Nadella compares AI data center water use to restaurants

Wednesday 03 - 15:44

By: Dakir Madiha

Nadella compares AI data center water use to restaurants

Satya Nadella addressed growing scrutiny over the environmental impact of artificial intelligence infrastructure during a keynote at a major developer conference in San Francisco. He positioned the rapid expansion of cloud and AI data centers as essential for technological progress while arguing that new designs are reducing pressure on local resources, including water and electricity.

Azure, Microsoft’s cloud platform, now operates more than 500 data centers across 80 regions worldwide. The company says this marks its largest-ever global infrastructure footprint and reflects an accelerated buildout over the past 18 months. That pace, according to company leadership, exceeds the expansion achieved during the first decade of the platform’s existence.

The expansion has triggered public opposition in several communities hosting new facilities. Protesters gathered outside the conference venue in San Francisco, voicing concerns over energy consumption, water usage, noise, and light pollution linked to large-scale data centers. Some community groups argue that rapid infrastructure growth risks straining local utilities and natural resources, particularly in areas already facing environmental stress.

Nadella pointed to a flagship facility in Wisconsin as an example of improved sustainability design. The site uses a closed-loop cooling system that is filled once and then recirculates water, which the company claims reduces ongoing consumption to levels comparable to a neighborhood restaurant. The campus also uses high-density GPU architecture powered by hundreds of thousands of advanced chips connected through extensive fiber networks.

Microsoft has also pledged broader environmental targets, including reducing overall water use across its data centers by 40 percent by 2030 and returning more water to the environment than it consumes. However, environmental groups remain cautious, arguing that large land acquisitions and legacy cooling systems still raise concerns about long-term resource demand as AI infrastructure continues to expand.