Walden Robotics, a Cambridge-based physical AI company, launched from stealth with $300 million in funding co-led by Toyota and Deviation Capital, valuing the startup at $1.1 billion. Founded in January 2026 by MIT professor Russ Tedrake and robotics veterans from Toyota Research Institute, Amazon, and Stanford, Walden builds general-purpose robots trained on large behavior models (LBMs)—AI for physical tasks analogous to LLMs for language. Within weeks of launch, robots were deployed in production at a Toyota factory in North America.
Walden's architecture differs from single-task industrial robots: its robots train on diffusion models adapted to physical control, allowing them to learn new tasks through real-world practice rather than weeks of reprogramming. By February 2026, only one month after spinning out, robots were performing dexterous, repetitive work—loading, unloading, kitting—on 8-hour production shifts. This speed-to-production is the key claim: other well-funded humanoid makers like Figure AI and Agility Robotics have not yet matched this deployment velocity from launch to live manufacturing.
Investors include NVIDIA, Boeing, Samsung Ventures, Prologis Ventures, CoreWeave, and financial backers Calibrate Ventures, NextView, Squarepoint Capital, and Menlo Ventures. Toyota's participation runs deep—three Toyota entities led the round, signaling internal commitment to the technology and data that flow from live deployment. CEO Tedrake emphasized kaizen and jidoka: continuous improvement with human oversight, not replacement.
For manufacturing automation teams, the signal is unambiguous: large behavior models are moving from research labs into factory floors. The $18.5 billion industrial robotics market (projected to hit $44 billion by 2030) is shifting toward AI-driven robots that learn on the job, and Walden's production proof-of-concept inside Toyota gives the company credibility most 2026 robotics startups don't yet have. The capital raise backs deployment, not just R&D.