Eka Robotics Raises $150M at $1.45B Valuation for Touch-Sensing AI
Robotics startup Eka raises $150 million at a $1.45 billion valuation for force sensing AI that lifts 1,500 objects an hour with 95% reliability in tests.
Summary
On October 20, 2026, Eka Robotics closed a $150 million Series A led by a16z and F-Prime at a $1.45 billion post-money valuation. Founders Pulkit Agrawal, an MIT professor, and Tuomas Haarnoja, a former Google DeepMind researcher, met in graduate school at the University of California, Berkeley. Their flexible, force-sensitive hand and vision-force-action model use visual estimates and contact forces for the final millimeter of grasping, supporting Rodney Brooks’s argument that robotic AI needs touch. Physics simulation creates scalable reinforcement learning and thousands of hours of robot experience in minutes, rather than relying on videos from gig workers or operators.
Eka says its arm processes about 1,500 items weighing up to 1 kilogram per hour, versus 600 for a human, succeeds more than 95% of the time, and adapts across objects without retraining. Its speed could damage the robot, yet it slows before contact and handles a raspberry without crushing it. University of Illinois Urbana-Champaign professor Kris Hauser praised its adaptability and repeated attempts. An anonymous frontier-lab roboticist compared the experience with reliable legged locomotion’s arrival five to ten years ago, while a16z’s Raghu Raguram expects superhuman performance across many dexterous tasks within months.
Eka’s model handles physical execution while general models plan, unlike top-down systems such as Astra, described as ChatGPT’s latest LLM, which can control unfamiliar robots but slowly and expensively. Secretive industrial pilots cover pick-and-place and wire manipulation, Eka’s first two verticals. The company must demonstrate yearlong assembly-line reliability and value beyond pick-and-place. Its gripper costs under $1,000 and is rated for 5 million duty cycles, versus the 20 million standard for industrial arms.
Positives
- $150 million from a16z and F-Prime gives Eka capital to commercialize its force-sensitive robotics architecture.
- About 1,500 items per hour more than doubles the roughly 600-item human rate cited by Eka.
- More than 95% claimed reliability combines high throughput with adaptation across many objects without retraining.
- Thousands of simulated training hours can be generated in minutes, reducing dependence on costly human demonstrations.
- Under $1,000 for the gripper could support economical deployment if Eka proves industrial durability.
Risks & concerns
- 5 million rated duty cycles remain one-quarter of the 20 million-cycle benchmark set by standard industrial arms.
- Secretive industrial pilots provide limited evidence that laboratory performance will survive yearlong assembly-line operation.
- Pick-and-place and wire handling do not yet establish that Eka’s methods generalize to more complex industrial work.
- High-speed movement gives the arm enough force to damage itself if its control algorithm fails.
- Astra and other general models remain slow and expensive when operating robots in unfamiliar environments.