The hidden environmental cost of “smart” tech
AI is everywhere right now.
It’s being sold as a climate hero: optimising energy, reducing waste, saving time, saving money — saving the planet, apparently.
Here at Zero Waste Llama, I need to start with an apology.
It didn’t even occur to me, at first, to look at the real environmental cost of using AI to adapt our photography — for example in a recent post, 🌊 The Two‑Minute Magic: How a Fraserburgh Promenade Walk Sparked a Tiny Act of Coastal Kindness.
In fact, it was a couple of members of the Fraserburgh – Brochers and Proud of it Facebook group who brought this to my attention.
I thought I had learned to pause whenever something is marketed as a “solution” without talking about its waste, energy, or extraction footprint. Clearly, I hadn’t paused enough.
Initially, I was incensed that anyone would think I used AI to produce my writing. Then I stopped reacting… and actually looked into it.
So let’s ask the awkward — but necessary — question:
Is AI actually green… or just very good at greenwashing?
This isn’t an anti‑AI rant.
It’s a zero‑waste reality check.
Estimated reading time: 7 minutes
How I ended up here
About a month ago, one of my ZWLSolutions clients asked me to complete a series of online training courses. These covered a range of AI tools and their practical use within business settings.
I studied. I learned. I read about AI being used to “solve problems” across the world — from mining and geological data analysis to customer relations and AI chat boxes.
I started thinking about how AI could help my business… and this blog.
Eventually, I thought I’d found a neat answer to a problem I often face: anonymising people in photographs I’ve taken, without resorting to stock images or clunky collages. AI seemed like the clean, elegant solution.






How wonderful, I thought.
But then…
The problem with “digital = clean”
There’s a persistent myth that anything digital is low‑impact.
No smoke, no bins, no plastic packaging — so it must be sustainable, right?
Except AI isn’t some fluffy cloud floating in the sky. It runs on very physical infrastructure:
- Energy‑hungry data centres
- Vast cooling systems using freshwater
- Constant hardware upgrades
- Minerals mined from real landscapes, by real people
Zero‑waste thinking teaches us this early on:
If you can’t see the waste, it’s probably just been moved somewhere else.
Energy: efficiency vs actual reduction
Yes, AI can make processes more efficient.
But efficiency does not automatically mean less energy use overall.
- Training large AI models consumes huge amounts of electricity
- Running them at scale adds continuous demand
- As AI becomes cheaper and faster, it gets used more — not less
This is the classic rebound effect:
efficiency lowers cost → usage increases → total impact grows
That’s not a glitch.
That’s how growth‑based systems work.
Water: the invisible resource
Here’s something rarely mentioned in “green AI” marketing:
AI uses a lot of water.
Data centres need constant cooling. That cooling often relies on freshwater — sometimes in regions already experiencing water stress.
No labels.
No disclosure.
No such thing as a transparent “digital water footprint”.
From a zero‑waste perspective, that’s a huge red flag.
Hardware: extraction, e‑waste, repeat
AI depends on:
- Rare earth elements
- Lithium, cobalt, and copper
- Global mining supply chains with significant environmental and social costs
As a geologist, I’m well aware of the impacts involved in extracting these materials.
And because AI models are constantly getting “bigger and better”, hardware becomes obsolete quickly. Which means:
- More extraction
- More shipping
- More e‑waste
Calling AI “clean” while ignoring this is like calling fast fashion sustainable because it’s sold online.
“Green AI” claims — llama‑checked 🦙
🟢 Claim: “AI helps cut emissions”
❌ Reality: Sometimes, in very specific cases — often outweighed by total system growth
🟢 Claim: “Our AI runs on renewable energy”
❌ Reality: Usually based on averages, offsets, or partial accounting
🟢 Claim: “AI is more sustainable than human labour”
❌ Reality: Efficiency isn’t sustainability if overall consumption keeps rising
Zero Waste Llama rule:
If it only works on paper, or in isolation, it’s not a solution — it’s a distraction.
This isn’t anti‑tech. It’s anti‑nonsense.
Let’s be clear:
AI can support environmental work when used carefully and sparingly.
But zero‑waste thinking asks different questions than tech marketing:
- Do we actually need this?
- What problem is it really solving?
- What does it replace — and what does it add?
- Who bears the hidden costs?
Not every problem needs AI.
Sometimes the most sustainable option is to:
- Use less
- Build smaller
- Slow down
- Or change behaviour — not software
That’s not regression.
That’s maturity.
What would genuinely climate‑aligned AI look like?
From a Zero Waste Llama point of view:
- Full transparency
Energy, water, hardware, lifecycle impacts — no vague “green” claims - Smaller, slower, purpose‑built systems
Not giant general models doing everything everywhere - AI as a support tool, not a replacement for responsibility
No outsourcing climate action to algorithms - Clear net benefit
If an AI system emits more than it saves, it doesn’t pass the sustainability test
The takeaway
AI isn’t the villain.
But unchecked AI growth isn’t the hero either.
Sustainability isn’t about doing things more efficiently —
it’s about doing less harm, within real planetary limits.
If AI wants to be part of a zero‑waste future, it needs to:
- Tell the whole story
- Accept limits
- Stop pretending growth = progress
And yes — expect a llama to keep asking uncomfortable questions…
especially when it gets called out. 🦙♻️

Key Takeaways
- The environmental impact of AI includes high energy consumption, water use for cooling, and the reliance on rare earth elements.
- AI often misrepresents itself as a green solution, leading to greenwashing without addressing real sustainability concerns.
- Efficiency in AI does not equate to reduced overall energy use, resulting in increased total impact over time.
- For AI to be sustainable, it must engage in full transparency, smaller purpose-built systems, and add clear net benefits without harmful growth.
- Zero Waste Llama advocates for re-evaluating the actual necessity of AI tools and prioritising less harmful alternatives.

