我们认为，Mercor的爆发式增长标志着AI招聘由“效率工具”向“能力基础设施”跃迁，打开了招聘行业新的商业化与估值范式。其核心并非自研底层大模型，而在于以 AI 招聘为入口构建的强产品化能力与数据闭环，通过“招聘效率提升 + 专家数据变现”的双引擎模式，验证了 AI 招聘向平台化、数据化演进的可行路径。对照中国市场，尽管Mercor模式难以整体复制，但AI对招聘流程的系统性改造已被本土头部平台验证。展望中长期，我们认为AI招聘的核心价值体现在三条主线：1）模型能力提升驱动存量招聘市场的深度替代；2）大模型演进催生专家数据与人类反馈的增量市场；3）能力可调度的人力资源平台形态打开长期想象空间。在此背景下，看好具备真实业务数据沉淀、模型反馈闭环与平台化潜力的招聘公司，AI应用有望成为其中长期估值抬升的核心催化。
We believe that the explosive growth of Mercor marks the transition of AI recruitment from an "efficiency tool" to a "competency infrastructure," opening up new commercialization and valuation paradigms in the recruitment industry. Its core is not self-researched foundational large models, but rather the strong product capabilities and data closed-loop built around AI recruitment as an entry point. Through the dual-engine model of "improving recruitment efficiency + monetizing expert data," it has validated the feasible path for AI recruitment to evolve into a platformized and data-driven model. Compared to the Chinese market, although the Mercor model is difficult to replicate in its entirety, the systematic transformation of AI in the recruitment process has been proven by domestic leading platforms. Looking ahead in the medium to long term, we believe the core value of AI recruitment is reflected in three main lines: 1) the improvement of model capabilities driving in-depth substitution in the existing recruitment market; 2) the evolution of large models giving rise to incremental markets for expert data and human feedback; 3) the form of human resource platforms with schedulable capabilities opening up long-term imagination space. Against this backdrop, we are optimistic about recruitment companies with accumulated real business data, model feedback loops, and platformization potential, with AI applications likely becoming the core catalyst for their long-term valuation increase.
▍报告缘起：Mercor在17个月内实现估值约40倍增长，映射出AI招聘商业模式的重塑及价值重估逻辑。
▍Origin of the report: Mercor achieved an estimated 40-fold valuation growth in 17 months, reflecting the reshaping and revaluation logic of the AI recruitment business model.
成立于2023年的美国AI创企Mercor，在两年内实现从AI招聘平台向高价值专家调度网络的快速跃迁，估值由2024年9月的约2.5亿美元升至2025年10月的约100亿美元，ARR在不足两年内提升至近5亿美元，并已服务OpenAI、Anthropic、Google等头部AI实验室。
Founded in 2023, the U.S.-based AI startup Mercor has rapidly transitioned from an AI recruitment platform to a high-value expert scheduling network within two years, with its valuation rising from approximately $250 million in September 2024 to $1 billion in October 2025, and its ARR increasing to nearly $500 million in less than two years, having already served leading AI labs such as OpenAI, Anthropic, and Google.
我们认为，Mercor的核心价值并非体现在是否自研底层大模型，而在于其突出的产品化能力与由此构建的数据闭环飞轮，一方面，Mercor以AI招聘为能力入口，对招聘、评估、任务执行及结果反馈等关键环节进行全流程数字化与结构化处理，使高端人才筛选与能力评估具备可量化、可复用的产品形态；另一方面，依托高端招聘与RLHF等真实任务场景，平台持续沉淀模型表现、专家判断与实际产出之间的对应关系，形成他人难以复制的反馈数据闭环。
We believe that the core value of Mercor does not lie in whether it independently develops foundational large models, but rather in its outstanding productization capabilities and the data feedback loop it has built. On one hand, Mercor uses AI recruitment as an entry point to digitally and structurally process key stages such as recruitment, evaluation, task execution, and result feedback, making high-end talent screening and capability assessment available in a quantifiable and reusable product form. On the other hand, by leveraging high-end recruitment and real-world task scenarios like RLHF, the platform continuously accumulates the correlation between model performance, expert judgments, and actual outputs, forming a feedback data loop that is difficult for others to replicate.
在此基础上，即便不自建模型，Mercor仍能够通过上述闭环实现高度精准、差异化的能力匹配，并持续优化其评估与调度体系。进一步看，该模式使Mercor具备向面向复杂任务的自用调度型人力资源平台演进的长期空间，有望推动AI招聘由“效率工具”升级为“能力基础设施”，亦构成其获得显著估值溢价的底层逻辑。
On this basis, even without building its own models, Mercor can still achieve highly precise and differentiated capability matching through the aforementioned feedback loop, and continuously optimize its evaluation and scheduling systems. Looking further, this model gives Mercor long-term potential to evolve into a self-scheduling human resources platform tailored for complex tasks. It is expected to elevate AI recruitment from an "efficiency tool" to an "ability infrastructure," which also constitutes the underlying logic for its significant valuation premium.
▍中国市场分析：落地土壤已具备，尽管Mercor模式仍受结构性约束，但AI招聘路径的商业价值已被验证。
▍China Market Analysis: The foundation for implementation is in place, although the Mercor model still faces structural constraints, the commercial value of the AI recruitment path has been validated.
我们认为，中国在大模型产业进展、高技能人才供给、灵活用工规模及生成式 AI 合规框架等方面已具备支撑类Mercor模式发展的基础条件：
We believe that China has the foundational conditions to support the development of Mercor-like models in terms of advancements in large model industries, the supply of high-skilled talent, the scale of flexible labor, and the regulatory framework for generative AI:
供给端：根据 OECD 数据，中国研发人员规模已由2019年的约480万人提升至2023年的约724万人，为高质量专家供给提供基础；
Supply side: According to OECD data, China's R&D personnel force has increased from approximately 4.8 million in 2019 to about 7.24 million in 2023, providing a foundation for high-quality expert supply;
需求端：AI训练与模型对齐相关需求持续释放，根据 Sapien.io，中国AI训练数据集市场规模有望由2023年不足3亿美元增长至2032年的约23亿美元；
Demand side: Demand for AI training and model alignment continues to grow. According to Sapien.io, the market size for AI training datasets in China is expected to expand from less than $300 million in 2023 to about $2.3 billion by 2032;
技术端：国产大模型中文语境下通用能力持续迭代，AI视频面试、智能筛选等能力已进入规模化应用阶段；
Technology side: Domestic large models have continuously iterated their general capabilities in Chinese contexts, with AI video interviews and intelligent screening capabilities entering the stage of large-scale application;
合规端：《生成式人工智能服务管理暂行办法》的实施及算法备案制度逐步落地，企业引入AI招聘工具的制度不确定性显著下降。但从现实落地看，招聘信任机制、用工形态认知及数据合规成本等因素使得Mercor模式在国内难以整体复制。尽管如此，头部平台已在 AI 招聘效率提升、模型反馈机制与真实业务数据沉淀方面形成可验证的商业闭环，表明AI对招聘流程的系统性改造在中国具备真实需求与可持续价值。
Compliance side: The implementation of the "Interim Measures for the Administration of Generative AI Services" and the gradual adoption of algorithm filing systems have significantly reduced institutional uncertainties for companies introducing AI recruitment tools. However, from a practical implementation perspective, factors such as recruitment trust mechanisms, awareness of employment forms, and data compliance costs make it difficult to fully replicate the Mercor model in China. Nevertheless, leading platforms have already formed verifiable commercial closed loops in terms of improving AI recruitment efficiency, model feedback mechanisms, and accumulating real business data, indicating that AI-driven systematic transformation of the recruitment process has genuine demand and sustainable value in China.
我们认为，尽管终态平台形态存在差异，AI招聘作为提升人力配置效率的核心路径已在中国得到验证，有望为具备平台化与数据积累优势的公司打开中长期结构性增长与估值抬升空间。
We believe that despite the differences in the final state platform forms, AI recruitment as a core path to improving human resource allocation efficiency has been validated in China, and it is expected to open up medium-to-long-term structural growth and valuation appreciation space for companies with platform and data accumulation advantages.
▍中国类Mercor模式市场空间测算：AI招聘提效、RLHF专家数据服务及可调度型人力资源平台有望合计带来约1300亿元潜在市场空间。
▍Market Size Estimation for China's Mercor-like Model: AI recruitment efficiency improvement, RLHF expert data services, and dispatchable human resource platforms are expected to collectively bring about approximately 130 billion yuan in potential market space.
基于Mercor已验证的AI招聘与专家数据服务两大业务形态，我们从行业视角对其在中国市场的潜在空间进行分层测算：
Based on the two validated business forms of AI recruitment and expert data services of Mercor, we measure the potential space in the Chinese market from an industry perspective:
1）AI招聘：有望通过模型精度提升实现对存量招聘市场的“深度替代”，根据我们测算，中国AI招聘业务在存量招聘市场中的可替代空间有望由2025年约220亿元提升至2028年约379-421亿元区间，并构成最具确定性的现实落地场景；
1) AI Recruitment: It is expected to achieve "deep substitution" in the existing recruitment market through model accuracy improvement, according to our calculations, the potential substitutable space of China's AI recruitment business in the existing recruitment market is expected to increase from approximately 22 billion yuan in 2025 to approximately 37.9-42.1 billion yuan in 2028, and it constitutes the most certain real-world implementation scenario;
2）RLHF等专家数据服务：根据我们测算，随着大模型能力向复杂推理与专业决策演进，RLHF等专家数据服务市场规模有望由2024年约18-53亿元增长至2028年约27-80亿元区间，该部分需求为完全增量型市场，具备独立于传统招聘周期的增长逻辑；
2) RLHF and other expert data services: According to our calculations, as the capabilities of large models evolve towards complex reasoning and professional decision-making, the market size of RLHF and other expert data services is expected to grow from approximately 1.8-5.3 billion yuan in 2024 to a range of 2.7-8 billion yuan by 2028. This portion of the demand represents a completely incremental market, with growth logic independent of traditional recruitment cycles;
3）在更长期维度上：若灵活可调度的人力资源平台模式逐步成熟并实现规模化运行，根据我们测算，2028年对应的潜在市场空间有望达到466-1283亿元区间。
3) On a longer-term horizon: If the model of flexible and schedulable human resources platforms gradually matures and achieves large-scale operation, according to our calculations, the potential market space in 2028 could reach a range of 46.6-128.3 billion yuan.
▍风险因素： ▍Risk Factors:
招聘行业AI应用发展不及预期，用户增长不及预期的风险；巨头入场导致在线招聘行业竞争恶化；宏观经济增长不及预期导致招聘行业发展不及预期；技术研发和创新不及预期；产品和服务拓展不及预期等。
The development of AI applications in the recruitment industry falling short of expectations, user growth not meeting expectations; major players entering the market leading to intensified competition in the online recruitment industry; macroeconomic growth falling short of expectations causing the recruitment industry to underperform; underperformance in technology R&D and innovation; expectations not being met in the expansion of products and services, etc.
▍投资策略。 ▍Investment Strategy.
在宏观处于温和复苏、Beta修复的背景下，我们认为在线招聘板块具备值得关注的结构性机会；同时AI对招聘流程的持续重构正在推动商业模式重塑与价值重估。我们认为，招聘平台AI落地的核心Alpha来自于技术储备与真实业务数据的持续沉淀。
Against the backdrop of macro 温和复苏 and Beta 修复, we believe the online recruitment sector possesses structural opportunities worth attention; at the same time, the continuous restructuring of the recruitment process by AI is driving the reshaping of business models and revaluation of value. We believe the core Alpha from AI implementation in recruitment platforms comes from technical reserves and the continuous accumulation of real business data.
Comments