The Coming Power Hunger of Artificial Intelligence

By Sanjeev Oak

Artificial Intelligence is reshaping the digital economy, but at a steep cost: electricity. A Goldman Sachs report warns that data centres could drive a 165% surge in global power demand by 2030, challenging India’s energy security and climate ambitions.

Artificial Intelligence is not just rewriting algorithms—it is rewriting electricity bills. The Goldman Sachs report, How AI Is Transforming Data Centers and Ramping Up Power Demand, places a stark number on the future: global data centre power demand is set to rise by 165 per cent by 2030. What looks like the silent hum of servers today could soon thunder across national grids.

The Scale of the Surge

The report estimates that data centre electricity demand will grow from around 35 GW in 2023 to nearly 92 GW by 2027. By 2030, the figure could cross 165 GW. This is not just growth—it is an infrastructural upheaval.

“Power demand from data centres is forecast to rise by 165 per cent by 2030.”

Occupancy levels in U.S. markets are already near saturation, set to peak beyond 95 per cent by 2026. Behind the buzzwords of machine learning and large language models lies a simple truth: AI is a power-hungry enterprise.

India’s Parallel Challenge

India is not insulated from this wave. In fact, it is at the crest of it. According to government data, India’s data centre capacity stood at 850 MW in 2023, with projections suggesting a leap to 1,500 MW by 2026. The Ministry of Electronics and IT has identified Hyderabad, Pune, Chennai, and Noida as emerging hubs.

Yet, the question is not just about square footage of server racks—it is about the watts that feed them. India’s annual electricity demand grew by 8 per cent in 2023–24, already straining grids during summer peaks. The addition of AI-driven data centres will further push demand, requiring careful integration with renewable capacity.

“India’s AI ecosystem cannot grow in isolation—it must grow in tandem with its power infrastructure.”

The Infrastructure Gap

Goldman Sachs estimates that $720 billion in grid spending will be needed globally by 2030 to meet this demand. For India, the figure translates into billions in transmission upgrades, land for sub-stations, and faster renewable deployment.

India has set a target of 500 GW renewable capacity by 2030, but integration remains a bottleneck. Transmission corridors lag years behind commitments. This is where AI’s electricity hunger intersects uncomfortably with climate ambition.

If AI data centres are fed by coal-based power, India’s climate narrative could falter. If they are paired with solar, wind, and emerging green hydrogen, they could accelerate the energy transition.

Policy Crossroads

The AI moment forces Indian policymakers to think holistically:

1. Energy–Tech Synergy: AI growth must be mapped onto renewable expansion. Data centre approvals should be tied to clean power purchase agreements.

2. Grid Modernisation: Smart grids, storage, and flexible transmission must be prioritised. Without them, renewable capacity will remain stranded.

3. Cooling and Water Use: Beyond electricity, AI data centres consume massive water volumes for cooling. In water-stressed regions like Maharashtra, policy must regulate sustainable practices.

4. Regional Equity: If most centres cluster around southern and western states, grid pressure could deepen inequalities. Balanced regional distribution is essential.

“The digital revolution cannot be decoupled from the physical one—defined by transformers, corridors and cooling towers.”

Global Context, Local Stakes

The International Energy Agency warns that global data-centre electricity consumption could double by 2030, driven largely by AI. For India, the stakes are two-fold: becoming a global AI hub while ensuring energy security.

The irony is sharp. AI promises efficiency, optimisation, and smarter grids. Yet its own existence could choke the very grids it is supposed to optimise—unless planning gets ahead of the curve.

The Road Ahead

India’s AI policy discourse must move beyond sandboxes and ethics boards into kilowatts and gigawatts. Without power, AI will remain a paper tiger. With clean power, it can become the engine of a green digital economy.

Goldman Sachs’ report is less a forecast than a warning: prepare your grids, or watch AI stall. For India, the choice is clear—integrate, invest, and innovate, or risk watching the data revolution dim under the weight of power cuts.

Companion Infographic Concept

Title: AI’s Growing Appetite for Power

Visual Elements:

1. Bar Chart: Global data-centre power demand – 2023 (35 GW), 2027 (92 GW), 2030 (165 GW).

2. India Focus: Current data-centre capacity (850 MW, 2023) vs projected (1,500 MW, 2026).

3. Pie Chart: Share of AI in data-centre electricity use – 14% (2023) → 27% (2027).

4. Iconography:

   • Server racks glowing → power plugs → electricity grids.

   • Renewable icons (solar panel, windmill) beside AI servers, showing potential clean integration.

5. Pull-Quote Box: “India’s AI ecosystem cannot grow in isolation—it must grow in tandem with its power infrastructure.”