官方介绍，Pika AI Selves 是一个由你「孕育、培养并放手」的 AI 分身，成为你的一个活生生的延伸。 它们拥有丰富多面的个性、持久记忆，甚至连「花生过 敏」这种细节都可以设定 —— 一切由你决定！ 它能替你在群聊里发照片、给你的宠物鱼做一款视频游戏、在你忙着做事情的时候替你打电话给妈妈……「可能性就像星空一样无限。」 当大多数的 AI 厂商都在忙着打造更多 AI 工具的时候，一家以制作 AI 视频著称的公司居然开始制作「第二个你」了。 近日，Pika 推出 AI Selves 产品， 宣称可以生成「AI 版的你」。 机器之心编辑部 网友 Aakash Gupta 称， 「Pika 的 AI Selves，或许是今年 AI 领域最具野心的一次品类跃迁，但几乎没人讨论：为什么率先做到这一点的，会是一家 AI 视频公 司？」 数据显示， 当前几乎所有大型科技公司都在竞速构建自主 AI 智能体，整个市场正以 46% 的年复合增长率扩张，预计到 2030 年将达到 520 亿美元规模。 但市面 上几乎所有智能体，都是基于文本展开：文本输入、文本输出、完成任务、自动化流程。 也有网友认为，「这不就是《黑镜》吗 ...

图片来源： The MAD Podcast Z Highlights Sebastian Borgeaud 是 Google DeepMind 的 Gemini 3 预训练负责人，同时也是开创性论文 RETRO 的合著者，在 AI 前沿模型研发与系统构建领域具备深厚专业 积淀。 2025 年 12 月 18 日，他在首次播客访谈中揭秘了这款今年 AI 领域里程碑式前沿模型的研发逻辑，分享了模型背后并非单纯依赖算力提升的系统构建思 路。 Gemini 3的巨大提升是庞大团队通力协作、融合无数改进与创新的成果，其基于Transformer的混合专家架构， 核心是将计算量使用与参数规模分离开 来 。 规模是预训练中提升模型性能的重要因素，但并非唯一， 架构和数据创新的重要性如今可能更甚 ，且预训练领域在长上下文能力、注意力机制等方 面有诸多值得关注的发展方向。 行业正从 "数据无限"向"数据有限"范式转变 ，合成数据需谨慎使用，模型架构改进能助力模型用更少数据实现更好效果，同时评估在预训练中至关 重要且极具难度。 的朴素。所以我很好奇你的看法，从某种程度上来说，事情真的这么简单吗？ Sebastian Bourge ...

Company Overview - Tuya is transitioning from a behind-the-scenes tech platform to a leader in real-world AI, focusing on smart assistants, companion robots, and connected devices [1] - The company operates an AI cloud platform that enables brands and developers to create connected products without building the entire software infrastructure [2] Market Position - As of September 2025, Tuya had over 1.62 million registered developer accounts across more than 200 countries, showcasing its extensive global reach in the smart device ecosystem [3] - Tuya's market capitalization is approximately $1.2 billion, with a stock price range of $1.86 to $4.63 over the past 52 weeks [7] Financial Performance - Tuya reported a GAAP net profit of $15 million in Q3 2025, a significant improvement from losses the previous year, with gross margins reaching 48.3% [8] - The company holds about $1.026 billion in cash and has zero debt, which constitutes roughly 85% of its market cap, indicating a strong financial position [9] Product Innovations - At CES 2026, Tuya introduced a new Software-Hardware-Platform Physical AI ecosystem, indicating a shift towards AI that interacts with the real world [4] - The company unveiled "Hey Tuya," a multi-agent AI life assistant, and "Aura," its first AI pet companion robot, demonstrating its commitment to integrating AI into daily life [5][6] Investment Opportunity - Tuya's stock is seen as undervalued, with a disconnect between its price and fundamentals, making it an attractive investment for those looking for opportunities in the AI sector [8][9] - For a small investment of $100, investors can acquire approximately 45 shares of a profitable, cash-rich AI platform that is currently overlooked by the market [10]

Group 1 - OpenAI CEO Sam Altman praised the rapid development of China's full-stack technology in the AI field, highlighting its impressive progress and global attention [1] - Full-stack technology includes a complete system from underlying chips and hardware infrastructure to operating systems, algorithms, and application scenarios, with China nearing or achieving international advanced levels in several key areas [1] - Chinese companies have made breakthroughs in full-stack technology across three dimensions: deep integration of hardware and software, global impact of open-source ecosystems, and deep penetration into vertical industries [1][2] Group 2 - Alibaba's Feitian enterprise platform has achieved a 30% increase in GPU utilization and a threefold optimization in inference efficiency, significantly reducing the development cycle for intelligent Q&A systems [1][2] - The Qwen series of open-source models from Alibaba has surpassed Meta's Llama in global downloads, creating a "Qwen anxiety" in Silicon Valley, with Nvidia's CEO noting Qwen's growing market share [2] - ZTE's 5G-A full-stack solution has reduced 4K video transmission latency to 50 milliseconds, supporting real-time subtitle generation for the 2025 CCTV Spring Festival Gala [2] Group 3 - OpenAI faces a strategic dilemma between maintaining a closed-source approach and the risk of losing developer ecosystems or shifting to an open-source strategy, leading to a gradual opening of some technology modules starting in late 2025 [2] - Despite performance advantages in optimization and algorithms, the gap between U.S. mainstream models and Chinese developments is rapidly narrowing, with a large and active global developer ecosystem forming around Chinese open-source models [3] - OpenAI is adjusting its strategy to maintain competitiveness, including pursuing a new funding round with a target valuation of $100 billion and exploring diversified revenue sources, such as introducing an advertising model in ChatGPT [3]

Core Viewpoint - OpenAI is significantly reducing its capital expenditure expectations, facing increasing concerns about its ability to generate sufficient revenue to cover costs, with a new total spending target of $600 billion by 2030, down from a previously stated $1.4 trillion [1][3][5]. Group 1: Financial Projections - OpenAI expects total revenue to exceed $280 billion by 2030, a substantial increase from $13.1 billion in 2025, with contributions from consumer and enterprise businesses being roughly equal [1]. - The adjusted gross margin for OpenAI is projected to decline to 33% by 2025, down from 40% in 2024, as the costs associated with running AI models are expected to quadruple by 2025 [2][6]. Group 2: User Growth and Competition - ChatGPT currently supports over 900 million weekly active users, an increase from 800 million in October of the previous year, with both weekly and daily active user numbers reaching historical highs [7][8]. - OpenAI is facing competitive pressure from companies like Google and Anthropic, prompting it to enter a "red alert" state in December to focus on improving its chatbot functionalities [9]. Group 3: Financing and Valuation - OpenAI is nearing completion of a new funding round, aiming to raise over $100 billion, which would value the company at approximately $830 billion, marking one of the largest private funding rounds in history [11][12]. - Major investors in this funding round include Nvidia, which may invest up to $30 billion, along with strategic investors like SoftBank and Amazon [13].

Core Insights - The launch of ByteDance's video generation model Seedance 2.0 has gained global popularity among creators due to its multimodal reference system and native audio-video synchronization capabilities [3] - The AI large model industry is experiencing a new wave of innovation, with numerous new models being released both domestically and internationally, including Alibaba's Qianwen 3.5 and OpenAI's GPT-5.3-Codex [3][4] - The rapid development of AI requires a comprehensive innovation ecosystem that integrates various sectors, including finance, healthcare, manufacturing, and transportation, to realize its full potential [4] Industry Developments - The AI large model sector is witnessing a surge in new applications and models, characterized by increasing differentiation and practicality, with many models breaking out of traditional boundaries [4] - Shanghai has strategically positioned itself in the AI landscape by fostering an environment conducive to innovation, including support for startups and a focus on real-world applications [5][6] - The city has achieved a record of five AI companies going public in one month, indicating its competitive edge in the AI sector [5] Innovation Ecosystem - The development of any innovative industry in Shanghai necessitates a systematic and interconnected framework to stimulate vitality and ensure sustainable growth [6] - The concept of "AI for science" and "AI for maths" highlights the interdependence between AI advancements and progress in other scientific fields, further emphasizing the need for a collaborative ecosystem [4]

受此消息影响，智谱股价2月以来连续大幅度上涨。至2月12日，智谱正式开源发布新一代基座模型GLM-5，印证此前市场传言，同时宣告 据内部人士对《智客星球》栏目透露，目前GLM-5已率先在海外版上新，在已公布的海外版价格中，Coding plan订阅价格提高30%— 60%，API调用价格提升67%—100%。这是国产大模型近期以来的首次大幅提价，显示出国产大模型的技术能力和市场竞争力正在快速提 "一位AI圈行业人士在与《智客星球》栏目沟通中直言。 而这一系列能力的发布，也进一步提升了资本市场对于公司的信心，进一步促使股价及估值提升。 "为中国经济点赞——企业家之夜"活动，不仅见证了智谱作为国内AI大模型企业代表的荣耀时刻，也汇聚了小马智行 的彭军、强脑科技的 韩璧丞、汇天的赵德力等前沿科技应用杰出代表，也见证了世界矿业龙头的紫金矿业 创始人陈景河，潍柴动力党委书记、董事长马常海， 中国飞鹤董事长冷友斌，阿那亚创始人兼CEO马寅，以及珍酒李渡集团、华致酒行董事长吴向东等影响力行业领域。 一如知名人文财经观察家秦朔所言， "十五五"规划建议提出"引领发展新质生产力" 。无论AI大模型，还是自动驾驶、脑机接口，都是 ...

Core Viewpoint - The market is experiencing a rotation where value stocks, cyclicals, and international markets are gaining traction while mega-cap technology stocks have pulled back, a common occurrence during bull markets [1][12]. Group 1: Market Dynamics - The Magnificent 7 stocks have traded sideways to modestly lower since last November due to concerns over AI overinvestment and valuation multiples, yet their fundamental positioning remains strong [2][12]. - AI is not the sole growth driver for these companies; they are also central to long-term trends in cloud computing, digital advertising, e-commerce, and more, with AI acting as an accelerator rather than a primary driver [3][12]. Group 2: Cloud Computing Strength - Companies like Amazon, Alphabet, and Microsoft are seeing a re-acceleration in cloud computing growth, driven by strong demand for AI-related compute capacity, with hyperscalers reporting sold-out capacity [5][6]. - Valuations for these companies have become more reasonable following recent market consolidation, making them attractive for investors seeking exposure to AI infrastructure [7][12]. Group 3: Individual Company Insights - Meta Platforms is highlighted for its compelling valuation, trading at less than 22x earnings with projected long-term earnings growth of around 20%, while also benefiting from AI-driven improvements [9]. - Nvidia remains a key player in the AI infrastructure space, trading at mid-20s on forward earnings with a PEG ratio near 0.5, indicating strong growth relative to valuation [10]. - Apple has maintained a stable position without aggressive AI infrastructure spending, showing renewed strength in iPhone demand and a rapidly expanding services segment, indicating resilience in its ecosystem [11]. Group 4: Future Outlook - The recent trading pattern of the Magnificent 7 suggests a healthy consolidation rather than a breakdown, with expectations that mega-cap technology may reassert leadership as cloud demand and AI adoption continue to grow [12][13]. - The long-term competitive positioning of these dominant companies remains intact, suggesting they are well-positioned to lead future market advances if the broader bull market persists [14].

OpenAI正在重新调整其扩张预期，向投资者表示到2030年的总算力支出目标约为6000亿美元，较CEO Sam Altman此前宣称的1.4万亿美元基础设施承诺大幅缩水。 2月20日据CNBC报道，OpenAI预计2030年总收入将超过2800亿美元，较2025年的131亿美元大幅增 长，消费者和企业业务贡献基本持平。 而据The Information报道，OpenAI调整后的毛利率在2025年已下滑至33%，该公司仍预计到2030年现 金流才能转正。 该公司的编程产品Codex周活跃用户已超过150万，与Anthropic的Claude Code形成直接竞争，后者在过 去一年中获得了大量用户。 与此同时，OpenAI正在敲定一轮可能超过1000亿美元的大规模融资，其中约90%来自战略投资者。据 CNBC证实，英伟达正商讨投资至多300亿美元，该轮融资对OpenAI的投前估值可能达到7300亿美元。 风险提示及免责条款 用户增长与竞争压力并存 市场有风险，投资需谨慎。本文不构成个人投资建议，也未考虑到个别用户特殊的投资目标、财务状况或需要。用户应考虑本文中的任何 意见、观点或结论是否符合其特定状况。据此投 ...

OpenAI is projecting that its revenue will grow at a fast clip in the next few years and exceed $280 billion in 2030, according to a person familiar with the matter. The revenue forecast reflects OpenAI’s strong momentum in subscription sales for its AI software to consumers and businesses. OpenAI also recently began testing advertising for certain users, creating a new potential moneymaker for the company. Most Read from Bloomberg OpenAI Chief Financial Officer Sarah Friar recently said the company’s ...