Microsoft's Emissions and Water Use Shoot Up as It Goes All in on AI Datacenters

Waste Water

As Microsoft ramps up efforts to conquer the artificial intelligence industry, its latest sustainability report shows that its water use and carbon emissions have shot up dramatically.

The software giant increased its water use from 6.4 million cubic meters in 2022 to 7.8 million cubic meters in 2023.  And its emissions jumped from around 12 million metric tons of carbon in 2020 to about 15 metric tons last year, mostly stemming "from the construction of more datacenters" and indirect carbon footprint from "building materials, as well as hardware components such as semiconductors, servers, and racks."

Microsoft has lofty goals to become carbon negative by 2030 — pledging to use greener tech, materials and processes — while pursuing aggressive water replenishment projects. But that's going to be tough as it continues its pursuit of AI dominance, including a high-profile partnership with OpenAI and immense new energy needs.

Energy Vampire

This is the same problem confronting the rest of the AI industry: building water- and energy-hungry datacenters to fuel the AI boom comes at the expense of our already overburdened planet.

CNBC reported that Goldman Sachs projects natural gas consumption will make up 60 percent of the energy source powering datacenters, with the rest made up from renewable energy.

The need for resources to run datacenters is so immense that the AI industry may require an energy breakthrough in fusion or another technology, OpenAI CEO Sam Altman recently opined. Other experts have suggested that the AI industry could be forced to lean on nuclear power — as exemplified by Amazon's recent $650 million acquisition of a datacenter powered by a nuclear power plant in Pennsylvania.

And Microsoft is also getting into the act with recent moves signaling its serious interest in nuclear power, such as a job listing to help build up "nuclear energy infrastructure."

The industry as a whole needs to come up with ways to reduce its carbon and water footprint — and fast. It's not just Mother Nature either, but communities — including some in Arizona and Iowa — that are already balking at the rise of datacenters and their thirsty ways.

More on AI datacenters: Machine Learning Expert Calls for Bombing Data Centers to Stop Rise of AI