Baidu unveiled Apollo World, a world model built to train autonomous vehicles such as its robotaxis, at its AI Day Apollo event on Tuesday. According to Baidu, the model can generate about three years of real-world testing per day and was trained on 217.48 million miles of data from its Apollo Go robotaxi fleet. The model reflects the growing importance of physical AI for industries such as autonomous vehicles, but it also highlights one of the main challenges associated with world models: expense. Omdia analyst Lian Jye Su said autonomous driving is "the most reasonable application of world models because the automotive industry is one of the more mature ones," adding that gaming is another opportunity for vendors to democratize world model usage. The article notes other world model creators focused on autonomous driving, including Google DeepMind's Waymo World Model, built on the Genie 3 foundation model and converting video logs into 3D simulations; Waabi, which uses its proprietary world model to train its AI-first virtual trucking fleet; and Nvidia's Cosmos open source world model. Despite growing popularity, a major challenge is cost due to the massive compute power required for training and real-time operation, with most world models requiring one GPU per active user. Su said the models are worth the investment, citing strong foundational data and accurate prediction and generation.