Runway co-founder and CEO Cristóbal Valenzuela argues that scaling next-frame prediction turns video models into general-purpose world simulators that implicitly learn physics, object permanence, gravity, occlusion, and interactivity from pixels alone, and that these will soon stream as a real-time "operating system" he calls Continuum. He says scaled video models have passed the real-vs-fake Turing test, and he demos Continuum, a video-model-based operating system for interactive simulation, shipping in 8-12 months. The talk walks through the arc from Runway's 2019 film "The Orange" to today's photoreal generated video, the emerging properties that appear as models scale, and a live demo of Continuum as an interactive, resizable, agent-driven interface. Valenzuela's central technical claim is that the same simple idea cinema discovered about 120–130 years ago — stitch frames fast enough and you create the illusion of movement — taken to sufficient scale, produces something far beyond pretty video. He says Runway's evaluation benchmarks are based on physics, geometry, and how objects behave in simulated worlds, and that scaling laws believed to apply to language models also apply to video, with Runway ramping compute behind its models.