FULL STORY

Machenta Emerges From Stealth With $40M and Instant Robot Learning

London robotics startup Machenta, founded by Imperial professor Edward Johns, emerged from stealth with a $40M seed round, then launched its Instant Robot Learning technology and expanded hiring.

2026-10-27 ~ 2026-10-29 · 2 episodes · 25 posts

Episode 1 · London robotics startup Machenta exits stealth with $40M seed round (2026-10-27, 21 posts)

On October 27, Machenta, a London robotics startup founded by Imperial College robot learning professor Edward Johns, exited stealth and announced a $40M seed round led by Creandum and Union Square Ventures, with valuation undisclosed. According to @lukasmziegler, it is the largest robotics seed round in UK history. The company's flagship is "Instant Robot Learning": a new robot foundation model (VDA, Vision-Demo-Action) trained entirely in simulation with zero real-robot data — after a single demonstration by an ordinary person, the robot learns the task on the spot, with no hours or days of extra data collection or training, and adapts on its own when the environment changes.

Confirmed

  • Machenta exited stealth with a $40M seed round led by Creandum and USV; valuation undisclosed; per @lukasmziegler, the largest robotics seed round in UK history.
  • Core product is an Instant Robot Learning foundation model (VDA, Vision-Demo-Action): one-shot task learning from a single demonstration, zero real-robot data, pure simulation training, self-adaptation to environment changes.
  • Founded by Ed Johns, head of Imperial's Robot Learning Lab, with a team from Imperial; @mangahomanga notes the model had been in development for several months. @ZeeshanZiaML notes results are still described as preliminary.
  • Per USV's @rebeccakaden, the investment thesis is that intelligence lies not in the scale of prior knowledge but in the speed of learning new tasks; Machenta uses fully synthetic data. Eric Rosen (via @EdJohns) praised the team's prior few-shot manipulation work, citing InstantPolicy and DINOBot.

Why it matters

  • If one-demo learning holds up, it could dramatically lower the data and training barrier to robot deployment, compressing the gap between buying a robot and putting it to work to a single demonstration.
  • While most general robot model makers race for massive real-world data, Machenta takes a contrarian pure-simulation route — a differentiation advantage if it transfers to reality.
  • The UK's largest robotics seed round itself signals the scale of capital betting on this direction.

1 more related posts →

Episode 2 · Machenta Unveils Instant Robot Learning and Expands Hiring (2026-10-28, 4 posts)

Robotics startup Machenta has launched Instant Robot Learning, a approach where robots learn tasks from a single demonstration instead of massive data collection. The London-based team is expanding hiring, including a Community Lead role to build its community from scratch.