A Chinese green tech company recently made history by launching the world’s first commercially deployed, fully integrated 800-volt DC power system in a massive artificial intelligence (AI) data center.
Shanghai’s Envision Energy announced the technology on Monday, November 2nd. She currently powers live computing workloads at the company’s Galaxy AI data center campus in Ulanqab, Inner Mongolia, China.
The platform combines solid-state transformers, battery storage and intelligent controls to deliver grid power to computing devices. According to company officials, this could reduce energy losses, reduce infrastructure costs and support more powerful AI servers in the same space.
Kevin Huang, senior vice president of Envision Energy, noted the potential of the technology. “AI data centers require power systems that coordinate renewable energy, energy storage and the grid to meet computing needs,” Huang explained.
Operating AI servers
As AI data centers expand, they require more power to operate. Global data center electricity demand could increase by 160 percent by 2030, according to reports. Additionally, research suggests that data centers alone could require up to 68 gigawatts (GW) of capacity worldwide by 2027, roughly equivalent to California’s power grid.
However, providing this energy efficiently remains a major challenge. Traditional data center power systems use multiple levels of conversion to deliver power from the grid to servers. These stages require additional equipment, cables and space.
To address this challenge, Envision has developed a new system that can convert power directly from a 10-kilovolt medium-voltage input to an 800-volt DC system. It reduces the need for additional equipment and copper cabling and leaves more space for computer hardware.
To maintain a stable power supply, the company integrated solid-state transformers with battery storage. The batteries can also respond to sudden changes in power demand within seconds and provide backup power for several minutes.
At the same time, Envision’s Dubhe Controller coordinates both the transformers and batteries via its Dubhe Energy Foundation Model. This allows the system to respond to rapid changes in power demand within milliseconds.
More performance in sight
Envision’s Galaxy AI data center campus became operational in August 2026. It has a planned capacity of three gigawatts (GW). Once fully developed, the facility is expected to support up to one million AI accelerators operating simultaneously.
Huang said the system uses renewable energy to power AI workloads. “This is a critical step in transforming power computing coordination from a strategic vision to an operational reality,” he concluded in a press release.
In the future, the company plans to obtain more than 80 percent of the campus’s electricity via direct renewable energy connections. The power architecture could serve as a reference for other data centers as the industry moves to 800-volt DC infrastructure.
“The system is already in commercial use ahead of widespread industry adoption and represents a reference for the industry’s transition to 800V DC,” the company said. It also aims to expand the technology through the Mission Gobi initiative. The goal is to provide five GW of green AI data center capacity in desert regions worldwide by 2030.
