‘Sustainable AI’ is an Oxymoron

‘Sustainable AI’ is AI that limits the usage of resources or uses ‘green’ resources for AI training and inference. Unfortunately, there is nothing sustainable about today’s AI landscape. Public demand is booming; in 2025, monthly tokens just from Google more than doubled and hit 1.3 quadrillion tokens in October².

And the AI vendors are not slowing down. It’s estimated that ChatGPT5 uses more than 8x the energy than ChatGPT4³. Microsoft expects water usage at its data centers (used for cooling) to double, if not more, from 2020 to 2030⁴. In the US, data centers are expected to triple their share of total energy consumption and it’s already ‘wreaking havoc on US power systems’⁵. 

Why is this so important now? Across the US, communities will face energy shortages, water scarcity and increased pollutants to support these massive data centers and AI demand. 

The US Department of Energy (DOE) warns that blackouts could become ‘100x more frequent by 2030’ due to AI demand⁶. Carbon dioxide emissions from data centers are expected to more than double from 2020 to 2030⁷. And renewable energy projects, like offshore wind, are being deprioritized and defunded⁸.

Some communities are already experiencing the negative side effects from a data center moving in next door. Amazon’s data centers in northern Spain plans to use more electricity than the entire region currently consumes⁹. In December 2024, Amazon submitted a request to the local government in Spain to increase the data centers’ water consumption because ‘climate change will lead to an increase in global temperatures and the frequency of extreme weather events, including heat waves’⁹. 

The good news; AI vendors know that scarcity is a problem. They are investing in making their models use less energy and work on smaller chips. But we can’t wait around for them to fix the problems they create. In the next blog post, we’ll help you understand and prepare for this unsustainable world of AI. 

Sources:

1. Roel Dobbe, AI Safety is Stuck in Technical Terms - A System Safety Response to the International AI Safety Report https://arxiv.org/pdf/2503.04743
2. Google processed nearly one quadrillion tokens in June, doubling May’s total https://the-decoder.com/google-processed-nearly-one-quadrillion-tokens-in-june-doubling-mays-total/
3. ChatGPT 5 power consumption could be as much as eight times higher than GPT 4 https://www.tomshardware.com/tech-industry/artificial-intelligence/chatgpt-5-power-consumption-could-be-as-much-as-eight-times-higher-than-gpt-4-research-institute-estimates-medium-sized-gpt-5-response-can-consume-up-to-40-watt-hours-of-electricity
4. Microsoft pledged to save water, in the ai era, it expects water use to soar  https://www.nytimes.com/2026/01/27/technology/microsoft-water-ai-data-centers.html
5. AI is already wreaking havoc on global power systems https://www.bloomberg.com/graphics/2024-ai-data-centers-power-grids/
6. Department of Energy Releases Report on Evaluating U.S. Grid Reliability and Security https://www.energy.gov/articles/department-energy-releases-report-evaluating-us-grid-reliability-and-security
7. AI and climate change https://www.iea.org/reports/energy-and-ai/ai-and-climate-change
8. The Trump administration pauses wind projects off New England, New York and Virginia https://www.npr.org/2025/12/22/g-s1-103320/offshore-wind-power-pause-trump-administration-new-england-new-york-virginia-national-security
9. Revealed: Big tech’s new datacentres will take water from the world’s driest areas https://www.theguardian.com/environment/2025/apr/09/big-tech-datacentres-water