NEURA Robotics

Robotics · AI · Humanoid Robots

Launch

NEURA Robotics launches Neuron AI, a multisensory foundation model for robots

The model already works across NEURA's full robot lineup, letting customers teach a skill once and deploy it across entire fleets.

• NEURA Robotics unveiled Neuron AI on October 14, 2026 in Metzingen, Germany, a multimodal foundation model the company calls Physical AI for its robots.
• Founded in 2019, NEURA builds its own robots, hands and sensors alongside the AI, and raised a €120M Series B in 2025 followed by a Series C of up to €1.2 billion in June 2026.
• Neuron AI's Perception-Language-Action architecture fuses 3D vision, touch, audio, robot state and language with memory, so a robot can tell a lock is truly open by sound and resistance alone.
• The model is trained on real robot data from NEURA's training halls, simulation-generated variations, and human movements captured via sensor suits, then runs across five robot form factors.
• At NVIDIA's GTC conference in Berlin next week, NEURA's 4NE1 humanoid will perform an eleven-minute laundry task fully autonomously to demonstrate the model live.
• The launch caps a wave of 2026 European physical-AI funding, including Sereact (€93M), RobCo (€89.2M), Humanoid (€133M), THEKER (€73M) and Trener Robotics (€26M), roughly €414M combined outside NEURA's own raise.
• Neuron AI is a bet that purpose-built, multisensory models can outperform the vision-language systems adapted from text-based AI that currently dominate robotics research in the US and China.