Artificial intelligence is reshaping modern life at breakneck speed — but its progress comes with a hefty energy bill. Every AI-generated response, virtual assistant command, or image creation consumes substantial electricity, much of it drawn by sprawling data centres. Whether you’re watching a movie, scrolling through social media, shopping online, or transferring money digitally, your activity runs through these massive computing hubs — and they’ve become some of the most energy-intensive facilities on Earth.

Imagine running a small business and using AI tools like ChatGPT, Gemini, or Copilot to plan your marketing. You might ask 15 questions about campaign ideas, generate 10 versions of an image, and produce three short promotional videos. Together, those actions alone would consume around 2.9 kWh of electricity — roughly equal to running a TV for 29 hours or a microwave continuously for three hours. That’s the invisible energy cost of everyday AI interactions.

Also Read | Zoho’s Ulaa Browser Challenges Chrome and Safari With Privacy-First Features

Now, multiply that by thousands of users and servers operating non-stop across global data centres. These facilities power everything from streaming services and banking systems to AI model training. Tech giants like Amazon operate over 100 data centres worldwide, each hosting nearly 50,000 servers that run 24/7.

How Much Electricity Do Data Centres Really Use?

A single 100 MW data centre run by a major tech firm can consume as much electricity as 100,000 households. Advanced “hyperscale” data centres designed for heavy AI workloads can require power equivalent to 2 million homes.

Also Read | Toyota’s Woven City: The Smart City of the Future

According to the International Energy Agency (IEA), total global electricity demand from data centres could more than double by 2030—from 415 TWh to 945 TWh, surpassing Japan’s annual consumption. Energy bills already make up nearly 65% of a data centre’s operational cost, primarily for computing and cooling systems.

Currently, data centres account for 1.5% of global electricity use, a figure expected to surge to 10% by 2030. Over the last five years, energy demand from data centres has grown by about 12% annually, driven largely by the AI boom and the race to build bigger, denser, and faster facilities.

Generative AI workloads alone consume about eight times more power than conventional computing tasks. To make matters worse, these AI models often have short lifespans — older versions are phased out quickly, rendering much of the initial energy investment obsolete. Each newer generation of models typically demands even greater computational power and, consequently, more energy.

In India, the generative AI market is projected to expand at a 28% CAGR from 2023 to 2030. As adoption accelerates, so will the country’s appetite for data centres — and electricity.

Regional Power Trends: Where the Energy Is Going

Globally, the United States leads data centre power consumption with a 45% share, followed by China (25%) and Europe (15%). As of September 2025, there are roughly 10,730 data centres worldwide.

By country, the top 10 are:

U.S. (4,049)

U.K. (485)

Germany (482)

China (379)

France (316)

Canada (284)

India (271)

Australia (265)

Japan (205)

Italy (204)

Despite having just 2.5% of global data centres, India generates nearly 20% of the world’s data, much of which is still processed overseas — revealing a glaring infrastructure deficit.

Most Indian data centres are concentrated in Mumbai, thanks to its proximity to undersea cables and global network access. India’s total capacity currently stands at 1,263 MW, compared to 15,930 MW in the U.S. and 3,800 MW in China. But the sector is expanding rapidly — capacity has tripled between 2019 and 2024. If growth continues, India could reach 4,500 MW by 2030, consuming about 3 percent of national electricity, up from less than 1 percent today.

Government Push and Policy Support

Recognising the importance of data infrastructure, the Indian government has opened the sector to 100% foreign direct investment (FDI), attracting global technology players. Investments in Indian data centres have already neared $60 billion over the last five years and are expected to exceed $100 billion by 2027, according to CBRE.

Recent policy initiatives — such as granting infrastructure status to data centres, implementing RBI’s data localisation guidelines, and enacting the Digital Personal Data Protection Act (2023) — aim to strengthen domestic data processing.

Also Read | Chennai Unveils ₹2.5 Lakh Crore Mobility Plan with Single-Ticket App

At the state level, incentives like tax breaks, reduced power tariffs, subsidised land, and fast-track approvals are fuelling regional hubs. Noida has emerged as a top choice under Uttar Pradesh’s progressive data policies, while Hyderabad continues to attract investment due to supportive governance and reliable electricity supply.

Challenges: Power, Emissions, and Cooling

Despite the rapid momentum, India’s data centre expansion faces several major hurdles.

Ageing Power Grids

Continuous, high-capacity electricity supply is essential for data centres, yet India’s transmission infrastructure remains outdated in many regions. Upgrading these systems — or building new lines — can take four to eight years, potentially delaying projects.

Emissions and Energy Sources

Most Indian data centres still rely on coal-based power, significantly adding to carbon emissions. Training a single large AI model can release as much CO₂ as five gas-powered cars produce in their lifetimes — and consume water equivalent to a small nation’s yearly use. The IEA warns that CO₂ emissions from data centres could climb from 180 million tonnes to 300 million tonnes by 2035. Transitioning to renewable energy and battery storage is critical, and falling battery costs may make this shift more achievable.

Cooling Efficiency

AI servers generate enormous heat. Currently, India’s nearly 1 GW of operational data centres primarily use air cooling, which accounts for 40% of their total energy use. While liquid cooling can cut energy needs by up to 90%, adoption remains limited — only about 10 MW of capacity uses it — due to high setup and maintenance costs.

Also Read | BSNL Plans 5G Rollout in Key Metro Cities in 2025

Outlook: India’s Energy Imperative in the AI Era

India’s data centre industry is still young, valued at about $8 billion today. But by 2033, it’s projected to hit $24.78 billion, growing at a 13.37% CAGR. Major global players — including Adani, Amazon Web Services, Equinix, and STT GDC — are betting on India’s potential as the next data infrastructure hub.

As India moves closer to becoming a global AI powerhouse, the real challenge lies in powering that growth sustainably. The future of AI innovation will depend not only on advanced chips and models but also on how efficiently and cleanly the world can keep the lights on.