Infineon Technologies is making strides in the tech industry by developing the industry’s first 12kW battery backup unit (BBU) specifically designed for AI data centres. This innovative solution comes in response to the increasing power demands of the next generation of AI accelerators, which can consume over 1000A of current, necessitating higher power capabilities to prevent power outages and potential data losses.
The BBU roadmap outlined by Infineon includes a range of power solutions, starting from 4 kW and scaling up to the impressive 12 kW battery backup unit. This roadmap is underpinned by a proprietary architecture that leverages a combination of silicon and GaN power devices to deliver optimal performance and reliability.
One of the key features of the Partial Power Converter (PPC) architecture is its ability to safeguard sensitive AI hardware from voltage spikes, surges, and other power irregularities. By effectively filtering and conditioning the power supplied to the data centre equipment, the BBU ensures a stable and secure operating environment for critical AI systems.
With the introduction of the 12 kW system, which integrates multiple 4 kW power converter cards equipped with Infineon’s PSOC microcontrollers, 40 V and 80 V OptiMOS, and EiceDRIVER Gate Drivers, data centre operators can benefit from enhanced power capabilities and system resilience.
Furthermore, the modular design of the system allows for easy scalability and maintenance. By enabling the paralleling of multiple cards on a motherboard, the system can achieve higher power levels while ensuring operational continuity. In the event of a card failure, the system can continue to function at a reduced capacity, minimizing downtime and enhancing overall reliability.
According to Adam White, Division President Power & Sensor Systems at Infineon, the development of energy-efficient battery backup units is crucial for ensuring uninterrupted power supply to AI systems. These units are engineered to deliver exceptional performance, flexibility, and efficiency, catering to the diverse power requirements of AI servers. By leveraging a comprehensive understanding of semiconductor materials such as silicon (Si), silicon carbide (SiC), and gallium nitride (GaN), Infineon aims to set a new standard in powering AI data centres.