What a quadruped robot actually costs to run in the UK: every line, and the ones the brochure leaves out

Only Unitree publishes list prices, and only for the consumer and education class: US$1,600 for the Go2 Air, US$2,800 for the Go2 Pro and US$4,500 for the Go2 X, excluding tax and freight, with the EDU on request (unitree.com list prices, read on 22 September 2026). The industrial class, Unitree B2, DEEP Robotics X30, Boston Dynamics Spot and ANYbotics ANYmal, and the wheel-legged DEEP Robotics Lynx M20, are quoted per project, and none of their makers publishes a UK price. Any sterling figure you see for those platforms is a reseller’s or a quotation’s, not the manufacturer’s, and it should be read with its scope beside it.

Two things happen to a published dollar price before it is a UK cost. VAT at the standard rate applies on import, and customs duty depends on the commodity classification of the goods; freight and insurance are added on top. So a list price in dollars excluding tax is the start of the calculation, not the end of it, and a UK buyer should budget the landed cost, not the sticker.

Fourteen lines. A quotation that shows fewer is leaving some for you to discover. The table gives each line, what drives it, and whether any maker publishes a figure for it.

Very little, and it is the one running cost you can work out from a spec sheet. Unitree publishes the B2’s battery as 2,250 Wh (unitree.com, read on 22 September 2026), so a full charge from empty is 2.25 kWh before charger losses. At an illustrative 25 pence per kWh, that is about 56 pence per full charge; at 30 pence it is about 68 pence. A robot that charges twice a night is therefore spending a little over a pound a night on energy, whatever the tariff does. The X30 and ANYmal makers publish endurance but not capacity in watt-hours, and Spot’s specification sheet gives runtime rather than capacity, so the same sum cannot be done for them from public figures. Energy is a rounding error in this budget; the point of the calculation is to show that the costs that matter are elsewhere.

Divide the annual cost of the deployment by the patrols it actually completes, and compare that with what the same coverage costs today. The structure is simple; the discipline is filling it with measured numbers rather than brochure ones.

Three of those six numbers, completion rate, operator hours per patrol and false alarms, cannot be known from any specification and are the reason a pilot comes before a purchase. A pilot that measures them turns the structure above into a costed case; a purchase that assumes them turns it into a hope. That is what a structured pilot programme is for, and the routes to a trial without buying are in our guide to trialling a quadruped in the UK.

Outright purchase is the normal starting point and usually the cheaper route over several years once a task is proven and stable. A subscription, where the maker or a partner offers one, bundles hardware, support and updates into a monthly operating cost, which suits a first deployment, a fast-changing platform, a seasonal need or a budget that is operating rather than capital; it is rarely cheaper over the life of a stable deployment. Not every platform is offered on subscription in the UK, and the terms, what the fee covers, fair use, damage and early termination, decide the comparison more than the headline rate. Run the comparison after the pilot has told you what the robot actually delivers; Robotics-as-a-Service explained sets out the model.

Sometimes, for the project rather than the purchase. Innovation grants, R&D support and regional business schemes may contribute towards a suitable trial where the project is eligible. Eligibility depends on location, sector, applicant type, project stage, deadlines and the schemes open at the time, approval is at the funder’s discretion, and funding is never guaranteed, so it is assessed before a pilot is scoped rather than assumed into the budget. A grant finder such as Subvention can identify which schemes you may qualify for.