Skild AI Inc., a fast-rising startup that makes software to help robots learn to complete tasks, has secured about $1.4 billion in a new funding round that values the company at more than $14 billion, more than triple what it was worth just seven months ago.
The Series C round was led by SoftBank Group Corp., with participation from Nvidia Corp., Macquarie Group Ltd., 1789 Capital and Jeff Bezos’ private investment firm Bezos Expeditions, co-founder and Chief Executive Officer Deepak Pathak told Bloomberg.
Several previous backers, including Lightspeed Venture Partners, Felicis Ventures, Sequoia Capital and Coatue Management, also joined the round. Other investors included Samsung Electronics Co., LG Electronics Inc. and Salesforce Inc. To date, Pittsburgh, Pennsylvania-based Skild has raised about $2 billion, Pathak said. Last June, it was valued at around $4.5 billion, Bloomberg reported.
Founded in 2023, Skild is part of a growing crop of startups that are drawing interest from top investors who are enthusiastic about AI’s potential to improve robotics. Although several robotics firms are focused on specific task areas, Skild is pursuing an AI-powered robotic “brain.” The goal is not to optimize for a single workflow, but to build a system that can adapt across environments and tasks, by training its systems much like humans learn: watching and practicing.
“There is no ‘Internet of Robots,’” Pathak said in an interview. “You cannot produce a brain for robots without having data, so this is where we’ve had a main focus from the very beginning, to build a general brain for robotics — one brain for any robot, any task, any scenario.”
The Skild Brain software can be loaded onto any standard graphics processing unit without any need for custom architecture, according to Pathak. The software is trained on large libraries of human videos and practicing simulations. Actions and mistakes generate new data in the real world, creating a feedback loop that improves performance. The system blends internal signals like motion around the joints and force with outside perception, such as vision, to enable the robot to understand itself and its environment at the same time, the company said.
“If one of the arms breaks down, it still continues to do the task,” said Abhinav Gupta, Skild’s president and co-founder, about the adaptability of the technology. “If one of the legs breaks down, it doesn’t fall down. It gives it a safety perspective that these robots never had before.”
The new funding will go toward expanding availability of the Skild Brain, improving training and deploying more robots across more environments, Pathak said.
The software is compatible with various robot types, from quadrupeds to robotic arms and hands, in settings such as hospitals, homes, warehouses and construction sites. It’s also designed to help humanoid robots better perform complex physical actions, from leaping over obstacles to grasping objects for tasks like loading the dishwasher, cleaning up and cooking eggs.
Skild said it went from zero to tens of millions of dollars in revenue over a few months in 2025. The startup is working with more than eight clients, according to a person familiar with the matter. The company declined to name its partners.
One of Skild’s customers dispatched a service robot at LaGuardia Airport in New York City to measure air quality, said the person, who declined to be identified discussing information that isn’t public. LG CNS Co. — which provides IT services in South Korea — has said it is partnering with Skild on humanoid solutions. (LG’s electronics division last week unveiled a household robot.)
Skild’s latest funding round and soaring valuation point to a larger pattern of major tech companies opting to ramp up investments in startups rather than make large acquisitions. Samsung views its investment in Skild as a way to keep visibility into the startup and its talent without a major commitment, Bloomberg has reported. And Nvidia has also backed other companies in the space, including Figure AI and Serve Robotics.
Skild is not the only upstart attempting to help make robots more intelligent and situationally aware. Physical Intelligence, a Samsung-backed startup last valued at $5.6 billion late last year, is operating in much the same lane.
Across the board, the industry is tackling some stubborn design and engineering obstacles, particularly for humanoid robots intended for consumer use. Battery life, mobility and safety continue to constrain large-scale deployment. Tesla Inc., for instance, has faced ongoing challenges in bringing its Optimus robot to market.
Craig Le Clair, a vice president and principal analyst at Forrester Research, said the robotics industry is “currently in the GPT-moment,” shifting from single task machines to general purpose robots.
“Just as large language models learned to speak by reading the internet, new vision-language-action models are learning to move by watching millions of hours of video,” he said. These so-called VLA models “allow for low-level motor skills — walking, folding laundry — to become easier,” he added.
Skild employs more than 100 people, many of whom have joined from companies including Meta Platforms Inc., Tesla, Nvidia, Amazon.com Inc. and Alphabet Inc.’s Google.
The idea for the startup began in early 2022, months before the release of ChatGPT. By then, the co-founders — who worked together at Meta’s robotics lab and later as professors from the Robotics Institute at Carnegie Mellon University — felt they had hit a ceiling in academia. Pathak’s lab at CMU was showing real-world demos at conferences but at the same time, computing power was limited and expensive to use at scale.
“We could see this kind of writing on the wall, specifically because robotics had always been about videos,” said Gupta, adding they believed it was the right time to launch a startup that could help with scaling. “People would post amazing videos of robots doing cute or hard things,” he said, but live presentations were rare. “Videos are easy to shoot because if it works one out of 10 times — you show that video and say this is how my thing works. But showing live demos in conferences is where things fail.”
Pathak and Gupta began speaking with executives and frontline workers at oil rigs, hospitals, factories and other demanding environments to understand where robotics consistently failed.
When the team purchased a commercially available humanoid robot from Unitree in February 2024, they installed their software and achieved meaningful results within a day, according to the founders. Skild emerged from stealth in the summer of 2024 and released public videos of its systems at work for the first time the following year. At Nvidia’s GTC keynote in October, CEO Jensen Huang announced that Skild’s software would be used for robotic automation at Nvidia’s upcoming GPU factory in Houston.
Skild also has aligned itself with a broader push for advanced automation in the US. The founders believe robotics will create more jobs down the line and fill more than 1 million available roles in the country in dangerous, repetitive or undesirable jobs.
“There will be a scenario where in the long, long future, robots can do almost any work that humans can do, but that’s a good thing,” Pathak said. “It is not a bad thing. We have time to prepare for this. It’s not happening overnight. It’s not happening in a few years.”
A message from Advisor Perspectives and VettaFi: Discover something new! Click here to register for our upcoming webcasts.
Bloomberg News provided this article. For more articles like this please visit
bloomberg.com.
Read more articles by Samantha Kelly
