GenAI and the Environment

This is Part 2 of our mini-series on GenAI and its impact on energy and the environment.

Location, Location, Location

The state in which the AI companies run their servers and build their data centers greatly impacts how harmful the energy is to the environment. This is because different states get their energy from different ratios of sources.

But GenAI Will Solve the Climate Problem, Right?

The AI boom since late 2022 has incentivized companies like Microsoft, OpenAI, Meta, and Google to use more electricity. Executives for those companies often claim that AI will eventually develop the breakthrough technology needed to solve global warming, but so far, no such breakthrough exists or is on the horizon. Still, carbon emissions grow.?

According to Wired, Google’s “energy consumption doubled from 2019 to 2023.” But Google was quick to pass on the blame, arguing that “Reducing emissions from our suppliers is extremely challenging, which makes up 75 percent of our footprint.” This was referring to the?“manufacturers of servers, networking equipment, and other technical infrastructure for the data centers—an energy-intensive process that is required to create physical parts for frontier AI models.”[1]

Microsoft hasn’t fared better, seeing increases in both megawatts of energy consumed and in liters of water used to cool its data centers.

The Wall Street Journal shares a few additional unpleasant factoids:[2]

• Emissions from the global build-out of data centers between now and 2030 could equal about 40% of the entire U.S. economy’s annual emissions, Morgan Stanley estimates.?
• The demand is already extending the lives of fossil-fuel power plants and setting back U.S. climate progress and optimism. Meta Platforms recently said its emissions last year were about 70% above 2019 levels. Microsoft’s jumped 40% in the three-year period through June 2023. Google’s surged nearly 50% in the four years through December. Chips, materials and power for data centers, plus overall business growth, were big drivers.?
• Tech companies are increasingly looking to nuclear to meet their huge clean-power needs. Microsoft has agreed to buy power for 20 years after Constellation Energy spends about $1.6 billion to restart Pennsylvania’s Three Mile Island, the site of the country’s worst nuclear power accident. Earlier this year, Amazon paid $650 million for a nuclear-powered data center.
• Over time, tech companies say AI could help them limit emissions by, for example, helping utilities more efficiently deploy renewable power. But it isn’t clear such benefits would come close to making up for the soaring emissions. “It’s a massive increase that may not be justified by the productivity gains from AI,” said Gernot Wagner, a climate economist at Columbia Business School.

If AI is going to lead to a technological breakthrough to solve the climate crisis, it needs to happen soon.

[1] https://www.wired.com/story/ai-energy-demands-water-impact-internet-hyper-consumption-era/

[2] https://www.wsj.com/tech/ai/big-tech-is-rushing-to-find-clean-power-to-fuel-ais-insatiable-appetite-31f91330?st=CRbJ9L&reflink=share_mobilewebshare

The following students from the University of Texas at Austin contributed to the editing and writing of the content of LEAI: Carter E. Moxley, Brian Villamar, Ananya Venkataramaiah, Parth Mehta, Lou Kahn, Vishal Rachpaudi, Chibudom Okereke, Isaac Lerma, Colton Clements, Catalina Mollai, Thaddeus Kvietok, Maria Carmona, Mikayla Francisco, Aaliyah Mcfarlin