Eka Robotics, founded by MIT professor Pulkit Agrawal and former DeepMind researcher Tuomas Haarnoja, has raised $150 million in Series A funding led by venture funds a16z and F-Prime. The round values the company at $1.45 billion post-money, a sign of investor confidence in its physical AI model for robotic manipulation.
Unlike common vision-language-action models, Eka uses a vision-force-action architecture paired with a custom force-sensitive gripper. The company says its robot can pick up about 1,500 objects per hour, compared with roughly 600 for a human, and does so reliably more than 95% of the time. The system reportedly relies on physics simulation and reinforcement learning to generate training experience at scale.
Roboticists who saw the robot were struck by its performance. One called the experience emotional, comparable to when robots first achieved reliable legged locomotion, while a professor noted a "wow factor" and the robot's ability to adapt to many objects without retraining. The founders remain secretive about implementation details for competitive reasons, but argue the model bridges the gap between generality and performance that has limited modern robotics.