In the field, one of the most common—and usually shallow—critiques I hear about AI goes like this:
“AI uses too much energy, water and GPU power.”
What concerns me more is that this view is often held by people responsible for major technology decisions, while those decisions are driven more by hype than by real engineering understanding.
The irony is that some of the same voices criticising AI resource use are the ones running it blindly, with no smart routing, no model segmentation, no usage controls and no clear sense of when AI is actually needed.
The real problem is not AI.
The real problem is using AI without the right strategy.
With smart routing, model segmentation and detached systems, we can cut resource use by up to 90%, because the large model is only brought online when it is genuinely required.
Using AI without strategy is like:
🚛 Hauling a single rock with a full-size truck.
🏎️ Moving house in a Ferrari.
🛡️ Driving an armoured car to the office.
It is also like running a day-long ESG workshop, then ordering hundreds of printed copies of the minutes to hand out afterwards.
Talking about sustainability does not mean we are actually practising it.
The same applies to AI.
Powerful technology should not be thrown at every task.
Run a small model for simple work.
Run a large model only for complex problems.
Use deterministic systems when AI is not needed.
AI is not automatically wasteful.
Weak AI architecture, poor governance, uncontrolled usage and hype-driven decisions are what actually create the waste.
Sustainable AI does not mean using less intelligence.
It means using the right intelligence, at the right time, for the right job.
