JATILUHUR - BDx Data Centers says construction is under way on a six-building build-out of its CGK4 AI data-centre campus in Jatiluhur, planned for 640MW of IT capacity.

BDx detailed the development in a Sept. 22 project announcement, nearly two months after West Java Governor Dedi Mulyadi, Purwakarta Regent Saepul Bahri Binzein and company executives attended a groundbreaking ceremony at the site on July 28.

The work is a further build-out of an existing CGK4 site rather than the establishment of an entirely new campus. BDx had previously described CGK4 as scalable to 500MW; its current CGK4 facility information lists the Jatiluhur campus as a phased 640MW development with 640MW of planned IT load.

Construction has begun on Building 1, the first of six planned buildings. BDx says it will deliver 120MW of IT capacity in phases, with only the first contracted phase expected to enter service in early 2027. The wider development is expected to take about three years, with buildings commissioned sequentially.

CGK4 is backed by 845MVA of secured grid capacity from state electricity utility PLN. BDx says the campus is about five kilometres from Jatiluhur Dam and is pursuing lower-carbon power options, including a hydroelectric arrangement with Perum Jasa Tirta II; it has not described that hydroelectric supply as already operating.

The campus was certified under NVIDIA's DGX-Ready Colocation Data Center program in April 2025. BDx says Building 1 will use direct-to-chip liquid cooling engineered to support rack densities of up to 500kW for high-density AI computing.

BDx Indonesia is a joint venture involving BDx Data Centers, Indosat Ooredoo Hutchison and Lintasarta. A US$320 million loan facility secured in April is supporting BDx's regional expansion and investments in high-voltage grid capacity at CGK4 and the company's CGK5 campus in Suryacipta.

Agus Hartono Wijaya, chief executive of BDx Indonesia, said the project would support domestic access to advanced computing. “It allows local enterprises, government agencies, and developers to run advanced AI workloads domestically,” he said.