截图来自社交平台X 硅谷大模型厂商正纷纷化身"基建狂魔"，疯狂融资以扩张算力。 当地时间1月6日，马斯克于2023年创立的大模型公司xAI宣布超额完成E轮融资，共筹集200亿美元。 参与本轮的投资者有，与马斯克关系密切的"老朋友"Valor Equity Partners和卡塔尔投资局，还有 Stepstone Group、Fidelity Management&Research Company、MGX和Baron Capital Group等xAI的重要合 作伙伴。 xAI特别强调，英伟达和思科是本轮的战略投资者，"它们继续支持xAI快速扩展我们的计算基础设施， 并建设全球最大的GPU集群"。 纽约时报称，即使按照xAI此前设定的150亿美元融资目标计算，公司估值也将达到2300亿美元，超额 认购后，公司估值显然将超过这一已十分巨大的数字。 目前的顶级大模型厂商中，Anthropic估值约3500亿美元，OpenAI估值则高达8300亿美元。此外，目前 最受市场瞩目的潜在IPO的明星公司，还有由马斯克旗下高达1.5万亿美元的SpaceX，以及，估值约1600 亿美元的云厂商Databricks，这些公司被认 ...

Core Insights - Dr. Yang Likun, a prominent figure in the AI field, is leaving Meta to start his own company, marking a significant turning point for both Meta and the AI industry [2] - Yang Likun is known for his groundbreaking work in convolutional neural networks, particularly the LeNet architecture, which revolutionized computer vision [2][4] - Meta is undergoing a strategic shift in its AI approach, facing internal disagreements and challenges in keeping pace with competitors like OpenAI and Google [5][6] Background of Yang Likun - Born on July 8, 1960, in France, Yang Likun developed an early interest in electronics, later earning an electrical engineering diploma in 1983 [3] - He completed his PhD in computer science in 1987, focusing on early forms of neural network training using backpropagation [3][4] - His work at AT&T's Bell Labs led to the development of convolutional neural networks, significantly impacting image processing and recognition [4] Meta's Strategic Changes - Meta is restructuring its AI strategy, investing $14.3 billion in Scale AI and appointing CEO Wang Tao to lead a new department [5] - The restructuring reflects deeper strategic divides within Meta, as Yang Likun has expressed skepticism about large language models, which the company is prioritizing [5][6] - The departure of Yang Likun highlights ongoing challenges within Meta's AI division, including a recent reduction of approximately 600 positions [6] Industry Implications - Yang Likun's new venture will focus on "world models," which aim to understand environments through video and spatial data rather than just text [5] - The AI industry is experiencing intense competition, with differing opinions on the path to achieving artificial general intelligence (AGI) [6]