
A robotics intelligence layer for metal fabrication
Report from 12 days ago
Forge Robotics builds a vision and control system that lets welding robot arms see real parts, find weld seams and features, and adapt the robot’s motion in real time instead of following a hard‑coded path. In practice, they mount a camera and compute on the robot, generate a sub‑millimeter 3D map, propose a weld path for operator approval in simulation, then execute while continuously correcting based on live vision YC page and company site. The company is in pilot/early‑customer stage, emphasizing modular cobot cells for faster installs; public materials claim higher first‑pass yield and less setup/teaching, but these are still being validated in customer deployments YC launch, Portershed interview, Forge site.
Top-down context:
The robotic welding market is roughly USD 7–10B today, which is the most directly relevant TAM for a vision/control supplier in welding cells; the broader global welding market is about USD 24–25B and underpins long‑term demand Fortune Business Insights – robotic welding, Fortune Business Insights – welding.
Bottom-up calculation:
There are ~4M industrial robots operating worldwide, with about 20.7% used for welding/soldering, implying ~0.8–0.9M welding robots that could be retrofitted or replaced over time IFR, WIPO application share. Dividing a representative USD ~7.4B robotic welding market by ~828k installed welding robots shows average spend per robot of ~USD 9k across hardware/software/services today (actual cell purchases vary widely) Fortune.
Assumptions: