Skild AI Launches Foundation Model That Could Redefine Robotics
images: Skild.Ai
Kit Mueller, Pittsburgh Startup News
Today, Skild AI unveiled a foundational artificial intelligence model designed to run on nearly any robot, from assembly-line machines to humanoids. The breakthrough represents what industry observers are calling the most significant advancement in robotics AI since programmable automation.
The model, called Skild Brain, enables robots to perform tasks requiring spatial reasoning and real-time adaptation that have long eluded robotics engineers. In demonstrations, Skild-powered robots climbed stairs, maintained balance after physical impacts, and manipulated cluttered objects with human-like adaptability.
Unlike existing systems designed for specific tasks, Skild's foundation model works across radically different hardware platforms, offering the first credible path to general-purpose robotic intelligence at commercial scale.
"We're developing generally intelligent decision-makers that surpass today's digitally focused AI by applying those same principles to the physical world," said Deepak Pathak, Skild's CEO and cofounder.
The technology addresses robotics' fundamental data problem. While language models benefit from internet-scale text data, physical AI requires action data that doesn't exist at scale. Skild's solution combines massive simulation with crowd-sourced learning from deployed robots—each machine feeds experience back to improve the shared intelligence. The company claims its model is trained on 1,000 times more data points than competing approaches.
Founded in 2023 by former Carnegie Mellon professors Pathak and Abhinav Gupta, Skild raised $300Mil at a $1.5Bil valuation in July 2024. The round was led by Lightspeed Venture Partners, Coatue, SoftBank Group, and Jeff Bezos through Bezos Expeditions, with participation from Sequoia, Amazon's venture funds, and others. Bloomberg and Financial Times report the company is now in talks for additional funding at a $4Bil+ valuation.
The rapid valuation increase reflects both technical breakthrough and competitive intensity. The robotics AI space has seen massive capital deployment: Figure AI raised $675Mil at $2.6Mil, Physical Intelligence secured $400Mil at $2Bil, and Tesla is developing humanoid factory automation.
Skild is pursuing a fundamentally different strategy. While competitors like Boston Dynamics optimize integrated hardware-software systems and Figure AI builds complete humanoid robots, Skild is betting on becoming the software platform that powers robots made by others.
"Rather than building the robot, we're building the brain that can go into any robot," said Abhinav Gupta, Skild's president.
This platform approach offers potential advantages: faster scaling without manufacturing constraints, broader market applicability, and operating system-like economics if widely adopted. The risks include proving generalization across diverse hardware platforms and competing against vertically integrated solutions.
Early commercial traction suggests the strategy may be working. LG CNS, the IT arm of South Korea's LG Group, has partnered with Skild for industrial applications. Other logistics and industrial partners are running pilot programs, according to the company.
The market opportunity is substantial. More than 1.7 million jobs are available than there are unemployed workers, according to the U.S. Chamber of Commerce. Manufacturing faces particular challenges, with experts projecting 2.1 million unfilled jobs by 2030, according to the National Association of Manufacturers.
"A GPT-3 moment is coming to the world of robotics," said Stephanie Zhan, a partner at Sequoia Capital. "It will spark a monumental shift that brings advancements similar to what we've seen in the world of digital intelligence, to the physical world."
The founders' academic credentials reinforce investor confidence. Pathak and Gupta have a combined 25 years of experience in robotics and AI, over 90,000 citations, and have been credited for breakthroughs in self-supervised robotics and adaptive robot learning. The team includes engineers from Meta, Tesla, Nvidia, Amazon, and Google.
Critical to widespread adoption, Skild Brain includes built-in safety constraints that limit robot force output when interacting with humans—addressing one of the primary barriers to robot deployment in human environments.
Whether Skild achieves the "ChatGPT moment" for robotics depends on execution at scale. The robotics field has numerous well-funded companies that failed to translate laboratory success into commercial products. But the combination of proven AI talent, substantial funding, early commercial partnerships, and platform strategy positions Skild to capture significant value if general-purpose robotics reaches widespread deployment.
The company's trajectory—from academic research to potential $4 billion valuation in two years—demonstrates how quickly foundational AI research can translate to commercial scale when properly capitalized and executed.
Skild's launch may mark the beginning of foundational infrastructure for the physical economy, with implications comparable to cloud computing and mobile platforms for digital services.