Studio Aitch is carbon neutral by default. Every project is offset as part of the process, at no additional cost. Working with Carbon Neutral Britain and Ecologi, each project supports verified carbon reduction and removal, alongside longer-term work restoring habitats and protecting ecosystems
Since 2022 it's included planting 1,494 trees, and avoiding 204+ tonnes CO2e.
This isn’t just about planting trees. It’s about backing projects that restore biodiversity, reduce emissions and protect environments under real pressure, from coral reefs to the Amazon rainforest. That same thinking carries through the way the work is delivered.
Refurbished equipment is prioritised over new wherever possible, reducing unnecessary demand on raw materials and avoiding short upgrade cycles. Our comms run on Honest Mobile, a carbon negative B Corp working towards a cleaner, fairer telecoms industry. It’s a small shift, but part of a wider effort to reduce impact wherever it can be reduced.
Digital tools, including AI, are used with care. They’re not weightless. Every prompt and generated asset draws on energy and infrastructure behind the scenes. They’re used where they add value, not by default, and the impact is factored into the wider offsetting approach.
Closer to home, everyday decisions still count. Walking or using public transport where possible, reducing waste, and supporting local initiatives.
A typical short text prompt can use a fraction of a watt-hour of electricity (around 0.0003–0.0004 kWh), while generating a single image is generally higher, often in the region of a few watt-hours (roughly 0.002–0.01 kWh), depending on the model and resolution. For context, boiling a full kettle in the UK is typically around 0.1 kWh, so even image generation sits well below that, but it adds up with scale and repetition.
Water use varies more widely. Some estimates suggest that a handful of AI interactions may use the equivalent of a small bottle of water (around 500 ml) when factoring in data centre cooling and energy production, although this depends heavily on the system and infrastructure used. These figures are based on recent academic and industry research (including MIT, Carnegie Mellon, and Hugging Face) and are intended as indicative rather than exact.