启明创投发布AI十大展望：首批可担绩效“AI员工”将上岗

Core Insights - The 2025 World Artificial Intelligence Conference (WAIC) hosted by Qiming Venture Partners highlighted significant advancements in AI technology and applications, predicting a resonance cycle in entrepreneurship and investment in the AI sector over the next 12-24 months [1] Group 1: AI Models and Agents - Qiming Venture Partners predicts that the context window of 2 million tokens will become standard for top AI models within the next 12-24 months, driving the development of more sophisticated context engineering [3] - A general video model capable of generating, reasoning, and understanding tasks in video modalities is expected to emerge in the same timeframe, revolutionizing video content generation and interaction [3] - AI agents are anticipated to evolve from "tool assistance" to "task execution," with the first true "AI employees" entering enterprises to participate in core processes such as customer service, sales, operations, and R&D, thus shifting from cost tools to value creation [3][4] Group 2: AI Infrastructure - In the AI chip sector, more domestically designed and manufactured GPUs are expected to begin mass delivery, alongside innovations in new-generation AI cloud chips [4] - The demand for computational power is projected to increase by 1 to 2 orders of magnitude in token consumption over the next 12-24 months, with optimization techniques becoming crucial for reducing token costs [4] Group 3: AI Applications - The transition in AI interaction paradigms is expected to accelerate, leading to the emergence of AI-native super applications as user reliance on mobile screens decreases and natural interaction methods gain importance [4] - Vertical scene AI applications are seen as having significant potential, with more startups leveraging industry knowledge to quickly match products with market needs, creating differentiated competition against larger firms [5] - The AI BPO (Business Process Outsourcing) model is projected to achieve commercialization breakthroughs, shifting from "delivery tools" to "delivery results" with a pay-per-result approach in standardized industries like finance and customer service [5] Group 4: Embodied Intelligence - The application prospects for embodied intelligence are considered promising, with predictions that robots will first achieve practical applications in picking, handling, and assembly, accumulating valuable operational data to drive model capability iterations [5] Group 5: Investment Landscape - Qiming Venture Partners has invested in over 100 AI projects, covering the entire AI industry chain, including AI platforms, autonomous driving, enterprise services, vertical industries, foundational software, healthcare, and consumer applications [5]