Path Robotics—AI Welding Robots Focused on Automation for Small and Medium-sized Metal Fabrication Shops
Founded: Andrew Lonsberry, Alex Lonsberry · Path Robotics
Key Fields
FIELD STAMPSOrigin
Founders Andrew and Alex Lonsberry grew up in an Ohio manufacturing family, with both their grandfather and father working as welders. In 2018, while working on a project at the Carnegie Mellon University Robotics Institute, they discovered that the U.S. welding labor shortage exceeded 400,000, yet all existing welding robots required manual teaching, making them cost-prohibitive for small shops. The brothers decided to leave academia to found Path Robotics, with the goal of enabling robots to automatically adapt to weld seam deviations through visual perception, thereby lowering the cost of welding automation to a level affordable for small businesses.
Milestones
Turning Points
- Shifting from academic prototypes to small-factory scenarios was the key step in confirming that visual adaptive welding could generate a stronger willingness to pay than general-purpose robots.
- Focusing on agricultural machinery, truck frames, and structural steel—industries with high weld repeatability—transformed AI welding from a demo piece into a sellable product.
- Switching from equipment buyouts to a per-inch welding fee lowered the barrier for small shops and allowed recurring revenue to surpass one-time hardware sales.
- The failure of the mixed-line welding pilot forced the development of the real-time point cloud reconstruction module, paving the way for the Rove mobile welding system.
Failures & Pitfalls
- The first delivery was rejected by the customer due to failures in identifying reflective surfaces and irregular bevels, resulting in a $150,000 loss on the order.
- In 2021, due to insufficient service staff and delivery delays, 14 units experienced frequent downtime within three months, causing the NPS to drop into negative territory.
- In 2023, the mixed-line welding pilot at a truck frame factory achieved only 55% of human efficiency, leading the customer to cancel the project.
- In 2025, rising hardware costs and longer sales cycles put pressure on cash flow, forcing a 15% workforce reduction and a halt to regional expansion.
关键成功要素
- Replacing one-time teaching with continuous online learning to allow robots to adapt to the actual weld seam deviations of each workpiece.
- Narrowing the target market to three scenarios with high weld repeatability: agricultural machinery, truck frames, and structural steel.
- Replacing full-unit buyouts with a per-inch welding fee to lower the decision-making barrier for small shops and build recurring revenue.
- Integrating vision, welding torches, and robotic arms onto a mobile base to cover multi-station, small-batch welding needs.
- Self-developing edge computing units for real-time weld tracking to avoid reliance on public cloud networks.
Lessons
- The difficulty of AI welding lies not in lab-based weld quality, but in perception collapse caused by on-site workpiece tolerances, reflections, and irregular bevels.
- Small shops aren't buying robots; they are buying finished welds. The billing model must align with the customer's value metrics.
- Cash flow crises in hardware startups often stem from uncontrolled service radii and delivery costs, rather than a lack of orders.
- Building a national sales network before achieving PMF in a narrow scenario only amplifies the collapse of delivery and service quality.
- Failures in mixed-line automation pilots can expose fundamental flaws in perception architecture, which is more valuable than continuing to optimize individual algorithms.
Core Data
- 2020 Revenue:$3 million (Company disclosed, as of 2026, not independently verified)
- 2021 Revenue:$12 million (Company disclosed, as of 2026, not independently verified)
- 2024 Revenue:$24 million (Company disclosed, as of 2026, not independently verified)
- 2021 Series B Funding:$56 million (Company disclosed, as of 2026, not independently verified)
- 2025 Layoff Percentage:15% (Company disclosed, as of 2026, not independently verified)
- Price per Mobile Welding System:$220,000 (Company disclosed, as of 2026, not independently verified)
- Hardware Cost per Unit:$180,000 (Company disclosed, as of 2026, not independently verified)
- 2024 Mobile System Deployments:80 units (Company disclosed, as of 2026, not independently verified)
- Team Size:200 people (Company disclosed, as of 2026, not independently verified)
Competitors / Peers
Path Robotics' main competitors in the North American AI welding space include the mobile welding verification project from Boston Dynamics and Ferrari, the Canadian company Novarc Technologies which packages collaborative arms with weld tracking software, and the 'no-teach' welding function packages launched by traditional welding robot giants FANUC and Yaskawa. Compared to the fully automated production line solutions from FANUC and Yaskawa—which often cost hundreds of thousands of dollars and rely on system integrators—Path Robotics' per-inch billing model is better suited for small and medium-sized metal shops with fragmented orders, though its hardware margins and service cost control are significantly weaker than those of traditional giants.