Core Viewpoint - The article emphasizes the critical role of material innovation in driving the next generation of AI computing power, highlighting the shift from traditional silicon-based materials to advanced materials that can support higher performance and efficiency in AI applications [2][52]. Group 1: Core Computing and Logic Chip Materials - Advanced channel materials are essential for semiconductor transistors, directly influencing the speed, power consumption, and integration of chips [4]. - AI chips require channel materials with high mobility, high switching ratio, high stability, low power consumption, low leakage current, and ultra-thin thickness [6]. - Various materials such as MoS₂, black phosphorus, InGaAs, germanium, carbon nanotubes, and high-mobility oxide semiconductors are being explored to meet these requirements [6][15]. Group 2: Gate and Dielectric Materials - Gate and dielectric materials are crucial for controlling the conduction of channel carriers and directly affect the switching speed, power consumption, and reliability of AI chips [16]. - HfO₂ and its doped variants are highlighted for their low leakage current and high dielectric constant, making them suitable for advanced logic chips [17][18]. Group 3: Substrate Materials - Substrate materials provide physical support and thermal management for semiconductor chips, impacting the performance and reliability of AI chips [21]. - Materials like silicon carbide (SiC) and gallium oxide (β-Ga₂O₃) are noted for their superior thermal conductivity and breakdown voltage, making them suitable for AI power modules [22][23]. Group 4: New Storage and Computing Materials - Non-volatile storage materials such as phase change materials and resistive switching materials are essential for AI applications, offering high speed and low power consumption [25]. - Neuromorphic computing materials, including memristor materials, are being developed to simulate neural networks and reduce inference energy consumption [26]. Group 5: Advanced Packaging and Integration Materials - Substrate and interconnect materials are critical for enhancing signal transmission speed and reducing power loss in AI chip packaging [29]. - Thermal management materials, such as diamond composites, are essential for effective heat dissipation in high-performance AI devices [30]. Group 6: New Computing Paradigm Hardware Materials - Photonic computing materials, like lithium niobate, offer significant advantages in speed and energy efficiency, positioning them as key technologies for future computing [34]. - Quantum computing materials, including superconductors and diamond nitrogen-vacancy centers, are crucial for developing quantum computing hardware [38]. Group 7: Perception, Sensing, and Connectivity Materials - Intelligent sensing materials are vital for AI sensors, enhancing detection accuracy and response speed [40]. - Wireless communication materials, such as high-frequency low-loss PCB materials, are essential for 5G and AI applications [43]. Group 8: Energy and Thermal Management Materials - Active thermal management materials can dynamically adjust thermal properties, improving the efficiency of AI devices [44]. - Energy materials, including GaN and SiC power devices, are critical for enhancing the efficiency of AI server power supplies [46]. Group 9: Investment Logic Analysis - Investment opportunities in AI materials focus on leveraging material innovation to surpass traditional silicon limitations, aligning with national strategies for semiconductor supply chain security [52]. - Key investment areas include advanced logic and storage materials, packaging and thermal management materials, and frontier materials for emerging computing paradigms [53]. Group 10: Summary - The article provides a comprehensive overview of the material innovations driving the AI computing revolution, emphasizing the importance of these advancements in achieving self-sufficiency in semiconductor technology and reshaping global competition [55].