small bets. Yet neither announcement fully resolves India’s core problem: land acquisition timelines, inconsistent state-level power availability, and a fibre backbone that still has serious last-mile gaps in secondary cities where cheaper land actually exists. Named players matter here. Vague ecosystem optimism does not.
Is the Hyperscaler Dominance Argument Overstated?
Critics of the hyperscaler-alarm view argue that foreign investment in India’s AI data centre race is net positive, not threatening. The logic runs like this: Microsoft’s $3 billion India commitment announced in early 2026 creates local jobs, builds domestic AI skilling capacity through its Azure partnerships, and taxes remain payable under Indian law. Some analysts at Nasscom and KPMG have pointed out that Indian cloud adoption is still so nascent that any capacity addition, regardless of who owns it, closes a real supply deficit. That argument deserves a serious hearing before it gets dismissed.
But it falls short on the fundamental ownership question. When AWS, Google Cloud, or Microsoft Azure own the physical infrastructure layer, Indian enterprises pay dollar-denominated compute bills in perpetuity. The value extracted from Indian AI workloads flows to Seattle and Redmond, not to Mumbai or Chennai. India has watched this movie before with telecom equipment — Ericsson and Nokia built the 4G networks, Indian operators paid licensing and maintenance costs for decades. Owning the AI data centre layer is categorically different from renting capacity on someone else’s sovereign infrastructure, and conflating the two is an analytical error.
What Must Change Before India Loses This Window?
The AI data centre race requires India to solve three non-negotiable problems simultaneously, and policy announcements alone will not cut it. First, DoT and the Ministry of Power must jointly mandate 24×7 renewable energy access corridors for hyperscale facilities — the current open-access framework is too slow and state-specific to support gigawatt-scale deployments. Second, TRAI needs to finalise its data centre connectivity framework, which has been in consultation limbo since 2026 began, so that licensed operators can price and provision dedicated dark fibre at scale without regulatory ambiguity strangling investment decisions.
Success in the AI data centre race looks like this by the end of 2026: at least three Indian-majority-owned operators running facilities above 100 MW each, a functional green power corridor connecting major data centre clusters in Pune, Chennai, and Noida to dedicated renewable sources, and a TRAI-approved framework allowing telcos to monetise fibre and edge compute as bundled infrastructure products. Jio Platforms has the network reach to anchor this vision. Bharti Airtel’s data centre subsidiary, Nxtra, has the enterprise relationships. What both need is a policy environment that stops treating data centre licensing as a zoning problem and starts treating it as a national infrastructure priority.
The Mobile Times Verdict
India will not win the AI data centre race by welcoming every foreign dollar and calling it a strategy. Ownership of compute infrastructure determines who captures AI’s economic surplus for the next three decades. India has the demand, the engineering talent, and two or three domestic conglomerates with genuine scale ambitions. What it keeps lacking is the regulatory urgency that matches the commercial moment. If DoT, TRAI, and state power ministries cannot align on AI data centre infrastructure as a strategic national asset before 2026 closes, India will spend the following decade buying back access to its own data at hyperscaler rates.
Sources: TRAI ↗ | ITU ↗ | GSMA ↗ JLL Data Centre Report 2026; CBRE Asia Pacific Infrastructure Outlook 2026; Nasscom Cloud Adoption Survey 2026; KPMG India Digital Infrastructure Analysis 2026; Reliance Industries Investor Presentation 2026; AdaniConneX capacity announcements; Microsoft India Investment Declaration 2026; TRAI Data Centre Connectivity Consultation Paper 2026
People Also Ask
- Who is leading India’s AI data centre infrastructure build-out? Reliance Jio, AdaniConneX, and Nxtra by Bharti Airtel lead domestic capacity. Foreign hyperscalers Microsoft, AWS, and Google Cloud are committing the largest single investment tranches through 2026.
- Why is power supply a problem for AI data centres in India? AI data centres require uninterrupted, preferably renewable, power at gigawatt scale. India’s current open-access renewable energy framework is too fragmented and state-dependent to reliably support facilities above 100 MW without significant policy reform.
- How can Indian telecom operators benefit from the AI data centre race? Operators like Airtel and Jio can monetise owned fibre, edge compute nodes, and colocation capacity as bundled infrastructure products, provided TRAI finalises a connectivity framework that allows flexible, large-scale dark fibre provisioning to enterprise clients.
