Jakarta, INTI - Jatiluhur is set to host one of Indonesia’s AI data centers, with a planned capacity of 640 megawatts (MW). Located approximately five kilometers from the Jatiluhur Dam, the AI Campus 2 (CGK4) project developed by BDx Data Centers was officially inaugurated by West Java Governor Dedi Mulyadi on September 22, 2026.
The facility has secured an electricity supply of 845 MVA from state-owned electricity company PLN. This capacity forms part of more than 1.2 GVA in total power supply that BDx has secured for its AI-ready data center campuses across Indonesia.
BDx Data Centers CEO Mayank Srivastava said the project represents a long-term investment in infrastructure needed to support the country’s digital and AI development. Construction of the campus is planned over three years, with buildings expected to become operational in phases.
“This campus will serve as an anchor for West Java’s growing AI ecosystem, providing scalable, high-density capacity designed to bring advanced AI workloads into production faster while offering customers a reliable path as demand increases,” Mayank said in a statement on September 28, 2026.
According to Mayank, the campus will receive electricity from PLN as part of Indonesia’s national power grid while actively exploring low-carbon energy options. These efforts include potential hydropower utilization in collaboration with the Jatiluhur water management authority (PJT) to support AI infrastructure growth in the region with cleaner and more sustainable energy.
Construction began with Building 1, which is planned to provide 120 MW of IT capacity in phases. The first contracted phase is expected to begin operations in early 2027. The facility will feature direct-to-chip liquid cooling technology designed to support up to 500 kilowatts per rack for current and future generations of AI accelerators.
The data center campus expansion is supported by a US$320 million loan facility secured in 2026. The financing includes the development of CGK4 in Jatiluhur as part of BDx’s broader expansion of AI infrastructure in Indonesia.
BDx Indonesia CEO Agus Hartono Wijaya said the availability of domestic AI infrastructure could enable local companies, government institutions, and developers to run advanced AI workloads within Indonesia. The development is also expected to create skilled employment opportunities and generate long-term economic value for West Java.
“This is a concrete step for Indonesia’s digital economy,” Agus said.
BDx’s Indonesian platform includes AI Campus 1 (CGK3/3A), AI Campus 2 (CGK4), and AI Campus 3 (CGK5), with more than 1.2 GVA of secured power capacity across its AI-ready campuses in Indonesia.
AI Data Centers Require Water for Cooling
The development of data centers also raises concerns over water consumption. Although data center water use remains relatively small nationally compared with sectors such as agriculture, the pressure can be more significant at the local level because water consumption is concentrated in specific areas.
UGM Professor and software engineering researcher Ridi Ferdiana cited data showing that data centers in the United States account for around 0.15 percent of total water consumption, equivalent to approximately 1.7 billion liters. This is significantly lower than agriculture, which accounts for more than 70 percent of water use.
“The issue is not the national figure, but its implications for the surrounding local area. Data center water consumption tends to be concentrated in certain locations, making the local impact much greater even when its national share is relatively small,” Ridi told Tirto on September 28, 2026.
A large-scale data center, for example, can consume up to approximately 19 million liters of water per day for cooling. Ridi said this volume is roughly equivalent to the daily water needs of a city with a population of between 10,000 and 50,000 people.
In Jatiluhur, water demand is particularly important because the reservoir also serves as a source of drinking water and irrigation. The reservoir supplies 81 percent of drinking water for Jakarta and surrounding areas while supporting irrigation for approximately 242,000 hectares of rice fields with two planting cycles per year.
For illustration, a 40 MW data center using evaporative cooling at an average rate of approximately 1.8 liters of water per kilowatt-hour could evaporate around 27 million liters of water per day. This is equivalent to approximately 0.3 cubic meters per second.
While that volume is significantly lower than the reported flow of around 16 cubic meters per second serving Jakarta and surrounding areas, the localized impact remains an issue that needs to be considered.
“The good news is that BDx reportedly uses direct-to-chip liquid cooling, which is generally 5–10 percent more water-efficient than conventional techniques, or around 25 million liters per day,” Ridi said.
Conclusion
The Jatiluhur AI data center represents a major expansion of Indonesia’s AI-ready digital infrastructure, supported by large-scale power capacity and advanced cooling technology. However, its development also highlights the need to balance growing AI computing demand with responsible water and energy management, particularly given Jatiluhur Reservoir’s important role in drinking water supply and irrigation.
Read more: AI Data Centers Drive Rising Water Demand, With 1 MW Requiring City-Scale Water Consumption