The Skild Brain — a unified robotics foundation model for any robot, any task
HQ: Pittsburgh, PA, United States | Founded: 2023 | Employees: Not disclosed (LinkedIn ~102k followers) | Stage: Late Stage VC (Series C, $14B+) | Website: https://www.skild.ai
Skild AI was founded in 2023 in Pittsburgh by Deepak Pathak and Abhinav Gupta, two of the most respected names in embodied AI research — both Carnegie Mellon professors and former Meta AI research leads whose prior work (including path to the 'Grounded Decoding' and world-model research lineages) shaped the field. Their founding bet: the next big AI platform isn't a robot company, it's a foundation model that works across all robots — the 'Android moment' for robotics, where hardware commoditizes and the model layer captures the value.\n\nThe company built the Skild Brain, an omni-bodied foundation model trained on massive, diverse robot data to control any robot for any task — manipulation, locomotion, and beyond — without robot-specific retraining. The approach attracted record capital: a $14.5M seed co-led by Lightspeed and Sequoia (2023), a $300M Series A in July 2024 (Lightspeed, Coatue, SoftBank, and Bezos Expeditions), and a $1.4B Series C announced in January 2026, led by SoftBank with NVIDIA participation, valuing the company at $14B+ — roughly tripling its valuation, and one of the largest rounds in robotics history.\n\nThe commercial traction backed the hype: Skild says it grew from zero to about $30 million in revenue 'in just a few months' in 2025 and is 'growing exponentially,' with deployments across warehouses, factories, and homes, and partnerships across the robotics ecosystem. With ~$2.1B+ raised, SoftBank and NVIDIA backing, and a platform strategy that touches every robot maker, Skild AI is the leading bet that a single 'brain' will power the physical-AI era.
By September 2026 Skild had crossed a $100M annualized revenue run-rate within ten months of its first commercial deployment, with 60+ paying customers across warehouses, factories, data centers, food preparation, and security/inspection; mobility accounts for ~10% of revenue and its Fetch-solutions business ~4%. Marquee deployments include NVIDIA and Foxconn's Blackwell GPU server-assembly lines in Houston, Sumitomo Wiring Systems (using the S1 model for wire-harness manufacturing), and Mitsui & Co. partnerships spanning general-purpose robots and commercial kitchens. The company acquired Zebra Technologies' robotics division (formerly Fetch Robotics) in April 2026, opened its first non-US office in Bengaluru in February 2026, and launched the S1 flagship model in August 2026.
Skild AI embodies the 'software eats hardware' thesis of physical AI: instead of building its own robot, it builds the Skild Brain — a foundation model that turns any robot into an intelligent one. The team (CMU professors Pathak and Gupta, with deep Meta AI research roots) convinced SoftBank, NVIDIA, Lightspeed, Coatue, and Bezos Expeditions to back them to the tune of ~$1.8B — a $300M Series A and a $1.4B Series C at a $14B+ valuation just 18 months after launch. Early revenue (~$30M in the first months of 2025, growing exponentially) and deployments across warehouses, factories, and homes validate the platform approach. The bull case is owning the 'Android of robotics' — the model layer every hardware maker uses; the bear case is that robot foundation models are a young, crowded field (Physical Intelligence, Figure, 1X, NVIDIA's GR00T) and the capital required to stay at the frontier is enormous.
Browse: Home | Blog | About | All Startups A-Z | View full profile on StartupWiki