How AI Is Supercharging Our Digital Carbon Footprint

Every time you use ChatGPT, generate an image with AI, or get a recommendation from an algorithm, energy is consumed. A lot of it. Artificial intelligence is the fastest-growing source of digital emissions on the planet — and it’s only just getting started.

Training a single large AI model can emit as much carbon as five cars over their entire lifetimes. Running that model at scale — serving millions of queries daily — consumes energy continuously. And the world is building more AI infrastructure faster than at any point in history.

The scale of AI’s energy appetite

Training vs inference: two different problems

There are two distinct energy costs in AI: training and inference. Training — teaching a model from scratch on vast datasets — happens once but consumes enormous energy. Inference — running the trained model to answer queries — happens billions of times daily at lower cost per query, but at a scale that makes the cumulative impact enormous. Every AI-generated image, every chatbot response, every personalised recommendation adds to the running total.

The water no one talks about

AI data centres don’t just consume electricity — they consume water at extraordinary rates to cool their hardware. Research from late 2025 found that AI systems could consume water equivalent to the world’s entire annual bottled water production. In water-scarce regions like Arizona and Nevada, where many AI clusters are being built, this is already causing local tensions over resources.

What can individuals do?

• Use AI intentionally, not habitually. Ask yourself whether you need an AI tool for a task, or whether a simpler tool — or your own thinking — would do.
• Prefer text over image generation. Generating images and video with AI is significantly more energy-intensive than text-based queries.
• Support regulation and transparency. Demand that AI companies disclose their energy and water usage publicly, not just in selective sustainability reports.
• Choose providers with verifiable clean energy. When possible, favour AI tools built by companies with genuinely renewable-powered infrastructure.

AI is not inherently bad for the environment — it has the potential to accelerate climate solutions across energy, agriculture and transport. But that potential is only realised if the industry is powered cleanly and built thoughtfully. At DigitalGarb, we believe in asking harder questions of the technologies we adopt — and holding their makers accountable for the full cost of what they build.