Group 1 - The market is experiencing a significant demand for AI talent, with major internet companies actively seeking both top-tier and foundational employees in the AI sector [2][12][20] - Companies like Alibaba and Tencent are ramping up recruitment efforts, with Alibaba's AI To C business opening hundreds of positions and Tencent's AI assistant also launching a large-scale hiring initiative [2][3] - High salaries are becoming a common trend in the AI industry, with positions like deep learning researchers offering salaries exceeding 1.76 million yuan annually, and many roles seeing salaries above 500,000 yuan [4][5][13] Group 2 - The rapid development of AI is causing traditional programmers to feel threatened, prompting them to seek opportunities in AI-related fields, with salary increases of around 30% being common for successful transitions [10][11][18] - The demand for AI talent is projected to grow significantly, with McKinsey predicting a sixfold increase in the need for skilled AI professionals in China by 2030, leading to a talent gap of 4 million [18][20] - Companies are facing challenges in finding qualified candidates, as many applicants lack the necessary skills and experience, highlighting a mismatch between educational training and industry needs [23][25] Group 3 - The AI industry is characterized by a competitive landscape, with companies competing for talent and offering attractive compensation packages, including stock options and high salaries [28][31] - Despite the promising outlook for AI, there are concerns about potential market bubbles and the sustainability of certain AI startups, as evidenced by layoffs in some companies [29][30] - The evolving nature of the AI job market is leading to a greater emphasis on interdisciplinary skills, with a need for professionals who understand both AI technology and its application in various business contexts [25][27]

以下文章来源于飞跑的鹿 ，作者RunningLu 飞跑的鹿 . 公司研究 | 产业挖掘 | 价值投资 | 长期主义每日分享上市公司及产业链 | 市场调味剂 01 产业链全景图 03 什么是DeepSeek DeepSeek 是一家人工智能技术公司，中文即"深度求索"。 它家的模型在自然语言处理和代码 生成等领域相当能打，像DeepSeek-v3在多个数学基准测试和代码能力测试中，就超越了众多 竞品。 真正让DeepSeek火出圈的是2024年12月26日，这家公司宣布上线并同步开源的 DeepSeek-V3 模型。它以 1/11的算力、仅2000个GPU芯片 训练出性能 超越GPT-4o的大模型 。其 总训练成 本只有557.6万美元 ，而 GPT-4o的约为1亿美元 ，使用 25000个GPU芯片 。 双方的成本至少 是10倍的差距。 在性能上，DeepSeek-V3在数学、代码能力和中文知识问答方面还超过了ChatGPT-4o， 以下即为两者的模型参数对比： 04 为什么被称为AI界的拼多多 据公司介绍，在数学、代码、自然语言推理等任务上， DeepSeek-R1 性能比肩已经OpenAlo1 正式版，这 ...

Group 1: AI and Technological Innovation - DeepSeek's launch marks a significant milestone in AI, enabling low-cost, large-scale, localized deployment, breaking the computational barriers set by the U.S.[10] - The cost of computational power is expected to decrease by 90% annually, leading to a global productivity surge and transformative impacts across industries[12]. - The 2025 private enterprise symposium signals a major push for AI technological innovation in China, emphasizing the importance of private enterprises in this sector[13]. Group 2: Market Trends and Asset Allocation - The Chinese equity market is undergoing a revaluation of AI technology assets, with significant inflows from global hedge funds into Chinese tech stocks, particularly in the Hong Kong and A-share markets[20]. - The global manufacturing PMI reached 50.1 in January, indicating an expansion phase, which is expected to boost demand for commodities[26]. - Gold is experiencing a physical shortage, with significant flows from London to New York, driven by geopolitical tensions and increasing demand from emerging markets[28]. Group 3: Economic Policies and Currency Dynamics - The U.S. has imposed a 25% tariff on steel and aluminum, with plans for similar tariffs on automobiles, initiating a global trade conflict[17]. - The Chinese yuan is expected to maintain a defensive stance around 7.35 to 7.40 against the dollar, with potential for appreciation as U.S. tariffs stabilize[32]. - Short-term government bond yields in China have risen from 0.90% to 1.50% over two months, reflecting market pressures and a shift in investment flows towards equities[31].

Core Insights - The article discusses the challenges faced by large model companies in securing funding and reducing costs, highlighting the competitive landscape introduced by quantitative firms like Jiukun and their collaboration with Microsoft to replicate DeepSeek-R1 [1][2][13] - It emphasizes the growing interest of quantitative firms in the AI space, driven by the need for algorithmic support and the accumulation of high-performance chips [5][6][12] Group 1: Industry Dynamics - Quantitative firms are increasingly entering the large model sector, with Jiukun's successful replication of DeepSeek-R1 indicating a shift in competitive dynamics [1][12] - The article notes that the success of figures like Liang Wenfeng has led to a rush among investors to find similar opportunities, despite the inherent difficulties in replicating such success [13][16] - The fear of missing out (FOMO) is prevalent among investors, as they worry about missing investment opportunities in top-tier projects [14][15] Group 2: Company Strategies - Jiukun has established multiple internal laboratories to support its quantitative research, indicating a strong foundation for its entry into the AI field [5][6] - The accumulation of over 10,000 A100 chips by companies like Huansheng demonstrates the importance of computational power in the quantitative finance sector [7][8] - The article suggests that while many quantitative firms are entering the large model space, most are still in the early stages and may not achieve the same level of success as leading firms [16][17]

孙不熟专栏，在这里读懂城市、学会买房。 以下文章来源于城市战争 ，作者孙不熟 城市战争 . 为何都不在广东创业？ 来源 | 城市战争 （ID：sunbushu123） 最近在网上看多一个说法：中国"AI三杰"都是广东人，但都没有在广东创业和发展。 公开资料显示，这三位蜚声全球的AI大牛都很年轻，其中两个是8 0后、一个是9 0后，一 个在杭州上大学和创业，一个在北京上大学和创业，另一个在美国MIT任教。 作者 | 孙不熟 梁文锋：广东湛江人 以一己之力干崩美股 De e ps e e k是春节期间火遍全球的一款国产大模型，成功地将大模型的训练成本断崖式降 低，被称之为"AI界的拼多多"，全球为之震惊，甚至有媒体称他"以一己之力干崩了美 股"。 梁文锋是De e p s e e k的创始人，1985年出生于广东湛江吴川的偏远农村，1 7岁就以优异 成绩考上浙江大学信息与电子工程系，并在浙大拿下硕士学位。 毕业后的梁文锋没有选择像很多同学那样进大厂当码农，而是选择自主创业，从事当时 并不被看好的全自动量化交易，就是使用数学模型和计算机方法来指导股票交易。 网友口中的"AI三杰"指的是De e ps e e k的创 ...

聪明投资者 . 聚焦优秀投资人和企业家，甄选高质量的内容，追求可累进的成长。更多内容可下载"聪明投资 者"APP，官网：www.cmtzz.cn 一名爱好量化投资的程序员。 排版 | 关鹤九 责编 | 艾暄 来源 | 聪明投资者 （ID：Capital-nature） 01 第一次见梁文锋，是2018年的6月份，幻方量化杭州总部。 以下文章来源于聪明投资者 ，作者永远好奇的 2018年的量化市场刚刚显露今天的格局，机构调研时挂在嘴边的那些公司的名字，包括 幻方、九坤、明汯们，背后承载的规模也仅仅是今天的零头。 当时成立3年的幻方管理客户资金约45亿，自营盘约10个亿。已经是量化的第一梯队。 作为掌门人的梁文锋一直隐形在幕后，很长一段时间业界都以为公司核心高管是另外两 位。 托一个朋友的福，围观了这次罕见的深度调研。 梁文锋走进小会议室面对面坐下时，捧着一个保温杯，穿着深蓝色的工装绒棉衬衫。很 瘦削，有点拘谨，活脱是上个世纪90年代的工程师模样。 瞬间有了跟一则趣闻对号入座的即视感：据说他买了新房却一直醉心于开发策略而无心 装修，所以在房间里支了帐篷睡觉。 同事说他除了编程，没有什么其他的爱好。 提前被安利了梁 ...

湛江85后 杭州另一面。 作者 I 吴琼 报道 I 投资界PEdaily 大年初一，广东湛江吴川米历岭村，迎来一波又一波打卡人群。 这是爆红全球De e pSe e k的创始人梁文锋老家。早年间成长于广东，后来考入浙江大学。 硕士毕业后，梁文锋与几位浙大同学开始创业之路。2 02 3年，他带领团队成立 De e pSe e k，如今席卷全球科技圈。 De e pSe e k被视为拉开中国科技爆炸的序幕，成为今年第一个现象级中国企业。甚至有投 资人透露，"找他们的人踏破门槛了"。 当De e pSe e k在海外大杀四方，央视春晚会上，一群人形机器人登上舞台扭秧歌，背后的 宇树科技正是中国具身智能标志性独角兽。追溯下来，他们都来自杭州。 缔造新年最火独角兽 梁文锋，1 9 8 5年出生于广东湛江吴川，父母都是小学语文老师，这也使得他从小就成绩 优异。"湛江发布"提及他一段成长往事：梁文锋性格很文静，但不是书呆子，在学习上 很有自己的"一套方法"，初中时就学完了高中数学，甚至开始学大学数学。 2 0 0 2年，梁文锋以吴川一中"高考状元"的成绩考上浙江大学，本科就读电子信息工程专 业，之后又继续攻读浙江大学信息与 ...

以下文章来源于南风窗 ，作者朱秋雨 南风窗 . AI小厂崛起。 作者 | 朱秋雨 来源 | 南风窗 （ID：shangyejingxiang） 2024年末，一个中国AI小厂，凭借过硬的技术，获得了全球铺天盖地的关注。 圣诞节过后，海外社交媒体以及技术论坛Github都在讨论一个最新发布的开源大模型， DeepSeek-V3。它被外国网友冠以名号——"来自东方的神秘力量"。 多个评测报告里，DeepSeek-V3在世界开源模型之中处在第一梯队，超过扎克伯格的 LLaMa 3.1。拿它比GPT-4o以及Claude 3.5两个最顶尖大模型也毫不逊色，甚至， 其在数学推理、代码生成和长文本处理等指标上，表现更强。 这还不是中国AI公司DeepSeek（中文名：深度求索）全部的"拿手好戏"。更让美国硅 谷等同行摸不着头脑的是，DeepSeek公布的53页技术报告显示，其训练顶尖大模型只 用了2048张H100的GPU集群，花费53天，共计耗费557.6万美元。有专业人员指 出，同等水平之下，世界AI大厂至少要用1.6万张以上的GPU，有的甚至需要10万张 GPU并行训练。 OpenAI早期成员安德烈·卡帕西感慨，D ...