The Cambridge startup behind a physics-trained robotic arm is reportedly now a unicorn, roughly a year after its seed round
Robots have long been great at lifting car doors and terrible at picking up a grape. Eka Robotics thinks it has closed that gap, and investors appear to agree.
The Cambridge, Massachusetts startup has reportedly raised a $150 million Series A at a $1 billion valuation, according to TechCrunch. The pitch is a robotic arm that can grab a wide range of objects with motion smooth enough to look almost human.
Eka was founded in 2025 by Pulkit Agrawal, a professor at MIT, and Tuomas Haarnoja, a former researcher at DeepMind.
The company closed an $11 million seed round on May 9, 2025. Andreessen Horowitz and MFV Partners backed that round.
If the reported Series A figures hold, the new round is more than 13 times the size of the seed. That jump happened in roughly a year.
Eka employs approximately 25 people as of April 2026. Divide a $1 billion valuation by that headcount and you get about $40 million per employee.
The startup’s public debut focused on everyday manipulation tasks like grasping and sorting objects.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
The core of Eka’s technology is what it calls a Vision-Force-Action (VFA) foundation model. The robot combines what it sees with what it feels through force, then decides how to move.
Many robotics companies train their systems on human demonstration data. That means recording people performing tasks, then having the robot learn by copying.
Eka takes a different route. It leans on simulation-based reinforcement learning and physics reasoning. The robot practices in a virtual world governed by physics, learns from trial and error, and carries that understanding into the real one.
The payoff, according to the company’s positioning, is dexterity with fragile and varied items. Its range runs from light bulbs to food products.
Robotic manipulation is one of the hardest unsolved problems in automation. That gap matters for industries like food processing and logistics, which handle enormous variety with much of the work still depending on human hands.
The simulation-first approach also has a scaling argument baked in. Human demonstration data grows only as fast as people can record it. Simulated practice can, in principle, run continuously across many virtual environments at once.
The first signal will be commercial deployments. Pilots with food processors, warehouses, or manufacturers would test whether the VFA model holds up across thousands of real-world picks.
The second signal is hiring. With roughly 25 staff, a $150 million war chest gives Eka room to expand engineering and hardware teams quickly.