Kandi acquires Xinchu to enter AI data-center backup power

This is a small deal in dollar terms, but it points to a critical chokepoint in the AI stack: power quality and backup. As AI data centers push grids to their limits, operators are looking for battery systems that can handle rapid, high‑amplitude load changes with millisecond response times. Xinchu’s focus on UPS lithium packs, high‑rate discharge cabinets and intelligent BMS tuned for AI data centers positions it squarely in that niche. Kandi is buying its way into this segment just as AI infrastructure buildouts worldwide start to collide with power bottlenecks. ([globenewswire.com](https://www.globenewswire.com/news-release/2026/06/29/3318735/0/en/kandi-technologies-acquires-controlling-stake-in-xinchu-new-energy.html))

For the AGI race, robust, efficient backup power is part of what makes ever‑larger training clusters politically and technically viable. Outages or power‑quality issues that take thousands of GPUs offline are hugely expensive and can derail long training runs. By combining Xinchu’s products with its own manufacturing and supply chain, Kandi is betting that AI‑driven data centers, plus 5G and cloud, will support a durable market for specialised batteries and BMS software.

It also highlights how value is diffusing into adjacent sectors: you don’t need to build models to make money from AGI; you can sell the electrons, cabinets and control systems that keep the GPUs humming. Expect more traditional industrial and automotive suppliers to reposition themselves as AI‑infrastructure plays.