Three Chinese Companies Join Forces to Target One Million Hours of Embodied-AI Data

QbitAI reports that three Chinese companies have formed a partnership targeting what it describes as the world’s first one-million-hour embodied-AI dataset. The proposed collaboration is intended to connect data collection, simulation, model training, and evaluation into a closed pipeline. However, the supplied headline and abstract do not identify the companies or provide a delivery schedule, dataset composition, quality criteria, benchmark results, or access terms. The one-million-hour figure should therefore be treated as a stated target, not a completed or independently verified achievement.