AI against water scarcity: More than a drop in the ocean

"AI – alright for the climate?" we recently asked with regard to the impact of artificial intelligence (AI) on climate change. Upon closer inspection, it becomes clear that the widespread use of AI could make energy supply more efficient and reduce CO₂ emissions by hundreds of megatons. An exciting forecast for investors interested in sustainable investment themes such as climate protection and decarbonization.

AI Data Centers Are Thirsty for Water

But this is only one facet of the tension between AI and sustainable development. The future technology is not only considered an energy guzzler but also extraordinarily thirsty.

A typical 100-megawatt data center, the foundation of AI, consumes about 1.1 million gallons (4.164 million liters) of water daily to cool the processors. This is according to a study from the USA published by the Lawrence Berkeley National Laboratory in 2016. Assuming an average US household consumes about 300 gallons (approximately 1,136 liters) of water, the daily water consumption of a data center corresponds to that of 3,667 US households (see graphic below). Given the rapid development of technology, consumption in the States could grow tenfold in the future.

Water Consumption of AI Data Centers in the USA (in Gallons)

Data centers are often located in regions that suffer from partial water scarcity. Therefore, efforts to increase the water efficiency of data centers are significant. This includes reusing water in a closed loop or optimizing cooling processes. However, the real potential for more efficient use of the precious resource water lies in what the data centers produce: AI.

The field for AI applications in the fight against water scarcity is surprisingly broad. Although many solutions are still in their early stages, we already identify four sub-areas where the use of AI could make a difference. Investors interested in the sustainable investment theme of water and solutions against water scarcity can position themselves accordingly early on.

1. Physical AI includes AI systems that interact with the physical world, often through robotics and automation. In water supply, for example, AI-controlled valves and pumps can dynamically adjust water flow and pressure in real-time. This optimizes distribution, minimizes water losses, and improves energy efficiency in pumping stations. Furthermore, AI-supported systems can autonomously control and optimize various processes in water treatment plants, such as the dosing of chemicals for coagulation and flocculation, aeration in biological treatment, and filtration rates. The goal is to ensure optimal water quality and reduce chemical use.
   
   Company Example: The American water treatment specialist Xylem offers pump systems in its "Smart Pumping Solutions" division that integrate motors, hydraulics, and control systems with embedded intelligence. These systems can dynamically adjust the speed and flow of water to real-time conditions, optimize energy consumption, and prevent blockages. 
   
   
2. Predictive AI uses data to predict future events, patterns, and behaviors. Historical water usage data, weather patterns, population growth, and economic indicators can be analyzed with AI to predict future water demand – but also disasters such as droughts or floods. This allows utilities to optimize water allocation and manage reservoir levels more effectively. Farmers, in turn, can apply water and fertilizers when truly needed, thanks to predictive AI. Other applications include predictive maintenance of pumps, motors, filters, and water quality forecasting.
   
   Company Example: The US-based industrial cleaning company Ecolab uses AI with its Water Quality IQ system. This system analyzes operational and historical data to predict changes in water quality in industrial processes. Real-time alerts can also optimize chemical use.
   
   
3. "Seeing" AI: Computer vision enables AI systems to recognize and interpret visual data. Examples include cameras mounted on remotely operated vehicles, drones, or autonomous underwater vehicles. Using computer vision, they can inspect hard-to-reach pipelines, reservoirs, and other structures for leaks, cracks, corrosion, and structural damage – often in real-time. Computer vision can also be used for visual analysis of water quality or to detect contamination with waste.
   
   Company Example: Xylem again offers various inspection tools for pipes and sewers. Although not exclusively AI-supported in every case, computer vision is increasingly used for the automatic detection and classification of defects based on visual data.
   
   
4. Generative AI focuses on creating new data, designs, or solutions. Generative algorithms can design novel and optimized water distribution networks, considering factors such as population density, topography, existing infrastructure, costs, pressure requirements, and resilience to disruptions. Generative AI can also be used to modernize existing infrastructure by helping plan the integration of new technologies.
   
   Company Example: Engineering and consulting firms like Tetra Tech are likely to integrate generative AI early into their design, planning, and simulation services for water infrastructure projects.

Spreading the Opportunities and Risks of AI for the Investment Theme of Water

As mentioned earlier, some of these applications are in their early stages. From an investment perspective, it makes sense to spread the risks in this area to cushion potential setbacks in the portfolio. The sustainable thematic fund "Swisscanto (LU) Equity Fund Sustainable Water" implements this by combining investments in the sub-themes of water efficiency, water infrastructure, and water protection. The goal is always to benefit from the long-term potential of the investment theme of water and applications against water scarcity. AI could prove to be an important driver.

Three Key Points to Remember About AI in the Fight Against Water Scarcity:

Investment theme «Water»: Insights