Beyond the Lego Era: Building Foundation Models for Physical Motion
The robotics industry is moving out of its 'Lego era' as foundation models like General Intuition’s begin to master spatial and temporal reasoning. Massive new investments are fueling the transition from basic actuators to generalized AI agents.
Robotics is currently undergoing what experts call its "Lego era"—a period characterized by the assembly of disparate components like actuators, sensors, and grippers, but lacking a unifying intelligence. That is changing rapidly. A new wave of startups, including General Intuition and Travis Kalanick’s Atoms, are raising billions to build the "foundation models" for motion, effectively giving robots a sense of space and time.
General Intuition, recently valued at $6 billion, is focused on training AI agents that don't just follow pre-programmed paths but understand the physics of their environment. Unlike Large Language Models (LLMs) that predict the next word, these robotic foundation models predict the next physical movement. This allows a robot to generalize tasks: if it learns how to open one type of door, it can apply that logic to any door, regardless of the handle's shape or the door's weight.
Meanwhile, companies like Atoms are looking to modernize the world’s industrial base through large-scale AI deployment. The infusion of capital—including a massive $1.7 billion round for Atoms led by a16z—signals that the market believes we are nearing a "ChatGPT moment" for robotics. The goal is to move away from single-use machines and toward generalized platforms capable of tidying a home, working a construction site, or managing a warehouse with minimal human intervention.
Source: TechCrunch
