Skild AI hits $100M revenue run rate 10 months after launch

Skild AI told Bloomberg it has crossed a $100 million annual revenue run rate ten months after its first commercial deployment, with its software now running on hundreds of robots at more than 60 companies, up from eight earlier this year.

The Pittsburgh startup, founded in 2023 by Carnegie Mellon Robotics Institute professors Deepak Pathak and Abhinav Gupta, is selling S1, a robotic foundation model it introduced in an August 18 research post as an in-context learner: an operator records a single egocentric video of a task and hands it to the robot as a prompt, without any weight updates or fine-tuning. According to Skild's own numbers, S1 hits 96% success on tasks seen in training and 66% on entirely unseen ones, versus 9% for a language-prompted baseline. "Learning by experience, and not preprogramming, is the step change that has happened in robotics," Pathak says.

Deployments span cafes in Japan to warehouses in the U.S., and Skild is working with NVIDIA and Foxconn to run the Skild Brain on dual-arm manipulators assembling NVIDIA's Blackwell systems. The commercial ramp lands eight months after a $1.4 billion round led by SoftBank with NVentures, Bezos Expeditions, Samsung, LG and others, which valued Skild above $14 billion, more than triple its price seven months earlier.

The $100M figure is a run-rate the company disclosed to a single outlet, not audited ARR, and Skild has not broken out how much of it is software licensing versus integration work. It arrives in a week that already brought a $100M Series A for palletizing-robot startup Maven and fresh funding across the robotics beat we track, a reminder that capital into physical-AI platforms is arriving faster than production data on how well they generalize.