I’ve been concerned for a while about how much energy and water are being consumed by the AI tools we’ve all come to rely on. So I decided to write this article as a reflection of my findings so far. I hope you find it an interesting read that provokes some thought.
Preface
I’ll be upfront—I used both Google NotebookLM, Canva AI and ChatGPT to help pull this together. They’re powerful tools, and like many of you, I’m still figuring out how to use them thoughtfully.
I genuinely believe AI can offer enormous benefits. But I also believe we shouldn’t be reinventing the wheel. If what you’re doing already works, use AI to enhance—not replace—what’s good. For example: if you’re spending hours generating the “perfect” image, when stepping outside and snapping a photo would have done the job, go for the photo! Even a simple Google search now uses AI when really, a basic search is often all that is needed.
In a world where energy demand is soaring and drought patterns are becoming more extreme, making smarter, more deliberate choices about how we use AI is one way we can contribute to a smaller environmental footprint. Data centres also need to step up. Maybe it’s time we start building them in places where floodwater can be captured, or at the very least, ensure that water used for cooling is recycled.
AI Environmental Impact Calculator
I honestly believe it’s important to be able to have an idea of the impact of the Ai we are using in our own personal space, whether at work, home or school, I have created a simple calculator that includes some of the more popular AI programs, so you can get a sense of what your own consumption might look like. You simply choose the program and enter how long you have been using it for. There is a sliding scale so that if it was something that required a lot more prompting and iterations, you can slide the scale further to the right. If it’s a simple question, then you can slide it to the left or just leave it centred. For a better idea, enter in your total hours for the week! You can find the link below
I hope you find the article useful—and that the calculator offers a few lightbulb moments. It’s not perfect, but it’s a great way to see which tools are heavy hitters when it comes to power and water use—and which ones are lighter on the planet.
“I hope that the calculator offers a few lightbulb moments.”
Try out the calculator here AI ENVIRONMENTAL IMPACT CALCULATOR
The AI Boom – What’s at Stake?
AI has exploded into every corner of our lives, and with it comes a footprint we don’t always see.
According to current research:
- Global data centers used around 415 terawatt-hours (TWh) in 2024. That’s expected to more than double by 2030.
- A single ChatGPT session (10–50 prompts) may consume up to 2 litres of water for server cooling.
- Microsoft saw a 34% increase in water use linked to AI; Google, a 20% rise.
- AI models require up to 10x the energy of a traditional search query.
The London School of Economics warns that AI policy must embed environmental sustainability—not just ethics and bias—and calls out “AI exceptionalism” for sidelining ecological responsibility.
Green Tech or Greenwash?
Like most technologies, AI comes with both promise and pressure. Below is a snapshot of the most commonly discussed advantages and drawbacks when it comes to its environmental footprint.
| ✅ Pros | ❌ Cons |
|---|---|
| AI can optimise energy systems, reducing waste in grids and logistics. | Training large models consumes huge energy, emitting tons of CO₂. |
| Smaller, task-specific models (e.g. NotebookLM) use less power. | Inference (daily usage) continues to draw significant energy & water. |
| Neuromorphic computing offers up to 1000x efficiency for edge tasks. | Most AI runs in data centers reliant on fossil fuels or vague offsets. |
| Big tech firms are investing in carbon-neutral and renewable energy. | Water usage for cooling is increasing sharply, often in dry regions. |
| AI enables sustainability tracking in supply chains and operations. | Greenwashing risk – many eco-claims lack transparency or verification. |
| Companies like Google and Microsoft are publicly committing to green AI. | Hardware turnover causes e-waste, with rapid chip obsolescence. |
*Neuromorphic computing is a branch of AI that mimics how the human brain works. Instead of relying on massive amounts of energy to run big language models like ChatGPT, neuromorphic systems use tiny electrical pulses (called spikes) to process information—just like our brains do.
Bottom line: Sustainability claims are encouraging—but without transparency and public benchmarks, it’s hard to validate them.
What Can We Do?
Individuals
- Use AI intentionally—not habitually. Fewer prompts = fewer watts.
- Choose simpler tools for simpler tasks.
- Avoid over-generating images or text when manual or creative alternatives exist.
- Try the [AI Impact Calculator] to see your own footprint.
CEOs and CMOs
- Audit internal AI use: What tools are being used, and why?
- Choose providers with verified sustainability metrics, not just branding.
- Ensure AI use aligns with climate and ESG goals—don’t let tech policy outpace green commitments.
- Create a culture of mindful AI use across teams—from marketing to operations.
- Bake AI impact analysis into your procurement process.
Data Centre Developers & Infrastructure Partner
If you’re building the future of compute, help build the future of responsibility too.
- Pitch with Purpose: Water usage and cooling strategies should be part of your investor and client conversations.
- Showcase Efficiency: Highlight how your facility uses renewables, recycles heat, or captures grey water.
- Be a Systems Thinker: Propose sites in areas where excess water can be reused or floodwater harvested.
- Elevate your social licence to operate: Corporate social responsibility still matters. It should lead, along with ethics. Stakeholders expect environmental stewardship, not just performance.
A Smarter Path Forward
AI is here to stay. But we have a choice: build a future where intelligence equals impact—or one where intelligence is also informed, intentional, and environmentally sound. Let’s not wait for regulation. Let’s lead with insight and integrity.
Research & References
- University of California, Riverside – Making AI Less Thirsty: Water Footprint of AI Models
- International Energy Agency (IEA), 2024–2025 energy demand forecasts
- Microsoft & Google Environmental Sustainability Reports, 2023–2024
- London School of Economics (LSE)– AI & Climate Governance, 2024
- LSE You Tube: https://youtu.be/F_-f5NS0pEc?si=8J2-RTIPup-oFRZd
- University of Sydney – Neuromorphic Computing for Energy-Efficient AI, August 2024
- Time Magazine, Vox, Nature Machine Intelligence, Wired, and The Washington Post AI energy features
Written by Fi Lucas with assistance from Notebook LM, ChatGPT 4.o and Canva AI.
Recent Comments