Eka Robotics Raises $150 Million to Develop VFA Model

Eka Robotics announced Oct. 20, 2026, that it raised $150 million in a Series A round. The Robot Report reported Oct. 24 that the company will use the funding to expand its vision-force-action (VFA) model for robots. Andreessen Horowitz and F-Prime led the round.

Eka Robotics says its robots can manipulate objects outside the training distribution about 1,500 times per hour. Whether that claim holds for tasks involving changes in the target objects will help determine whether the technology can reduce the need to configure task-specific actions. Because the approach transfers experience generated in simulation to physical robots, confirming that performance is maintained in real-world environments and that the system is safe will help establish its scope of application.

The significance of the $150 million raise lies not only in the size of the funding but also in how Eka Robotics aims to make manipulation more general through its training approach. The company says it combines reinforcement learning with a VFA model that links vision and force to action, then transfers experience generated in simulation to physical robots. The approach connects not only improvements to the robots themselves but also the computing and training environment used to generate experience with real-world actions. The company’s stated rate of about 1,500 manipulations per hour should be assessed alongside its claim that the objects are outside the training distribution. The practical value of the VFA model will depend not only on the speed of a single task but also on whether it can maintain performance when objects or environments change. The company’s statement that simulation can generate thousands of hours of experience in a few minutes is a design claim about how far training can scale without relying on data collection from physical robots. How well simulation reflects real-world contact and friction, and how performance changes after transfer to physical robots, will require separate verification. We covered the same funding round on Oct. 20 in Eka Robotics raises funds at a $1.45 billion valuation for physical AI models. This report concerns the same round, not a separate fundraising effort. Alongside the valuation, examining the manipulation model described in this article suggests the investment targets not only robot hardware but also the combination of models and simulation used to teach physical manipulation. The impact of this approach on industrial robot adoption will depend on how much it can reduce the need for dedicated machines or individually programmed routines whenever tasks change. However, public information does not specify the conditions for the 1,500-manipulation figure, the range of objects, the success rate, or assessments of continuous operation and safety on physical robots. Going forward, reproducibility of tasks, performance after transfer from simulation to physical robots, and operational results in workplaces will need to be evaluated separately.