Path Robotics

Two brothers who learned to weld with their father in an Ohio garage went on to earn PhDs in AI-related fields and then set out to build technology that makes physical things rather than software that stays on a screen. That is how Path Robotics describes its own origin, and the company now sells robotic welding cells whose behavior comes from a purpose-built welding AI. Its pitch rests on a labor argument: it points to a projected shortage of skilled welders in American manufacturing by 2029, an average welder age of 55 and annual turnover of about 30 percent.

A welding model that learned from its own fleet

Path's AI stack has two layers. The Weld World Model is a neural network trained on multimodal weld data gathered over eight years from deployed machines and an internal data farm, which the company puts at tens of millions of welded inches. Obsidian is a reinforcement learning agent trained inside that model and then run at the edge, where it reads sensor data and decides seam by seam how to weld. Data from before, during and after each weld flows back to improve the model.

Sensors built to survive the arc

The sensor stack is proprietary. Its current generation uses two cameras and four lasers to give 360 degree awareness around the torch and produces sub-millimeter point clouds. A neural network filters out false reflections as the weld happens, so the cell can see imperfect parts and adjust rather than repeat a fixed path.

Cells for ships, poles and heavy equipment

The company positions its adaptive and mobile welding cells for high-mix work, large parts and variable fit-up. It names four industry groups: defense (hulls, armored panels, vehicle frames), utilities and energy (poles, substations, battery storage), data centers (structural frames, e-houses, cooling units, server racks) and heavy industry (mining equipment, trailers, loader arms and buckets). Its news feed reports a $600M production agreement with HII for Navy shipbuilding capacity, signed in August 2026, and a $300M Series E led by Georgian at a $2.45B valuation on November 10, 2026. Its headline claims, marked as estimates, are four times the productivity, 30 percent lower cost, no capital expenditure and round-the-clock remote support.

Key facts

People: Andy Lonsberry (co-founder and CEO), Alex Lonsberry (co-founder and CTO) Funding: $610M raised at a $2.45B valuation, including the $300M Series E announced November 10, 2026 Website: https://www.path-robotics.com