OpenAI has announced the discontinuation of public app and API access for Sora, the model that originally ignited the generative video boom. This decision, officially unveiled on March 24, 2026, signals a strategic reallocation of development resources toward fundamental research. Moving forward, Sora is set to evolve beyond a simple content creation tool to focus on "World Simulation"—the pursuit of learning and reproducing the physical laws of the real world.

This pivot is largely driven by the explosive demand for computing resources required for high-fidelity video generation. OpenAI has opted to prioritize its hardware toward robotics and building the intelligence necessary for AI to act autonomously in physical spaces. The goal is to move the technology from merely creating visually realistic videos to understanding the underlying mechanics of the world at a physics-based level.

In terms of service, older Sora features are already being retired as advanced generation capabilities are integrated into the ChatGPT ecosystem. In regions like the United States, this consolidation is already underway to streamline the user experience. However, the immediate result is a reduction in the environment where general users can freely generate video content.

As competitors like Google with Veo 2 and DeepMind with Genie 3 enter a new era of "simulation-generation" hybrids, OpenAI is doubling down on AI as a physics engine for Artificial General Intelligence (AGI). The industry is now closely watching how this strategic shift will impact the development of next-generation multimodal AI and autonomous robotics.

OpenAI has announced the discontinuation of public app and API access for Sora, the model that originally ignited the generative video boom. This decision, officially unveiled on March 24, 2026, signals a strategic reallocation of development resources toward fundamental research. Moving forward, Sora is set to evolve beyond a simple content creation tool to focus on "World Simulation"—the pursuit of learning and reproducing the physical laws of the real world.

This pivot is largely driven by the explosive demand for computing resources required for high-fidelity video generation. OpenAI has opted to prioritize its hardware toward robotics and building the intelligence necessary for AI to act autonomously in physical spaces. The goal is to move the technology from merely creating visually realistic videos to understanding the underlying mechanics of the world at a physics-based level.

In terms of service, older Sora features are already being retired as advanced generation capabilities are integrated into the ChatGPT ecosystem. In regions like the United States, this consolidation is already underway to streamline the user experience. However, the immediate result is a reduction in the environment where general users can freely generate video content.

As competitors like Google with Veo 2 and DeepMind with Genie 3 enter a new era of "simulation-generation" hybrids, OpenAI is doubling down on AI as a physics engine for Artificial General Intelligence (AGI). The industry is now closely watching how this strategic shift will impact the development of next-generation multimodal AI and autonomous robotics.