The green washing era of Microsoft seems to have come an abrupt end, like many tech companies now knee-deep in this thing called “AI.”
Once upon a time, Microsoft pledged to replenish more water than it used by 2030, but a recent report from NYT suggests that Microsoft is internally projecting that its water requirements will double by 2030, all in the name of artificial intelligence.
The NYT report says that it saw Microsoft documents suggesting that its annual water needs would more than triple within this decade, to a massive 28 billion liters in 2030. Microsoft used 7.9 billion liters in 2020 and 10.4 billion in 2024.
Microsoft and other cloud companies like Google, Oracle, and Amazon all use billions of liters of water to cool their data centers. Servers churning out compute 24/7 require masses of electricity for power and water for cooling, and electricity to power the cooling and for water to cool the power powering the water cooling. You get the idea. It’s partially why Microsoft and other companies have been exploring nuclear power as an alternative, as the peaks and troughs of solar and wind power often don’t meet the 24/7 uptime AI demands. It might be decades before nuclear infrastructure is realized for this type of project, though.
In any case, Microsoft claims that it has improved efficiencies since the NYT documents, noting that it “only” expects water use to increase 150% by 2030 instead. Microsoft also recently revealed a new Maia server chip, designed entirely around efficiency, to improve some of these metrics.
However, it’s unfortunate reading then that Microsoft’s water usage requirements are expected to disproportionately impact regions that are already suffering droughts. Areas like Arizona and Texas in the United States, Spain’s Aragon region, hyperscale sites in India, as well as Indonesia — are regions particularly hard hit by water shortages, while also playing host to the hyperscalers.
Water usage by data centers is a comparatively miniscule amount of the world’s total consumption, particularly when compared to things like agriculture. However, Microsoft itself has identified the severity its water use can have on tight localities.
U.S. President Donald Trump called out Microsoft specifically for its data center impacts on local communities, prompting Microsoft President Brad Smith to share a large 5-point plan toward becoming a “community-first AI company.”
Given how easy it was for Microsoft to run away from the climate work it was doing, the pledges were met with a non-trivial amount of ridicule on social media.
Microsoft has generally been cast in a negative light in recent years for its wanton AI push. Artificial intelligence platforms are destroying entire industries, all in the name of providing what is effectively a slightly-faster search engine. So far, the social costs of AI don’t seem to match up with the benefits, a concern that Microsoft CEO Satya Nadella himself echoed in a new year personal blog. Nadella noted that cloud companies could lose “societal permission” to hyperscale if tangible benefits aren’t delivered soon.
So far, artificial intelligence has delivered primarily hype and little else. It hasn’t led to any major scientific breakthroughs as promised, it hasn’t boosted efficiency in a particularly meaningful way. It has only served to eliminate entry-level jobs with low-quality automations and imitations, while flooding the internet with industrial-scale memeslop and disinformation.
As the negative impacts continue to snowball, time is running out for Microsoft and its artificially intelligent peers to justify all of this.
Follow Windows Central on Google News to keep our latest news, insights, and features at the top of your feeds!
