Imperial spinout Machenta raises a $40m seed to teach robots new tasks from a single demonstration
The London physical AI company, co-founded by Imperial professor Edward Johns, says its model learns a new task from one demonstration and is pre-trained entirely on simulated data.
Machenta, a physical AI spinout from Imperial College London, has raised $40 million in seed funding in a round led by Creandum and Union Square Ventures. Imperial announced the round on 27 October 2026 and described it as the largest seed raise ever by a UK robotics startup.
The company is building a robot foundation model that, it says, lets robots learn a new task from a single demonstration. The money will be used to scale that model and to grow the team at a new office and lab in King's Cross, London.
What Machenta is building
Most AI-powered robots today learn a task by watching humans demonstrate it many times, a process Imperial says often takes days. The resulting robot can do that task, but cannot pick up new skills without developers updating the model. Machenta's pitch is the opposite: a robot that is shown a task once by its end user and can then repeat it.
"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 and Professor of Robot Learning in Imperial's Department of Computing. "Humans don't need to see the same thing over and over before they can do it."
According to the university, the team has recently shown a single model instantly learning varied tasks such as packing and tidying, across different robots including humanoids.
Simulation instead of human data
The second part of the approach concerns pre-training. Many robot models are pre-trained on large real-world datasets, often recorded from human demonstrations. Machenta's model is pre-trained entirely on simulated data, which Imperial says could spare developers the practical and privacy problems of collecting large volumes of human data. Tech.eu summed up the idea as "physical vibe coding".
From lab to company
The research behind Machenta started six years ago in Johns's Robot Learning Lab at Imperial. When the lab's techniques began to show promise, Johns and his PhD student Vitalis Vosylius founded the company with colleagues from the lab. Business Insider reports that the startup was founded late last year and has 11 employees, that its valuation was not disclosed, and that it has no plans to commercialise its technology in the near term.
Business Insider also notes the limits that remain. Very precise tasks are still a challenge for the model, the company acknowledged to the publication.
"Very few teams in robotics can point to a body of work like this, which has repeatedly changed what people thought was possible," said Kathi Wilhelm, a partner at Creandum. Professor Mary Ryan, Imperial's Vice-Provost for Research and Enterprise, said Machenta's early success "signals an opportunity for the UK to become a global leader in AI-enabled robotics".
A crowded, well-funded field
Machenta enters a market where the largest competitors are American and very richly valued. Business Insider lists Physical Intelligence, valued at about $11 billion, and Skild AI, valued at more than $14 billion, among the companies working on general-purpose robot intelligence. Against that, a $40 million seed is a modest war chest, but it is large by European early-stage standards.
In Europe, investors have been backing physical AI at increasingly early stages. Europreneurs Media has covered Enact Intelligence's $30m pre-seed in Zurich and Ubitium's €14m seed for a physical AI processor. Creandum, which co-led Machenta's round, also led the seed round of Basel's Atlas Bio, announced on the same day.
What it means for European founders and investors
The round shows that university research with a long track record can still command large seed cheques in Europe, even before revenue. Investors are not only paying for a product; they are paying for six years of published work and a team that built it. For academic founders, that is a reminder that a visible research record can be a fundraising asset in itself.
The co-lead is also notable. Union Square Ventures is a New York firm, and its participation alongside Creandum shows that US investors are willing to back European robotics early, rather than waiting for later rounds.
For investors, the open question is timing. Machenta has said it does not plan to commercialise soon, so this is a bet on technology leadership in robot learning, with revenue some way off. The simulation-first approach could lower data costs if it works at scale, but it will have to prove itself against rivals with far larger budgets.
For the UK, the deal adds to a run of AI-related investment news this autumn, including the government's AI package. The test will be whether Machenta can keep growing in London as it scales.
Spotted an error? Email contact@europreneursmedia.com and we'll correct it. See our corrections policy and editorial standards.
