Eka Robotics raises $150M at $1.45B valuation for force-sensing robot hands

The startup offers an artificial intelligence model for robotic manipulation that works from force data as well as camera images. Eka calls it a vision-force-action model. Force sensors in a custom gripper tell the model how firmly the hand is holding an object, which lets a robot change its grip partway through a task if something begins to slip. Eka’s announcement cites jobs as varied as packing strawberries and assembling server racks for graphics processing units.

Andreessen Horowitz and F-Prime, a Fidelity Investments affiliate, led the $150 million Series A round. Obvious Ventures and E14 Fund, which co-led Eka’s seed round in May 2025, also participated. PitchBook data shows the seed round raised $11 million. Investors have now put about $161 million into the company.

Wired reported in April on a demo of an earlier version of the system at Eka’s office in Cambridge, Mass., where its grippers screwed in light bulbs and packed chicken nuggets onto a moving conveyor. The current robots pick objects that fall outside their training data at about 1,500 operations per hour, according to the company. The human baseline Eka uses for comparison is about 600.

Eka told TechCrunch that its robots handle objects weighing up to 1 kilogram and succeed on more than 95% of picks, and that pilots with industrial customers it has not named are underway on pick-and-place and wire-handling work. The gripper costs less than $1,000 and is rated for 5 million duty cycles for now. Eka wants to reach 20 million, the level standard industrial arms are built for.

Kris Hauser, a professor at the University of Illinois Urbana-Champaign, told TechCrunch that a demo of Eka’s robot had a “wow factor” next to other robot demonstrations he had seen recently and that the robot “was able to adapt to a large number of different objects without retraining.” He also questioned whether the system could run on an assembly line for a year without a failure. Whether it generalizes beyond pick-and-place is still open, Hauser said.

Pulkit Agrawal, a professor at the Massachusetts Institute of Technology who leads its Improbable AI Lab, started Eka last year with Tuomas Haarnoja. Haarnoja came to the company from Google DeepMind, where his team trained small humanoid robots to play soccer.