AI can be a lot of fun. The incredible visual art, “self”-portraits, poems, movie scripts and movies themselves can be truly captivating. And even when it’s terrible, it still has entertainment value. Plus the feedback teaches the AI model how to improve.
Enchanting landscapes are readily available with a quick prompt and a couple of keyboard clicks. Instantly, you can be transported into an ethereal forest lush with giant flowers, mythical creatures, a pink sky and maybe a purple ocean.
The experimental nature and profitability of AI is unmistakable. It is becoming increasingly common and widely used, stirring excitement and controversy.
While it clearly holds immense value, it also comes at a significant cost. We’re not just talking about the loss of jobs, income and wellbeing of people being replaced by AI.
Generating anything from a blog to one those digitalized landscapes comes at the expense of of the real natural world. AI is resource-heavy, especially at the magnitude it is relied upon.
Down the drain
AI requires massive amounts of energy for data processing and data, not to mention millions of gallons of water needed to cool data centre machinery.
Servers in data centers generate a lot of heat that must be vented or cooled, otherwise they can shut down or sustain damage.
According to a 2024 NPR report, the average data centre uses 300,000 gallons of water a day to keep cool, roughly equal to water use in 100,000 homes.
Data centers’ electricity consumption is projected to reach 1,000 terawatts, roughly Japan’s total consumption, by 2026.
Data in the dark
Jesse Dodge is a research scientist at the Allen Institute for Artificial Intelligence in Seattle. She was part of a group of 10 prominent researchers that wrote a conference paper on the impact of AI wrote in 2022.
It reads: “Data scientists today do not have easy or reliable access to measurements of [greenhouse gas impacts from AI], which precludes development of actionable tactics”.
Since the article’s release, AI applications and users have exploded, yet the public is still largely left in the the dark about those data, Dodge added.
At the same time, AI use accounts for carbon emissions from non-renewable electricity, and as mentioned before, the consumption of millions of gallons of fresh water. Indirectly, this elevates impacts from building and maintaining the power-intensive equipment AI needs to function.
As AI continues to expand and touch every part of our lives, from choosing a new hair colour to planning a custom trip to Mexico, tech companies cite many ways AI can enhance and reduce energy consumption.
This can be especially true in medical fields, where AI has proven extremely effective. For example, AI systems have helped radiologists identify breast cancer that would have otherwise been undetectable in its early stages
Clearly, this is a powerful example of how, when used intentionally and with purpose, AI is a gift.
Sustainable AI?
Many legislators, regulators, activists, and different international organizations are determined to put policies and protective measures in place to prevent the risks of AI from crushing the benefits.
“The development of the next generation of A.I. tools cannot come at the expense of the health of our planet,” Massachusetts Senator Edward Markey stated in a 2024 Press Release.
Markey along with other senators and representatives introduced a bill that would require the federal government to study AI’s current environmental footprint. Based on the data, the government would develop a standardized system for reporting future impact.
The European Union’s AI Act will also require “high-risk AI systems” such as “foundation models” that power ChatGPT and similar platforms) to report:
- Energy consumption
- Resource use
- And other impacts during a systems’ lifecycle
In addition, the International Organization for Standardization has publishes documentation and guides for “responsible AI”. This is to help AI users make more deliberate and educated decisions about how to use and engage with AI.
The ISO ongoingly “develops standards for measuring energy efficiency, raw material use, transportation, and water consumption, as well as practices for reducing A.I. impacts throughout its life cycle, from the process of mining materials and making computer components to the electricity consumed by its calculations” (Yale Environment).
Problems at play
Data centres are considered the warehouses of the digital world, containing and storing the data that makes up the cloud.
Cybersecurity Ventures predicts that the total amount of data stored in the cloud will reach 100 zettabytes this year. This accounts for:
- Public clouds operated by vendors and social media companies (Apple, META, Google, Microsoft, X, etc.)
- Government-owned clouds that are accessible to citizens and businesses
- Private clouds owned by mid-to-large-sized corporations
- Cloud storage providers
In other word, that is 50% of the world’s data at that time, reflecting a 25% jump from how much was stored in the cloud in 2015.
A 2025 survey by AFCOM, a data centre industry body, found that only 46% of its members’ facilities can meet their needs by their existing cooling systems. Unfortunately, more than one-third reported persistently running out of cooling capacity.
This begs the question: How will companies manage with new, more advanced computer chips (considered critical to AI) that are predicted to generate between 5 and 10 times as much heat?
Design a new chip, get creative and lean on more sustainability practices. That sounds like a joke but it isn’t.
Doing better
In the same AFCOM survey, 73% of respondents plan to utilize renewable energy, including the 27% who are already at it.
59% of respondents believe solar is gaining the greatest popularity in data centres, followed by 28% who believe wind is picking up the greatest speed.
Tech company Lenovo pioneered the Lenovo Neptune water cooling technology that ranks #1 on the Green500 list for innovative designs powered by the tech giant NVIDIA’s graphic processing units (GPU). Lenovo claims it allows computing capabilities to accelerate without over heating in high-heat, multi-GPU environments.
The Lenovo Neptune promises to help clients to see up to “40% reduction in power consumption and a 3.5x improvement in thermal efficiencies compared to traditional air-cooled systems”.
NVIDIA has also developed “AI Superchips”, such as the NVIDIA Grace™ CPU that are said to “delivers over 2x more performance at the server level and 3x better energy efficiency compared to leading x86 systems”.
Responsible AI
While AI itself isn’t new to world, the level it is depended on and it’s advanced capabilities are — and there are no signs of it slowing down.
Much research is to be done to continue studying how AI affects on sociocultural and environmental levels. There certainly isn’t a black and white answer as to how AI can both benefit or damage us.
It’s worth asking yourself, however, how necessary it is to hop on ChatGPT and ask it to write an email or a social media caption for you. How badly do you need AI to turn a photo of you into a cast member of your favorite TV show?
Is it worth the energy expenditure — the water and electricity needed to process millions of other similar, and often self-explanatory, prompts?
You can always ask a friend who writes or illustrates. In the thick of a constantly evolving world, hang on to your critical thinking skills and creativity. It’s in there and while convenience is tempting, sometimes cutting corners simply isn’t rewarding.
For more stories like this, click here.
