薛正华梳理AI全球投资案例，展示资本的前沿布局动向。如在具身智能与机器人领域，Figure AI已完成 6.75亿美元B轮融资，估值接近27亿美元。 | 從圖名榜 | 行业 | 融房肥炭 | 盒製 | 时间 | 投资方及更点 | | --- | --- | --- | --- | --- | --- | | Databricks | 殖蛋唇瓣是/AJ平台 | 156 | 100亿表元 | 2025年1月 | Thrive Capical@股,Andreesson Horowitz、DST Global导参与,信值达420亿美元 | | Sale Superintelligence | AI 安全 | 种子接 | 10 亿美元 | 2025 4-2 H | OpenAl 联合创始人,liga Sutskaver 创立,建冶史上最大林子枪之一。 | | Anthropic | 生成式A | LIG | 35 亿美元 | 2025 年1月 | Lightspeed 创投,份值815 坚决元,平年后备值暴演组 1830 名受完, | | Isomorphic Labs | AL生物医药 | A. 46 | 6 亿美元 ...

Core Viewpoint - The AI large model battlefield has concluded with a few winners, specifically Alibaba's Qianwen 3, DeepSeek, ByteDance's Doubao, and Tencent's Yuanbao, which are considered the first tier [1]. Industry Development - The current stage of AI development is comparable to the mobile internet in 2011, a period ripe for explosive application opportunities [2]. - Many current AI offerings are deemed "pseudo AI," as true AI integration into work and business has not yet occurred [2][6]. - The true integration of AI will be realized when it becomes an inherent part of work and life, rather than a buzzword [2][3]. Data as a Resource - The valuable "gold mine" in AI is data, as "no data, no AI" [7]. - Nvidia is viewed as a tool for enabling AI, while the large models represent the automated tools necessary for data extraction [8]. - The most valuable AI data is private data in vertical scenarios, particularly "dirty data" that has not been previously mined [10]. Company Outlook - Companies that have accumulated unique private data in their past work scenarios are favored for investment, as this data can be used to train smaller models to enhance customer experience and improve product offerings [12]. - It is estimated that 90% of current AI services are pseudo AI, merely adding an AI suffix to existing projects [13].