Skild AI's CEO on building one brain for every robot
My full Sources Live interview with Skild AI CEO Deepak Pathak from Davos.
Deepak Pathak has spent 15 years trying to solve one of the hardest problems in AI: getting machines to move through and manipulate the physical world. As a professor at Carnegie Mellon, he worked on research showing how robots could be trained in simulation and perform in the real world almost immediately.
Pathak is the CEO and co-founder of Skild AI, which recently raised $1.4 billion at a valuation of over $14 billion from SoftBank, Nvidia, Jeff Bezos, Disruptive, and others. The company’s pitch is deceptively simple and absurdly ambitious: one general-purpose brain for any robot and any task. It published research last year showing the same model could operate a humanoid or a dog-like robot, and even adapt in real time when limbs were removed.
Last month, I sat down with Pathak in front of a live audience at Brunswick Home on the sidelines of the World Economic Forum in Davos, Switzerland. The conversation that follows gets into why Pathak thinks the entire robotics industry is approaching data wrong, why Tesla’s self-driving dataset won’t help Optimus generalize, and why he believes physical intelligence — not language — is the real foundation of AGI. It also covers Skild’s emerging enterprise business, which went from zero to tens of millions in revenue in 2025.
You can watch the full interview on YouTube. Thanks to Disruptive for making it possible.
The following transcript has been lightly edited for length and clarity:
