Machenta, a London robotics start-up spun out of Imperial College London, emerged from stealth on October 27, 2026 with a $40 million seed round co-led by Creandum and Union Square Ventures, Tech.eu reported. The company says it is the largest seed round yet for a UK robotics company. Its model is trained entirely in simulation, with no real-world data, and learns a new task after seeing it performed once.
Why it matters: the field is spending heavily on real-world robot data, from teleoperation farms to first-person video. Machenta is a well-funded test of whether simulation alone can do the job.
Key Facts
- Round: $40M seed, co-led by Creandum and Union Square Ventures (Oct 27, 2026)
- Claim: largest seed round for a UK robotics company
- Founders: Edward Johns (Imperial professor) and Vitalis Vosylius; about 12 staff
- Method: one-shot task learning; trained entirely on simulated data (company claim)
Who founded Machenta?
Edward Johns, a professor of robot learning at Imperial, and his PhD student Vitalis Vosylius. The work began about six years ago in Johns’s lab. The team has about a dozen people in King’s Cross, London. Johns calls the method “physical vibe coding”.
What can the model do?
According to the company, tasks such as packaging, assembling and tidying after a single demonstration, without further training. The Next Web reported that one model has worked across different robots, including humanoids, and that precise tasks and complex movements remain hard. No independent evaluation has been published. The money will go on scaling the technology and hiring.
For the simulation-versus-real debate, see sim-to-real explained and DexForce’s simulation bet.
Frequently Asked
What is Machenta?
A London robotics start-up spun out of Imperial College London. It builds a robot AI model trained entirely in simulation that learns new tasks from one demonstration.
How much did Machenta raise?
$40 million in seed funding, co-led by Creandum and Union Square Ventures, announced on October 27, 2026.