Physical AI teams burn 3–6 months and up to $150K just setting up hardware before training begins. Anvil Robotics wants to kill that bottleneck. The company raised $6.5M (led by Matter Venture Partners) and has shipped 100+ OpenARM units to NVIDIA, Path Robotics, Intrinsic AI (Google), and startups across the US and Southeast Asia.
What $5,000 Gets You
OpenARM is a bimanual manipulator — two 7-DOF arms, CNC-machined and sheet-metal built, 633mm reach per arm, 1500mm gripper-to-gripper wingspan, 4.1kg payload held for 60+ seconds. Arm length, payload, and actuators are all swappable. Four 60fps cameras included. Crunchbase called it “Legos for robots” — the modularity is the point.
Full Open-Source Stack With ROS2 and Teleoperation SDK
Everything ships open-source on GitHub. The software stack: 500Hz ROS2 controllers, 7-DOF inverse kinematics solver, a teleoperation SDK with headless Quest APK, and data pipelines that export directly to MCAP and LeRobot formats for model training. Developers can control arm movements, read sensor data, and record demonstration trajectories through the API. Plugs into ROS2, FoxGlove, and LeRobot out of the box.
Founder Mike Xia pitched Anvil as “the AWS for robotics.” At $5K per kit with a fully open stack, 100+ units in the field, and NVIDIA on the customer list — that framing is hard to argue with.
You Might Also Like
- Rivians Founder Raised 500m to Build Factory Robots not Humanoids Mind Robotics Explained
- Nvidia Dreamdojo Teaching Robots to Think by Watching 44000 Hours of us
- Ricursive Intelligence Just Raised 335m in 4 Months and i Think Nvidia Should be Paying Attention
- Nvidia Personaplex Just Made Every Voice ai Pipeline Feel Obsolete
- Ant Group Alipay ai pay aq Health Just hit 100 Million Users Each and Thats a big Deal
Top AI Product 
Leave a comment