Eka is worth $1.45B thanks to a physical AI model that has experts raving

Why it matters

Eka Robotics, founded by MIT professor Pulkit Agrawal and former Google DeepMind researcher Tuomas Haarnoja, has closed a Series A round led by a16z and F-Prime, valuing the company well into unicorn territory. The startup tackles a persistent weakness in robotics: reliable grasping. Its approach pairs a force-sensitive gripper with a vision-force-action model, which learns the relationship between force inputs and actions rather than relying primarily on vision, as most vision-language-action models do. Co-founder Agrawal argues that vision only roughly locates an object, while the actual grasping is accomplished through forces. Training relies on large-scale reinforcement learning in a physics simulator, letting the robot exceed what human demonstration data could teach it. Experts quoted by TechCrunch, including University of Illinois Urbana-Champaign professor Kris Hauser, say the robot adapts to new objects without retraining. The company frames its contribution as closing the gap between generality and performance, handling physical manipulation while general models handle planning.