Vention opens Physical AI Lab in Montreal

Vention's new lab is not just a research facility but a data engine. The company's existing deployment of over 28,000 machines and 6,000 factories provides a continuous stream of real-world industrial manipulation data, which CEO Etienne Lacroix notes was previously an underutilized asset. This data is crucial for post-training physical AI models, addressing the challenge that "Any physical AI foundation model is data-hungry."

The lab's focus on closing the loop between academic research and production feedback is central to its approach. With Dr. Jimmy Li, a McGIll researcher, leading the effort and Dr. Joelle Pineau as external advisor, Vention aims to accelerate the path from research results to factory-floor deployment. The company is already working with a large automotive OEM on complex tasks, and its GRIIP pipeline, launched in early 2026, is set to be open-sourced.

The 400% increase in physical AI revenue signals market traction, but the real test is whether Vention can make these capabilities reliable and economical for broad adoption. As foundation models mature, the complexity and cost of deploying robots are expected to drop, potentially widening the range of manufacturers that can automate. The lab's success will hinge on its ability to demonstrate that physical AI can handle the variability and cost constraints of real production lines, not just perform in a lab setting.