2026.02. 13 本文字数：1558，阅读时长大约3分钟 作者 | 第一财经 刘晓洁 封图 | AI生成 北京时间2月13日，OpenAI强有力的竞争对手Anthropic宣布完成G轮融资，共筹集300亿美元，投 后估值达3800亿美元（约合人民币2.6万亿元）。 这是目前大模型行业第二大规模的融资，仅次于OpenAI 去年历史性的400亿美元交易，也使得 Anthropic成为全球第二家2万亿级AI独角兽——OpenAI的最新估值为5000亿美元。 去年9月，Anthropic才宣布完成 130 亿美元的F轮融资，估值为1830亿美元，这轮融资后Anthropic 估值也成功实现翻倍。按这一估值计算，Anthropic在全球上市公司中能排进Top30。 Anthropic本轮融资的明星投资者名单不少，官方公告显示，这轮融资由新加坡政府投资公司 (GIC) 和 Coatue 领投，联合领投方包括 DE Shaw Ventures、Dragoneer、Founders Fund、ICONIQ 和 MGX。 Anthropic的博客提到，Claude Code 的年化收入已超过 25 亿美元，自 2026 ...

当前，人工智能正加速从技术探索走向产业应用。2月11日，由湖北大数据集团投资建设的AI工厂在武 汉小洪山数据产业园正式建成并投入运营。AI工厂是集团打造的数智化支撑核心设施，将通过集约 化、标准化的生产方式，降低人工智能应用开发门槛，为千行百业数智化转型提供服务支撑。 一、破解开发难题：AI工厂因何而生 长期以来，人工智能应用开发面临三重门槛：算力获取难、数据合规成本高、算法开发周期长。传统模 式以项目制、分散化为主，技术栈不统一，人才要求高，数据与算力资源重复投入，产出质量参差不 齐，导致开发周期动辄数月、投入成本高达数百万元，规模化落地困难重重。 而AI工厂将人工智能开发的三大核心要素——算力、数据、算法深度整合、系统打通，提供全栈式AI 服务。 管理，保障模型服务稳定可靠；场景运营中心则是作为AI工厂的应用展示与体验窗口，提供重点场景 的全方位演示，实现能力可感知、效果可呈现。 二、看得见的生产线：四大中心与三大车间 在AI工厂内，三大核心生产车间协同运转、流水作业——数据车间将原始数据加工为高质量数据集， 模型算法车间负责通过高质量数据集训练模型，智能体车间利用模型、工作流、知识库等能力打造不同 场景 ...

Core Insights - Ant Group has released the world's first trillion-parameter thinking model, Ring-2.5-1T, based on a hybrid linear architecture, achieving leading performance in long text generation, mathematical reasoning, and agent task execution [1][2] Performance Metrics - In terms of generation efficiency, Ring-2.5-1T reduces memory footprint by over 10 times compared to the previous model in long text generation scenarios exceeding 32K, while improving generation throughput by over 3 times [1][2] - The model demonstrates deep reasoning capabilities, achieving gold medal levels in self-assessments for the International Mathematical Olympiad (IMO 2025) with a score of 35 and the Chinese Mathematical Olympiad (CMO 2025) with a score of 105 [1][2] Compatibility and Applications - Ring-2.5-1T is easily adaptable to agent frameworks such as Claude Code and the OpenClaw personal AI assistant, supporting multi-step planning and tool invocation [1][2]

Group 1 - The core accusation against SpaceX involves alleged secret investments from Chinese investors through offshore entities, which may pose a national security risk [2][3] - The senators' letter cites media reports and court testimonies claiming that Chinese capital is funneled through special purpose vehicles in tax havens to bypass regulations [3] - The recent acquisition of Musk's AI company xAI by SpaceX is highlighted as a significant risk factor, suggesting that the integration of technologies could create a "composite threat" to U.S. military and civilian infrastructure [5] Group 2 - SpaceX is noted for its critical role in U.S. national security, handling 80% of military satellite launches and supporting NASA's lunar missions [6] - The political context includes previous accusations against Musk regarding his business ties with China, which have been framed as serious national security threats [8] - The ongoing investigation reflects deeper political struggles in the U.S., where national security concerns are being used as a tool in partisan conflicts, potentially undermining market fairness and the stability of core technologies [10]

智能体和具身智能是通向通用人工智能（AGI）的必经之路！ 今日（2月13日），港股智谱盘初快速拉升，涨超20%，股价再创新高。过去5个交易日，该股最大涨幅达 142%。2月12日凌晨，智谱发布新一代模型GLM-5。火遍全球的神秘模型"Pony Alpha"揭开面纱。业界称，这 是继DeepSeek后，国产AI大模型春节档的又一爆款。 智能体与具身智能的融合，是AGI的必经之路，也是打开具身智能市场容量的必要环节。近期，机器人也在预 热当中。宇树科技创始人王兴兴谈及具身智能赛道时表示，未来热度或是现在的1000倍，远超移动互联网。 智能体彻底爆发 今日早盘，智谱盘初快速拉升涨超20%，股价一度达到492港元/股，再创上市以来新高，与上周五（2月6日） 的收盘价203.2港元/股相比，最近一周累计涨幅达142%。今日消息面显示，此前Coding Plan连续包月、连续包 季、连续包年套餐全部售罄。 真正令智谱爆发的还是它的大模型。近日，智谱正式推出新一代旗舰模型GLM-5，主攻编程与智能体能力， 官方称已实现开源领域最优表现。GLM-5参数规模由上一代的355B扩展至744B，激活参数从32B提升至40B。 智谱 ...

"全球大模型第一股"智谱继续冲A。 近日，中国证监会官网更新了智谱公司的IPO辅导进展信息，智谱撤回了其于2025年4月提交的辅导备案，并办理了新的辅导备案登记。 智谱涨价的底气，来自GLM-5的技术突破。这款模型即匿名模型"Pony Alpha"，在海外早已"未发先火"。其面向复杂系统工程与长程Agent任务的开源基 座模型，在大型编程任务中提供对标顶尖闭源模型的生产力，能帮助程序员大幅提高开发效率。 最新辅导备案报告显示，智谱拟冲刺科创板。同时，其IPO辅导机构变更为两家券商，由原来的中金公司变更为国泰海通证券和中金公司两家。 2月13日，智谱港股走势持续上扬，盘中最高涨超22%，截至发稿涨幅近17%，市值近2100亿港元。 | 智谱 < V ાં વ | | --- | | 02513 | | 470.000 今开 422.000 最高 492.000 最低 408.000 | | 16.92% 68.000 换手 1.27% 总手 281万 金额 13.00亿 | | 更多 总值 2095亿 市盈TTM-47.44 市净 -31.09 | 行情的直接催化剂，是智谱2月12日发布的一则涨价声明：因市场需 ...

从定位上看，Codex-Spark 并不是为了替代现有的 Codex，而是补齐其在"即时交互"场景中的短板：在过去，Codex 更擅长长时间运行的复杂任务，而 Codex-Spark 的目标则非常明确——把人与模型之间的交互延迟压缩到接近"无感"的程度。 这一发布同时也是 OpenAI 与 芯片初创企业 Cerebras 合作的重要阶段性成果。为了减少对英伟达芯片的依赖，上个月 OpenAI 签署了一项金额超过 100 亿 美元的协议，使用 Cerebras 的硬件以提升其模型的响应速度，而 Codex-Spark 被视为这项合作落地的第一个技术里程碑。 为实时而生：Codex-Spark 的核心是"速度" OpenAI 发布新模型，专为实时编码而生 昨晚，OpenAI正式发布了GPT-5.3-Codex-Spark的研究预览版本。这是一款从 GPT-5.3-Codex 主模型中"裁剪"而来的精简版本，同时也是 OpenAI 首个专 门围绕实时编码（real-time coding）场景设计的模型。 在官方定义中，Codex-Spark 是一个"专为实时使用 Codex 而设计的模型"，它支持进行针对性编辑、 ...

Core Insights - Anthropic has successfully completed a Series G funding round, raising $30 billion with a post-money valuation of $380 billion, making it the second-largest funding round in the large model industry after OpenAI's $40 billion deal last year [1][3] - The funding round was led by Singapore's GIC and Coatue, with participation from notable investors including Blackstone, Goldman Sachs, and Microsoft, which plans to invest up to $5 billion [4][6] - Anthropic's revenue has reached an annualized figure of $14 billion, growing over tenfold in the past three years, driven by its Claude model, which has gained significant traction among Fortune 10 companies [4][6] Company Performance - The number of Claude customers spending over $100,000 annually has increased sevenfold in the past year, with over 500 customers now spending more than $1 million annually [6] - Claude Code, Anthropic's AI programming tool, has surpassed $2.5 billion in annualized revenue, doubling since early 2026, with enterprise subscription users growing fourfold [7] - Anthropic is recognized as a strong competitor to OpenAI, with its core team originating from OpenAI and significant backing from tech giants like Amazon and Google [7] Competitive Landscape - OpenAI is accelerating its product and funding strategies in response to competition, with plans for a new funding round in 2026 targeting $100 billion, potentially increasing its valuation to $830 billion [8][9] - Major tech companies, including Amazon, Google, Microsoft, and Meta, are projected to spend a total of $660 billion on data centers and chip development by 2026, raising the capital expenditure threshold for AI model development [9] Future Outlook - The recent funding will enable Anthropic to expand its infrastructure and ensure Claude's availability across all customer regions, utilizing various hardware for training and operation [10]

Core Viewpoint - The "Hangzhou Dialect Systematic Collection Project" aims to develop an AI capable of speaking authentic Hangzhou dialect, addressing the challenges posed by the dialect's unique phonetic characteristics and the declining number of fluent speakers among the younger generation [1][2][3]. Group 1: Project Overview - The project is led by Zhejiang University and Alibaba's AI Data team, with the goal of preserving and transmitting the Hangzhou dialect through modern technology [1][3]. - The project is currently recruiting volunteers who can speak natural and authentic Hangzhou dialect, particularly from the old urban areas of Hangzhou [2][3]. Group 2: Linguistic Challenges - Hangzhou dialect, part of the Wu language family, retains ancient Chinese phonetic features, including seven tones and tonal changes in connected speech, making it difficult for AI to learn [2]. - The project emphasizes the need for data collected from native speakers to ensure the AI does not produce a distorted version of the dialect [2]. Group 3: Cultural Significance - The project highlights the urgency of preserving the Hangzhou dialect, as fluent speakers are becoming increasingly rare, particularly among the younger population [3][4]. - Participants in the recording process will contribute to a high-quality database that will serve future research, technology training, and cultural preservation [3]. Group 4: Participation and Future Prospects - Volunteers can record their speech using a mobile app, with flexible recording times and compensation for their contributions [3]. - The richness of the collected data will determine the speed of the AI model's development, which aims to generate extensive Hangzhou dialect text and speech for various applications [3].

Core Argument - The article discusses the profound impact of artificial intelligence (AI) on the job market, suggesting that the U.S. is unprepared for the potential disruptions it may cause to employment and economic stability [1]. Group 1: Historical Context and Current Trends - The establishment of the U.S. Bureau of Labor Statistics (BLS) aimed to measure labor conditions and create fair outcomes amidst industrial changes, highlighting the importance of data in understanding economic realities [5][6]. - The BLS has documented significant job growth in various sectors, such as a 907% increase in mobile food service jobs since 2000, indicating a dynamic labor market [6]. - However, the BLS is limited in its predictive capabilities, particularly regarding the impact of emerging technologies like AI on the workforce [7]. Group 2: AI's Impact on Employment - AI is rapidly transforming job functions, enabling tasks to be completed more efficiently than ever before, which raises concerns about job displacement [8][9]. - Predictions from industry leaders suggest that AI could lead to a 10% to 20% increase in unemployment rates and potentially eliminate half of entry-level white-collar jobs within the next decade [10]. - A Reuters/Ipsos survey indicates that 71% of Americans fear AI will lead to permanent job losses, reflecting widespread anxiety about the future of work [9]. Group 3: Economic Resilience and Job Creation - Economists argue that capitalism has a strong resilience, often leading to job creation following technological advancements, as seen with ATMs and software like Excel [8]. - The BLS forecasts a 3.1% employment growth rate over the next decade, which, while lower than previous years, still represents the addition of 5 million jobs [8]. Group 4: The Role of Policy and Corporate Responsibility - There is a growing concern that corporate leaders are prioritizing automation and efficiency over employee welfare, leading to potential mass layoffs [22][23]. - The article suggests that CEOs are under pressure to demonstrate the benefits of AI quickly, often resulting in job cuts rather than exploring ways to integrate AI while supporting their workforce [22][23]. - Proposals for policies such as retraining programs and a robot tax to support displaced workers are discussed, but there is skepticism about their implementation [33][28]. Group 5: Political and Social Implications - The political landscape is characterized by a lack of proactive measures to address the challenges posed by AI, with many lawmakers adopting a hands-off approach [26][27]. - The article emphasizes the need for a coordinated response to the potential upheaval caused by AI, suggesting that without intervention, the consequences could be severe for both the economy and society [30][31].