Skild AI raises $1.4B Series C as valuation reaches $14 billion
The Pittsburgh robotics software company, founded in 2023 by two former Carnegie Mellon professors, has grown from a $1.5 billion valuation to $14 billion in 18 months on a hardware-agnostic model that claims to run any robot without prior knowledge of its body form.
Skild AI announced a $1.4 billion Series C round on January 14, 2026, led by SoftBank Group, with participation from NVentures, NVIDIA's venture capital arm, entities administered by Macquarie Capital, Bezos Expeditions, Samsung, LG Technology Ventures, Schneider Electric, CommonSpirit Health, and Salesforce Ventures. The round values Skild at more than $14 billion and brings total funding to over $1.8 billion since the company's founding in 2023.
The founders
Skild was co-founded by Deepak Pathak, CEO, and Abhinav Gupta, President, both former professors at Carnegie Mellon University's Robotics Institute, among the most published research groups in robot learning and computer vision globally. Pathak's academic work includes foundational contributions to curiosity-driven exploration, a method for robots to learn from their environment without labelled data by treating novelty as a reward signal. Gupta's work spans large-scale visual learning and physical reasoning. The two did not start Skild from a product or commercial gap; they started from a research conviction that a unified, hardware-agnostic brain was the right architectural bet, and structured the company around it. Skild is headquartered in Pittsburgh, where both founders built their academic careers, with additional offices in San Francisco and Bengaluru.
The thesis
Skild is not building robots. Its stated strategy is to build the brain that operates robots made by others, positioning Skild Brain as a platform that hardware companies integrate rather than a proprietary stack tied to one body. Skild Brain is described as an omni-bodied foundation model; it can control any robot without prior knowledge of its body form, running across quadrupeds, humanoids, tabletop arms, and mobile manipulators on a single model. The platform approach offers different economics from vertically integrated humanoid companies; if the model generalises broadly, Skild captures value across every hardware platform rather than betting on one form factor. The risk is that generalisation at commercial scale across diverse hardware has not yet been demonstrated fully.
The core data problem Skild addresses is the absence of an internet of robotics. Unlike language models trained on text scraped from the web, robot models have no equivalent corpus. Skild trains on internet videos of humans performing tasks and on physics-based simulation across thousands of robot form factors, treating human behaviour as a proxy for the physical intelligence it wants to transfer to machines. Post-training data comes from live deployments, which continuously feed the model and compound its capability across the fleet.
The funding arc
Skild raised $300 million in a Series A in July 2024 at a $1.5 billion valuation, approximately $500 million in a Series B in mid-2025 at a $4.5 billion valuation led by SoftBank, and the January 2026 Series C at $14 billion. That is a roughly ninefold increase in valuation in 18 months, driven entirely by a software product with no proprietary hardware. Lightspeed, Felicis, Coatue, and Sequoia Capital expanded their positions in the Series C. SoftBank has now led two consecutive rounds. AI
Revenue and deployments
Skild grew from zero to approximately $30 million in revenue in a few months in 2025 across deployments in security and facility inspection, last-mile and point-to-point delivery, warehouses, manufacturing, data centres, and construction environments. LG CNS, the IT arm of LG Group, has partnered with Skild for industrial applications. The company has not disclosed named robot hardware partners beyond its general hardware-agnostic positioning or operational data from deployments at scale.
Maturity
$30 million in revenue from a standing start in one year is a real commercial signal, and the flywheel logic, more deployments generating more data generating a better model, is architecturally sound if the generalisation holds. The $14 billion valuation prices significant future adoption that has not yet been demonstrated. No hardware partner has been named at commercial deployment scale, and the claim that a single model can control any robot for any task remains a research proposition as much as a commercial one at this stage.
Have a robotics update Korthos should review? Send news, deployments, product releases, funding rounds, research, or media to tips@korthos.xyz or reach out on X at @agkorthos.