AI观察|“杀手级”应用继续缺失，AI大模型开始寻求硬件加持

Group 1 - The AI market has seen a decline in enthusiasm despite the continuous advancements in AI capabilities, with a consensus emerging that there is currently no "killer application" to drive significant investment returns [1] - The AI landscape is characterized by a "winner-takes-all" dynamic, where smaller AI startups face challenges as larger models like GPT-3 and deepSeek-R1 quickly overshadow their innovations [2] - The importance of hardware in AI development is emphasized, with the need for startups to establish "dual moats" in both software and hardware to remain competitive [3] Group 2 - The competition in the AI industry has shifted from solely algorithmic battles to a deeper engagement in software and hardware collaboration, indicating a need for AI to integrate into real-world applications for sustainable growth [4] - Significant investments are being made in AI hardware, as seen with OpenAI's acquisition of a hardware design company for $6.5 billion and Meta's collaboration with Oakley to develop AI-powered smart glasses [3][4] - Domestic companies like Yushu Technology are also focusing on optimizing hardware and integrating advanced algorithms, securing substantial funding to enhance their capabilities in the AI sector [4]