The rapid growth of artificial intelligence (AI) is reshaping industries, but it is also placing increasing pressure on global resources, particularly water and energy, as the world marks Earth Day.

A growing body of research shows that the infrastructure powering AI, from data centers to chip manufacturing, requires vast amounts of electricity and water, raising concerns about long-term sustainability.

As someone with an engineering background, innovation is always the goal. Progress is expected. But there is also a fundamental principle: systems must balance. In the simplest sense, like an equation, inputs and outputs must eventually equate. If they do not, something breaks.

According to the World Economic Forum, citing research from Global Water Intelligence and Xylem, the global AI economy already consumes around 23 cubic kilometers of water annually, with demand projected to more than double to over 54 cubic kilometers by 2050. That would require an additional 31 cubic kilometers of water each year, equivalent to thousands of liters of freshwater for every person on Earth.

The demand is driven largely by data centers, which use water to cool high-performance computing systems. Even with efficient cooling systems, a single hyperscale data center can consume up to 171 million liters of water annually.

Water consumption is also tied to chip manufacturing and power generation, forming what experts describe as a broader “AI economy” that is expanding rapidly.

“Water stress is already a global challenge, and demand from the growing AI economy is adding to the pressure,” the World Economic Forum said.

The issue is compounded by existing water scarcity. The World Bank estimates that four billion people already live in water-stressed areas, and by 2030, global water demand could exceed sustainable supply by up to 40 percent. As many as 1.6 billion people may lack access to safe drinking water.

At the same time, the energy demands of AI are also rising sharply.

Data centers are projected to consume up to 3 to 4 percent of global electricity by 2030, according to the European Commission. The rapid expansion of AI infrastructure is forcing policymakers to balance technological growth with environmental and societal impacts.

“[AI growth] is expected to create major challenges for decision-makers balancing technological competitiveness with environmental resilience,” the Commission said.

The scale of demand has raised questions about how sustainable the AI boom can be if current trends continue.

Research from the Massachusetts Institute of Technology shows that generative AI systems require significantly more computing power than traditional workloads, leading to higher electricity consumption and carbon emissions. Training and deploying large AI models can consume vast amounts of energy, while everyday use adds further strain.

Cooling systems for these operations also rely heavily on water.

“It is not just the electricity you consume, there are much broader consequences that persist based on actions we take,” said Elsa Olivetti, a professor at MIT.

Environmental groups have also raised concerns about the broader impact of AI expansion.

Greenpeace said the rapid buildout of AI infrastructure is increasing demand for energy, water, and raw materials, while raising questions about who benefits from the technology and who bears the environmental costs.

“The AI boom is being sold as inevitable progress… but the question is who pays the environmental bill,” the group said.

Communities in several countries have begun pushing back against data center projects, citing concerns over water use, energy demand, and local environmental impact.

Despite these concerns, researchers and policymakers are exploring ways to make AI infrastructure more sustainable.

One approach involves rethinking data centers not just as energy consumers, but as part of a circular system. New studies suggest that waste heat generated by data centers could be used for applications such as carbon capture and water purification.

This could allow data centers to become “carbon-negative” or “water-positive,” offsetting some of their environmental impact.

Other proposals include improving water efficiency through advanced cooling systems, increasing the use of renewable energy, and developing closed-loop water systems that recycle and reuse water.

The World Economic Forum has also outlined strategies to “waterproof” the AI economy, including reducing water loss in infrastructure, expanding water recycling, and building partnerships between industries and communities.

“Leaders who integrate water security into the AI value chain will shape the next century of growth,” the forum said.

The technology promises productivity gains and economic growth, but its environmental footprint is becoming harder to ignore.

On Earth Day, that balance is increasingly seen as critical.

Because in the end, no matter how advanced the system is, the equation still has to work.

If left unchecked, the same systems powering the digital economy could strain the very resources needed to sustain it.