Alibaba has unveiled an upgraded open-source AI model, Wan2.2-S2V, enabling the creation of "film-quality avatars" that can speak, sing, and perform. This advancement incorporates 14 billion parameters in a Mixture-of-Experts architecture, generating synchronized videos at 720p from single images and audio. Following significant competition from DeepSeek, Alibaba plans to invest $53 billion in AI and cloud over three years to bolster its market position. However, recent reports indicate only 7% revenue growth amid fierce competition.