"AI is resource-intensive, especially when rolled out at scale. It uses a ton of electricity and water, drives new forms of e-waste, and complicates carbon accounting. For mission-driven companies-especially those built on sustainability-that creates a real tension. So we asked ourselves a deceptively simple question: What's our AI footprint? We didn't know the answer. There was no standard. No export from a large language model. No tool."
"Partnering with our longtime friends at Gravity, a carbon and energy accounting platform, we developed a science-informed method for estimating AI emissions-factoring in compute time, server power, and grid emissions. It's not perfect (no model is). But it's a practical start that gives us real visibility into the footprint we're creating. Our projected 2025 AI-related carbon footprint is 17.8 metric tons of CO2e, equivalent to roughly 6% of our 2024 business travel emissions (299 metric tons of CO2e)."
Artificial intelligence consumes electricity, water, and creates new e-waste, increasing emissions and complicating carbon accounting for sustainability-focused companies. Grove Collaborative measured its own AI footprint by developing a science-informed estimation method with Gravity that factors compute time, server power, and grid emissions. The projected 2025 AI-related footprint is 17.8 metric tons CO2e, about 6% of the company's 2024 business travel emissions (299 metric tons CO2e). The estimate establishes a baseline for monitoring growth and targeting reductions. Grove disclosed the estimate during NYC Climate Week and plans to include AI emissions in annual sustainability reporting in 2026. The methodology will be open-sourced.
Read at Fast Company
Unable to calculate read time
Collection
[
|
...
]