For years, the environmental conversation around cloud computing focused almost entirely on electricity. Carbon footprints, renewable energy targets and power usage ratios dominated every sustainability report from every major cloud provider.

Water barely got a mention. That is starting to change and India is where the shift is becoming visible.

Reuters reported last week that Amazon is facing active scrutiny from Indian authorities over its data centre water consumption, a development that coincides with the company declaring "water positive" status in India a year ahead of its own internal target. The timing of both announcements tells you something important: the pressure to disclose was real and it arrived before any formal regulation required it.

India is in the middle of an aggressive data centre build-out. According to real estate consultancy JLL, the country's data centre capacity is projected to grow by 77 percent by 2027, reaching 1.8 gigawatts of installed capacity. Amazon, Google, Microsoft and Meta are all expanding infrastructure there, drawn by a growing digital economy, a large developer base and government policy actively encouraging technology investment.

The problem is that data centres are water-intensive by nature. Cooling systems, which keep server hardware from overheating, use significant volumes of water through evaporative cooling towers. A single 1-megawatt facility can consume up to 25.5 million litres of water annually, roughly equivalent to the daily water needs of 300,000 people.

India's data centre sector consumed approximately 150 billion litres of water in 2025. By 2030, that figure is projected to more than double to 358 billion litres. That expansion is happening in a country where large parts of the population already face seasonal water stress and where agricultural water demand competes directly with industrial and urban use.

As of early 2026, India has no sector-wide water performance standards for data centres and no mandatory reporting requirements for water usage. The scrutiny Amazon is now facing appears to be arriving through regulatory and parliamentary channels before formal rules are in place, which is precisely when voluntary disclosures tend to matter most.

The Indian Ministry of Electronics and Information Technology confirmed in a parliamentary response earlier this year that data centre cooling requirements vary depending on the technology deployed and that the industry is "adopting advanced cooling mechanisms." That language is careful. It stops short of setting performance targets but it signals official awareness of the problem.

Amazon's response to the scrutiny was to go public with a detailed account of its water position in India. The disclosures are worth reading carefully because they make a meaningful distinction that most coverage missed.

Amazon does not use water to cool its data centres in India. The company's facilities there rely on air-side cooling, drawing in outside air rather than evaporative water-based systems. This is a legitimate operational difference from many competing facilities and explains in part how Amazon was able to claim "water positive" status.

Globally, Amazon reported that its data centres used 2.5 billion gallons of water in 2025, with a water usage effectiveness rate of 0.12 litres per kilowatt-hour, compared to an industry average of 0.84 litres per kilowatt-hour. That is a meaningful efficiency gap. The company also disclosed that it returned approximately two-thirds of its total water withdrawal back to communities through infrastructure investment projects.

For India specifically, Amazon says it achieved water positive status through a combination of reduced facility water use and external initiatives including watershed restoration and efficient irrigation support.

These are substantive disclosures. The question worth sitting with is how much of this would have been published without external pressure.

India is not the only jurisdiction where data centre environmental reporting is becoming a governance issue rather than a voluntary exercise.

In the European Union, the updated Energy Efficiency Directive now requires data centres to report water usage effectiveness metrics alongside power usage figures and to demonstrate active steps toward optimization. Facilities above a certain size threshold must disclose these metrics publicly. The EU's AI Act adds another layer by requiring transparency around the infrastructure hosting high-risk AI workloads, which inevitably includes the cooling and energy systems supporting them.

In the United States, the picture is more fragmented. The previous administration moved to streamline permitting for data centre projects and eased some environmental review requirements. But individual states, particularly those with water scarcity concerns, are increasingly active. Arizona and New Mexico have both pushed back on large data centre projects citing aquifer depletion.

India's path is likely to run through a combination of parliamentary pressure, state-level land and water allocation decisions and eventually formal sector standards from MeitY. The pace will depend partly on how visibly the sector grows and whether local communities begin to raise concerns about water access in the areas where large facilities are being built.

Most AI newsletters summarize headlines. MavSource is different.

We aggregate updates from all major AI newsletters, podcasts, company news, AI labs, public and private company activity, GitHub projects, funding rounds, earnings calls, and investor letters — hundreds of sources every day. Then we summarize what matters, analyze emerging trends, and add our own founder commentary so you understand why a development may matter — not just what happened.

One 5-minute email, every morning. Built for investors, founders, and operators who want to understand AI as a business, technology, and market trend — not just another news cycle.

The daily email is free. It's also the entry point to a deeper intelligence product covering watchlists, public-company read-throughs, startup trackers, and investor-letter tracking.

Subscribe Now For Free

There is a practical dimension here that most technology readers do not factor into how they think about the services they use.

Every file stored in the cloud, every query run through a search engine, every video streamed and every voice assistant response has a physical resource cost attached to it. A single conventional web search consumes roughly half a millilitre of water. A sustained interaction with a generative AI tool can consume up to 500 millilitres across the session. Multiply those figures across billions of daily interactions and the aggregate is enormous.

Amazon's efficiency figures are genuinely good relative to the industry average. But the category itself is resource-intensive and the fastest-growing workloads, particularly AI inference and training, are the most demanding both in electricity and in the cooling required to run the hardware that supports them.

The companies building AI infrastructure are aware of this. Amazon's VP of Global Data Centre Operations noted that for AI workloads specifically, the company is deploying closed-loop liquid cooling systems that recirculate fluid rather than consuming fresh water. That is the right direction. It also signals that the current industry efficiency numbers will face pressure as AI workloads scale.

The significance of what happened in India last week is not really about Amazon's water figures, which turned out to be relatively defensible. It is about the nature of the conversation itself.

A national government formally directing a hyperscaler to account for its physical resource consumption, before regulations requiring that disclosure exist, is a meaningful moment. It suggests that the implicit bargain where tech companies build freely in exchange for jobs, investment, and digital infrastructure is being renegotiated, at least at the margins, to include environmental accountability.

That precedent will travel. The next government to raise these questions will do so with Amazon's response as a reference point, and with the knowledge that the scrutiny produced detailed public disclosures that would not otherwise have appeared. For every cloud company planning large infrastructure in water-stressed regions, that is the clearest signal yet that voluntary reporting is no longer a safe long-term position.