A massive artificial intelligence infrastructure project planned for Batang, Central Java, is drawing scrutiny over its potential water consumption, as the region grapples with recurring droughts and land subsidence. The Zankore AI Factory, developed by Indosat Ooredoo Hutchison with Ooredoo Group, Nokia, and NVIDIA, is slated to begin construction in the first half of 2027. Its first phase, with a capacity of about 200 megawatts, is part of a broader 1-gigawatt development promoted as a step toward Indonesia's ambition to become a regional AI hub.
But advanced AI infrastructure requires not only enormous amounts of electricity. Its servers also need large volumes of water for cooling, and the project's water demand could be staggering. According to a 2025 study by University of California researchers, Indonesia has the highest projected cooling-water consumption among the countries examined, at about 1.9 liters for every kilowatt-hour of electricity used by servers. Based on those projections, Zankore's 200-MW first phase could evaporate between 4.8 million and 9.1 million liters of water per day, assuming continuous operation. At that rate, the facility could consume an amount equivalent to the 75.46 million liters of clean water prepared by the Central Java government during the 2026 drought crisis in just eight to 16 days. At full 1-GW capacity, the same volume could be consumed in roughly two to three days.
The urgency is heightened by recent climate shocks. Central Java declared a drought alert between June 5 and September 28, 2026, affecting 2.34 million people in 784 villages, including areas of Batang. The provincial government subsequently prepared 75.46 million liters of clean water for residents. Meanwhile, the northern coast of Central Java is already struggling with land subsidence. Indonesia's Geological Agency and researchers at the Bandung Institute of Technology have recorded subsidence of around 6 to 20 centimeters per year in the Pekalongan-Batang area. As the ground sinks, soil becomes less able to store water, while seawater can move farther inland and worsen tidal flooding. These changes are difficult to reverse and could eventually force communities to relocate.
Zankore represents an investment of about US$4.59 billion, or Rp82 trillion, more than the five-year investment target of US$4.23 billion for the Batang Special Economic Zone. The government has also invested Rp409.7 billion from the national budget to develop raw-water infrastructure for the Batang industrial zone, including Zankore's. That system has a capacity of 285 liters per second, or about 24.5 million liters per day. If Zankore relies on this source, its 200-MW first phase could consume more than 30% of the infrastructure's daily capacity. At 1 GW, its estimated demand could exceed the entire capacity.
The economic benefits deserve closer scrutiny. Data centers are highly capital-intensive but generate relatively few jobs. A 2026 study based on thousands of US data centers found that such facilities typically employ fewer than 100 permanent workers. The question is therefore not simply how much investment enters Batang, but how much value stays there, and who bears the environmental costs. Zankore is 49% owned by Ooredoo, 37% by Indosat, 11% by NVIDIA, and 3% by Nokia. This means 63% of its shares are held by companies from Qatar, the United States, and Finland. Much of its equipment and AI chips will also be imported. In economic terms, this creates a risk of a negative externality: profits accrue to shareholders, while part of the environmental cost, particularly water consumption, could fall on the local community.
As of now, Zankore has not disclosed which cooling system it will use. The NVIDIA chips that will power the AI infrastructure can use evaporative cooling towers, while alternatives such as seawater-based cooling could reduce freshwater consumption. Such systems, however, may create other environmental risks, including impacts on marine ecosystems. An AI factory is designed to operate continuously. Its servers cannot simply shut down because water becomes scarce. That is precisely why the government must understand its resource demands before the facility becomes fully operational.
At minimum, Zankore should be required to publicly disclose its projected water consumption, water sources, cooling technology, and contingency plans during droughts. Authorities should also establish clear rules for allocating water between industrial users and local communities. The same scrutiny should apply to other large data center projects, including the proposed AI data center in Jatiluhur, West Java. Indonesia's ambition to become an AI hub should not come at the expense of communities already struggling with water insecurity and sinking land.
16
