Eseandre Mordi
⬤ Physical Intelligence closed a $600 million funding round at a $5.6 billion valuation to develop its universal robot control system. Backed by Alphabet's CapitalG and other major investors, the startup is building a single AI model that can run different types of robots—whether it's a robotic arm, hand, or mobile unit—without needing custom software for each one. The company launched in 2024 with a team of former Google DeepMind researchers and robotics experts who specialize in AI that reads camera feeds and sensor data, makes sense of messy real-world spaces, and controls motors with precision.
⬤ The company's control model is being tested on robotic arms handling everyday tasks like folding clothes, making hot drinks, and putting together boxes. The same core system powers all these different jobs, proving it can handle both home and warehouse environments without starting from scratch each time.
⬤ Physical Intelligence recently rolled out a vision model trained through reinforcement learning—robots get rewarded for doing things right and gradually figure out better ways to work. This approach more than doubled their efficiency. The robotic arms now run for about three hours straight, picking and placing items in roughly three minutes and assembling boxes in around 2.5 minutes.
⬤ The funding signals growing interest in robotics companies building flexible, AI-powered control systems instead of single-purpose machines. As these models get better at understanding complex spaces and handling longer tasks, practical automation for both homes and factories is getting closer to reality.
Eseandre Mordi
