Machenta raises $40M seed to train robots without real-world data
What's the deal? MachentaDealroom has a profile for this one. Try Dealroom →, a London-based robotics startup founded by researchers from Imperial College London, has raised a $40 million seed round to build an AI model that teaches robots new physical tasks after watching them performed just once. Creandum and Union Square Ventures co-led the round. The company declined to disclose its valuation.
Who's behind it? Machenta was founded in 2025 by Edward Johns, Professor of Robot Learning at Imperial College London, and his former PhD student Vitalis Vosylius, now CTO, together with colleagues from Imperial's Robot Learning Lab, where the one-shot learning work began six years ago. Imperial calls the round the largest seed ever raised by a UK robotics startup. "The team at Machenta has spent years breaking through this extremely hard problem, long before there was a company around it," said Creandum partner Kathi Wilhelm. The company is moving into a new office and lab in King's Cross, London.
Why it stands out: The deal ranks in the 99.6th percentile of all-time seed rounds for robotics startups in the UK, across a sample of 271 — a strikingly large bet for an 11-person team still focused on research.
What's the endgame? While rivals race to amass real-world data, Machenta trains its model entirely in computer-generated environments, without what chief executive officer and cofounder Edward Johns called a "single pixel of real-world data." Once trained, the model can direct a robot to tidy a table or pack a box after a single demonstration, Johns said, with no further training.
The funding will expand the team and train the model with more computing power and simulated data, aiming to handle more precise and complex tasks.
What could go wrong? Machenta has no near-term plans to commercialise, and it acknowledges that precise tasks and complex movements remain challenging. It also faces rivals with far deeper pockets: Physical Intelligence is valued at roughly $11 billion, FieldAI is raising at a $10 billion valuation, and Skild AI, valued at over $14 billion, already sells technology with similar capabilities.
By the numbers: Skild has reported $100 million in annual recurring revenue from more than 60 customers and in August unveiled a model it said could perform unseen tasks of up to 10 minutes after a single video demonstration — a direct challenge to Machenta's core pitch.
The signal: The round is a high-stakes wager on synthetic data as the answer to robotics' data bottleneck, not proof that it can replace real-world experience. Whether a small team betting against the data-hungry giants can keep pace remains the open question.
Read more: Business Insider, Imperial College London, Tech.eu
Image credit: Machenta
