Skild AI Raises $1.4B, Reaching $14B Valuation in Physical AI Bet
Skild AI raised nearly $1.4 billion in a funding round led by SoftBank, valuing the robotics AI company at more than $14 billion as it scales a unified foundation model for robots.
Skild AI has raised close to $1.4 billion in new funding, pushing the Pittsburgh-based robotics AI company to a valuation exceeding $14 billion. The round was led by SoftBank Group and included participation from NVIDIA Ventures, Macquarie Capital, Jeff Bezos, and a wide range of strategic and institutional investors, underscoring growing confidence in foundational AI models for the physical world.
The financing positions Skild AI as one of the most highly valued companies in the emerging Physical AI sector. Rather than building robots themselves, the company is focused on developing a general-purpose robotics foundation model, known as the Skild Brain, designed to operate across virtually any robotic body and task.
Building a General Brain for Robots
At the core of Skild AI’s strategy is the idea of an omni-bodied intelligence. Unlike traditional robotics systems that are tightly coupled to specific hardware, the Skild Brain is designed to control robots without prior knowledge of their exact physical form. The model can operate humanoids, quadrupeds, mobile manipulators, tabletop arms, and other machines that can move.
Skild AI says this approach allows robots to perform a wide range of activities, from everyday household tasks such as cleaning, loading a dishwasher, or cooking simple meals, to physically demanding work like navigating unstable terrain or handling heavy payloads. The company’s long-term vision is that any machine capable of motion could eventually be operated by the same underlying AI brain.
A central challenge in robotics AI is the lack of large-scale, standardized training data. Unlike language models, which can be trained on vast amounts of text, robots do not have an equivalent “internet of robotics.” Skild AI addresses this gap by pre-training its model on a combination of human video data and large-scale physics-based simulations, allowing it to learn general physical behavior without being tied to a single robot design.
Adaptation Without Retraining
One of Skild AI’s key technical claims is that the Skild Brain can adapt in real time to unexpected changes without retraining or fine-tuning. This includes scenarios such as damaged limbs, jammed wheels, increased payloads, or being deployed on an entirely new robotic body.
According to the company, this adaptability is driven by in-context learning. When the model encounters a new environment or embodiment where its initial actions fail, it adjusts its behavior based on live experience. Skild AI describes this as a major departure from conventional robotics approaches, which often require extensive retraining for each new scenario.
“The Skild Brain can control robots it has never trained on, adapting in real time to extreme changes in form or environments,” said Deepak Pathak, CEO and co-founder of Skild AI. He added that forcing the model to adapt rather than memorize is critical to building intelligence that works reliably in the real world.
Rapid Growth and Commercial Focus
Skild AI is also reporting rapid commercial traction. The company said it grew from zero to approximately $30 million in revenue within a few months in 2025 and is deploying its technology across multiple enterprise settings. Current use cases include security and facility inspection, warehouse operations, manufacturing, data centers, construction, and delivery tasks.
While consumer robotics remains a long-term goal, Skild AI is prioritizing enterprise and industrial deployments, where robots can be rolled out at scale and generate continuous data to improve the model. The company believes this creates a reinforcing data flywheel, allowing the Skild Brain to improve with every deployment regardless of hardware type or task.
Abhinav Gupta, co-founder and president of Skild AI, said this generality is essential for building intelligent systems that can operate safely and dynamically alongside humans. He described omni-bodied learning as a foundational requirement for bringing advanced AI into everyday physical environments.
Investors Signal Strategic Importance
The breadth of the investor group reflects both commercial and strategic interest in Physical AI. In addition to financial investors, the round included strategic backers such as Samsung, LG, Schneider, CommonSpirit, and Salesforce, pointing to potential applications across manufacturing, healthcare, and enterprise automation.
SoftBank Investment Advisers described Skild AI as foundational infrastructure for the future of robotics, while other investors emphasized the long-term economic and strategic significance of solving intelligence for the physical world.
Founded in 2023, Skild AI operates across Pittsburgh, the San Francisco Bay Area, and Bengaluru. With its latest funding, the company plans to scale training of its foundation model and expand deployments, aiming to establish a shared intelligence layer for robots across industries as Physical AI moves closer to mainstream adoption.
