Could battery monitoring become the next challenge for AI infrastructure? A compact IC design focuses on reliability, lower power draw, and simplified system architecture.
Nuvoton Technology Corporation Japan has announced two battery monitoring integrated circuits (BMICs), the KA49703A and KA49713A, for battery backup units (BBUs) used in AI servers. The devices are scheduled to enter mass production in September 2026 and are designed to support high-voltage battery systems by using daisy-chain communication technology previously deployed in automotive applications.
As AI servers continue to power data centres running around the clock, reliable backup power has become increasingly important. BBUs rely on multiple lithium-ion cells that require continuous voltage monitoring to ensure safe operation and maximise battery life. The latest BMICs are intended to simplify battery management by reducing wiring complexity and the number of supporting components required in high-voltage systems.
The ICs support daisy-chain communication across as many as 55 devices, allowing battery-monitoring systems to scale while reducing the need for additional microcontrollers and isolation components. According to the company, this approach lowers component count, saves board space and simplifies system design. Both devices are housed in a 7mm × 7mm QFP-48 package and offer voltage measurement accuracy of ±2.9mV for improved cell monitoring.
One variant, the KA49713A, also includes a dedicated transportation and storage mode that reduces shutdown current to below 0.1µA. The feature is intended to minimise battery discharge while battery packs are transported or stored, helping extend storage periods and reduce the need for recharging before deployment. Such capability could be particularly relevant for large battery systems used in AI server backup units and energy storage installations, where logistics can contribute to battery degradation.
Click here for the official announcement.
