Problem

Most AI remains confined to cyberspace, while the physical world is information-rich and difficult to automate. Skild addresses the robotics industry’s fragmentation: robots are typically limited by their embodiment and by the narrow set of tasks for which they have been programmed. The economic opportunity is substantial because automating physical work can generate significant returns while reducing reliance on people for repetitive, hazardous, or constantly supervised labor.

The clearest near-term use case is security and inspection. Skild targets unstructured and dangerous environments where robots can conduct continuous inspections instead of requiring people to perform manual checks. Other applications include mobile manipulation and repetitive, dexterous packing tasks.

Product / Service

Skild is building the “Skild Brain,” a unified, omni-bodied intelligence layer intended to control any robot for any task. Its models learn low-level capabilities such as navigation, grasping, handover, and precise manipulation, then expose those capabilities through an API so application developers can build robot workflows without handling every detail of the physical environment themselves.

The company trains its robot foundation model with accelerated simulation and data-generation infrastructure, including NVIDIA Omniverse, Isaac Lab, and Cosmos. Skild says the model can adapt to new robot embodiments and skills with zero or minimal post-training; NVIDIA reports that it can reach 60–80% task performance within hours of data collection. The intended benefit is to make robotic deployments more general, faster to configure, and less dependent on bespoke programming for each robot and task.

Market

Skild competes in physical AI, embodied AI, and the emerging market for general-purpose robot foundation models. Its direct and adjacent competitors include Figure AI, Sanctuary AI, Physical Intelligence, Agibot, Boston Dynamics, and Apptronik. The competitive distinction is that Skild is pursuing a reusable intelligence layer across robot types and tasks, while several alternatives are more closely associated with a particular humanoid or robotic platform.

The company is well funded and appears to be moving beyond pure research: it announced a $300 million Series A in July 2024 and a $1.4 billion Series C in January 2026 at a valuation above $14 billion. The supplied evidence does not provide audited revenue, customer counts, or deployment volumes, so Skild cannot be definitively characterized as mature commercial software or as pre-revenue. A third-party estimate puts annual revenue at approximately $30 million, but that figure should be treated as indicative rather than verified.