The Cost of Becoming an AI Superpower

India is rapidly investing in positioning itself as a global hub for artificial intelligence. Large investments, government incentives, and international AI summits all signal the same ambition: to place the country at the centre of the emerging AI economy.

Major industrial groups are committing enormous resources to this goal. Reliance has announced plans to invest around $110 billion into AI infrastructure, while the Adani Group has pledged $100 billion by 2035 to develop AI-enabled data centres ¹.

To support this ambition, India has begun easing regulations and introducing tax incentives to attract technology companies.

One proposal offers a tax holiday until 2047 for foreign firms that set up data centres in the country ².

Companies providing data centre services from India to related global affiliates may now benefit from a 15% safe harbour profit margin, replacing the earlier reliance on general transfer-pricing rules ³. Together, these policies are designed to encourage companies to build and operate digital infrastructure in India.

The promise behind these developments is compelling. AI infrastructure is portrayed as the engine of the next economic transformation, one that will generate jobs, accelerate growth, and place India among the world’s leading technological powers.

But beneath these announcements lies a quieter reality: artificial intelligence is not weightless. It depends on a rapidly expanding network of physical infrastructure that consumes vast amounts of electricity, water, and land.

And that infrastructure comes with demands that are only beginning to surface.

The Gateway for India

India has played only a limited role in the global artificial intelligence boom so far, largely because it lacks large-scale semiconductor manufacturing ⁴. As a result, the country’s most viable entry point into the rapidly expanding AI economy lies in building the infrastructure that powers it, particularly large data centres.

This shift is already visible in cities like Mumbai, one of the most densely populated cities in the world. Mumbai has rapidly emerged as one of the largest data centre hubs in the Asia-Pacific region.

By early 2025, the city’s data centre capacity had crossed 4 gigawatts, placing it behind only Shanghai and Tokyo in the region. As artificial intelligence, cloud computing, and streaming services continue to expand, the demand for such infrastructure is expected to grow even further ⁵.

At the same time, the global geography of this infrastructure is beginning to shift. In several countries in the Global North, new data centre projects are facing increasing resistance from local communities ⁶. 

As opposition grows in these regions, companies are increasingly looking toward the Global South as a new frontier for expansion, where regulatory systems are still evolving and public accountability can be weaker ⁷.

These facilities require enormous amounts of electricity and water to operate, particularly for cooling ⁸. In some regions, data centres have already contributed to rising electricity costs and increased strain on water supplies in areas that are already drought-prone ⁹. 

Reflecting the growing global anxiety over freshwater depletion, the United Nations has warned that the world may be moving beyond a simple water crisis toward what it describes as water bankruptcy ¹⁰.

Against this backdrop of growing debate over AI’s environmental footprint, India recently hosted the AI Impact Summit, the fourth in a series of global gatherings on artificial intelligence and notably the first to be held in a Global South country ¹¹. In principle, such a forum could have created space to address the environmental and social consequences of rapidly expanding AI infrastructure.

Yet much of the event appeared to centre on geopolitical positioning and the announcement of massive corporate investments. Discussions of sustainability and climate responsibility were present, but they often seemed overshadowed by the larger emphasis on positioning India as a destination for global AI infrastructure ¹².

The Resource Demands Behind Artificial Intelligence

While AI is often imagined as an intangible technological system, the facilities that power it depend heavily on physical resources, particularly electricity and water.

Data centres require a constant and uninterrupted power supply, operating around the clock to prevent service disruptions. Their energy demand has grown steadily. By 2025, data centres accounted for roughly 0.5 percent of India’s total electricity consumption, a figure expected to more than double by the end of the decade as new facilities are built ¹³.

Electricity represents the highest operational cost for data centres, often accounting for 60–70 percent of total expenses. Operators rely on backup diesel generators during power disruptions, while renewable sources such as solar and wind are increasingly explored but remain difficult to use as a sole power source due to their variability and the requirement for continuous uptime ¹⁴.

Alongside electricity, water plays a critical role in cooling servers that generate large amounts of heat. Water-based cooling systems are often used because they are more energy-efficient than air cooling ¹⁵.

As a result, data centre water consumption has grown rapidly. By 2025, facilities in India were estimated to consume around 150 billion litres of water annually, a figure projected to more than double by 2030 ¹⁶. A large data centre with a capacity of around 100 megawatts can require roughly two million litres of water per day.

These pressures are particularly significant because many data centre hubs are located in cities such as Mumbai, Chennai, Hyderabad, and Bengaluru, where water demand is already high. India holds only about 4 percent of the world’s freshwater resources while supporting nearly 18 percent of the global population, making water management an ongoing challenge ¹⁷.

Operators therefore face a trade-off between electricity and water consumption. Cooling systems that minimise water use often require more electricity, while water-based systems reduce energy demand but depend on reliable local water supplies. As AI workloads grow more computationally intensive, this balance becomes increasingly difficult to maintain ¹⁸.

Despite these challenges, India still lacks a comprehensive national policy governing the environmental performance of data centres. While several states have introduced policies to attract investment, sustainability requirements related to energy efficiency and water use remain limited ¹⁹. 

Development Beside Inequality

In Mumbai, these developments are unfolding in a city already defined by stark contrasts. The city hosts some of the country’s most advanced digital infrastructure, while at the same time nearly one in five residents lives below the poverty line. Informal settlements lacking reliable access to electricity and safe water exist alongside rapidly expanding technology districts ²⁰.

Globally, the environmental impacts of large industrial projects often fall most heavily on communities with the least political and economic power. Data centres require vast amounts of resources and decisions about where they are built are rarely neutral. Land acquisition, infrastructure development, and the allocation of resources shape the geography of such projects.

In countries marked by deep social and economic inequalities, such decisions rarely remain neutral. They shape who benefits from development and who is asked to bear its costs. Communities already living at the margins often find themselves absorbing the environmental and social burdens of projects designed to serve distant markets. 

In India, this imbalance is particularly visible. Slum settlements that struggle for reliable electricity and clean water exist just kilometres away from hyperscale data centres that consume more power in a single day than entire villages use in weeks ²¹.

When billionaires such as Bill Gates describe India as a “laboratory to do things” ²², such a remark invites an uncomfortable question: whose communities will become the testing ground for this technological experimentation?

As resistance to data centres in the Global North is growing, companies may increasingly look toward regions in the Global South where regulatory frameworks are still evolving. If that pattern continues, marginalised communities in countries like India could find themselves absorbing the costs of a technological transition largely driven elsewhere.

The dynamics of the AI economy already reveal similar patterns. Across the Global South, thousands of data workers are employed to clean and moderate AI training datasets, filtering out violent, hateful, and disturbing content from the internet ²³. Reports have shown that this work often exposes workers to psychologically harmful material while providing limited protection or support. Many take up these jobs simply because few other employment opportunities are available ²⁴. 

Taken together, these developments point to a broader concern sometimes described as AI colonialism ²⁵. The infrastructure, labour, and environmental costs of building the AI economy risk being concentrated in parts of the world with fewer resources and weaker safeguards.

As India continues its push to become a global hub for artificial intelligence, the question is not only how the technology will advance, but also who will ultimately bear the costs of that progress.

Echoes of an Old Logic

Decades ago, a controversial internal memo written by economist Lawrence Summers, then chief economist of the World Bank, suggested that there was an economic logic to relocating pollution-intensive industries to poorer countries.

He suggested “economic logic behind dumping a load of toxic waste in the lowest wage country is impeccable and we should face up to that.” ²⁶

The memo was leaked and widely criticised, becoming associated with what critics called toxic colonialism; the shifting of environmental harm to places with fewer resources and weaker political power ²⁷.

Today, the echoes of that logic are visible in the geography of global technology infrastructure. As communities in parts of the Global North increasingly question the environmental costs of large data centres, countries eager to participate in the digital economy may find themselves welcoming these same infrastructures.

Artificial intelligence itself remains widely misunderstood. For many, it simply means chatbots or generative tools that produce text and images. Yet these systems are only one part of a much larger technological ecosystem that is gradually being integrated into logistics, finance, governance, surveillance systems, and public services. At the same time, regulatory frameworks for these technologies remain incomplete across much of the world ²⁸.

As India expands the infrastructure that will power this new technological landscape, the country faces a quiet but significant question. Will this expansion produce systems that serve the needs of its people and its environment, or will the costs of this transformation fall most heavily on communities that benefit from it the least?

For a country that ranks among the world’s largest economies yet continues to grapple with deep social and regional inequalities, this question reflects a broader debate about the meaning of development itself. The direction of technological growth raises fundamental concerns about whose interests are prioritised, who bears the environmental costs, and who is left behind.

Within a global economic system that often amplifies existing disparities, artificial intelligence may have simply become another force accelerating uneven forms of growth. In this context, the question becomes not only how quickly technological systems expand, but also development for whom.

For now, construction continues, and the long-term consequences of this digital transformation are yet to unfold.

References

1. Reuters, Misra, S., & Dugar, U. M. (2026, February 19). Reliance, Adani drive India’s AI push with plans to invest $210 billion. Reuters. Retrieved March 10, 2026, from https://www.reuters.com/world/india/ambanis-reliance-will-invest-110-billion-ai-2026-02-19/
2. Reuters & Pandya, D. (2026, February 1). India gives 20-year tax holiday to foreign firms using local data centres. Reuters. Retrieved March 10, 2026, from https://www.reuters.com/world/india/india-gives-20-year-tax-holiday-foreign-firms-using-local-data-centres-2026-02-01/
3. PIB. (2026). Budget 2026–27 Sets the Stage for India as a Global Hub for Cloud and AI Infrastructure. https://www.pib.gov.in/PressReleasePage.aspx?PRID=2227953&reg=3&lang=2
4. IBEF. (2024).  https://www.ibef.org/research/case-study/india-s-semiconductor-push-building-a-robust-chip-manufacturing-ecosystem
5. Earth Journalism Network. (2025). https://earthjournalism.net/stories/mumbais-data-center-dreams-run-on-coal-and-inequality
6. Digital Peace & Migliorini, D. (2026). https://digital-peace.org/how-grassroots-opposition-data-centers
7. Vogel, D., & Kagan, R. (2002). Dynamics of Regulatory Change: How Globalization Affects National Regulatory Policies.
8. Vogel, D., & Kagan, R. (2002). Dynamics of Regulatory Change: How Globalization Affects National Regulatory Policies.
9. Rodríguez, E., & New York Times. (2025, October 20). AI Data Centers Create Fury From Mexico to Ireland – The New York Times. nytimes. Retrieved March 10, 2026, from https://www.nytimes.com/2025/10/20/technology/ai-data-center-backlash-mexico-ireland.html
10. United Nations. (2026). World enters era of ‘global water bankruptcy’. https://news.un.org/en/story/2026/01/1166800
11. The Guardian. (2026). Tech billionaires fly in for Delhi AI expo as Modi jostles to lead in south. https://www.theguardian.com/technology/2026/feb/18/delhi-ai-expo-modi-jostles-lead-south
12. Deccan Herald & Bose, M. (2026). AI Global Summit ignores massive water toll of data centres, greens red-flag leaders. Mrityunjay Bose Read more at: https://www.deccanherald.com/india/maharashtra/ai-global-summit-ignores-massive-water-toll-of-data-centres-greens-red-flag-leaders-3909859
13. BBC & Inamdar, N. (2025, November 9). Google, Meta, Amazon: India’s data centre boom confronts a water challenge. BBC. Retrieved March 10, 2026, from https://www.bbc.com/news/articles/cgr417pwek7o
14. Earth Journalism Network, Day, M., Baratt, L., & Sushmita. (2026). Amazon’s Hidden Data Centers in India and Beyond. https://earthjournalism.net/stories/amazons-hidden-data-centers-in-india-and-beyond
15. Bacurin, L. (2026). The Man Behind the Curtain: The Hidden Data Centers Powering the AI Revolution. Digital Peace. https://digital-peace.org/environmental-impact-ai-data-centers/
16. Tripathi, V. (2026). Scaling India’s Data Centre Ecosystem. Council on Energy, Environment and Water. https://www.ceew.in/publications/how-is-data-centre-infrastructure-in-india-shaping-power-and-water-use
17. World Bank Group. (2023). How is India addressing its water needs? https://www.worldbank.org/en/country/india/brief/world-water-day-2022-how-india-is-addressing-its-water-needs
18. Tripathi, V. (2026). Scaling India’s Data Centre Ecosystem. Council on Energy, Environment and Water. https://www.ceew.in/publications/how-is-data-centre-infrastructure-in-india-shaping-power-and-water-use
19. BOOM & Rizwan, H. (2026). Interview: In India, A Data Centre Can Be Built Without Any Environmental Assessment https://www.boomlive.in/decode/interview-in-india-a-data-centre-can-be-built-without-any-environmental-assessment-30754.  https://www.boomlive.in/decode/interview-in-india-a-data-centre-can-be-built-without-any-environmental-assessment-30754
20. Earth Journalism Network, Day, M., Baratt, L., & Sushmita. (2026). Amazon’s Hidden Data Centers in India and Beyond. https://earthjournalism.net/stories/amazons-hidden-data-centers-in-india-and-beyond
21. Earth Journalism Network. (2025). https://earthjournalism.net/stories/mumbais-data-center-dreams-run-on-coal-and-inequality
22. Times of India. (2024). Bill Gates calls India a laboratory to try things. Times of India. https://timesofindia.indiatimes.com/world/us/bill-gates-slammed-for-calling-india-laboratory-to-try-things/articleshow/115907231.cms
23. The Guardian. (2026, February 5). ‘In the end, you feel blank’: India’s female workers watching hours of abusive content to train AI. The Guardian. Retrieved March 10, 2026, from https://www.theguardian.com/global-development/2026/feb/05/in-the-end-you-feel-blank-indias-female-workers-watching-hours-of-abusive-content-to-train-ai
24. Rest of World, Wangari, S., & Vaidyanathan, G. (2025). How Big Tech hides its outsourced African workforce. https://restofworld.org/2025/big-tech-ai-labor-supply-chain-african-workers/
25. Khan, R., & LSE. (2025). From AI colonialism to co-creation: bridging the global AI divide. https://blogs.lse.ac.uk/medialse/2025/07/14/from-ai-colonialism-to-co-creation-bridging-the-global-ai-divide/
26. Khan, R., & LSE. (2025). From AI colonialism to co-creation: bridging the global AI divide. https://blogs.lse.ac.uk/medialse/2025/07/14/from-ai-colonialism-to-co-creation-bridging-the-global-ai-divide/
27. The Wire, Tiwari, V. K., & Siddiqui, K. (2026). Who Will Bear the Cost of AI Colonialism? The Wire. Retrieved March 10, 2026, from https://thewire.in/tech/who-will-bear-the-cost-of-ai-colonialism
28. Mind Foundry. (2026). AI Regulations around the World – 2026. https://www.mindfoundry.ai/blog/ai-regulations-around-the-world