Core Viewpoint - A significant number of restaurants are rapidly abandoning shopping malls, indicating a prolonged period of restructuring and value system reformation in the industry, expected to last at least two more years [2][5]. Group 1: Current Trends in Restaurant Operations - Many restaurant owners are experiencing declining business in shopping malls, leading to closures despite substantial investments, such as a reported investment of nearly 3 million yuan for a hot pot restaurant in Chengdu [1]. - The trend of leaving malls is not isolated; various brands, including "Nai Si Ye" and "Yin Tang," have recently announced their exit from major shopping centers due to low foot traffic and high rents [1][3]. - A report indicates that by 2024, 34.9% of shopping centers will see more closures than new openings, highlighting a growing vacancy issue [3]. Group 2: Challenges Faced by Shopping Malls - Shopping malls are facing a decline in foot traffic, with some reporting a nearly 40% drop in visitor numbers compared to the previous year, leading to a closure rate of over 30% for restaurant outlets [8]. - The oversaturation of shopping malls has diluted foot traffic, with nearly 6,700 shopping centers of over 30,000 square meters expected by the end of 2024, compared to fewer in the past [7]. - Many malls are still operating under outdated models, failing to adapt to changing consumer preferences for experiential and emotional value, which diminishes their attractiveness [8]. Group 3: Economic Factors and Rental Issues - The high rental costs in shopping malls are becoming increasingly unfeasible for restaurant operators, with average rents reported at over 27 yuan per square meter per day, showing only a slight decrease of 0.06% year-on-year [12]. - A restaurant owner calculated that to cover a rent of approximately 2.2 million yuan annually, daily sales must reach at least 12,200 yuan, necessitating a minimum of 82 customers per day [13]. - Despite declining foot traffic and rising vacancy rates, many malls are reluctant to lower rents, leading to a challenging operating environment for restaurants [12]. Group 4: Opportunities in the Market - The concept of "downward" expansion is emerging as a potential opportunity, with many restaurants considering opening in lower levels of shopping malls, which often have lower rents and stable foot traffic due to proximity to public transport [16]. - There is a growing interest in county-level markets, where shopping centers are experiencing significant foot traffic, such as a reported daily average of 60,000 visitors during the Spring Festival in a Jiangsu county mall [17][18]. - The underdeveloped nature of commercial real estate in lower-tier cities presents a unique opportunity for restaurant brands to establish themselves in less competitive environments, leveraging the growing consumer demand for quality dining experiences [18].

Core Viewpoint - The article discusses the phenomenon of long queues at popular restaurants in major Chinese cities, particularly focusing on the experience of dining in Beijing, highlighting the social dynamics and consumer behavior associated with trendy eateries [2][3][4]. Group 1: Restaurant Queuing Culture - In cities like Beijing, the concept of "delicious, cheap, and no queue" is deemed impossible, leading to a culture where people endure long waits for food [2][3]. - The article illustrates the extreme queuing situation at a popular restaurant "烤匠" (Kao Jiang), where customers may wait up to 9 hours to dine, showcasing the lengths people will go to for trendy food experiences [4][5][8]. - The experience of waiting in line often leads to mixed feelings about the food quality, with some customers expressing disappointment after long waits, indicating a potential disconnect between expectation and reality [17][53]. Group 2: Social Media Influence - Social media platforms like 小红书 (Xiaohongshu) play a significant role in driving the popularity of certain restaurants, creating a cycle of hype and disappointment as consumers flock to these locations based on online recommendations [39][41]. - The article notes that the marketing strategies employed by restaurants often lead to a "hunger marketing" trap, where the act of queuing becomes a social experience rather than just a means to enjoy a meal [49][50]. - Consumers are increasingly aware of the pitfalls of following social media trends, developing a critical eye towards the quality of food at highly publicized establishments [40][56]. Group 3: Economic and Cultural Context - The article highlights that in times of economic downturn, dining out becomes one of the few affordable forms of entertainment for young people in urban areas, leading to a surge in demand for trendy eateries [55][56]. - The phenomenon of queuing for food reflects broader societal trends, where young people seek validation and a sense of belonging through shared experiences, even if those experiences are often disappointing [57][58]. - The article concludes that as long as consumer tastes evolve, new restaurants will continue to emerge, perpetuating the cycle of queuing and social media hype [58][59].

比亚迪在港交所发布公告称，将新增发行 1.298 亿 股新 H 股，配售价为 335.2 港元 / 股，共计可融 资约 435.09 亿港元，扣除佣金和估计费用后，所得净额预计约为 433.83 亿港元，融资所得拟用作 集团研发投入、海外业务发展、补充营运资金以及一般企业用途。 比亚迪表示，这是过去十年全球汽车行业最大的股权再融资项目，阿联酋 Al-Futtaim 家族办公室以 战略投资者身份参与本次配售。Al-Futtaim 在中东和东南亚等地区拥有汽车、金融服务、房地产、 零售和医疗保健等产业。比亚迪这次股票发售也是 2021 年以来港交所同类售股中规模最大的一 次。 OpenAI 老对手 Anthropic 估值冲到 615 亿美元。 Anthropic 上周刚发完全球首个混合推理模型 Claude 3.7 Sonnet，这周就宣布完成 35 亿美元的 E 轮 融资，投后估值达到 615 亿美元；马斯克的 xAI 在去年 12 月 的估值介于 350 亿至 450 亿美元之 间。 Anthropic 是 OpenAI 前核心员工创立，2021 年时因对 OpenAI 在 AI 未来走向、安全风险以及商业 ...

OpenAI 的三则动态：新推理模型 o3 亮相、GPT-5 训练遇阻、负责搜索的新高管 离职。 周末，OpenAI 结束了为期 12 天的发布活动，压轴产品是推理大模型 o3，在顶尖的编程竞赛 CodeForces 中，得分可以排进前 200，达到顶级程序员的水平；在专门测试通用人工智能（AGI） 的基准中，o3 得分最高达到 87.5%，而 9 月发布的 o1 得分不到 50% 等。OpenAI 并没有解释为什 么 o3 性能有提升，但有一点比较明确：高性能计算模式下， o3 完成一个任务花费数千美元。不过 o3 还是一个期货。 据媒体报道，OpenAI 训练下一代模型 GPT-5（代号 Orion）远落后进度。目前 OpenAI 训练 GPT-5 已经 18 个月，从头训练了两次，每次预估花费 5 亿美元，但是性能提升远不及预期。OpenAI 希望 GPT-5 能达到博士生水平。问题在于 OpenAI 难以找到优质数据继续训练模型，他们选择用人直接 编写数据，但效率低下，并尝试用大模型生成数据，但质量难以保证。 据媒体报道，OpenAI 今年 5 月从 Google 挖来的高管希瓦库马尔·文卡塔拉曼（S ...

美国三位共和党议员希望马斯克当众议院议长。 美国国会周二公布临时拨款法案，为联邦政府提供运营资金到明年 3 月。这项法案遭到埃隆·马斯 克（Elon Musk）反对，他在社交媒体上发布 20 多条帖子批评，比如不应该为国会议员加薪等。如 果这项方案在当地时间周五（20 日）没有通过，美国政府部门得不到资金支持，就会停摆。 同时有三位共和党议员在社交媒体上建议马斯克担任众议院议长，"众议院议长不必是国会议员"。 此前特朗普提议马斯克担任美国政府效率部门（DOGE）联合主席，但该部门主要负责提案。如果 马斯克成为众议院议长，他将拥有实权。众议院议长也是美国总统第二顺位继承人，但马斯克不是 美国原生公民，无法担任总统。 ChatGPT 早期奠基者全部离开 OpenAI。 据媒体报道，OpenAI 早期员工阿列克·拉德福德（Alec Radford）从 OpenAI 离职。他参与过 GPT-1 到 GPT-4o 的研发，在 2016 年加入 OpenAI，是最早把 Google 提出的 Transformer 架构用到大规模 语言训练的研究者之一，也是确立了当前大语言模型的训练模式的人之一。"只有阿列克·拉德福德 离 ...

6 月多家新能源汽车品牌批发量为年内新高 在 Shein、Temu 的推动下，华南至美国空运价格较 2019 年翻倍 上海中海·顺昌玖里二期开盘即售罄，单价近每平方米 30 万元 特斯拉止住颓势，二季度销量超预期 关注《晚点财经》并设为星标，第一时间获取每日商业精华。 上半年国内新能源车销量陆续公布。比亚迪以超 34 万辆的月销量继续断层领先、上半年累计卖出 161 万辆新车，完成年度销量目标的 45%。特斯拉中国排 6 月销量榜第二，据乘联会数据约 7.1 万 辆，继续同比下滑 24%、但环比小幅增长。本月，特斯拉推出 5 年免息等优惠购车政策。 理想和蔚来找回节奏，因为给了消费者更有性价比的选项。前者在 L6 的带动下，月销量回暖至 4.78 万辆，同比增长 47%，接近去年 12 月 5 万辆的最高纪录。后者则在 BaaS 电池租用服务费降低以及 创始人李斌的卖力直播影响下，年内第二次月销超过 2 万辆、6 月达到 2.1 万辆，同比接近翻倍。 据摩根士丹利统计，5 月理想和蔚来国内市场份额不同程度增长，机构预计 L6 订单动力将继续支撑 理想在三季度份额回升 —— 顶着赛力斯给的压力。 作为另一家月销 ...

B 站单日增值百亿，不靠二次元 天齐锂业发公告，智利不同意它召开特别股东大会 关注《晚点财经》并设为星标，第一时间获取每日商业精华。 B 站单日增值百亿，不靠二次元 6 月 19 日，哔哩哔哩港股收盘大涨近 19%，市值一天内增长近百亿港元（约 90 亿元人民币），刷新 近 10 个月新高。6 月 20 日，B 站美股盘前又大涨超 8%。 有钱人流向阿联酋 绿茶餐厅押宝三线城市 股价大涨跟 B 站近期代理发行的一款新游戏《三国：谋定天下》有关。自上周开启公测以来，《三 国：谋定天下》维持在 iOS 游戏畅销榜前四，仅次于腾讯的 《地下城与勇士：起源》《王者荣耀》 和《和平精英》。 它很不二次元，B 站以往出圈的游戏包括《崩坏学园 2》《Fate/Grand Order》《碧蓝航线》等。而 《三国：谋定天下》是一款回合制策略（SLG）游戏，从名字到画风到玩法都跟二次元没什么关系。 这个类型的代表《三国志：战略版》，典型的用户画像是喜欢军事和历史、有一定消费能力的中年男 性，请过的代言人包括高晓松、张绍忠、金灿荣等。 但是 "三国题材" 很赚钱。《三国志：战略版》巅峰期的月流水接近 10 亿元。交银国际预计《三国 ...

吉姆·西蒙斯去世，半生在用数学搞清楚市场发生了什么 极氪上市，一台车撑起 2.8 万美元市值 学钢琴的孩子少了，雅马哈利润下滑 关注《晚点财经》并设为星标，第一时间获取每日商业精华。 吉姆·西蒙斯去世，半生在用数学搞清楚市场发生了什么 有两个投资团队几乎年年跑赢市场，几十年来为投资者带去上千亿美元回报。 一个团队精简、只有几个人参与直接决策。不动用高等数学和编程语言，只在看过公司基本面后挑最 好的时机买入。平均一年做不了一次交易决策。这是巴菲特和芒格的伯克希尔哈撒韦。 另一个团队雇了数百名数学和物理学博士，用大量数据训练模型、写算法交易，对他们而言一家公司 跟标了名字的筹码没什么区别，电脑一天自动做出成千上万次交易。这是吉姆·西蒙斯的文艺复兴科 技。 图片来源：西蒙斯基金会 当地时间 5 月 10 日，文艺复兴创始人、数学家、破译学家吉姆·西蒙斯在曼哈顿的家中去世，享年 86 岁。西蒙斯的发言人乔纳森·加斯塔尔特证实了他的死讯，但没有具体说明死因。 西蒙斯一生都在求解。三岁时喜欢心算倍数，二十岁从麻省理工学院毕业，然后去加州大学伯克利分 校读博 —— 他听说那里的微分几何最好。毕业后在哈佛和麻省理工学院任教一 ...