India's AI Ambition Has a Hidden Price Tag. Are We Looking at It?

A few months ago, most people weren't talking about data centres. Today, they are at the centre of India's AI ambitions.

Google is partnering with Adani to build what is being described as India's largest data centre campus in Visakhapatnam. Andhra Pradesh is preparing power infrastructure capable of supporting more than 15 GW of data centre demand. Haryana has signed a ₹20,000 crore agreement with Anant Raj to expand data centre infrastructure.

On paper, this looks like progress.

Billions in investment. New digital infrastructure. More cloud computing capacity. A stronger AI ecosystem.

The story sounds simple.

But the deeper you look, the more complicated it becomes.

Because while India is racing to build the infrastructure needed for the AI age, a question is quietly emerging in the background:

What exactly will it cost us?

The scale of what is being planned is enormous.

The Visakhapatnam region alone has attracted proposals from major players including Google, Reliance, Tata Consultancy Services, CtrlS and GMR. Andhra Pradesh estimates that 16 major data centre projects are already under development there.

These projects are expected to generate approximately 9.1 GW of IT power load.

To put that into perspective, that is more electricity demand than some countries consume.

And the state's long-term roadmap is even bigger.

The government is building a transmission network designed to support over 15 GW of data centre demand by 2034, with room for further expansion beyond that.

This isn't just about storing photos, emails, or business files anymore.

This is about powering artificial intelligence. AI is changing the economics of computing.

Traditional cloud workloads are demanding, but AI systems require significantly more computing power. Training large AI models involves thousands of specialised chips running continuously for weeks or even months. Every query, image generation request, or AI-powered service adds to the load.

The result is simple.

More AI means more servers. More servers mean more electricity and more electricity means more infrastructure.

India's data centre capacity has already grown rapidly, rising from roughly 520 MW in 2020 to around 1.5 GW in 2025. Estimates suggest this could reach between 4.5 GW and 6.5 GW by 2030.

That growth is not being driven by accident.

It is being driven by demand.

Businesses want cloud infrastructure. Governments want digital services. Startups want AI capabilities. Global technology companies want local computing power.

The opportunity is real.

Data centres can attract investment, improve digital infrastructure, strengthen India's technology ecosystem, and reduce dependence on overseas computing resources.

For a country that wants to become a major AI player, these facilities are not optional.

They are necessary.

But necessary does not mean cost-free.

One of the least discussed costs is electricity.

Today, India's data centres consume an estimated 10 to 15 terawatt-hours of electricity annually.

By 2030, that figure could rise to 40 to 45 terawatt-hours.

That is a massive increase in demand.

In Andhra Pradesh, officials are already planning new 765 kV, 400 kV and 220 kV transmission infrastructure specifically to support future data centre clusters.

Entire sections of the power grid are being redesigned around the needs of digital infrastructure.

And electricity is only part of the story.

Data centres also need cooling.

A lot of cooling.

Every server generates heat. Thousands of servers packed into large facilities generate enormous amounts of it. If that heat is not removed efficiently, equipment can fail.

Cooling requires water.

Lots of it.

According to estimates cited by industry studies, a 1 MW data centre can consume roughly 68,500 litres of water every day.

A 20 MW facility can require around 1.4 million litres daily.

That is equivalent to the daily water needs of approximately 27,000 families.

Across India, data centres are estimated to have consumed around 150 billion litres of water in 2024 alone.

As the sector expands, that number is expected to grow.

This raises difficult questions.

Not because data centres are inherently bad.

But because India is also one of the most water-stressed countries in the world.

Several regions already face groundwater depletion, seasonal shortages, and increasing pressure on water resources.

If digital infrastructure becomes a major consumer of water, how should those trade-offs be managed?

Who gets priority when resources become constrained?

Those questions rarely appear in investment announcements.

Yet they are becoming increasingly important.

Then there is another issue that receives even less attention.

Electronic waste.

AI systems evolve rapidly.

The chips powering today's most advanced models can become outdated within a few years.

Servers, GPUs, networking equipment and storage systems are regularly replaced to maintain performance.

Industry estimates suggest servers typically have a lifespan of only three to six years.

That creates a growing stream of electronic waste.

A study published in Nature Computational Science estimated that AI-related e-waste could reach between 1.2 million and 5 million metric tonnes globally by 2030.

Some projections suggest this could represent a thousand-fold increase compared to 2023 levels.

That waste contains valuable materials but it also contains hazardous components that require proper recycling and disposal.

Without adequate systems, the environmental burden simply shifts elsewhere.

This is where the conversation becomes more interesting.

India is not wrong to pursue data centres.

Far from it.

The world is entering an AI-driven era.

Countries that fail to build digital infrastructure may find themselves dependent on those that do.

The economic opportunity is substantial.

New investment is flowing into the country. Infrastructure is being built. Technology ecosystems are expanding. Local computing capacity can support startups, businesses, researchers and public services.

The benefits are real.

But so are the costs.

And the challenge is not choosing one side over the other.

The challenge is recognising both at the same time.

Because much of the public conversation currently focuses on the gains.

The billions being invested.

The companies arriving.

The facilities being announced.

The future being promised.

Far less attention is given to the resources required to make that future possible.

How much electricity will be needed?

Where will that electricity come from?

How much water will be consumed?

How will electronic waste be managed?

What standards should govern these facilities?

And perhaps most importantly, who is asking these questions before the infrastructure is already built?

India's data centre boom is not a story about technology alone.

It is a story about trade-offs.

The same facilities that power AI models, cloud platforms and digital services also consume energy, water and physical resources at an unprecedented scale.

Ignoring that reality would be a mistake.

But rejecting the opportunity entirely would be another.

The goal should not be to slow innovation.

The goal should be to understand its full cost.

Because every major technological shift comes with a price tag.

For decades, we measured progress by what technology could create.

The AI era may force us to pay equal attention to what it consumes.

India is betting big on data centres.

The investment is happening. The infrastructure is being built. The AI race has already begun.

The question is not whether this future is coming.

The question is whether we are preparing for everything that comes with it.