Skild AI Launches S1 and Claims $100M by 2026

By the end of 2026, funding for embodied AI has accelerated, particularly for control software and humanoid robotics. In this context, Skild AI has unveiled S1, a contextual learning model, claiming the ability to learn from a single example.

Key Facts: GPT-3, 175 Billion Parameters, and a Dated Publication

GPT-3, developed by OpenAI, has 175 billion parameters. Its scientific paper was published on May 20, 2020. An event presented as the moment ChatGPT occurred approximately 3.5 years after this publication.

By the End of 2026, Capital is Flowing into Embodied AI and Robotics

At the end of 2026, a sudden surge in funding was observed for physical AI and embodied AI. This movement particularly affects robotic brain software and humanoid robotics. However, it is not established whether this momentum is significant or sustainable. This push comes after a series of IPOs from Chinese robotics companies, as China is widely considered to be leading the sector in sales, hardware, and manufacturing.

Skild AI Unveils S1 and Claims One-Shot Learning

About three weeks ago, Skild AI launched S1, a contextual learning model dedicated to robotics. The company reports exceeding $100 million in revenue in 2026. It presents S1 as a foundational model capable of learning from a single example. According to Skild AI, the system can be trained on previously unseen tasks in just 10 minutes, using only a single video and without any adjustments.