Microsoft has continued to seek business from the fossil-fuel industry; documents related to its overall pitch strategy show that it has sought energy-industry business in part by marketing the abilities to optimize and automate drilling and to maximize oil and gas production

The idea that AI’s climate benefits will outpace its environmental costs is largely speculative, however, especially given that generative-AI tools are themselves tremendously resource-hungry

Microsoft has failed to reduce its annual emissions each year since then. Its latest environmental report, released this May, shows a 29 percent increase in emissions since 2020—a change that has been driven in no small part by recent AI development, as the company explains in the report. “All of Microsoft’s public statements and publications paint a beautiful picture of the uses of AI for sustainability,” Alpine told me. “But this focus on the positives is hiding the whole story, which is much darker.”

One slide deck from January 2022 that I obtained presented an analysis of how Microsoft’s tools could allow ExxonMobil to increase its annual revenue by $1.4 billion—$600 million of which would come from maximizing so-called sustainable production, or oil drilled using less energy.

An executive strategy memo from June 2023 indicated that Microsoft hoped to pitch Chevron on adopting OpenAI’s GPT-3.5 and GPT-4 to “deliver more business value.” A Chevron spokesperson told me that the company uses AI in part to “identify efficiencies in exploration and recovery and help reduce our environmental footprint.” There is the tension. On the one hand, AI may be able to help reduce drilling’s toll on the environment. On the other hand, it’s used for drilling.

“AI will solve more problems than it creates,” Willis told me. “A lot of the dilemmas that we’re facing with energy will be resolved because of the relationship with generative AI.”

Microsoft is reportedly planning a $100 billion supercomputer to support the next generations of OpenAI’s technologies; it could require as much energy annually as 4 million American homes

As Joppa told me: “This must be the most money we’ve ever spent in the least amount of time on something we fundamentally don’t understand.”

Blackwell” B200 AI chips will take at least three months longer to produce than was planne

the worker put aside his early doubts after the video call because other people in attendance had looked and sounded just like colleagues he recognized

Microsoft disclosed that its global water consumption spiked 34% from 2021 to 2022 (to nearly 1.7 billion gallons, or more than 2,500 Olympic-sized swimming pools)

ChatGPT gulps up 500 milliliters of water (close to what’s in a 16-ounce water bottle) every time you ask it a series of between 5 to 50 prompts or questions

Google reported a 20% growth in water use

“was literally made next to cornfields west of Des Moines.”