Learn about memory, storage, and energy requirements for On-Device AI inference services in this 24-minute conference talk from SK AI SUMMIT 2024. Explore the on-device AI software ecosystem and latest model trends while examining memory requirements through shared evaluation results of multimodal model inference on smartphones and laptops. Gain key insights into the performance, memory demands, storage requirements, and energy consumption across different computing units (CPU, GPU, NPU) for On-Device AI applications. Presented by Kyungsoo Lee from SK Hynix, a system application technologist with extensive experience in embedded/Linux development, who provides comprehensive insights and vision for SKH memory from a customer perspective while researching AI platform software with his team.