Meta has heavily recruited from Google's AI division DeepMind for its new superintelligence team, bringing in at least 10 previously unreported researchers since July, according to LinkedIn data reviewed by Business Insider. Meta's DeepMind hires include researchers who worked on Google's most powerful AI models, and even one whose work helped Google earn gold at the International Mathematical Olympiad.Meta has also hired heavily from Scale AI for its safety and evaluations team. At least six of the data l ...

机器之心报道 编辑：张倩、+0 十亿美元能买一栋别墅，但买不了梦想？ 最近，Meta 内部发生了一些有意思的事情 —— 一边是扎克伯格动辄上亿美金薪资招兵买马，高调组建超级智能团队；另一边是一些老员工宣布开启新的「冒险之 旅」，转投其他 AI 公司。 不过，还有很多选择离开的人，可能是因为钱没给够。在超级智能实验室成立之后，Meta 内部的待遇差距多次引发争议。 前 Meta 研究员 Rohan Anil（现 Anthropic）曾发帖称「非超级智能研究者待遇次等，像巨型社会实验」。 今天，有两位资深研究者宣布离开 Meta，一位是专注于强化学习的 Rishabh Agarwal（去向未定）；另一位是已经在 Meta 工作了 12 年、参与了 PyTorch 构建的 Bert Maher（确定加入 Anthropic）。 看来，除了小扎挖不到的人，还有一些他留不住的人。甚至有人嘲讽「钱买不到顶级研究员」。 有人认为 Meta 内部薪资差距（同事赚 1-2 亿美元）会杀死工作动力，导致更多离职。 当外部挖不来真正的梦想家，内部又因分配不均而人心浮动，这背后折射出的，可能是比薪酬更深层的结构性问题。 这让一些人联 ...

Forget Nvidia: This tech stock has the scale, data, and cash to turn today's AI spend into years of compounding value -- and shares look attractively priced.Meta Platforms (META -0.15%) is one of my top picks for investors seeking exposure to artificial intelligence (AI) without chasing speculation. The company already embeds AI in safety, ads, and even the user experience across Facebook, Instagram, WhatsApp, and Messenger. But there's another Meta app that has largely gone under the radar (for now) -- and ...

2025 年 8 月 25 日，美银全球研究部发布《全球股票持仓报告》（Global Positioning in Stocks），数据覆盖截至 2025 年 7 月 31 日的全球纯多头基金（Long-only Funds）持仓动向。 报告核心逻辑围绕 "基金持仓行为 + 主动敞口 + 三重动量" 构建分析框架，揭示全球机构资金在 AI 长期主题主导下的配置偏 好：半导体板块因 AI 景气度获巨额增持，工业与医疗板块遭显著抛售；地区层面，亚太（除日本）成为资金避风港，美国市场 则面临持续流出压力。此外，报告通过 "四大股票筛选标准"（拥挤型利好、拥挤型利空、欠配利好、欠配利空）识别出具有潜 在收益或风险的标的，为投资者提供决策参考。 一、全球资金流向：半导体成最大赢家，亚太（除日本）逆势吸金 1. 行业维度：AI 主题驱动半导体板块 "一枝独秀" 受 AI 长期投资主题回暖推动，上月全球纯多头基金对行业板块的配置呈现显著分化： 增持核心——半导体板块：上月全球半导体板块获纯多头基金净买入 272 亿美元，成为全球增持规模最大的行业，反映机构对 AI 产业链上游核心资产的长期信心。从细分领域看，AI 芯片、先 ...

Secret Fintech Payments Cloud $725 Million Facebook Class Action Settlement https://t.co/KhPzVJFclv https://t.co/3JmcarJraO ...

Group 1: Trade Negotiations - Chinese Vice Minister of Commerce, Li Chenggang, will visit Washington this week for trade negotiations with the U.S. [1] - Li Chenggang is expected to meet with U.S. Trade Representative Katherine Tai and senior officials from the U.S. Treasury [2] - The U.S. and China have previously extended the tariff truce period by 90 days to continue discussions on trade deficits, market access, and rare earth issues [3] Group 2: U.S.-India Tariffs - The U.S. plans to impose a 50% tariff on Indian goods starting from August 27 [4] - This tariff will apply to all imported Indian goods for consumption or storage [5] - The overall tariff rate on Indian goods will increase to 50% due to an additional 25% tariff signed by President Trump [6][7] - India has expressed that the U.S. tariff increase is "unfair, unjust, and unreasonable," and will take necessary actions to protect its national interests [8] Group 3: South Korea's Diplomatic Efforts - A South Korean delegation led by former National Assembly Speaker Park Byeong-seok visited China to strengthen communication [10] - The delegation aims to discuss economic cooperation with Chinese officials, including Commerce Minister Wang Wentao [10] - The visit coincides with the 33rd anniversary of diplomatic relations between China and South Korea, reflecting South Korea's emphasis on balancing its relations with both the U.S. and China [12][13] Group 4: South Korea's Investment in the U.S. - South Korean companies plan to invest approximately $150 billion in the U.S. [14][15] - This investment commitment was announced during a meeting between business leaders from both countries [15] - Notable attendees included Samsung Electronics Chairman Lee Jae-Yong and NVIDIA CEO Jensen Huang [16] Group 5: AI Investment in South Korea - South Korean President Lee Jae-myung announced plans to prioritize investment in artificial intelligence (AI) [20] - The government aims to increase research spending by nearly 20% by 2026, reaching a record 35.3 trillion won (approximately $252.3 billion) [21] - The government will launch policies for 30 major AI and innovation projects starting in the second half of 2025 [22] Group 6: Semiconductor Industry Developments - The U.S. Department of Commerce canceled a $7.4 billion semiconductor manufacturing research funding program initiated during the Biden administration [45] - This program aimed to strengthen the U.S. semiconductor supply chain and close the technology gap with Asian manufacturers [45] - The cancellation may be influenced by recent budget disputes in Congress and shifts in industrial policy priorities [45] Group 7: 3D DRAM Technology Breakthrough - Researchers from imec and Ghent University announced a breakthrough in 3D DRAM technology, successfully growing 120 layers of Si/SiGe on 300mm wafers [48][49] - This advancement addresses challenges in traditional DRAM processes and is expected to enhance storage density for AI and high-performance computing applications [48][49] - The successful development of 3D DRAM could significantly impact the semiconductor industry and meet the growing demand for high-performance storage [49] Group 8: RISC-V Industry Alliance in Jiangsu - The Jiangsu Province RISC-V Industry Alliance was established to link innovation resources in the RISC-V field [50][51] - The alliance aims to enhance collaboration among industry leaders, universities, research institutions, and industry associations [51] - Jiangsu has a robust semiconductor industry with over 380 companies and a scale exceeding 120 billion yuan [50] Group 9: ChipScale's GaN Semiconductor Production - ChipScale has begun mass production of 650V gallium nitride (GaN) power semiconductors in South Korea [53][54] - This technology offers higher power efficiency and faster switching speeds compared to traditional silicon materials [53] - The company aims to expand its market presence in various applications, including AI and data centers [54]

Core Viewpoint - Meta is aggressively recruiting AI talent while facing internal challenges, including the departure of key researchers and restructuring of its AI division [1][9][10]. Group 1: Recruitment and Talent Acquisition - Meta's CEO, Mark Zuckerberg, is personally involved in recruiting top AI researchers, offering salaries that can reach up to $100 million [7][8]. - As of mid-August, Meta has successfully recruited over 50 AI researchers from various companies, including more than 20 from OpenAI and at least 13 from Google [8]. Group 2: Departures and Internal Challenges - Rishabh Agarwal, a prominent researcher at Meta, announced his departure, citing a desire to take on different types of risks despite the attractive vision of the new Superintelligence team [2][3][4]. - Agarwal's resignation was influenced by the internal restructuring of Meta's AI division, which has led to a hiring freeze and a reduction in team size [9][10]. Group 3: Research Contributions - During his time at Meta, Agarwal contributed significantly to advancements in AI, including improvements in reinforcement learning models [12][16]. - His academic credentials include over 10,000 citations and a strong h-index of 34, indicating his influence in the AI research community [19].

一水 发自 凹非寺 量子位 | 公众号 QbitAI 小扎在这头疯狂挖人，结果家里的老员工纷纷跑路了？？ 最新消息， Meta万引强化学习大佬Rishabh Agarwal 即将离职，还留下了一篇让人浮想联翩的小作文： 这是我在Meta的最后一周。 决定不加入新的超级智能实验室并不容易，毕竟那里人才济济、算力爆棚。但在Google Brain、DeepMind和Meta度过了7年半之后， 我更想冒险去尝试一条完全不同的路。 Meta组建超级智能团队的想法非常引人注目，但我最终选择听从扎克伯格的建议："在这个瞬息万变的世界里，最大的风险就是不去冒 险。" 曾被Hinton劝退"不要做强化学习"，下一站未定 Rishabh Agarwals，一直以来从事强化学习和推理研究，谷歌学术论文被上万次引用，h-index也有34。 虽然表面上看起来双方是"和平分手"，但网友们还是从中嗅出了一丝不同寻常的味道： 把小扎的原话甩回他自己脸上，这操作绝了，瑞思拜！ 十亿可以为你买一栋房子，但买不到你的梦想。 不过猜测也好，吐槽也罢。对于Rishabh Agarwal的离职，谷歌、Meta的同事们都清一色地送上了祝福，而且还顺带 ...

作者 / 钱亚光 编辑 / 张 南 设计 / 柴文静 当地时间2025年8月20日，《华尔街日报》报道的一则关于Meta的消息震动全球科技圈——其新成立的人工智能部门正式暂停招聘，不仅停止面试新候选 人，还宣布将重心从疯狂挖角转向内部员工整合。 要知道，这家企业半年前才以"四年3亿美元天价薪酬包"从OpenAI挖走核心研究员，如今突然踩刹车，背后是3200亿美元全球AI总投入、超百万人才缺口 与纳斯达克指数单日大跌1.4%的复杂棋局。 这场持续近三年的全球AI人才争夺，终于暂告一段落，但它留下的行业洗牌、资本泡沫与人才重构难题，才刚刚开始浮现。 01 AI人才争夺三个阶段 这场席卷全球的AI人才争夺，并非一蹴而就，而是经历了"预热-引爆-全面战"三个阶段，最终以Meta暂停招聘画上阶段性句号。 预热期（2023-2024年）：理性竞争阶段 此时AI人才争夺仅集中于OpenAI、DeepMind等少数领先企业，薪酬水平虽高于传统科技岗位，但仍处于"理性区间"——顶尖AI工程师年薪约50万-80万美 元，尚未出现天价挖角现象。竞争焦点主要围绕"基础模型研发人才"，企业更倾向于内部培养而非外部挖人。 引爆期（202 ...

文/刀客doc 流量越来越贵了，但贵得不无道理。 如果你是一个经常在做北美投放的出海商家，应该已经感受到，过去两年里"便宜的社交广告"正在迅速消失。8月份，eMarketer出了一份 《2025年美国社交广告CPM预测》，让这种感受有了量化的注脚。 预测显示，2025年美国主流社交平台的流量价格都上涨，而且是一次全面的、系统性的调整。 到2025年第四季度，Meta旗下的Instagram成为首个CPM突破10美元的社交平台，正式坐上高价王座。而TikTok的广告价格更是一路狂飙，从 年初的7.12美元涨至年末的9.86美元，单年涨幅高达38.5%，涨势在主流平台中尤为激进。Pinterest同步进入高位区间，保持两位数增幅； Snapchat虽仍处在价格洼地，但也在缓慢爬升中。 | Q4 2022 | Q4 2023 | Q4 2024 | Q4 2025 | | --- | --- | --- | --- | | Instagram _ TikTok = Facebook = Pinterest = Snapchat | | | | 从"出多少展示"转向"触达多少真实意图"。平台不再鼓励盲目曝光，而是通过 ...