行业研究 证 券 研 究 报 告 计算机行业重大事项点评 政策落地，数据+AI 驱动要素价值释放 事项： ❑ 2026 年 2 月 7 日，国家数据局与工信部等多部门联合发布《关于培育数据流 通服务机构加快推进数据要素市场化价值化的意见》，提出到 2029 年底，数 据流通服务机构能力显著提升，流通交易形态更加多元，数据产品和服务更加 丰富，各类主体供数用数意愿持续增强，全社会数据流通利用水平明显提高。 计算机 2026 年 02 月 26 日 推荐（维持） 华创证券研究所 证券分析师：吴鸣远 邮箱：wumingyuan@hcyjs.com 执业编号：S0360523040001 行业基本数据 | | | 占比% | | --- | --- | --- | | 股票家数(只) | 337 | 0.04 | | 总市值(亿元) | 61,676.90 | 4.81 | | 流通市值(亿元) | 55,806.98 | 5.39 | 评论： 《计算机行业重大事项点评：CPU：供需格局优 化，国产龙头或迎价值重估机遇》 2026-01-29 《计算机行业重大事项点评： Agent ：海外 Clawdbot 引爆市场 ...

Core Insights - The alliance between founders and companies in Silicon Valley is deteriorating, replaced by lucrative offers and constant talent "transfers" [2][4] - Major tech giants like Meta, Google, and Nvidia have spent over $36 billion on "acquisition hiring" since mid-2025, directly poaching top talent rather than traditional acquisitions [2][5] - The concept of loyalty is collapsing, with founders leading the way in leaving their companies, and early employees becoming mere bargaining chips [2][12] Group 1: Talent Acquisition Dynamics - Compensation has no ceiling, with reports of Meta offering over $200 million total packages to top talent, making nine-figure offers commonplace [2][12] - "Legal poaching" has become mainstream, exemplified by Nvidia's $20 billion technology licensing deal with Groq, which is seen as a workaround for antitrust regulations [2][7] - The trend of "acquisition hiring" focuses on securing top talent and their technological contributions rather than acquiring products [5][7] Group 2: Impact on Company Culture - The massive salary disparities between new and existing employees are creating tensions within companies, as seen in Meta's attempts to mitigate the impact of high salaries on company culture [12][18] - The traditional narrative of loyalty and shared mission is being replaced by a transactional mindset, where founders prioritize personal financial gain over employee welfare [14][17] - The shift towards viewing teams as low-risk assets by investors reflects a changing landscape where talent acquisition is prioritized over product success [18][19] Group 3: The Future of Silicon Valley - The ongoing talent war is reshaping the industry, leading to more concentrated capital and income, as well as faster talent turnover [21] - The emergence of a new social contract for tech talent is anticipated, involving stricter confidentiality agreements and collective bargaining [20] - The long-term winners in this evolving landscape may be those companies that can foster team cohesion and create lasting value amidst the chaos [21]

Alexandr Wang cofounded Scale AI in 2016 to help companies put their raw data to work for AI and machine learning. In June 2025, Meta acquired 49% of the company in a deal that valued Scale at around $29 billion and Wang left Scale to become Meta's chief AI officer. See where he lands on the #Forbes250 America’s Greatest Innovators list. https://t.co/S1qgkoTqmt (📸: Ethan Pines for Forbes) ...

Core Insights - The rise of the profession of Prompt Engineer is attributed to the limitations of AI, which requires human guidance to interpret user needs and generate appropriate responses [1][2] - The profession gained popularity after the launch of ChatGPT in 2022, with significant salary potential and a lack of technical background requirements [2][4] - However, by early 2025, the role was deemed obsolete by industry experts, leading to a rapid decline in demand for Prompt Engineers [2][3] Group 1: Emergence and Popularity - The profession of Prompt Engineer emerged as a response to the need for human interaction with AI models, particularly after the introduction of ChatGPT [1] - In 2023, the role was considered one of the most attractive in the tech industry, with salaries reaching up to $335,000, and many companies planning to hire Prompt Engineers [2][4] - A survey indicated that nearly 29% of companies intended to hire Prompt Engineers in 2023, with about 25% expecting starting salaries to exceed $200,000 [2] Group 2: Decline and Obsolescence - By early 2025, the role of Prompt Engineer was labeled as "dead" by a top researcher at OpenAI, marking a swift decline in its desirability [2][3] - A Microsoft survey revealed that Prompt Engineers were among the least desired positions for companies to add in the next 12 to 18 months [2][3][18] Group 3: Job Responsibilities and Evolution - Initially, the responsibilities of Prompt Engineers were not well-defined, often resembling that of AI consultants, leading to high salaries based on unclear job roles [7][11] - As AI technology evolved, the role required a deeper understanding of product management and technical skills, transitioning from a simple prompt-writing task to a more integrated role involving product development [16][19] - The market is shifting towards hiring hybrid talents who can navigate both AI technology and product management, indicating a move from generalist to specialist roles [19][20] Group 4: Future Outlook - The demand for Prompt Engineers is expected to evolve, with a focus on vertical expertise in fields like healthcare, finance, and government, requiring 1-3 years of industry experience and programming knowledge [19][20] - The profession is seen as transitional, with the need for professionals who can adapt to the changing landscape of AI and its applications [19][20]

Core Insights - Venture capital funding for artificial intelligence (AI) and machine learning startups surged by 72% in 2025, marking a significant milestone where VC dollars for these ventures exceeded all other sectors combined for the first time [1][3]. VC Funding Overview - Global funding for AI startups reached $270.2 billion, representing 52.7% of the total $512.6 billion in venture capital investments [2]. - Despite the increase in funding, the overall number of VC deals declined for the third consecutive year, with 9,844 registered deals in the fourth quarter of 2025, the lowest since early 2020 [3]. Deal Dynamics - The trend indicates fewer deals but larger investments, similar to M&A activity, highlighted by SoftBank's $40 billion investment in OpenAI, the largest single investment in a private company [4]. - Other significant investments included Meta's $14.3 billion in Scale AI and Anthropic's $13 billion funding round at a valuation of $183 billion [5]. Regional Distribution - Of the $270 billion invested in AI by VC firms, 79.3% was allocated to North America, 13.6% to Europe, 5.7% to Asia, and only 0.5% to Latin America [6]. Industry Trends - Many heavily funded startups, such as Thinking Machine Labs and Safe Superintelligence, were founded by former OpenAI staff, indicating a concentration of AI expertise in elite startups [7]. - The share of AI in global VC deal value increased from 27.5% in 2023 to 40% in 2024, and further to 52.7% in 2025 [7]. Exit Value - The aggregate value of AI and machine learning exits was $242.4 billion, accounting for about 40% of all exit value, a significant increase from $73.6 billion and 22% in 2024 [8].

Core Insights - Meta Platforms has shifted its focus from the metaverse to artificial intelligence (AI), with significant investments planned for the future [1] Group 1: Investment and Expenditure - Meta plans to increase its capital expenditures significantly, with a projected spending of $135 billion on new data centers, marking an 87% increase from the previous year [2] - The investment will support the AI Superintelligence team, which includes a $14 billion investment in Scale AI for a 49% stake [3] Group 2: Profitability and Financial Impact - The increase in capital expenditures is leading to higher depreciation expenses, which rose by 20% in 2025 and are expected to increase further in 2026 [4] - Meta's operating margin decreased from 48% in Q4 2024 to 41% in 2025, although management anticipates overall operating income growth in 2025 despite margin pressures [5] Group 3: Growth Potential in AI - Meta has effectively utilized AI advancements to drive revenue growth, achieving a 24% revenue increase in Q4 due to strong user engagement and rising ad prices [6] - The integration of large language models (LLMs) with the machine learning recommendation system is expected to enhance user engagement by providing more tailored content feeds [7]

Group 1 - The article highlights the emergence of young billionaires in Silicon Valley, particularly focusing on the rapid rise of Steve Hao, co-founder and CTO of Cognition.AI, whose net worth has reached approximately $1.3 billion after a recent funding round that valued the company at $10.2 billion [2][28][29] - Cognition.AI's flagship product, Devin, is described as the world's first AI programmer, capable of autonomously completing software development tasks from start to finish, which has garnered significant attention and investment [5][8][12] - The company has successfully navigated competitive challenges, including acquiring the assets of Windsurf, a rival AI coding startup, which has enhanced Cognition's market position and product offerings [10][12][28] Group 2 - Cognition.AI's founders, all of whom are young Chinese entrepreneurs, have impressive backgrounds in competitive programming, with multiple International Olympiad in Informatics (IOI) medals among them, which has contributed to their collaborative success [13][16][18] - Following the launch of Devin, Cognition.AI experienced a significant increase in its valuation, from approximately $350 million to $2 billion within months, and further to $9.8 billion by August 2025 [22][28] - The annual recurring revenue (ARR) for Devin surged from about $1 million in September 2024 to $73 million by mid-2025, indicating strong market demand and customer acquisition, including major clients like Dell and Cisco [33]

Core Insights - Meta has developed a new generation large language model named "Avocado," which is claimed to be the company's most powerful pre-trained model to date, achieving significant advancements in knowledge, visual perception, and multilingual performance without fine-tuning [1] - The model "Avocado" reportedly offers a tenfold efficiency improvement over the Llama 4 "Maverick" version and a hundredfold improvement compared to the unreleased "Behemoth" version, attributed to higher quality data, infrastructure investments, and the adoption of "deterministic training" methods [1] - Meta's anticipated capital expenditures for AI are projected to surge to between $115 billion and $135 billion by 2026, making these efficiency gains crucial for managing costs while competing with rivals [1] Group 1 - Meta's CTO Andrew Bosworth described the newly formed team's model as "excellent," but noted that the technology is not yet fully mature and requires significant fine-tuning before it can be made available to users [2] - Bosworth acknowledged that 2025 will be a chaotic year for building infrastructure and ensuring computational resources, but he believes that the substantial investments are beginning to yield returns [2] - The company has faced setbacks in its AI development history, including delays in the release of the Llama 4 model due to performance issues, which led to a major strategic shift in its AI approach [2] Group 2 - CEO Mark Zuckerberg expressed a pragmatic and forward-looking view on the outputs from the Super Intelligence Lab, indicating that the initial models will be promising and will demonstrate the company's rapid progress [3] - The leak of the internal memo and subsequent confirmation at the Davos forum signal that Meta is attempting to convey a clear message about its AI research breakthroughs following a restructuring and record investments [3] - Both the "Avocado" model and the initial models described as "excellent" carry the hope for Meta to turn around its AI strategy and regain competitive ground [3]

Core Insights - Meta Platforms is optimistic about its new AI team and the upcoming launch of its core large model, Avocado, which has completed pre-training and is described as the company's most capable pre-trained foundational model to date [2][7] - The performance of Avocado has surpassed that of the best current open-source foundational models, and it matches top post-trained models in knowledge retention, visual perception, and multilingual capabilities, despite not yet completing the post-training phase [2][7] Group 1 - The internal memo indicates that Meta's AI model progress is optimistic but remains untested in the external environment, raising potential risks for the company [3][8] - Meta's previous AI model, Llama 4, underperformed, leading to a delay in its release and disappointment among developers regarding its actual performance [3][8] Group 2 - The setbacks in AI development prompted a significant restructuring of Meta's AI business, including the acquisition of Scale AI for $14.3 billion and the establishment of the Meta Superintelligence Labs led by Alexandr Wang [9] - Meta plans to increase its capital expenditure on AI, including computing costs, by approximately 73% in 2026, projecting a total of $115 billion to $135 billion [9] Group 3 - Avocado has demonstrated significant efficiency improvements, achieving a tenfold increase in computational efficiency compared to Maverick and over a hundredfold compared to Behemoth, which has not yet been released [4][9] - The efficiency gains are attributed to higher quality data acquisition, investment in model infrastructure, and the use of deterministic training methods, which are crucial for reducing energy consumption and costs in AI development [10] Group 4 - Recent public statements from Meta executives align with the positive tone of the internal memo, with CTO Andrew Bosworth highlighting similar efficiency improvements and CEO Mark Zuckerberg expressing confidence in the performance of upcoming models [5][10]

Core Insights - The article discusses the significant salary increases and talent competition in the artificial intelligence (AI) sector, highlighting how top talent is negotiating salaries in the millions, contrasting sharply with traditional job markets [5][7][41]. Salary Revolution - Nvidia's CEO Jensen Huang claims that his team has produced more billionaires than any other CEO, reflecting the company's stock price surge and the wealth accumulation of long-term employees [9]. - Salaries for AI research scientists can reach up to $489,000 at Meta, while OpenAI offers around $440,000, and Google's DeepMind can go as high as $893,000 for senior positions [14]. - AI product managers earn between $150,000 and $900,000, with Netflix setting a notable salary range of $300,000 to $900,000 for this role [15][16]. - The emergence of roles like "prompt engineers" and "Chief AI Officers" has seen salaries of $270,000 and $351,000 respectively, indicating a shift in job market dynamics [18][19]. - OpenAI's remote employees earn between $160,000 and $490,000, with a median base salary of $285,000, showcasing the high value placed on AI talent [20]. Talent War - Meta's aggressive recruitment strategy includes a $14.3 billion investment in Scale AI to attract its co-founder, Alexandr Wang [22]. - Mark Zuckerberg has taken a personal role in recruitment, creating a WhatsApp group for hiring discussions and directly reaching out to potential candidates [24]. - OpenAI has responded to Meta's poaching attempts by offering counter-offers to retain talent, with employees potentially receiving $1.5 million in stock as part of their compensation [32][33]. Economic Justification - The high salaries in AI are justified by the substantial costs associated with building AI models, with estimates indicating that training models like GPT-4 and Gemini 1.0 Ultra can exceed $79 million and $192 million respectively [42][45]. - The scarcity of top AI talent, with only about 5,000 individuals capable of designing advanced AI models, has created a seller's market, driving up salaries [46][47]. Employment Structure Changes - The article notes a dual-track employment structure in the tech industry, where AI elites command multi-million dollar offers while traditional roles face obsolescence [51][56]. - Companies like Meta are laying off employees while simultaneously offering lucrative salaries to AI researchers, indicating a shift in workforce dynamics [53][54].