MIT study quantifies AI’s true environmental footprint for the first time

Researchers have quantified the energy and emissions impacts of AI systems in unprecedented detail, revealing significant variability based on query type, model size, and power source location. This comprehensive analysis from MIT Technology Review marks the first data-driven examination of AI’s true environmental footprint, highlighting how the same AI query can have dramatically different climate impacts depending on when and where it’s processed. These findings arrive at a critical time as AI deployment accelerates globally without transparent reporting from major companies about their systems’ resource demands.

The big picture: AI’s energy consumption varies dramatically based on query complexity, model size, and content type, making it impossible to assign a single environmental impact figure to “using AI.”

• Text responses from larger AI models can consume up to 70 times more energy than smaller models answering identical prompts.
• Video generation typically demands hundreds of times more energy than producing text responses.
• Complex queries like creating detailed travel itineraries can require nearly 10 times the energy of simple requests like generating jokes.

Why this matters: The geographical location of data centers significantly influences the climate impact of identical AI operations due to regional differences in energy sources.

• Researchers found that querying a data center in West Virginia creates nearly twice the emissions of the same query in California, based on 2024 average data.
• These regional disparities highlight how critical data center location and energy sourcing decisions are for companies aiming to minimize AI’s environmental impact.

Behind the numbers: The six-month investigation by MIT Technology Review represents the most detailed assessment yet of AI’s energy footprint, though significant data gaps remain.

• Researchers focused on open-source models where technical details are publicly available, as companies running closed-source systems like ChatGPT and Gemini declined to provide energy consumption data.
• The findings suggest that without standardized reporting requirements, the true scale of AI’s environmental impact remains obscured as the industry expands rapidly.

Looking ahead: As AI becomes increasingly influential in reshaping society, work, and power infrastructure, understanding its resource demands will be crucial for sustainable planning.

• The research team documented their methodology and findings across multiple articles, including analyses of data center growth in Nevada and the relationship between AI expansion and natural gas consumption.
• Several technical innovations for improving AI energy efficiency were also explored alongside perspectives on nuclear energy’s potential role in powering AI infrastructure.