Eka Robotics Raises $150M at $1.45B Valuation to Scale Dexterous AI Robots

Eka Robotics, a Cambridge, Massachusetts-based robotics startup, has raised $150 million in a Series A funding round at a post-money valuation of $1.45 billion. The round was led by Andreessen Horowitz (a16z) and F-Prime, with continued backing from existing investors Obvious Ventures and E14 Fund.

Founded by Pulkit Agrawal and Tuomas Haarnoja, Eka Robotics is developing artificial intelligence models designed to help robots perform complex physical tasks with greater speed, precision and adaptability. The company says its robots can handle unfamiliar objects, insert wires and screw in light bulbs, positioning its technology for practical applications in industrial and manufacturing environments.

The new funding will support the expansion of Eka's Vision-Force-Action (VFA) model, enable robots to perform a wider range of tasks and accelerate customer deployments.

Eka Robotics' $150M Funding Round and Investors

The $150 million Series A values Eka Robotics at $1.45 billion post-money, making it a significant new entrant in the rapidly expanding physical AI and robotics sector.

Andreessen Horowitz and F-Prime led the round. Existing investors Obvious Ventures and E14 Fund also participated.

F-Prime is a venture investment firm affiliated with FMR, the parent company of Fidelity Investments. The investment reflects growing investor interest in AI systems that can operate beyond digital environments and interact directly with the physical world.

The capital will primarily support research and development, expansion of the company's robotics capabilities and faster deployment of its technology with customers.

Eka has not disclosed specific customer contracts, revenue figures or a detailed commercial deployment timeline in the funding announcement covered by the source.

What Is Eka Robotics Building?

Eka Robotics is developing what it calls a Vision-Force-Action model, or VFA. The system combines three inputs that help a robot understand and interact with its surroundings: visual information, force feedback and physical actions.

Rather than relying solely on demonstrations of human behaviour, the company trains its robots through trial and error in simulated environments using reinforcement learning. The objective is to help robots learn how physical objects respond to contact, pressure and movement before transferring those skills to real-world hardware.

This approach is intended to address one of robotics' persistent challenges: dexterous manipulation.

Tasks such as picking up unfamiliar objects, inserting wires into connectors or handling delicate components require more than recognising an object visually. Robots must also judge how much force to apply, adjust their movements and respond when an interaction does not go as expected.

Eka's VFA model is designed to combine these capabilities within a single system.

The company also says it has developed a custom robotic gripper equipped with more force sensors than a standard gripper. These sensors provide additional feedback during physical interactions, helping the robot respond to objects rather than relying on vision alone.

Eka Robotics Claims 1,500 Operations Per Hour

One of Eka Robotics' headline performance claims concerns the speed at which its robot can pick up a wide variety of unfamiliar objects.

According to the company, its system can complete approximately 1,500 object-picking operations per hour. The article compares this with a human baseline of around 600 operations per hour, putting the reported robotic throughput at roughly 2.5 times that benchmark.

The company also says its robots can perform different types of manipulation, including screwing in a light bulb and inserting wires.

These capabilities could be relevant to manufacturing tasks where workers or machines must repeatedly handle components of different shapes, sizes and physical properties.

However, the reported figures are company claims. The article does not provide independent benchmark results or enough testing details to establish whether the same performance can be maintained across different factory environments, shifts and operating conditions.

That distinction matters because industrial customers need more than high-speed demonstrations. They also require reliability, repeatability, safety and consistent performance under real production conditions.

Simulation and Reinforcement Learning Drive Robot Training

Eka says simulation plays a central role in developing its robotic systems.

According to the company, simulation can generate thousands of hours of robot experience in minutes. Reinforcement learning allows the robot to improve through repeated attempts, learning which actions produce the desired physical outcomes.

Once trained, these skills can be transferred from simulated environments to physical robotic hardware.

The approach is designed to reduce the time and effort needed to teach robots new tasks. If it works reliably at scale, it could help manufacturers introduce new robotic capabilities without requiring extensive task-specific programming or collecting large amounts of real-world demonstration data for every application.

Eka's approach differs from systems that primarily learn to imitate human movements from video demonstrations. The company argues that learning through physical interaction, force feedback and simulation can help robots develop the capabilities needed for more complex manipulation tasks.

The broader challenge is whether these learned skills transfer reliably to the varied conditions of actual industrial environments.

Who Founded Eka Robotics?

Eka Robotics was founded by Pulkit Agrawal and Tuomas Haarnoja, who met while pursuing graduate studies at the University of California, Berkeley.

Pulkit Agrawal is an associate professor at the Massachusetts Institute of Technology. He also co-founded SafelyYou, a computer vision company focused on reducing falls among people living with dementia.

Tuomas Haarnoja previously worked at Google DeepMind on reinforcement learning for small humanoid robots, including research involving robotic soccer. He also co-created the Soft Actor-Critic algorithm, a reinforcement learning method used in robotics and other control problems.

The company says its broader team brings experience from organisations and institutions including MIT, UC Berkeley, DeepMind, Boston Dynamics, Microsoft, Carnegie Mellon University, the University of Pennsylvania, Northeastern University and the University of Oxford.

This combination of academic research and robotics experience is central to Eka's effort to translate advanced AI techniques into practical physical systems.

Physical AI Funding Is Accelerating

Eka Robotics' funding comes amid growing investment in companies developing AI systems for robots and other physical applications.

The article points to several well-funded competitors. Physical Intelligence has been reported at an $11 billion valuation, while Generalist was reported at a $3 billion valuation following a funding round of nearly $200 million. Skild AI raised $1.4 billion in January at a reported valuation above $14 billion.

Eka's $1.45 billion valuation is considerably lower than those reported figures, although differences in funding stages, business models and commercial progress mean the companies are not directly comparable.

Barcelona-based robotics startup THEKER also raised an $85 million Series A in June, while Dream Labs, founded by former NVIDIA robotics researchers, announced a $90 million seed round on October 20, 2026.

The broader industrial market provides another reason for investor interest. The International Federation of Robotics reported that factories installed 603,000 industrial robots in 2025, an increase of 11%, taking the global operational stock above five million robots.

Meanwhile, Crunchbase data cited in the article counted $47.4 billion in physical AI venture funding across 521 deals during the first half of 2026. That category also includes areas such as autonomous vehicles, aerospace and drones, rather than robotics alone.

Can Eka Robotics Move From Demonstrations to Production?

The next major milestone for Eka Robotics is demonstrating that its robots can deliver sustained performance in real production environments.

High throughput in a controlled setting is an encouraging starting point, but manufacturers must also consider downtime, maintenance, safety, integration costs and the ability to repeat tasks consistently over long periods.

A named customer operating Eka's robots on a production line for months would provide a more concrete indication of commercial readiness than a short demonstration.

The company's $150 million Series A gives it additional resources to develop its VFA model, expand the range of tasks its robots can perform and accelerate deployments.

Whether Eka can turn its reported manipulation capabilities into dependable factory operations will be a key test of its technology and its $1.45 billion valuation.