Eka Robotics Raises $150 Million at $1.45 Billion for a Force-Sensing Robot Model With Unnamed Customers

Eka Robotics, a Cambridge, Massachusetts startup, raised a $150 million Series A at a $1.45 billion post-money valuation on Oct. 20, led by Andreessen Horowitz and F-Prime Capital, TechCrunch reported.

None of the performance figures has been tested by an outside party. They come from Eka, its customers are unnamed, and its announcement page lists no customers, deployments or release dates.

What the model is

Eka calls its system a vision-force-action model, set against the vision-language-action models used by most robot-learning labs. It takes force readings as well as camera images. SiliconANGLE says sensors in a custom gripper report how firmly the hand holds an object, so the robot can adjust its grip mid-task if something slips.

Training uses a physics simulator and reinforcement learning at scale, TechCrunch reported. The founders say they can generate thousands of hours of robot experience in minutes. The article does not give the model a name.

The founders are Pulkit Agrawal, a Massachusetts Institute of Technology professor who leads its Improbable AI Lab, and Tuomas Haarnoja, formerly of Google DeepMind. Eka's own page says: "We are building intelligence for the physical world in its native language: force."

The claims, and who has checked them

Eka says the system picks up about 1,500 objects an hour, objects up to 1 kg, against about 600 for a human. It says reliability is above 95%. The gripper costs under $1,000 and is rated for 5 million duty cycles.

1. Eka system1500 objects per hour
2. Eka's human baseline600 objects per hour

Source: TechCrunch and SiliconANGLE reporting of Eka Robotics figures, accessed 2026-10-21

Claim	Figure	Independent check
Throughput	1,500 objects an hour	None reported
Pick success	Above 95%	None reported
Gripper rating	5 million cycles	Target is 20 million
Customers	Industrial pilots	None named

The throughput works out to one pick every 2.4 seconds, against one every 6 seconds for the human baseline. SiliconANGLE says the figures come from the company, not independent testing.

The gripper's current rating is a quarter of the 20 million cycles that standard industrial arms are built for, which is Eka's target.

What an outside roboticist said

Kris Hauser, a professor at the University of Illinois Urbana-Champaign, saw a demo and said it had a "wow factor". He said the robot "was able to adapt to a large number of different objects without retraining".

Hauser also raised two open questions: whether the system can run for a year on a real assembly line, and whether it generalises beyond pick-and-place. The pilots cover pick-and-place and wire handling, and Agrawal plans two verticals, picking and placing and wire handling.

SiliconANGLE adds that Wired reported on an April demo in Cambridge, where grippers screwed in light bulbs and packed chicken nuggets onto a moving conveyor.

Money and headcount

Eka's seed round was $11 million in May 2025, per PitchBook as cited by SiliconANGLE, and total funding is about $161 million. Obvious Ventures and E14 Fund also took part in the Series A. Raghu Raghuram, a general partner at a16z, co-led it and said the system already exceeds superhuman performance on some tasks.

The data aggregator Caplight lists 24 employees. On that count, a single source, the valuation is about $60 million per head.

SiliconANGLE says the new money goes to widening the model's task range and speeding customer deployments. Eka has not named a first customer or a shipping date.

For other early-stage rounds the site has covered, see Mach9's $34 million Series A and Legora's $550 million Series E.