Thames Water deploys satellite AI in Swindon to target 100m litre leak loss

TL;DR:

• Thames Water has signed a 13-month deal with UK firm Origin Tech to use satellite imagery and AI to detect underground leaks across Swindon and surrounding Wiltshire, after a successful trial.
• The previous trial identified 800 leaks and saved an estimated 8.7 million litres of water a day — enough to fill more than three Olympic-sized swimming pools.
• The companies are targeting up to 100 million litres a day of leak reduction, against an Environment Agency benchmark that 19% of UK water is lost before reaching customers.

The satellites pass over Swindon roughly every six days and produce a map of points where leakage is suspected, Origin Tech co-founder John Marsden told the BBC. Thames Water engineers then use the map to direct field crews — a process Marsden said is at least five times faster than unaided ground search for “invisible leaks that are really difficult to find”.

The first 13 months will test the case at scale

The 13-month deal scales the partnership beyond the initial trial. In that trial, several of the leaks ran at more than 10 litres per second — enough water to fill a bath every eight seconds, in Thames Water’s framing. Shane Gloster, Thames Water’s head of leakage, said the partnership lets the company work “more efficiently in our pursuit of bringing down water loss” by giving engineers a faster and more accurate indication of network problems that are invisible at street level.

The strategic context is unusually sharp for a regional utility partnership. The Environment Agency’s 19% UK water-loss figure has been a fixture of every Ofwat price review for two decades; satellite-AI detection is one of the few interventions that has produced credible verified savings at scale rather than incremental improvements on Victorian-era pipe replacement. The Thames Water decision to extend beyond trial sits against a backdrop of regulator pressure on the company over service performance and the wider question of whether private investors continue funding the network’s modernisation.

Looking forward

For UK SME readers, this is a concrete UK utility-AI deployment with measured ROI, not a vendor pitch. The pattern — third-party satellite data, AI processing, integrated with the utility’s existing field-operations workflows — is replicable across Severn Trent, Anglian Water and the Scottish and Welsh networks, all of which face comparable leakage benchmarks. Watch for whether other water companies sign similar Origin Tech deals before the next Ofwat AMP cycle, and whether Ofwat begins counting satellite-AI deployment specifically in operational-performance reviews. The bigger question for the policy debate is whether AI-enabled efficiency gains let UK water firms hit leakage-reduction targets without the bill increases they have already trailed.