Wind, Sun, and Silicon: China's New AI Bet

Envision Group switched on its Galaxy Campus AI data center in Inner Mongolia on Aug. 6, 2026, running entirely on wind and solar power

19 Aug 2026

Envision Group, a Chinese renewable energy company, has commissioned the first phase of its gigawatt-scale Galaxy Campus data centre in Ulanqab, Inner Mongolia. The facility, spanning 120,000 square metres, became operational at 120 megawatts on August 6 and is fully allocated to two domestic technology companies. Its entry into artificial intelligence infrastructure marks an unusual move for a company built on clean power generation rather than computing.

What sets the site apart is its power source. Instead of drawing from the public grid, Envision connected the facility directly to on-site wind and solar generation through its own AI Power System. The company says the setup is "designed to deliver industry-leading token-generation capacity while establishing a new model for gigawatt-scale renewable AI infrastructure." There is no diesel backup and no grid dependency; only renewable electricity feeds the site's processors around the clock.

Beyond this first phase, the company's plans extend further. Envision intends to expand Galaxy Campus beyond 2 gigawatts of capacity, enough to support up to one million AI accelerators. If that target is reached, Ulanqab, a region already known for strong wind resources, would become one of the largest hubs for AI computing anywhere in the world. Companies seeking sustainable, high-performance computing capacity would gain a rare option combining scale with verified low-carbon credentials.

The implications reach beyond China. Regulators worldwide are placing growing pressure on data centres to curb their carbon footprints, prompting large technology buyers to search for credible low-carbon alternatives. Galaxy Campus suggests that renewable energy companies could become significant, if unexpected, participants in that supply chain. Rather than simply consuming power, firms like Envision may increasingly generate it, and in doing so decide where and how the next generation of AI systems is built and run. Whether that shift accelerates will depend on how quickly power availability, rather than chip supply, becomes the binding constraint on AI expansion.