Robotics startup Skild AI secured $1.4 billion in new funding to build a通用智能 'brain' for diverse machines. The company's approach centers on a single AI model adaptable across various robot form factors. CEO Deepak Pathak, a Carnegie Mellon computer science professor, co-founded the Pittsburgh-based firm to move beyond single-task automation.

Pathak's vision tackles a core robotics challenge: most AI models are brittle and tied to specific hardware. By training a generalized intelligence through observation and simulated practice, Skild aims to create a scalable software layer. This mirrors strategies at firms like Google DeepMind, which also pursue foundational models for embodied AI. The funding signals strong investor appetite for platforms that can power factory arms, drones, and home bots with one core system.

The new capital will accelerate data collection and model training, critical hurdles for physical AI. Pathak recently detailed the learning methodology on Bloomberg Tech. The key question remains whether Skild's approach can deliver a truly versatile intelligence that generalizes safely to unpredictable real-world environments, a hurdle for the entire industry.