Meta 收购 Scale AI 补强数据能力，引发客户流失与监管争议 Meta's Acquisition of Scale AI Bolsters Data Capabilities but Triggers Client Exodus and Regulatory Concerns wo[Table_Title] Research Report 16 Jun 2025 Technology 姚书桥 Barney Yao 吴叡霖 Louis Ng barney.sq.yao@htisec.com louis.yl.ng@htisec.com [Table_yemei1] Flash Analysis [Table_summary] 事件： Meta 于 2025 年 6 月宣布以约 148 亿美元收购 Scale AI 49%的非投票股份，同时 Scale 创始人 Alexandr Wang 将加入 Meta，负责"超智能（superintelligence）"实验室的建设。交易完成后，Scale AI 将继续独立运营，但其与微软、谷歌 等客户的合作前景存疑。 点评： 补齐 AI 训练链条，强化数据闭环能 ...

前天的文章《 深度对话 Benchmark 合伙人：AI 的增长打破了 SaaS 的 3322 规则 》里曾提到了扎克 伯格为何大力投入 AI 发展 Llama 模型，核心就是 扎克伯格不希望再出现一个由他人掌控、他完全无 法左右的平台（这里主要指苹果） 。 因此此次对 Scale AI 的投资（收购）也是这个战略的延续，扎克伯格看中的显然不是 Scale AI，而是 包括创始人 Alexandr Wang 在内的一次人才收购。最近不断爆出 Meta 正在全球以 7- 9 位数的薪酬招 募顶级 AI 人才，展示了扎克伯格不能输掉 AI 战争的决心。 对于 Scale AI 来说，随着创始人 Alexandr Wang 带领一批人才加入 Meta，其未来可能会面临很大挑 战，有可能会失去 Meta 一些竞争对手的合作订单，比方说 Google、OpenAI、Anthropic 等。Scale AI 有的投资人认为这就是一次退出，对 Scale AI 的未来不是那么的看好。 不过对于一些新加入这个领域的创业公司来说，应该是一个好消息，目前最热门的 AI 招聘类产品显然 也会受益于此，比方说 Mercor 以及我昨 ...

Core Viewpoint - Meta is restructuring its AI team to enhance product development speed and flexibility, dividing it into two main teams: AI Products and AGI Foundations [2][3] Group 1: Organizational Changes - The AI Products team will focus on consumer-facing applications like Facebook, Instagram, and WhatsApp, as well as a new independent AI application [2] - The AGI Foundations department will work on broader technologies, including improvements to the Llama model [3] - The restructuring aims to grant teams more autonomy while minimizing inter-team dependencies [3] Group 2: Competitive Landscape - Meta is striving to keep pace with competitors like OpenAI and Google, launching initiatives like "Llama for Startups" to encourage early-stage companies to utilize its generative AI products [3] - Despite initial success, Meta's reputation in the open-source AI field has declined, with significant talent loss from its foundational AI research team, FAIR [4][7] Group 3: Talent and Leadership Issues - A significant number of key researchers from the Llama project have left Meta, raising concerns about the company's ability to retain top AI talent [7][23] - The departure of Joelle Pineau, a long-time leader at FAIR, has highlighted internal issues regarding performance and leadership [8][13] Group 4: Financial Commitment and Future Plans - Meta plans to invest approximately $65 billion in AI projects by 2025, with the aim of enhancing its AI capabilities [22] - The company is expanding its data center capacity, including a new 2GW facility, to support its AI initiatives [22]

Core Insights - Meta has restructured its AI teams into two main groups: an AI product team led by Connor Hayes and an AGI Foundations team co-led by Ahmad Al-Dahle and Amir Frenkel, aiming to enhance product development speed and flexibility [1][2] - The restructuring comes amid increasing competition in the AI space from companies like OpenAI and Google, as Meta seeks to maintain its relevance [2][3] - The departure of key personnel from Meta's AI research division, FAIR, has raised concerns about the company's ability to retain top AI talent and its competitive position in the market [3][4] Team Structure and Focus - The AI product team will focus on consumer-facing applications, including AI features in Facebook, Instagram, and WhatsApp, as well as new independent AI applications [1] - The AGI Foundations team will concentrate on broader technological advancements, such as improving the Llama model [1][2] - FAIR remains independent but has seen a multimedia team transition to the AGI Foundations team [1] Talent and Leadership Changes - Meta has experienced significant talent loss, with 11 out of 14 original authors of the Llama model leaving the company, many joining competitors like Mistral [3][4] - Joelle Pineau, who led FAIR for eight years, recently resigned, raising questions about the future direction of the research team [4][6] - The leadership change in FAIR has been accompanied by a shift in focus towards product-oriented AI projects, sidelining exploratory research [14][15] Competitive Landscape - Meta's initial lead in open-source AI models has diminished, with competitors like DeepSeek and Qwen gaining traction [4][19] - The recent launch of Llama 4 has faced criticism for being rushed and lacking transparency, further impacting Meta's reputation in the AI community [10][19] - Despite substantial investments in AI, including a projected $65 billion by 2025, Meta lacks a dedicated reasoning model, which is becoming increasingly important in the AI landscape [16][19] Future Outlook - Meta's commitment to AI research remains, with plans to enhance collaboration between FAIR and the GenAI team to accelerate decision-making [16] - However, internal dynamics suggest a shift towards prioritizing profitability over foundational AI research, leading to concerns about the long-term viability of FAIR [16][17] - The ongoing talent exodus and competitive pressures indicate that Meta may struggle to reclaim its former leadership position in the AI sector [19]

Group 1 - The core discussion at LlamaCon 2025 focused on the transformative impact of AI on the boundaries between documents, applications, and websites, as articulated by Satya Nadella [5][6] - Nadella emphasized that modern AI acts as a "universal converter," understanding user intent and enabling a shift from "tool-oriented computing" to "intent-oriented computing," enhancing user experience [6][7] - Nadella identified the current AI wave as a significant technological platform shift, necessitating a complete overhaul of the technology stack to optimize for AI workloads [7] Group 2 - Nadella noted that approximately 20% to 30% of Microsoft's internal code is now generated by AI, indicating a broad application of AI in software development beyond mere code completion [7][8] - Zuckerberg projected that by 2026, half of Meta's development work will be completed by AI, showcasing the growing reliance on AI in the tech industry [8] - The dialogue also highlighted the strategic value of both open-source and closed-source models, with Nadella advocating for a flexible approach that supports both [9][10] Group 3 - The concept of "distillation factories" was introduced as a key area for future development in the AI ecosystem, with both CEOs agreeing on the importance of infrastructure and toolchains for model distillation [10][11] - Nadella pointed out the trend towards multi-model applications and the necessity of standardized protocols for seamless collaboration among various AI models [10] - Zuckerberg acknowledged Microsoft's unique advantages in supporting multi-model collaboration infrastructure, reinforcing the significance of the "distillation factory" concept [10]

特别声明：以上文章内容仅代表作者本人观点，不代表新浪网观点或立场。如有关于作品内容、版权或其它问 题请于作品发表后的30日内与新浪网联系。 来源：虎说天下 曾作为老牌科技强国的日本，面对中美在 AI 领域日益拉大的差距，其本土发展一度陷入困境。两个月 前，日本政府投入 10 亿日元研发的 AI 工具，测试阶段误判率竟高达 60%。 如今，日本开始将目光投向中国，借助中国 AI 的开源力量推动本土发展。据《日本经济新闻》报道， 阿里通义千问大模型已成为日本 AI 开发的重要基础。例如，日本企业 ABEJA 发布的推理模型，正是 基于千问 QwQ-32B 开源模型开发，在逻辑推理方面表现出色。大量日本新兴企业也做出了同样选择。 为何日本企业青睐中国大模型而非美国？关键在于中国顶尖大模型走开源路线，而美国多数模型闭源， 通过技术壁垒获取利益，对日本并不开放。以阿里通义千问为例，2023 年至今已开源 200 多个模型， 全球下载量超 3 亿次，衍生模型数量突破 10 亿，碾压美国 Llama 模型，登上全球第一开源模型宝座， 受到全球开发者和企业欢迎。 这一现象不仅是日本 AI 发展的转向，更标志着全球科技格局的深 ...