Connecting the Unconnected is a monthly column by the Digital Empowerment Foundation (DEF) that explores how technology can drive inclusion and governance in India. The column focuses on how the digital divide impacts communities differently and advocates for equitable, citizen-informed solutions that ensure technology empowers rather than excludes.
As India prepares for the AI Impact Summit 2026 and the aspiration to support the trillion-dollar digital economy, there is no doubt that the country is a vigorous runner in the AI race. Artificial intelligence has become a political and economic inevitability and is now seen as essential to strategic autonomy, data sovereignty, global competitiveness, and investment attractiveness. The message is clear: India cannot afford to sit this race out.
On paper, the momentum looks quite impressive. Official statements point to India ranking among the top three globally in AI competitiveness, and a recent Deloitte report also suggests the AI market could grow to US$20-22 billion by 2027 at nearly 30% annually.
But behind this optimism sits a harder truth that is not being discussed as much. India’s AI data centre ambitions are greater than its energy system can realistically support, and it is also in direct tension with the country’s commitment to achieving net zero by 2070.
India generates close to one-fifth of the world’s data, but hosts only about 3% of global data centre capacity. Now this imbalance has become the standard justification for the domestic build-out, and if paired with national security issues, the situation has almost created political urgency for some, leaving very little appetite and willingness to slow down.
Yet AI data centres are not just any other industry that came up in the last few decades; they are among the most energy-intensive forms of infrastructure in the economy. A single hyperscale AI facility can consume as much electricity as a city of 1,00,000 households, as per the International Energy Agency. Reports also suggest that some of the largest data centre parks planned around Mumbai are expected to draw power equivalent to nearly two million homes.
Amidst all this, industry forecasts indicate that India’s data centre power demand will rise to more than 2 GW by 2026, and by 2030, the capacity is expected to balloon to over 8 GW, which is a fivefold jump that raises serious questions about how this demand will be met.
Now, these figures can be easily ignored by pointing to India’s installed capacity of nearly 500 GW, implying that data centres will only account for a small fraction of total power. However, this framing could be misleading. Roughly half of India’s installed capacity comes from renewables such as solar and wind, which do not generate electricity continuously. Large-scale energy storage, which is essential for making renewables reliable, is still missing, and despite announcements in the 2024 Budget, a national storage policy has yet to materialise.
Data centres, as we know, cannot operate on intermittent power. They require uninterrupted, high-quality electricity, day and night. In practice, this means reliance on coal- and lignite-based generation, which still anchors India’s grid. Even these plants operate at an average plant load factor of about 70%, not at full capacity. Experts believe that once reliability is accounted for, a 10 GW data centre load amounts to roughly 6% of India’s dependable round-the-clock power supply, not the 2% often cited. This is a significant share, especially when India is also electrifying transport, cooking, and industry at scale.
This begs an uncomfortable question: will India rely on coal to fuel its AI expansion?
The trajectory suggests that it already is to some extent. While India’s energy mix is gradually diversifying, coal production continues to rise in response to growing power demand. As per the Ministry of Coal’s year-end review 2025, the country witnessed its highest ever coal production in the year 2024-25.
The coal production for 2024-25 increased by 4.98% over the previous year to 1047.523 million tonnes. The coal supply also increased from 973.01 million tonnes in FY 2023-24 to 1025.33 million tonnes in FY 2024-25, with a growth of about 5.38%. With increased domestic coal production, the coal imports in the country during 2024-25 fell by 7.9% as compared to the previous year. As per the ministry, there has been abundant and uninterrupted coal supply to the consumers as per their ‘requirement’ in this period, and there have been no reports of coal shortages in the country. It also added that the coal demand is expected to increase to about 1.5 billion tonnes by 2030.
In parallel, India asked utilities in 2023 not to retire coal-fired plants until at least 2030 due to rising demand. In several cases, earlier closure decisions have also been reversed. In Mumbai, coal plants run by Tata and Adani were slated for shutdown as part of emission reduction efforts only for those plans to be rolled back when authorities concluded that electricity demand was rising too fast. A recent investigation report by SourceMaterial and the Guardian revealed that the main reason for this was the emergence of data centres.
This reflects that every additional megawatt required for AI infrastructure strengthens the argument to keep coal’s presence for longer, until our renewable energy options are reliable enough.
Meanwhile, global giants such as Google and Amazon Web Services are continuously committing billions of dollars for expansion, alongside Indian conglomerates like Reliance, AdaniConneX, Tata Consultancy Services, and Bharti Airtel. State governments, competing for these investments, are also offering subsidised land, electricity duty waivers, and industrial tariffs. With energy accounting for roughly 65% of data centre operating costs, cheap and reliable power becomes the overriding priority, and carbon intensity a secondary concern.
The strain is not limited to electricity. Data centres are also heavy water users, particularly for cooling, and they are concentrated in cities already facing acute water stress. Mumbai, Bengaluru, Chennai, and Hyderabad routinely experience water shortages during the summer months, yet they remain the focal points of expansion due to their geographical positioning.
This is unfolding in a country where nearly two-thirds of households still experience some form of energy poverty and where basic access to reliable power remains a development challenge in many parts of the country. Reports have also revealed that India will likely experience significant evening power shortages by 2027 (20-40 GW), even if all the thermal & hydro capacity currently under construction comes online as planned. As a result, during periods of grid stress, households and small businesses are likely to bear the impact through outages and rationing, while AI data centres remain insulated by contractual protections and policy incentives.
None of this is an argument against AI itself. India cannot afford to opt out of a technology that will shape economic and strategic power in the decades ahead. But the current reality is concerning. AI infrastructure is scaling far faster than India’s ability to provide clean and reliable power. The gap is being filled by coal in the meantime, which continues to expand despite its well-documented links to health hazards, displacement, and carbon emissions, even as affected communities protest against increased mining.
On the other hand, Coal India Limited is also exploring the possibility ofsetting up data centres in its decommissioned mines, repurposing redundant mines to meet India’s rising demands for digital infrastructure, which can again pose the question of whether we are replacing one problem with another.
On a positive note, last month, the Ministry of Power released the Draft National Electricity Policy (NEP) 2026, which acknowledges many of the structural tensions with India’s power infrastructure and places emphasis on large-scale deployment of energy storage, grid modernisation, and gradual reduction of cross-subsidisation, all of which are critical if India is to power energy-intensive digital infrastructure without locking itself into a coal-heavy future. It also signals a longer-term transition through nuclear, storage-backed renewables and smarter distribution systems that could, in theory, reconcile rising electricity demand with climate commitments.
Whether this framework becomes a genuine course correction or remains aspirational will depend entirely on implementation. Until then, India’s AI ambitions, rising power demand, and net zero goals are set to remain in tension.
Irrespective of the final outcome, it is imperative that industrial transitions are managed in a manner that safeguards communities already undergoing energy transitions and experiencing resource-related vulnerabilities, ensuring that the process results in no unintended or collateral adverse impacts.
Maitri Singh works in the Research and Communications division at the Digital Empowerment Foundation, focusing on just transition and the environmental impacts of AI data centres in India. She also leads the Secretariat of DEF’s Digital Just Transition Taskforce, which convenes experts at the intersection of climate and digital to inform policy, research, and advocacy.
Views expressed are the author’s own.