Does China Have Enough Water For Its Thirsty AI Data Centers?

Authored by James Gorrie via The Epoch Times,According to China's Ministry of Water Resources, China has nearly one-fifth of the world's population but only about 6 percent of global freshwater resources.That presents planners in Beijing with a challenging problem: How to support 20 percent of humanity, more than 18 percent of global economic output, and an expanding digital economy, with a disproportionately small water resource?Things become even more complicated because China's most economically developed regions are not necessarily where water is most abundant.Now, Beijing wants to build substantially more artificial intelligence (AI) computing capacity. That raises another fundamental question: Can China expand its AI infrastructure without worsening its existing water crisis?China Is Betting Big on AI InfrastructureOver the next five years, China is preparing to expand its nationwide AI data center infrastructure by roughly 2 trillion yuan, or $295 billion. The plan would create interconnected computing hubs and rely heavily on domestic technology suppliers.Rystad Energy estimates China's installed data-center capacity reached about 32 gigawatts at the end of 2025 and could rise above 60 gigawatts by 2030. Data-center electricity consumption could reach 289 terawatt-hours by 2030, more than twice 2025 levels. AI and high-performance computing are expected to drive much of that growth.But electricity is only half the infrastructure equation. Since AI servers generate enormous amounts of heat, data centers must rely on cooling technologies that require substantial amounts of water.What's more, China's existing data centers are already using a lot of it.China's Data Centers Already Have a Water ProblemA 2025 study published in Applied Energy provides perhaps the clearest warning.Researchers estimated that China's data centers consumed approximately 15.7 billion cubic meters of water in 2022, equivalent to about 2.7 percent of China's total water consumption. Even more significant, the researchers found that 72 percent of China's computing capacity was located in regions experiencing severe water scarcity.The Chinese regime has been trying to move computing capacity away from the crowded eastern regions toward western provinces where electricity, particularly renewable electricity, is more readily available.The strategy makes sense from an energy perspective, but not so much from a water supply perspective. The same study found that China's "Eastern Data and Western Computing" strategy could increase water-footprint pressures in water-scarce western regions such as Gansu and Ningxia.The imbalance between China's water supply and rising cooling needs is obvious. Beijing is solving one bottleneck while creating another.Moving Data West Doesn't Move the WaterThis is the central problem with China's strategy. Western provinces have vast solar and wind resources and enormous areas available for data-center construction. But parts of the west are also among China's driest regions.That creates a fundamental trade-off because China cannot simply allocate more water to AI without hurting other key areas. Water used to cool an AI data center competes, directly or indirectly, with agricultural, industrial, municipal, and ecological needs.Agriculture accounted for about 62 percent of China's total water use in 2025, compared with 16 percent for industry and 15.8 percent for domestic consumption. Total national water use reached approximately 594.45 billion cubic meters. Agriculture's dominance matters because food security is a strategic priority for Beijing.On the flip side, China is improving efficiency. Water consumption per 10,000 yuan of GDP fell 4.5 percent in 2025, while water consumption per 10,000 yuan of industrial value-added declined 7.2 percent.Those gains are significant, but efficiency cannot eliminate scarcity.If AI computing grows rapidly enough, a data center can become more water-efficient even as the overall sector consumes more water.That's the danger.China Is Trying to Make AI GreenerBeijing has set a goal to source 80 percent of its data-center electricity from green energy by 2030, up from 11 percent in 2023. But analysts say AI's highly variable computing loads make it difficult to integrate data centers smoothly with renewable power systems.The regime is also encouraging higher efficiency in data centers by improving power and water usage effectiveness. That technology can make a major difference.The Applied Energy study found that simultaneous improvements in power efficiency, water efficiency, and electricity-grid decarbonization could reduce China's national data-center water footprint by 51.5 percent to 58.3 percent by 2050, compared with a business-as-usual scenario.The bad news is that these are scenarios, not guarantees.The AI Boom Could Outrun the Efficiency GainsChina's data-center industry is expanding as AI demand grows. The same technological advances that make AI more useful also increase demand for computing power.Rystad Energy expects Chinese data-center electricity consumption to grow at roughly 19 percent annually between 2025 and 2030. But those efficiency improvements may not offset the water consequences of rapidly increasing computational capacity.The result could be a familiar economic pattern of more efficient machines on the one hand, but more of those machines on the other.This is where China's water dilemma becomes more serious. It can reduce the amount of water required per unit of computing while still increasing total water consumption.What Happens If Water Becomes the Bottleneck?If water becomes scarce around major AI hubs, China could face pressure to restrict where new data centers are built. That could increase infrastructure costs and force companies to move computing capacity toward regions with better water availability.It could also increase competition among agriculture, cities, and manufacturing and technology companies. That matters because China's economy is already heavily dependent on water-intensive industries and food production.China can build reservoirs, pipelines, water-transfer systems, recycling facilities, and increasingly efficient industrial infrastructure. It can force industries to meet state-directed efficiency targets. But it can't manufacture freshwater at the scale necessary to meet the underlying demands of AI, industry, agriculture, and public use.Water is ultimately constrained by geography and climate and could become one of the least visible but most important constraints on China's AI expansion through the 2030s.Views expressed in this article are opinions of the author and do not necessarily reflect the views of The Epoch Times or ZeroHedge.