Bitera bets on in-town multi-storey data centers to capture Jakarta's AI inference demand
The operator, with over 30 MW of capacity, says low-latency city-center sites with up to 200 kW per rack are the right shape for inference workloads, as an analyst warns of concentration risk.
Indonesian data center operator Bitera is targeting the AI inference market with in-town, multi-storey facility designs in Jakarta, CEO Tedy Harjanto said in an interview published Monday. The company, whose capacity exceeds 30 MW after adding 10 MW since 2020, argues that low-latency sites near points of data consumption are the right topology for inference workloads, citing 100% power availability, 99.999% SLA, power densities up to 200 kW per rack, Tier III certification and ISO/IEC 27001 compliance.
The strategy reflects a broader infrastructure shift as inference demand pulls capacity from large suburban training sites toward urban multi-storey facilities. In the same publication, ICT Institute director Heru Sutadi warned that Jakarta's growing data center concentration could push up electricity and water costs, even as it reinforces the city's position as the country's digital economic center. Capacity and performance claims are vendor-reported and unaudited. For enterprises, growing local inference capacity lowers one of the adoption barriers for latency-sensitive AI workloads, including document processing and real-time video analytics, that previously faced limited domestic compute options.