SoftBank, NVIDIA, HSG, Jeff Bezos, etc. participated in the investment. Skild AI, a robotics startup, raised $1.4 billion to develop a general foundation model.

The robotics startup Skild AI has completed a $1.4 billion Series C financing round, with a valuation exceeding $14 billion. This financing round was led by Japan's SoftBank Group, and strategic investors such as NVentures under Nvidia, Macquarie Capital, and Bezos Expeditions of Amazon founder Jeff Bezos participated. Samsung, LG, Schneider Electric, and Salesforce Ventures were also among the investors.

In mid-January 2026, the robotics startup Skild AI announced the completion of a Series C financing round of approximately $1.4 billion, and the company's valuation exceeded $14 billion. This financing round was led by Japan's SoftBank Group, and strategic investors such as NVentures under Nvidia, Macquarie Capital, and Bezos Expeditions of Amazon founder Jeff Bezos participated. Samsung, LG, Schneider Electric, and Salesforce Ventures were also among the investors.

For readers interested in this field, this list of investors may seem familiar. Many of these investors also bet on another star robotics startup, Field AI, not long ago. Field AI is committed to building a "general robotic intelligent brain" that can be applied across different types of robots and adapt to various environments. At the same time, Skild AI clearly claims to build an AI-driven robotic "brain," and the two seem to have similar strategic directions.

At a time when the hardware form of robots is not yet finalized and application scenarios are still highly fragmented, capital has repeatedly gathered around a few companies that do more than just build robots almost simultaneously. Under the iron law of capital chasing profits, it also confirms to some extent that this startup, which has been established for less than three years, has chosen a promising track.

Company website: https://www.skild.ai

Any robot. Any task. One brain.

"Any robot. Any task. One brain."

When you open Skild AI's official website, the first thing that catches your eye is this ambitious slogan. On the official X account and in an interview with NDTV that one of the founders, Abhinav Gupta, participated in a few days ago, this slogan, which is regarded as the company's tenet, was repeatedly mentioned: "Any robot. Any task. One brain." This accurately summarizes the unique features of Skild AI, which are very different from most robotics companies.

Source: Skild AI official website

Deepak Pathak said bluntly in an interview: "In the past 70 years, there have been many robot demonstrations, but no robots have really appeared around us. That's because robots lack a brain." In his view, the fundamental reason why robots have been difficult to be widely deployed for a long time is the lack of a truly general "intelligent brain."

Therefore, Skild AI's core goal is not to build a specific robot, but to develop a set of foundation models that can be deployed on various robots. Whether it is a humanoid robot, a quadruped robot, an industrial robotic arm, or a mobile platform, this system can work across tasks and environments, providing power for the all-round perception intelligence of robots. Its core value lies in providing a sustainable and scalable data solution, enabling robots to adapt to the physical world by observing and learning like humans.

This is a very interesting direction. As we all know, the success of large language models depends on the huge data on the Internet behind them. However, Deepak Pathak pointed out a key pain point: "Where is the Internet for robots?" In reality, there is no ready-made "robot Internet" that contains a large amount of physical interaction data. Therefore, their unique solution is to convert the endless human video data on the Internet into the experience of robots. "Humans learn by observing, and robots should do the same."

Source: Skild AI official X account

Two "mentor-style" founders: From academic research to industrial implementation

Another interesting part of Skild AI's story is its founding team.

The company was founded by Deepak Pathak and Abhinav Gupta, both of whom are senior researchers in the fields of artificial intelligence and robotics. Deepak Pathak is the current CEO and has deep experience in the cross - research of artificial intelligence and robotics. Abhinav Gupta serves as the company's president and is also a scholar with profound academic achievements in the fields of AI self - supervised learning and robotics learning. Both co - founders previously taught and conducted research at Carnegie Mellon University, one of the earliest institutions in the world to conduct in - depth research on the combination of robotics and AI.

Former Carnegie Mellon University professors Deepak Pathak (left) and Abhinav Gupta (right). Source: Forbes

The technical philosophy of the current CEO, Deepak Pathak, was formed when he was pursuing his doctorate at the University of California, Berkeley. According to Forbes, Pathak developed a method to drive robot learning by stimulating "curiosity" to encourage artificial intelligence to explore more scenarios. The relevant research was published in 2017 under the title "Curiosity - driven Exploration by Self - supervised Prediction" and has been cited nearly 4,000 times.

Source: Preprint platform arXiv

If Pathak solved the problem of "how robots can actively learn," then Abhinav Gupta brought in the gene of "large - scale learning." As a senior scholar in the fields of computer vision and robotics learning, Gupta has long been committed to researching how to train AI with a large amount of unlabeled video data. This complementarity forms Skild AI's technological moat. One enables robots to have the ability to explore the physical world independently through a curiosity mechanism, and the other endows robots with the general common sense to understand the world by processing Internet - scale visual data.

Source: Abhinav Gupta's personal homepage at Carnegie Mellon University

They decided to establish Skild AI in 2023 and quickly launched it. This is not a startup aimed at "making quick money," but rather an attempt to turn a long - term thinking and research direction into reality. In their view, the verticalization dilemma of traditional robots is becoming more and more obvious. They are all specially designed for specific tasks, and this approach is difficult to solve the general physical reasoning and reaction abilities required by robots in unknown environments. They hope to truly break the data barriers in the robotics field. This vision has also attracted a group of robotics and artificial intelligence experts from top institutions such as Meta, Tesla, Nvidia, Amazon, Google, and top universities like Carnegie Mellon University, Stanford University, the University of California, Berkeley, and the University of Illinois at Urbana - Champaign to join the company.

Source: Skild AI official website

Skild Brain brings the "foundation model" into the physical world

If the previous concepts solve the problem of "how robots should learn," then the core product, Skild Brain, answers another more engineering - oriented question: How can this learning method be truly deployed in real - world robot systems?

According to the description on Skild AI's official technical blog, Skild Brain is not a control model trained for a single task or a specific robot form, but is positioned as a set of general intelligent systems that can be deployed on different robot bodies. Skild Brain follows a hierarchical architecture. A low - frequency high - level action strategy is responsible for understanding the environmental semantics and planning goals, providing input for the high - frequency low - level action strategy. Its underlying control ability is achieved through fully end - to - end motion control, which is completely driven by online vision and proprioception, realizing a true physical interaction closed - loop.

Source: Skild AI official website

This architecture ultimately endows Skild Brain with three disruptive technological features:

* Omni - bodied ability: Traditional robot algorithms are often "special - purpose for special machines," while Skild Brain proves that the same pre - trained model can drive quadruped robots, biped robots, and even robotic arms at the same time. Through large - scale training on diversified robot form data, the system can extract general physical laws across hardware. This means that the model is no longer limited by specific motor torques or foot - end structures, but has a certain "general motion intuition."

* Learning by Watching: Skild AI bypasses expensive manual demonstrations and directly allows the model to draw nutrients from hundreds of millions of human activity videos on the Internet. This technology can convert visual signals into the physical experience of robots, enabling robots to establish common - sense cognition of the physical world by watching how humans open doors and cross obstacles, thus achieving extremely strong zero - shot generalization ability.

* One Policy, All Scenarios: In Skild AI's actual tests, robots equipped with this system showed extremely strong robustness. Whether it is a smooth floor in a laboratory, a warehouse full of sundries, or a wild forest full of rocks and snow, Skild Brain can adjust its posture in real - time with the same strategy. This adaptability to unknown environments is the key for robots to move out of the laboratory and into various industries.

Source: Skild AI official YouTube account

Conclusion

Skild AI did not choose the easiest path to be verified. Instead, it directly bet on the most difficult and long - term problem in the robotics field: "generality." At a stage when the hardware is not yet finalized and the application boundaries are still constantly changing, this choice is highly risky but also a vision for the future. What Skild AI is trying may be a necessary prerequisite for the next stage of robot development.

In addition, whether general robots will really come still needs to be verified by time. However, it is certain that the focus of the industry is changing. For a long time in the past, discussions about robots mainly revolved around specific forms, single scenarios, or local performance. Now, more and more capital, researchers, and startup teams are starting to focus on a more fundamental question: Do robots need a truly general and transferable intelligent base?

References:

1.https://www.bloomberg.com/news/articles/2026 - 01 - 14/robotics - startup - skild - valued - above - 14 - billion - after - softbank - led - funding - round?embedded - checkout = true

2.https://www.forbes.com/sites/rashishrivastava/2024/07/09/this - 15 - billion - ai - company - is - building - a - general - purpose - brain - for - robots

3.https://www.businesswire.com/news/home/20240709306400/en/Skild - AI - Raises - %24300M - Series - A - To - Build - A - Scalable - AI - Foundation - Model - For - Robotics

4.https://www.youtube.com/watch?v = yesita2zN5c

This article is from the WeChat official account "HyperAI Super Neural". Author: Pa Daxingxing. Republished by 36Kr with permission.