A robot hand that only sees is guessing about weight. Eka Robotics has built one that feels, and investors just handed it $150M to scale the idea.
The Cambridge, Massachusetts startup raised a Series A at a reported $1.45B valuation, led by Andreessen Horowitz and F-Prime, a Fidelity Investments affiliate. Obvious Ventures and E14 Fund, which co-led a $11M seed in May 2025, returned. Total funding now sits near $161M.
Eka calls its approach a vision-force-action model. Force sensors inside a custom gripper report how firmly an object is held, so the system can adjust mid-task when something starts to slide, rather than failing and retrying.
Faster than a human baseline
The company says its robots now pick unfamiliar objects at roughly 1,500 operations per hour, against a human benchmark of about 600. Grippers handle items up to one kilogram with a success rate above 95%, cost under $1,000 each, and are rated for 5 million duty cycles, short of the 20 million typical industrial arms sustain.
Pilots are running with unnamed industrial customers on pick-and-place and wire handling, the sort of dull, repetitive work packing strawberries or assembling server racks for GPUs.
The open question
Kris Hauser, a professor at the University of Illinois Urbana-Champaign, told TechCrunch the demo carried real wow factor because the robot adapted across many objects without retraining. He also asked whether the system survives a year on an assembly line without failing, and whether it generalizes past picking and placing.
That is the right question. Manipulation demos have impressed for a decade; factory floors are less forgiving. Eka’s founders, Pulkit Agrawal of MIT’s Improbable AI Lab and Tuomas Haarnoja from Google DeepMind, are betting on force data being the difference.
