Is AI a climate risk?

We spend most of our time on the visibility and disposal side of IT, but the physical footprint behind the current AI build-out is becoming hard to ignore. A single AI query is reported to use several times the power of a standard web search, and major tech firms have seen their own carbon emissions climb as a result. Cooling that infrastructure also uses large volumes of water, often in areas already under water stress.

The quieter problem: dark data

A significant share of the data organisations store is thought to be "dark data" — collected but never actually used again. Storing it in high-performance, always-on storage is like leaving the lights on in a thousand empty rooms: real energy spent maintaining information nobody is reading. Moving inactive data onto lower-energy storage, and removing duplicate files through deduplication, are both straightforward ways to cut that load without touching anything anyone actually needs.

Where the real win is

The instinct while demand rises is to build more capacity. The more durable win is efficiency: cleaning out data nobody needs, extending the life of hardware that still works, and making sure retired equipment is genuinely reused rather than landfilled. AI can sit alongside sustainability goals, but only if energy and water are treated as real constraints rather than an afterthought.

See how we handle the reuse side of that equation.