Machenta, a London robotics company spun out of Imperial College London, has come out of stealth with a $40m seed round co-led by Creandum and Union Square Ventures (USV). The company announced the round on 27 October 2026.
Tech.eu reported that the round is the largest seed investment raised by a UK robotics company to date. No other investors were named, and the company did not disclose a valuation.
“Intelligence is not so much about what you can already do, it's about how quickly you can learn to do something new.”Edward Johns, Co-founder and CEO, Machenta
What Machenta does
Machenta is building a robot foundation model for what it calls Instant Robot Learning. The idea is that a robot can learn a new physical task after seeing it demonstrated once, without further training.
Most robotics companies train their models on hundreds or thousands of real-world demonstrations. Machenta instead trains its model entirely in simulation, using computer-generated environments and no real-world data. The company describes the model as one that has learned how to learn.
The company says it has shown the approach on packaging, assembly and tidying tasks, and that one model has learned everyday tasks across several types of robot, including humanoids, according to Tech Funding News. It also said fine-precision and complex motions remain difficult. Machenta did not cite a customer, and Tech Funding News described it as a research company with no near-term commercial plans.
On its blog, Machenta describes the system as a vision-demo-action model: it takes in what a robot sees and a single demonstration, and produces the actions needed to repeat the task. The company argues that more computing power improves its model in three ways, by scaling the model itself, the learner and the simulated world it trains in. It has not published independent benchmarks.
Alongside the funding news, Machenta published a technical post on its website explaining its approach.
The team
Machenta was founded in 2025 by chief executive Edward Johns, a professor of robot learning who runs the Robot Learning Lab at Imperial College London, and Vitalis Vosylius, his former PhD student. The company says its work builds on six years of research on one-shot learning at Imperial.
Tech Funding News reported that the company has 11 staff. Johns calls the approach "physical vibe coding", according to Tech.eu.
Kathi Wilhelm, a partner at Creandum, said very few robotics teams can point to a body of work like Machenta's.
What the money is for
Machenta will use the money to hire, buy more computing power and generate more simulated data, so its model can handle more precise and complex tasks. It is also moving into a new office and lab in King's Cross, London.
Tech Funding News named Physical Intelligence, FieldAI and Skild AI as competitors. Those companies also build general-purpose models for robots, but rely more heavily on real-world demonstration data.
Tech Funding News reported that the company has not set a public hiring target.
Company website: machenta.com