Skild AI Raises $1.4B for Omni-Bodied Skild Brain Robotics AI

Skild AI Funding Signals Scale Ambitions for Robotics Foundation Models

Skild AI announced its Series C financing of nearly $1.4 billion on January 15, 2026, led by SoftBank with participation from NVentures, Bezos Expeditions, and existing backers including Lightspeed and Sequoia. The round tripled the company's post-money valuation past $14 billion. Founded in 2023, Skild AI had previously raised $300 million in 2024 and reached roughly $30 million in revenue within months during 2025. The capital will accelerate training of the Skild Brain and expand deployments across enterprise use cases before consumer applications.

Investors see the omni-bodied model as a route to faster iteration than hardware-specific approaches. SoftBank managing partner Dennis Chang highlighted the shared vision for physical AI across robots, tasks, and environments. Additional strategic participants include Samsung, LG, Schneider Electric, and Salesforce Ventures. The round positions Skild AI among the highest-valued pure-play robotics AI companies at this stage.

Skild Brain Architecture Targets Cross-Embodiment Generalization

The Skild Brain uses hierarchical policies trained across thousands of simulated robot instances to achieve generalization without embodiment-specific fine-tuning. Pre-training draws from human demonstration videos scraped from the internet and large-scale physics-based simulation rather than real-world robot fleets. This approach addresses the absence of an “internet of robotics” data source that language and vision models enjoy.

In-context learning allows the model to adapt behavior in real time when introduced to new bodies or when hardware changes occur mid-task, such as loss of a limb or increased payload. Skild AI claims the system can control quadrupeds, humanoids, arms, and mobile manipulators without prior knowledge of their kinematics. Co-founder and CEO Deepak Pathak emphasized that the model is forced to adapt rather than memorize, mirroring biological intelligence.

NVIDIA partnerships provide GPU resources for accelerated training at scale. The company maintains offices in Pittsburgh, the San Francisco Bay Area, and Bengaluru to support model development and customer integration.

Current Deployments Span Inspection, Logistics, and Manufacturing

Skild AI has deployed its technology in security and facility inspection, last-mile delivery, warehouses, data centers, and construction sites. Mobile manipulation platforms handle grasping, handover, and navigation tasks exposed through APIs so integrators can focus on application logic. Autonomous packing demonstrations show precise, dexterous skills learned from video and refined in simulation.

Enterprise tasks serve as the initial revenue driver while the company scales toward home deployments. Revenue growth from zero to $30 million in a short period during 2025 indicates early commercial traction in unstructured environments where traditional scripted robots struggle.

Competitive Landscape and Investor Implications

Other groups pursuing general-purpose robotics AI include Qualcomm, Sanctuary AI, and X Square Robot. Skild AI differentiates through its explicit omni-bodied thesis and emphasis on in-context adaptation without retraining. The valuation reflects investor conviction that a single foundation model can serve multiple hardware platforms more efficiently than siloed development.

For fleet operators and integrators, the model promises reduced per-robot engineering costs and faster rollout across mixed hardware fleets. Continued scaling of simulation and video pre-training will determine whether real-world performance matches the claims made at the January 2026 announcement.

Path to Broader Adoption

Skild AI plans to use the new capital to expand model training capacity and grow live deployments. The company positions its technology as foundational for re-inventing American manufacturing through automation. Success will hinge on proving that hierarchical policies trained in simulation transfer reliably to physical robots across morphologies while maintaining safety in unstructured settings.

Longer term, the omni-bodied approach could accelerate the data flywheel where every deployment improves the shared model regardless of the underlying hardware. Investors will watch execution on the announced enterprise use cases as a leading indicator of consumer readiness.