For decades, we have scrolled for a quick search without thinking about the back-end mechanisms or processes involved. However, the rise of Artificial Intelligence (AI) has paved the way for a fundamental reasoning: Innovation has a price, and India’s energy ecosystem must be ready for the upcoming disruption. The instant answers and creative outputs we receive from AI are not magic — they are the result of vast amounts of energy being consumed, processed, and transformed.

The old search engine economy was frugal. A single web search consumed a mere 0.3 watt-hours of power, a tiny, almost negligible amount. The new AI economy is interactive and doesn’t just retrieve information; it reasons, drafts, and assists, making it a far more power-intensive process. A single AI prompt, on conservative estimates, consumes about 1.7 watt-hours — a six-fold increase over a standard search. This shift isn’t just about the per-prompt cost; it’s about volume. Unlike a simple search, AI is sticky. People don’t ask one question and stop; they engage in extended conversations, asking 10, 50, or even 100 prompts a day. This behavioural change is transforming the digital economy into a new kind of energy economy.

If we take a look at the numbers on a global scale, with over 5.5 billion people connected to the internet, we currently perform an average of 2.5 searches a person daily, totalling 13.7 billion searches. This rhythm defined the search era, but the agent era will multiply these interactions.

The potential energy consumption is staggering. If the world’s 5.5 billion internet users each ask just 10 prompts a day, AI would consume 34.5 terawatt-hours (TWh) of electricity annually — about 0.1% of today’s global demand. At 50 prompts a person, that number skyrockets to 172.5 TWh. And if we reach 100 prompts a day, AI would demand 345 TWh annually, a striking 1.2% of current global power usage.

For a nation like India, with its over 900 million internet users, the picture is equally compelling. At 10 prompts a person daily, AI would consume 5.6 TWh a year, about 0.3% of India’s current generation. If that usage climbs to 100 prompts a day, it would rise to 55.9 TWh, roughly 3% of national power. While manageable, this growth underscores the need for strategic planning.

This is not a crisis but a choice. As AI becomes our default interface, several critical changes are unfolding. One, the sheer volume of tasks is exploding; a single search becomes a series of tasks, from comparing flights to drafting emails, dramatically increasing total energy consumption. Two, the load is concentrating; AI centralises demand in massive data campuses, each consuming hundreds of megawatts and running 24/7, creating a concentrated new load on the grid. Three, data centres are projected to double; by 2030, global data centre electricity use could nearly triple, with AI as the primary driver, potentially accounting for over a third of that demand in a 100-prompt world.

This moment is an opportunity for leadership, not a cause for fear. We must act with a new social contract in mind, which balances technological ambition with environmental responsibility. We must build for abundance, investing in clean, firm, 24/7 power and co-siting new data centres with renewable energy sources. This strategy turns AI’s energy hunger into a driver for a clean energy boom. We must also crush energy per prompt by using smaller, more efficient models for routine tasks and reserving larger ones for complex work, making efficiency gains the new standard. Another key step is to align with the grid, scheduling heavy inference tasks when clean power is abundant to reduce strain. We must localise benefits, ensuring that new data campuses anchor local benefits like transmission upgrades and domestic manufacturing. Finally, we need to tell the truth through radical transparency about the carbon footprint, water use, and energy intensity of AI, as trust is the foundation of a sustainable AI-first world.

The shift from search to agents is a transformative one. While today’s search economy is a rounding error on the grid, the AI economy, even at peak efficiency, will be significantly larger. This is not a reason to fear, but a reason to act. We must invest, optimise, and accelerate the clean energy transition, ensuring that AI becomes not just an energy load but the engine of green jobs and digital leadership for all.

Vineet Mittal is chairman, Avaada Group. The views expressed are personal