AI's Thirst: Data Centers Seek Desert Power

Microsoft's 2030 "carbon negative" pledge faces growing concerns as its carbon emissions surged by over 25% last year. The insatiable energy demand from AI data centers is straining power grids and driving up emissions. Chinese clean energy billionaire Zhang Lei has put forth a novel solution.

Zhang proposes building AI data centers in desert locations, operating on independent microgrids powered entirely by wind and solar energy. His company, Envision, has already established a proof-of-concept facility in Inner Mongolia, demonstrating the feasibility of running data centers around the clock on renewable power without compromising reliability.

This strategy directly challenges the energy approach of major US tech companies, which often rely on natural gas and carbon offset credits. Zhang criticizes this reliance, arguing that companies not fully embracing renewables are being disingenuous and risk creating energy poverty. He envisions gigawatt-scale data centers powered by off-grid renewables.

Envision's desert data center model also addresses water consumption concerns through advanced dry and closed-loop liquid cooling technologies. While requiring more electricity, this is offset by the abundance of cheap renewable energy in desert regions. The company plans to construct 5 gigawatts of such data centers by 2030.

This vision aligns with China's broader strategy to develop AI infrastructure in less populated inland areas utilizing renewable energy. Despite international political and trade tensions, Zhang sees global potential, including in arid regions of Spain, highlighting Chinese renewable technology as a worldwide civilizational output.