West Java reviews water requirements for BDx's suspended AI campus near Jatiluhur Reservoir

Anyone writing a date next to this campus in a deployment plan has water figures to choose from, and they do not agree. The governor's estimate of about 380 liters per second ^[12] is 12.7% of the 3,000 liters per second he said Purwakarta's water utility holds ^[17]. IESR's 384 liters per second at full scale ^[3] is about 14 times the 27 cited in the October 5 review discussion ^[18].

The nearby supply system gives a third way to size it. Permitted for 120 liters per second with about 70 in use ^[15], it has 50 liters per second of headroom ^[19]. That covers 27. The 384 is about 7.7 times that headroom ^[20]. Treat this as a sizing check and nothing more: TechRepublic reports that the figures appear to concern different supply arrangements or development stages, and that their relationship has not been fully clarified ^[16].

Water is one item inside a larger permit. The suspension was ordered over missing approvals, and the water question sits inside one of them. Officials believe a draw near 380 liters per second could be accommodated without disrupting existing supplies ^[13]. The environmental assessment is the document that must establish projected withdrawals ^[5], and the West Java government confirmed that the AMDAL has not been issued ^[2].

Teams tend to tell themselves that a secured grid connection and a stated date add up to a schedule. BDx's September 22 announcement covered the first: 845 megavolt-amperes of grid capacity from state utility PLN ^[7]. For the second, The Straits Times reported that BDx still targets early 2027 for its first contracted phase, subject to approvals ^[8]. Purwakarta's permitting decisions will determine when construction can resume and whether that timetable is achievable ^[22].

We think a team with a regional deployment tied to this campus should treat early 2027 as a conditional date and keep a second source of capacity open until the AMDAL and PBG are issued. The tradeoff is the cost of holding a fallback that may go unused. A two-axis sort sizes that cost: whether the approvals are issued, and whether the sources agree on one water figure. If both are yes, plan to the date. Where approvals are issued but the figure is open, plan to the date and ask how the dry-period withdrawal was assessed. When the figure is agreed and approvals are pending, plan to the permit decision. If neither holds, keep the fallback. On the record so far BDx sits in that last cell: the AMDAL and PBG have not been issued ^[2], and the water figures have not been reconciled ^[16].