According to PitchBook, as reported by the Financial Times, robotics start-ups raised nearly $48 billion in 2026. An increasing proportion of this funding is being used to film people at work in order to train the models.
Robotics and "physical AI" companies have raised nearly $48 billion since the beginning of 2026, according to PitchBook data reported by the Financial Times. What is changing is where this money goes: a growing share is financing the collection of a resource that the sector still does not have in sufficient quantities, recordings of people performing ordinary tasks.
"Robot gyms" and cameras in homes
Physical Intelligence, Skild AI, NEURA Robotics or Generalist combine external service providers and internal collection teams. Some have operators work on robotic arms in dedicated rooms; others install recording equipment in homes, offices and factories to capture movements, contacts and applied efforts. NEURA, which calls this data the scarcest resource in physical AI, plans five of its "NEURA Gyms" in Europe, the United States and China by the end of 2026, including RWTH Aachen and the Technical University of Munich.
Unlike text and images, this type of sequence has never been sucked up on the web: it has to be ordered and then annotated frame by frame. For the FT, the bottleneck in the sector is no longer the hardware or the architecture of the models, but the data.
Read the full article on the original source
Go to the source →