Generative AI is everywhere. In just a few years, it has gone from a niche technology to a household name, with ChatGPT now reaching ~800 million weekly active users (up from 300 million in late 2024) and handling 2.5 billion prompts per day. The promise is undeniable: smarter workflows, new creativity, and breakthroughs across industries. But here’s the part most headlines don’t tell you: this revolution comes with a steep sustainability bill. Behind every slick AI-generated text, image, or code snippet are massive data centers running 24/7 drawing power, consuming water, and emitting carbon at a scale that rivals entire industries. This is the hidden cost of the AI boom.

Why Energy Matters

Training modern AI models is no small task. It requires thousands of high-performance chips running for weeks in massive warehouses of servers. Every new generation of models grows in size and complexity and so does the electricity required to train them. The hidden cost is that it doesn’t stop at training. Once a model is built, every use consumes energy too.

A single AI query uses about five times more electricity than a standard web search. That might sound small, but with billions of queries daily, the numbers add up fast. In 2024, global data centers consumed about 415 terawatt-hours (TWh) of electricity nearly 1.5% of all global demand. By 2030, this figure could surge to 945 TWh, almost equal to Japan’s entire annual use.

The problem is made worse by user expectations of speed. Nobody wants to wait for an AI that “thinks.” To deliver instant answers, companies keep servers running at full throttle, often with extra capacity on standby for usage spikes. The illusion of an “infinite assistant in the cloud” hides the reality of a massive energy drain on the ground.

AI’s Hidden Thirst

Data centers don’t just eat electricity they drink water. Cooling thousands of servers is an enormous challenge. Many rely on water-based cooling systems, which consume large volumes of freshwater to prevent overheating.

Research shows that training and running AI models will require 4.2–6.6 billion cubic meters of water annually by 2027 which is more than half the total yearly water withdrawal of the UK. A single 100-megawatt data center can use up to 2 million liters of water a day for cooling, enough to supply more than 6,000 households.

To put it in perspective: a single extended AI chat session of 20–50 prompts is estimated to consume about half a liter of water in cooling and electricity generation. That’s the equivalent of a bottle of drinking water for each digital conversation.

This raises critical questions. Should water-stressed communities have to compete with AI for the same limited supply? And is the trade-off worth it when alternatives like liquid cooling, wastewater reuse, and AI-optimized cooling systems already exist?

💡 “Google’s water use in its data centers grew from 6.1B gallons in 2023 to 8.1B gallons in 2024 - a ~33% rise, equivalent to going from irrigating 41 to 54 golf courses annually.”

Carbon Costs on the Rise

Electricity and water are visible resource costs, but the climate toll is just as concerning. In the U.S., data centers already consume around 4.4% of national electricity (~176 TWh) and produce over 105 million metric tons of CO₂e annually. Unless grids decarbonize quickly, this share could rise to 9–12% of U.S. electricity by 2030.

That makes the carbon footprint of data centers comparable to the domestic aviation sector ; an industry long under scrutiny for its emissions. And this trend is accelerating. Electricity demand from AI is expected to account for up to 50% of future growth in power use across advanced economies this decade.

Unless electricity grids decarbonize quickly, the climate burden from generative AI could undo years of progress in reducing tech-sector emissions.

Expansion Without Limits

The AI boom is outpacing sustainability planning.

Global electricity demand from data centers is growing at an average 15% per year.
Shipments of GPUs for example the chips behind AI are now in the millions annually, fueling demand for ever-larger models.
By 2030, U.S. data centers alone could consume 580 TWh annually as much as 12% of U.S. electricity use.

And power is just part of the story. Each new hyperscale facility requires millions of square feet of construction, enormous amounts of steel and concrete, rare metals for chips, and backup diesel generators. The full lifecycle footprint stretches from mining to construction to operation.

Without careful planning, this infrastructure could lock in decades of high emissions and water use.

The Transparency Gap

One of the biggest challenges is how little we know about AI’s true footprint. Companies rarely disclose detailed figures for AI-specific energy, carbon, or water use. Instead, they report broad corporate-level data, often averaged across global operations and balanced by offsets. What little we know often comes from independent research, utilities, or investigative reporting.

This lack of transparency makes it nearly impossible for businesses, investors, or policymakers to evaluate the real environmental cost of AI services. Imagine trying to run your company’s sustainability strategy without knowing how much carbon your suppliers emit that’s the situation with AI today.

Some change is coming. The EU AI Act will require large AI models to disclose environmental impact, and international standards bodies are developing sustainability metrics. But until these frameworks are globally adopted, opacity remains the norm.

Investors and ESG: A Growing Flashpoint

For years, Big Tech was considered ESG-friendly. Compared to oil, steel, or aviation, its footprint seemed relatively low. But the AI boom is forcing investors to rethink.

Key concerns:

Can companies keep their climate pledges while AI demand accelerates?
Will local pushback in drought-stricken regions slow expansion?
Could governments impose new rules, taxes, or carbon costs on AI infrastructure?

The risks aren’t theoretical. Several European countries have already paused new data centers in places facing grid strain. U.S. regulators are studying whether power demand from AI could overwhelm regional utilities.

For ESG funds, the takeaway is unmistakable: AI’s sustainability cost is now a material risk factor. But there’s also opportunity. Companies that lead on “green AI” developing more efficient models, investing in renewable-powered data centers, and pioneering advanced cooling will differentiate themselves. Investors are watching closely for these signals.

So, What Next?

Generative AI isn’t going away and it shouldn’t. Its potential for productivity, creativity, and even climate science is enormous. But responsibility has to catch up with innovation.

That means:

Efficiency by design: Building models that use fewer resources.
Renewable power: Supplying data centers with clean energy, not offsets.
Smarter cooling: Liquid cooling, wastewater reuse, and waste-heat recovery.
Transparency: Publishing clear metrics on AI’s energy, water, and carbon costs.
Investor engagement: Treating sustainability as part of AI strategy, not a PR exercise.

What’s at Stake

The story of AI isn’t just about algorithms and breakthroughs. It’s about concrete, steel, electricity, water, and carbon. It’s about communities rationing water while nearby servers train new models. It’s about grids buckling under new power loads. It’s about whether the tech sector can meet its climate pledges while scaling one of the most resource-hungry technologies in history.

AI can either become a strain on the planet or a driver of new efficiencies. Which path we take depends on the choices made now, while the industry is still young. The sustainability cost of AI is no longer invisible and addressing it is now a responsibility every business and investor must share.

Explore ESG Solutions on our marketplace - OneStop ESG Marketplace.

Keep abreast of the top ESG Events on OneStop ESG Events.

OneStop ESG Educate: Your go-to source for top ESG courses and training programs tailored to your needs.

Stay informed with the latest insights on OneStop ESG News.

Discover meaningful career opportunities on OneStop ESG Jobs.