Eka Robotics Reaches $1.45B Valuation as Physical AI Model Achieves 95% Pick Success Rate

Eka Robotics Secures $150M Series A at $1.45B Valuation

Eka Robotics closed a $150 million Series A funding round led by Andreessen Horowitz (a16z) and F-Prime, valuing the startup at $1.45 billion post-money1

1

. Founded by MIT professor Pulkit Agrawal and former Google DeepMind researcher Tuomas Haarnoja, the company has developed a breakthrough approach to robotic manipulation that's drawing widespread attention from robotics experts. Raghu Raguram, the a16z general partner who co-led the round, told TechCrunch that Eka Robotics is "already exceeding superhuman performance for certain tasks," with expectations for superhuman performance across many dexterous manipulation in robotics tasks within months1

1

.

Vision-Force-Action Model Transforms Physical AI Capabilities

Unlike traditional vision-language-action models that dominate current robotics AI, Eka Robotics has built a vision-force-action model that fundamentally changes how robots interact with physical objects. The physical AI model focuses on the relationship between visual inputs, force feedback, and actions rather than relying primarily on vision and language understanding1

1

. "Vision is not the primary way in which you close this last millimeter problem," Agrawal explained. "Vision tells you roughly where the object is, what the pose is, but the actual grabbing, the actual searching is happening through forces"1

1

. This approach addresses long-standing criticism from robotics pioneers like Rodney Brooks, who have argued that generative AI robotics models will fail without properly representing force and touch.

Force-Sensitive Grippers Enable Delicate Object Handling

The company designed proprietary force-sensitive grippers that are more flexible than traditional robotic hands and include enhanced force sensors that react to contact in real-time1

1

. This capability allows Eka's robots to handle delicate items like raspberries without crushing them—a demonstration that left one frontier lab roboticist emotionally moved. "The experience of a big, heavy robot rushing towards my arm and then slowing down just in time and very gently touching me is a feeling that I am, as we speak, carrying this feeling in my body," the researcher told TechCrunch1

1

. The robot can pick up approximately 1,500 objects weighing up to 1 kg per hour, compared to about 600 for a human worker, while maintaining a reliability rate exceeding 95%1

1

.

Reinforcement Learning in Physics Simulation Drives Performance

Eka's breakthrough stems from its use of reinforcement learning in physics simulation rather than relying on human demonstrations collected by gigworkers or robot operators. The founders developed a physics simulator that models the multiple forces robots encounter when grasping objects, creating a reinforcement learning environment that runs at scale1

1

. The company claims it can generate thousands of hours of robot experience in minutes, allowing virtual robots to practice millions of variations in object position, friction, and contact before policies reach physical hardware2

2

. "Rather than being limited to the data that you're collecting from, say, human demonstrations, Eka is pushing through reinforcement learning this approach where the robot can surpass any kind of demonstrations that are given to it," said Kris Hauser, a professor at the University of Illinois Urbana-Champaign1

1

.

Bridging the Generality-Performance Gap in Robotics

Agrawal and Haarnoja, who first met in graduate school at the University of California, Berkeley, are leading researchers in reinforcement learning for robotics control. Their key insight was that this technique could scale through simulation rather than requiring extensive real-world data collection1

1

. "What EKA is bridging is the generality-performance gap, which plagues a lot of modern robotics," Agrawal argues1

1

. The robot demonstrated remarkable adaptability by handling a large number of different objects without retraining, exhibiting what Hauser called a "humanness, creepy humanness that it would try over and over again if it failed the first time"1

1

.

Implications for Manufacturing and Logistics Automation

Andreessen Horowitz's investment thesis centers on the economic opportunity in automating physical tasks that currently require human dexterity. Much of the world's physical economy—from manufacturing and electronics assembly to logistics and food production—still depends on people using their hands because the real world is variable2

2

. Industrial robots have transformed highly structured production where fixtures hold parts in precise positions, but enormous categories of physical work remain difficult to automate because parts move, materials deform, objects slip, and connectors don't align perfectly2

2

. The ability to combine human versatility with machine speed, reliability, and uptime represents what a16z describes as one of the largest opportunities in general-purpose robotics, particularly as labor constraints intensify across these industries2

2

.