Did you know AI’s Data Growth Drains Water Supplies, as 70% More Centers Emerge

 

Every time an AI chatbot generates a summary or designs an image, it sends a request to data centers, consuming precious water resources. These centers, essential for powering AI systems, require substantial amounts of water for cooling purposes and, indirectly, for the electricity that fuels them.

As major tech companies push for more powerful AI systems, the race to build larger data centers intensifies. With this expansion, the demand for water has reached new levels, adding to the strain faced by regions already grappling with water scarcity. The issue is not new, but as AI technology surges, the problem worsens.

Recent data from Bloomberg highlights a disturbing trend. Over the past three years, more than 160 new AI data centers have been established across the United States. These facilities are increasingly concentrated in areas already struggling with limited water resources, marking a 70% rise from the previous three-year period.

A similar pattern is visible worldwide. Arid countries like Saudi Arabia and the UAE are welcoming AI-related data centers, while China and India, facing severe water scarcity, have seen a greater share of new centers built in dry regions. This trend underscores the growing need for more data infrastructure and the environmental costs tied to it.

In efforts to support cutting-edge AI, many companies have targeted states or nations with abundant energy but often lack sufficient water supplies. This mismatch threatens local water sources, agriculture, and energy production, compounding the impact of climate change.

 

Experts, including Newsha Ajami from Lawrence Berkeley National Laboratory, warn that this issue is rapidly escalating. Water-related protests have already taken place in countries like the Netherlands, Uruguay, and Chile, where locals have voiced concerns about the toll that massive data centers are taking on their limited water supplies.

In the US, tech giants have chosen states like Texas and Arizona for their new data hubs, further exacerbating water concerns. These drought-prone regions are becoming more vulnerable to water scarcity as data centers continue to proliferate.

Despite these concerns, companies like Microsoft are reevaluating their growth plans. However, according to market intelligence firm DC Byte, more than 50 new data centers are expected to emerge in dry regions across the US by 2028, underscoring the relentless expansion.

AI's insatiable thirst for power directly correlates with its water consumption. According to a 2023 report by the International Energy Agency, a 100-megawatt data center consumes as much water as 6,500 homes, an alarming rate when considering the global push to scale up AI operations. By 2030, global data centers could consume up to 1,200 billion liters of water annually.

Cooling systems, which typically rely on evaporative methods, consume massive quantities of water. Around 80% of the water used by data centers evaporates, a stark contrast to residential water use, where only 10% evaporates. As water is lost through evaporation, much of it cannot be reused, further draining local supplies.

The environmental impact extends beyond cooling systems. A significant portion of water used by data centers comes from power plants located in water-scarce regions. The IEA estimates that indirect water consumption accounts for nearly 60% of the total water usage by these centers.

The growing demand for water presents a significant challenge for tech firms. Many companies, such as Amazon and Google, have pledged to achieve “water positive” statuses by 2030, meaning they intend to replenish more water than they consume. However, the regions best suited for renewable energy, like solar power, often face severe water shortages, presenting a difficult balancing act between power and water usage.

As AI data centers proliferate in water-stressed areas, some companies are experimenting with innovative cooling methods. Microsoft, for example, has developed a system that recycles water continuously within closed loops, reducing evaporation losses. Other firms are exploring immersion cooling, which involves submerging chips in liquids to avoid excessive water use.

Despite these efforts, evaporation-based cooling remains the norm. Reports suggest that by 2028, a vast majority of AI data centers will still rely on this water-intensive method. Companies like Amazon are working to minimize the impact of this method by using treated sewage water for cooling, aiming to lessen the demand on drinking water supplies.

Tech firms are also engaging in local initiatives, partnering with communities and nonprofits to restore and replenish water supplies. However, water advocates emphasize the need for greater transparency regarding data center water consumption, particularly at the individual facility level. Information on water usage is sparse, and companies often guard these details as trade secrets.

As global demand for AI infrastructure continues to rise, balancing technological progress with environmental responsibility has become more crucial than ever. The growing tension between AI’s water needs and the availability of clean water will require urgent attention from both tech companies and governments worldwide.