美国软件行业将会面临这样的局面：一些小型公司最终会变得一文不值，而巨头公司仍将拥有惊人的投 资者偏好。 同样关键的是，即使去年美股科技股上市有所回暖，据Dealogic分析的数据显示，去年上市的美股科技 股中，四分之三目前的交易价格已跌破其IPO发行价。截至美东时间周二收盘，2025年所有发行IPO的 美股科技股平均下跌了18%。Dealogic追踪的这些公司发行规模均至少达到5000万美元。 其中，Figma可谓热门科技公司发行IPO风险的最佳写照。这家软件公司于去年7月下旬上市，为公司和 投资者筹集了约12亿美元资金，成为自2021年以来规模最大的风险投资支持的科技IPO。Figma的股价 在上市首日飙升了250%，一度导致IPO承销商因此前对该股定价过低而面临巨大压力。Figma的成功更 引发了去年稍晚的一波美股科技股上市浪潮，紧随其后上市的包括StubHub、Klarna和Navan。 今年原本应该是预期中美股首次公开募股（IPO）的爆发之年。然而，对人工智能（AI）的担忧却破坏 了这一切，令许多科技企业在美国市场的IPO陷入停滞。 截至美东时间周二（24日）收盘，去年上市的美股科技股中，四分之三录得 ...

格隆汇2月25日丨中望软件(688083.SH)在互动平台表示，公司始终对于国内外优质研发设计类工业软件 标的保持密切的关注，并积极尝试通过收并购增强公司研发能力与核心技术储备。后续若有具体收购计 划，公司将依据相关法律法规及时履行信息披露义务。 与此同时，公司当下专注于研发设计类工业软 件领域，目前的关注重点主要为二、三维CAD以及CAE，因此暂未有通过并购进入EDA行业的计划。 ...

格隆汇2月25日丨东华软件(002065.SZ)在投资者互动平台表示，公司在过去一年已为国民经济各行业建 设了几十亿规模的算力中心（云中心），采用并集成华为、摩尔、英伟达等多家厂商的算力产品及平 台。公司与腾讯、阿里、字节跳动均有深入的互为客户式合作，围绕医疗、金融、未来城市、公安、大 企业等领域开发行业智能体，其中"张警官智能体"深受公安干警欢迎，同时公司还为上述企业提供网络 产品及运维解决方案，相关业务与对方AI智能体均实现完全兼容。 ...

每经AI快讯，2月25日，阿里继续开源千问3.5系列模型。本次开源三款中等规模的新模型，包括 Qwen3.5-35B-A3B、Qwen3.5-122B-A10B和Qwen3.5-27B。目前，Qwen3.5-Flash已上线阿里云百炼，每 百万Token输入低至0.2元。 ...

Core Viewpoint - Snowflake is anticipated to report fourth-quarter product revenue of $1.2 billion, which could serve as a catalyst for the stock to recover from recent declines in the software sector [1]. Group 1 - Analysts expect Snowflake's fourth-quarter product revenue to reach $1.2 billion [1]. - The company has been negatively impacted by the broader software selloff [1]. - Earnings reports are viewed as a potential trigger for the stock's recovery [1].

北京商报讯（记者 魏蔚）2月25日，阿里云百炼推出包含Qwen3.5、GLM-5、MiniMax M2.5、Kimi K2.5 四款开源模型API（应用程序编程接口）服务的Coding Plan（编程模型订阅服务）。用户订阅套餐后不 再受限于单一模型，可实现多模型自由切换。Coding Plan Lite基础套餐可享受每月最多1.8万次请求， Pro高级套餐可享受每月9万次请求。用户订阅后，可在Qwen Code、Claude Code、Cline、OpenClaw等 AI工具上无缝切换使用这些模型。 ...

Core Insights - Alternative assets have transitioned from niche allocations to essential components of institutional and wealth portfolios, prompting managers to explore diverse strategies and channels as competition increases and limited partners (LPs) demand more transparency [1] Group 1: Industry Trends - Private markets managers are currently operating in siloed data environments, and to fully leverage AI, they need coherent operating models rather than isolated tools [2] - The demand for transparency and efficiency in alternative asset management is driving the need for advanced operating models that minimize manual work [1][2] Group 2: Navatar's AI-Powered Model - Navatar has developed an AI-powered CRM operating model that features a single AI Deal Engine, which integrates various functions such as sourcing, diligence, capital deployment, and investor engagement [3] - The AI continuously captures intelligence and maintains institutional context, automating workflows across the private markets lifecycle without relying on manual updates [3] Group 3: Functionality and Applications - Navatar's AI is designed to serve as a firm-wide intelligence layer for both multi-strategy platforms and focused managers, capturing signals from daily workflows and transforming them into actionable insights [4] - In the origination phase, the AI maps networks and identifies early signals, helping teams prioritize opportunities based on historical data [5] - The model extends to capital formation and investor relations, interpreting engagement across various channels and linking it to deal activity for improved fundraising precision [6] Group 4: Security and Compliance - The AI operating model is built to meet the confidentiality and regulatory requirements of alternative asset managers, ensuring client data remains secure and is not exposed to public AI models [7] - The platform includes safeguards to enhance accuracy, completeness, and traceability in critical decision-making processes [7] Group 5: Market Applications - Navatar's platform supports various sectors within private markets, including private equity, venture capital, private credit, and real estate, by providing insights that enhance decision-making and risk evaluation [8] - The AI facilitates connections between technical, operational, and market data, allowing teams to better assess value creation opportunities [8]

Core Viewpoint - The article highlights that while quantum computing has potential, it is not yet a near-term growth driver for companies like IBM, which offers a more established investment opportunity with significant exposure to quantum innovation and a profitable AI-driven business [1][2][4]. Quantum Computing Strategy - IBM has been investing in quantum computing hardware and software for decades, aiming to achieve "quantum advantage" by the end of 2026 and develop a large-scale, fault-tolerant computer by 2029 [5][8]. - The first 120-qubit Nighthawk system is set for client use in late 2025, indicating progress in quantum capabilities [7]. Financial Performance - In Q4, IBM's software revenue increased by 14% year-over-year to $9 billion, making up nearly 46% of total revenue, showcasing a shift towards a software-led model [10]. - IBM generated $14.7 billion in free cash flow in fiscal 2025, with guidance for nearly $15.7 billion in 2026, providing flexibility for dividends and acquisitions [11]. AI Strategy - AI is a significant growth engine for IBM, with a cumulative generative AI business valued at over $12.5 billion, including over $2 billion in software and more than $10.5 billion in consulting [12]. - The company's AI stack, including the watsonx suite and Red Hat OpenShift platform, is designed to enhance productivity and flexibility across hybrid cloud environments [13]. Competitive Positioning - IBM's acquisition strategy, including a planned $11 billion acquisition of Confluent, aims to enhance its AI capabilities by providing access to live enterprise data [14]. - Management anticipates that nearly 50% of enterprise AI usage will occur in private clouds or on-premise data centers in the next three to five years, positioning IBM favorably in this market [15]. Valuation - IBM is trading at 19.2 times forward earnings with a 2.6% dividend yield, presenting a balanced exposure to quantum computing research, AI monetization, recurring software revenue, and strong free cash flows [17].

Core Insights - Insider buying is a strong bullish indicator, particularly relevant for assessing the future performance of stocks, especially in the AI sector [1][2] - The actions of insiders, who have better knowledge of the company's internal situation, can provide insights into the potential rise in stock prices [2] Company Overview: Salesforce - Salesforce has experienced a challenging market performance, with shares down 30% year-to-date as of February 19 [5] - The company has had limited insider activity recently, with only four transactions recorded: two buys and two sells [5] Insider Transactions - Harris Parker, co-founder of Salesforce, sold $31.6 million in shares in December 2025 as part of a pre-planned 10b5-1 plan, which is not indicative of the company's performance [6] - Neelie Kroes, a director, sold 3,893 shares for $929,276 in January 2026, but the reasons for such sales can vary [7] - On the buying side, Kirk David Blair, a director, purchased 1,936 shares for $500,722 in mid-December 2025, signaling confidence in Salesforce's future [9] - The most significant purchase was made by Mason Morfit, co-CEO of ValueAct Capital, whose firm bought $25 million worth of Salesforce shares on December 5, indicating strong confidence in the company's prospects [10] Insider Buying Summary - Excluding the 10b5-1 sales, net insider buying totaled approximately $24.6 million over the past three months, suggesting positive sentiment among insiders [11] - Upcoming fiscal results for 2026 are anticipated on February 25, with a product release planned for February 23, 2026, which will introduce new AI capabilities [11]

新浪科技讯 2月25日午间消息，阿里云宣布Coding Plan订阅服务新增支持Qwen3.5-Plus、Qwen3-Coder- Next、GLM-4.7、Kimi-K2.5共4款顶尖编程模型，用户无需更改订阅和配置即可自由切换。阿里云 Coding Plan还上线了优惠活动，即日起，新用户首购可享2折优惠，首月18000次请求额度仅需7.9元， 90000次请求仅需39.9元。 AI Agent是目前大模型的主要应用载体，业界已涌现Claude Code、OpenClaw等爆火的AI工具。但Agent 需要执行复杂任务，并进行多轮规划与工具调用，Tokens消耗量呈指数级增长，有数据显示，单个 Agent的算力消耗是传统Chatbot的100至1000倍 。为此，阿里云推出Coding Plan订阅服务，为用户提供 性价比更高、可自由选择的模型服务。 此前，阿里云百炼已上线全新的Qwen3.5-Plus，其API价格低至每百万Tokens 0.8元，用户仅需不到5% 的价格即可获得媲美Gemini 3的高性能。现在，阿里云Coding Plan的上新进一步降低了模型的使用成 本。（文猛） 新浪科技讯 2月 ...