James O'Donnell

Robotics funding surged to $6.1 billion for humanoid robots in 2025, quadrupling from 2024, driven by a fundamental shift in how machines learn. Rather than hand-coding rules for every scenario, researchers now train robots using large language models and other AI systems that learn from vast amounts of data, sensor readings, and trial-and-error in simulated environments. This approach, catalyzed by ChatGPT's success in 2022, has made the long-standing dream of adaptable, helpful robots seem achievable again after decades of incremental progress on narrow tasks.