TL;DR

Machenta, a physical AI company spun out of Imperial College London, has raised £30m ($40m) in a seed round led jointly by Union Square Ventures and Creandum, Imperial announced on 27 October. Imperial describes it as the largest seed round ever raised by a UK robotics startup. The company builds a robot foundation model designed to pick up a new task after seeing it done once.

The technology

Robots today usually learn a task by watching people repeat it many times, a process that can take days, and they need developers to update the model before they can do anything new. Machenta’s approach aims to cut that to a single demonstration, so that eventually the people using a robot could teach it new jobs on the spot.

The second claim concerns training data. Rather than pre-training on large volumes of real-world footage of humans, Machenta pre-trains its model entirely in simulation. Imperial says this could free developers from the practical and privacy problems of collecting human data at scale. According to the team, a single model has picked up varied jobs, packing and tidying among them, on several kinds of robot, humanoids included.

“Intelligence is not so much about what you can already do, it’s about how quickly you can learn to do something new,” said Professor Edward Johns, Machenta’s co-founder and chief executive, whose team at Imperial’s Robot Learning Lab did the underlying research.

From lab to company

The research started six years ago in that lab. Johns founded the company with his PhD student Vitalis Vosylius and other lab colleagues once the techniques showed promise. The funding will go on scaling the model and growing the team, which is moving into new King’s Cross premises with lab space.

Imperial’s vice-provost for research and enterprise, Professor Mary Ryan, said Machenta’s early success signals “an opportunity for the UK to become a global leader in AI-enabled robotics”.

Looking Forward

The claims here come from Imperial’s own announcement, and no independent benchmark is cited, so the one-demonstration promise has not been independently verified. In our view, the simulation-only training is the part to watch: if it holds up on real robots in real settings, it would sidestep the gathering of large human demonstration datasets, with the practical and privacy problems that brings. For UK robotics, a seed round of this size shows that investors will back university research at an early stage.

Sterling figures converted at the ECB reference rate of 27 October 2026 ($1 = £0.75).