Flex Capacitive Energy Storage System (CESS)

Balance short-duration energy pulses

A Flex CESS balances spikes during large electrical transients by absorbing and supplying energy as power fluctuates. This addresses the real-time management required of AI and HPC workloads when the response time of traditional uninterrupted power supply (UPS) systems is too slow.

The Flex CESS, designed with Musashi Hybrid SuperCapacitors (HSCs), offers:

• Efficient energy management by alternately supplying and absorbing energy in its internal HSCs
• A significantly longer lifespan than current battery energy storage systems, with ability to provide multimillion charge/discharge cycles
• A wide operating temperature range and inherent safety with cells able to operate in temperatures from -30°C to >70°C
• UL 810A certification and completion of UL 9540A testing for thermal runaway at the cell level

Lower energy costs

Peak shaving reduces reliance on the grid during periods of elevated demand when utilities may charge higher prices. Using a Flex CESS, data center operators can significantly lower costs by storing energy during off-peak hours then releasing it as needed to power AI and HPC workloads.

Manage GPU power spikes in AI-era data centers

With an innovative approach to GPU Vdroop management, the Flex CESS is particularly suited to the demands of AI and HPC workloads. A Flex CESS offers an effective alternative to traditional power management solutions, supporting infrastructure resilience through a unique combination of energy storage and management technologies. Parallelable up to eight units per rack power zone, the Flex CESS is part of a broad critical and embedded power portfolio that addresses data center power requirements from grid to chip.

Beneficios clave de aprovechar los servicios de centro de datos probados de Flex:

• Outperform traditional UPS systems at scale with real-time power stabilization
• Reduce the risk of performance throttling and system crashes
• Provide consistent performance throughout the lifespan of GPU deployments
• Store excess energy at lower power thresholds
• Optimize power usage to reduce overall energy costs and deliver it when peak performance demands it

Mitigate power fluctuations in data centers deploying AI and HPC workloads