Vertiv has expanded its Vertiv PowerUPS 9000 uninterruptible power supply (UPS) family with a new 1.8-megawatt model designed to support large-scale data centers, high-density AI environments, and other mission-critical applications.
The new system extends the PowerUPS 9000 platform’s power range to 1.8MW, building on existing models from 250kW to 1.25MW. The expanded lineup is now available across Southeast Asia, Australia, and New Zealand.
Built for dynamic AI workloads
The 1.8MW UPS incorporates advanced controls designed to address rapid power fluctuations generated by AI workloads. Its Battery Shield technology is designed to reduce unnecessary battery cycling during sudden load changes, while Input Power Smoothing helps create more predictable input power flow and reduce stress on upstream electrical infrastructure.
Vertiv said the system also uses advanced state-of-charge management to help maintain battery availability for backup operations while supporting stability in high-density computing environments.
The platform is designed around high power density to help data center operators accommodate increasing rack densities and limited white space. It also focuses on energy efficiency, serviceability, and flexible deployment alongside broader power and thermal infrastructure.
As GPU clusters can shift from low utilization to peak demand within milliseconds, AI workloads can create rapid power fluctuations that affect batteries, switchgear, generators, transformers, and utility infrastructure. Vertiv said the new UPS is designed to help manage these dynamic load profiles while maintaining power protection and operational continuity.
Vertiv tested its control technologies using a proprietary AI load simulator across different AI workload profiles, battery configurations, duty cycles, and fluctuation ranges. The company said testing showed stable input current behavior despite significant output variability.
In Singapore, Vertiv also collaborated with STT Singapore to test the technology under conditions designed to reflect real-world AI infrastructure requirements, including rapid load transitions and coordination between UPS systems, energy storage, and upstream power infrastructure.
