AI-Driven Robotics Revolution Sparks $6.1 Billion Investment Surge in Humanoid Machines
In 2025, investments in humanoid robotics surged to $6.1 billion, marking a fourfold increase from 2024. This transformation in robotic learning techniques began around 2015, focusing on trial-and-error simulation training. The introduction of ChatGPT in 2022 accelerated advancements by enabling large language models to anticipate robotic movements. Cynthia Breazeal of MIT unveiled the social robot Jibo in 2014, which ultimately failed despite raising $3.7 million through crowdfunding. By 2018, OpenAI's Dactyl managed a 60% success rate with Rubik's Cubes. Kanishka Rao's team at Google DeepMind developed the RT-1 and RT-2 models for training robots in human tasks, while Covariant's RFM-1 system produced interactive warehouse robots, and Agility Robotics' Digit, utilized by Amazon, can lift up to 35 pounds.
Key facts
- Humanoid robot investment reached $6.1 billion in 2025, quadruple 2024 levels
- Robotic learning shifted from scripted rules to trial-and-error simulation around 2015
- ChatGPT's 2022 launch accelerated robotics by enabling predictive action models
- MIT's Cynthia Breazeal introduced social robot Jibo in 2014 with $3.7 million crowdfunding
- OpenAI's Dactyl hand achieved 60% Rubik's Cube success using domain randomization
- Google DeepMind's RT-2 trained on internet images to interpret scenes and objects
- Covariant's RFM-1 enabled warehouse robots to communicate like coworkers
- Agility Robotics' Digit humanoid lifts 35 pounds and is deployed by Amazon, Toyota, and GXO
Entities
Artists
- Cynthia Breazeal
- Kanishka Rao
- Peter Chen
- Pieter Abbeel
Institutions
- MIT
- OpenAI
- Google DeepMind
- Covariant
- Agility Robotics
- Amazon
- Toyota
- GXO
- Apple
- Crate & Barrel
- Silicon Valley
Locations
- United States