TLDR: Amazon-backed robotics startup Skild AI has introduced ‘Skild Brain,’ a foundational artificial intelligence model designed to operate on a wide range of robots, from industrial machines to humanoids. This model enables robots to perform complex, human-like tasks and is trained using a combination of simulated episodes, human-action videos, and real-world data from robots running the system. The company recently secured $300 million in Series A funding, valuing it at $1.5 billion.
Skild AI, a robotics startup with significant backing from Amazon and Japan’s SoftBank Group, announced on Tuesday, July 29, 2025, the launch of ‘Skild Brain,’ a groundbreaking general-purpose artificial intelligence model. This innovative AI is engineered to be compatible with nearly any type of robot, spanning from traditional assembly-line machinery to advanced humanoids.
The ‘Skild Brain’ model empowers robots with enhanced cognitive and motor capabilities, allowing them to ‘think, navigate, and respond more like humans.’ This development marks a pivotal step in the broader industry push towards creating versatile humanoid robots capable of executing a diverse array of tasks, moving beyond the single-purpose machines commonly found in current factory environments.
Demonstration videos showcased robots powered by Skild Brain performing impressive feats, including climbing stairs, maintaining balance even after being shoved, and adeptly picking up objects in cluttered surroundings. A key feature highlighted by the company is the integration of built-in power limits, ensuring that robots do not exert unsafe levels of force during interactions.
Addressing a significant challenge in robotics—the scarcity of relevant training data—Skild AI’s co-founders, Deepak Pathak and Abhinav Gupta, explained their unique approach. Unlike language or vision AI, where vast datasets are readily available online, robotics lacks such extensive public data. To overcome this, Skild Brain is trained using a hybrid methodology: simulated episodes, human-action videos, and crucially, real-world performance data collected from every robot running the Skild system. Gupta described this as creating a ‘shared brain’ across all robots utilizing the system, allowing for continuous improvement and rapid acquisition of new skills across various industries.
Also Read:
- Robots Take Center Stage at China’s Premier AI Summit in Shanghai
- Micro1 Secures $500 Million Valuation in Series A Funding Amidst Surging AI Data Labeling Demand
Skild AI recently completed a Series A funding round, raising $300 million, which propelled the company’s valuation to $1.5 billion. Notable investors in this round include Menlo Ventures, Khosla Ventures, Sequoia Capital, Jeff Bezos, Amazon.com, and SoftBank Group. The company is already seeing adoption, with LG CNS, the IT solutions arm of LG Group, and other unnamed partners in logistics and industrial sectors leveraging this advanced technology.
