Runway as robotics infrastructure supplier

This turns Runway from a single end market AI video vendor into an upstream infrastructure supplier for physical AI. Instead of earning mainly from editors and marketing teams buying seats and generation credits, Runway can sell a base robot model and a simulator to robotics teams that need cheaper training data, faster policy testing, and fewer costly runs on real hardware.1 2 3 4

• The product split matters. Praxis-1 is an open weight robot action model that developers adapt to their own arm or mobile robot. GWM-1 for Robotics is the sandbox where they test policies and generate synthetic scenes before touching a real machine, which makes the offering look more like model licensing plus infrastructure than creative software.2 3
• The customer base is also different. An early partner like Standard Bots sells industrial robots and no code control software, and reached $24M in annualized revenue in 2025. That shows Runway is not just chasing research demos, it is trying to plug into commercial robot builders that need better training loops right now.4 5
• The timing lines up with where value is moving in robotics. Funding has poured into companies attacking the data bottleneck, from Skild AI raising a $1.4B Series C in January 2026 to AMD agreeing to buy World Labs for about $8.2B in September 2026. That makes Runway's video to robotics jump look like entry into a hot infrastructure layer, not a side project.1 6 7

If Runway can prove that video pretraining and simulation transfer cleanly into real robot behavior, robotics could become its highest leverage business line. The likely path is deeper licensing into robot OEMs, more simulator based workflow tools, and a broader position as the software layer that helps physical AI teams train before they deploy.1 2 3 4