Eka Robotics raises $150M at $1.45B valuation for force-sensitive AI manipulation model

What's Happening

• Eka Robotics secured a $150 million Series A funding round led by a16z and F-Prime, reaching a $1.45 billion post-money valuation.
• The startup developed a manipulation architecture combining a force-sensitive gripper with a vision-force-action deep learning model.
• The robotic system picks up 1,500 objects weighing up to one kilogram per hour with a 95% success rate, outpacing the human average of 600 objects.

Why It Matters

Most generative AI robotics rely on vision-language-action models that struggle with tactile feedback. Eka's shift to a vision-force-action schema allows robots to handle delicate items and adapt to new objects without retraining. This capability bridges the gap between digital AI reasoning and physical world execution.

The Context

• MIT professor Pulkit Agrawal and former Google DeepMind researcher Tuomas Haarnoja founded the company to solve robotic manipulation challenges.
• Traditional deep learning models struggle to teach robots how to pick up objects despite extensive video training data.
• The system processes language, force inputs, and actions to control the hardware, avoiding damage to the robot arm during high-speed movements.

What They're Saying

• The system exceeds human performance on specific tasks and will achieve superhuman capabilities in dexterous manipulation soon, a16z general partner Raghu Raguram told TechCrunch.
• University of Illinois Urbana-Champaign professor Kris Hauser noted the robot adapts to new objects without retraining and exhibits a human-like persistence when retrying failed grasps.

What to Watch

• The company plans to expand the system's capabilities to achieve superhuman performance in many dexterous manipulation tasks over the coming months.

The Bottom Line

Eka Robotics is commercializing a tactile AI model that gives industrial robots the speed and sensitivity to outperform humans in physical manipulation tasks.