就等梁文锋敲钟了

冯雨晨 2026-07-16 12:05

冯雨晨 2026/07/16 12:05

邦小白快读

全文速览

本文核心梳理了AI大模型企业DeepSeek筹备IPO、开启新一轮融资的最新行业动态，核心干货信息如下：

1. 最新进展：DeepSeek已启动内地IPO筹备，最快今年提交上市申请，同时开启新一轮融资，投前估值约710亿美元，约合人民币4800亿元，投资机构争抢份额但额度紧张，国资、头部产业资本和一线市场化机构纷纷入局。

2. 行业背景：上交所6月出台新规，支持尚未盈利的优质AI大模型企业登陆科创板，已有智谱、MiniMax等多家头部大模型企业推进A股IPO，行业迎来资本化浪潮。

3. 企业发展动向：DeepSeek此前拒绝外部融资保障控制权，此次融资为应对行业竞争，覆盖算力成本，同时企业已开始布局Agent、长文本、Coding甚至AI推理芯片等新业务方向。

本文透露国内AI大模型行业的最新发展动向，可为布局AI相关业务的品牌提供决策参考，核心干货如下：

1. 消费与行业趋势变化：AI大模型已经从技术展示阶段走向商业化落地，C端用户对AI提升生产力的需求大幅增长，推动大模型从回答问题转向执行任务，用户边界和商业空间被重新估值，行业迎来新的增长周期。

2. 产品研发方向参考：头部大模型企业已开始布局Agent、长文本、Coding、AI基建甚至AI推理芯片领域，品牌可参考这些方向提前布局相关产品与技术。

3. 竞争环境提示：当前资本大量涌入AI赛道，头部企业通过融资、IPO快速巩固竞争优势，行业竞争烈度大幅提升，品牌需要紧跟技术迭代节奏，才能跟上行业发展。

本文梳理了AI大模型行业最新政策与市场动态，可为布局AI相关业务的卖家提供参考，核心干货如下：

1. 政策利好：上交所已出台专门规则，支持尚未形成一定收入规模的优质AI大模型企业登陆科创板，打开了AI行业的融资与退出通道，AI赛道迎来政策扶持下的增长红利，相关上下游业务都将受益。

2. 市场机会：当前AI大模型商业化边界不断拓展，C端用户对AI提升生产力的需求爆发，Agent、Coding、AI基建、AI推理芯片等多个细分领域都存在新的创业与增长机会。

3. 风险提示：当前头部AI企业纷纷通过融资、IPO拉开竞争差距，行业进入资本与技术双驱动的新阶段，中小玩家竞争压力大幅提升，同时头部企业估值已处高位，需要警惕估值泡沫带来的相关风险。

本文透露AI大模型行业的最新发展动态，可为制造工厂把握商业机会、推进数字化转型提供参考，核心干货如下：

1. 新商业机会：当前头部AI大模型企业已经开始拓展业务边界，布局AI推理芯片研发，对芯片制造产能的需求将逐步释放，给芯片相关代工厂带来新的稳定订单机会。

2. 数字化转型启示：AI大模型已经成为提升全行业生产力的核心技术，工厂可借助AI大模型优化生产流程、产品设计环节，提升生产效率，抓住AI赋能制造业的发展趋势。

3. 合作机会：当前头部AI大模型企业快速扩张，业务延伸到AI硬件、AI基建领域，制造工厂可主动对接头部AI企业的生产研发需求，开展相关合作，切入AI赛道获得新增长空间。

本文梳理了国内AI大模型行业的最新发展趋势与客户痛点，可为AI相关服务商明确业务方向提供参考，核心干货如下：

1. 行业发展趋势：AI大模型已经从早期技术研发阶段进入商业化竞争阶段，资本大量涌入，头部企业纷纷启动IPO，行业集中度逐步提升，技术研发方向转向Agent、Coding、AI推理芯片等落地领域，行业增长动力充足。

2. 客户核心痛点：头部AI大模型企业研发需要承担巨额算力成本，同时需要稳定的激励机制留住核心人才，对资本服务、人才服务、算力基础设施服务都有旺盛需求。

3. 业务拓展机会：头部AI大模型企业拓展芯片研发、产品落地等新业务，需要对应的技术服务、供应链服务、人力资源服务等支持，服务商可围绕头部AI企业的需求推出针对性解决方案，拓展自身客户群体。

本文披露了AI大模型行业的最新资本动态与企业需求，可为面向AI企业服务的平台优化运营提供参考，核心干货如下：

1. 企业核心需求：AI大模型企业尤其是头部未盈利企业，对上市融资、低成本股权融资有强烈需求，同时需要配套的工商注册、资本对接等服务支持，上交所已经推出适配AI大模型企业的上市标准，吸引大批头部企业筹备A股IPO。

2. 平台运营方向：平台可将AI大模型作为重点招商赛道，针对AI企业早期估值高、暂未盈利的特点推出适配的招商与孵化服务政策，吸引优质AI企业落地，带动平台相关产业发展。

3. 风险规避提示：当前AI大赛道资本热度极高，部分头部企业估值已达数百亿美金，需要警惕行业过热带来的估值泡沫风险，平台要加强对入驻企业的资质审核，筛选真正具备核心技术实力的优质企业。

本文记录了国内AI大模型行业最新的产业与资本动向，可为AI产业研究者提供一手研究素材，核心干货如下：

1. 产业新动向：国内AI大模型行业已经从早期技术竞争进入资本+技术驱动的新阶段，头部企业纷纷启动IPO，依托资本投入扩大技术优势，行业格局逐步清晰，头部两家粤籍创始人的大模型企业有望占据国内大模型万亿元估值版图。

2. 政策新变化：国内监管层针对AI大模型行业出台了专门的科创板第五套上市标准，支持未盈利的优质AI大模型企业上市，解决了AI企业早期融资难的问题，将有力推动行业整体发展，为研究政策对硬科技行业的支持提供了新样本。

3. 治理新模式：头部AI大模型企业创始人通过特殊融资架构保障控制权，设置五年股权锁定期维持公司战略独立性，同时推出员工持股计划绑定核心人才，这种模式为AI初创企业的融资治理提供了新的研究方向。

返回默认

声明：快读内容全程由AI生成，请注意甄别信息。如您发现问题，请发送邮件至 run@ebrun.com 。

我是品牌商卖家工厂服务商平台商研究者帮我再读一遍。

Quick Summary

This article outlines the latest industry updates on Chinese AI large language model (LLM) firm DeepSeek, which is preparing for an IPO and launching a new round of financing. Key takeaways are as follows:

1. Latest developments: DeepSeek has kicked off preparations for an onshore IPO, and could submit its listing application as early as this year. It is also raising fresh capital at a pre-money valuation of approximately $71 billion (around 480 billion yuan). Stakes are heavily oversubscribed, with participation from state-owned capital, leading industrial investors and top-tier market-driven investment institutions.

2. Industry context: In June, the Shanghai Stock Exchange introduced new rules allowing unprofitable high-quality LLM companies to list on the STAR Market. Several leading domestic LLM players including Zhipu AI and MiniMax are already advancing A-share IPO plans, marking the start of a wave of industry capitalization.

3. Corporate strategy: DeepSeek previously rejected external financing to protect founder control. This new financing round will help it keep up with industry competition and cover high computing costs. The company has also started expanding into new business lines including AI agents, long context processing, code generation and even AI inference chips.

This article shares the latest development trends of China's LLM industry, providing strategic reference for brands布局ing AI-related business. Key insights are as follows:

1. Shifting consumer and industry trends: Large AI models have moved beyond technology demonstration to commercial落地. Demand from C-end users for AI-powered productivity improvements has grown sharply, pushing LLMs to evolve from question-answering to task execution. This has redefined the industry's user base and total addressable market, opening a new growth cycle for the sector.

2. Product R&D direction reference: Leading LLM companies are already expanding into AI agents, long context processing, code generation, AI infrastructure and even AI inference chips. Brands can reference these directions to布局 related products and technologies in advance.

3. Competitive landscape note: Large volumes of capital are flowing into the AI track, and leading players are rapidly consolidating competitive advantages via financing and IPOs, leading to much fiercer industry competition. Brands need to keep pace with technological iteration to stay aligned with industry development.

This article summarizes the latest policy and market updates for the LLM industry, providing reference for sellers布局ing AI-related business. Key insights are as follows:

1. Policy tailwinds: The Shanghai Stock Exchange has introduced dedicated rules allowing high-quality unprofitable LLM companies to list on the STAR Market, opening up financing and exit channels for the AI industry. The AI track now benefits from policy-supported growth, which will lift all related upstream and downstream businesses.

2. Market opportunities: The commercial boundary of LLMs continues to expand, with C-end demand for AI productivity improvements surging. Multiple niche segments including AI agents, code generation, AI infrastructure and AI inference chips offer new opportunities for entrepreneurship and growth.

3. Risk warning: Leading AI players are widening their competitive gap via financing and IPOs, and the industry has entered a new phase driven by both capital and technology. Small and mid-sized players face sharply higher competitive pressure, while valuations of leading firms are already at elevated levels, requiring caution against risks related to valuation bubbles.

This article shares the latest industry updates in China's LLM sector, providing reference for manufacturing factories to capture business opportunities and advance digital transformation. Key insights are as follows:

1. New business opportunities: Leading LLM companies are expanding their business boundaries and have started developing AI inference chips, which will gradually release demand for chip manufacturing capacity, bringing new stable order opportunities for chip-related foundries.

2. Insights for digital transformation: Large AI models have become core technology to boost productivity across all industries. Factories can leverage LLMs to optimize production processes and product design, improve operational efficiency, and capitalize on the trend of AI赋能 for manufacturing.

3. Collaboration opportunities: Leading LLM companies are expanding rapidly, extending their businesses into AI hardware and AI infrastructure. Manufacturing factories can proactively connect with leading AI firms to meet their R&D and production demand, establish collaboration, and enter the AI track to unlock new growth.

This article summarizes the latest development trends and client pain points of China's LLM industry, providing reference for AI-related service providers to define their business direction. Key insights are as follows:

1. Industry trends: Large AI models have moved from early-stage R&D to commercial competition, with massive capital inflow and leading players launching IPOs one after another. Industry concentration is gradually rising, R&D is shifting toward commercial落地 areas such as AI agents, code generation and AI inference chips, and the industry maintains strong growth momentum.

2. Core client pain points: Leading LLM companies bear enormous computing costs for R&D, and need stable incentive mechanisms to retain core talent, creating strong demand for capital services, talent services and computing infrastructure services.

3. Business expansion opportunities: As leading LLM companies expand into new businesses such as chip development and product commercialization, they require supporting technical services, supply chain services and human resources services. Service providers can develop targeted solutions aligned with the demand of leading AI firms to expand their customer base.

This article discloses the latest capital dynamics and corporate demand in the LLM industry, providing reference for platforms serving AI companies to optimize operations. Key insights are as follows:

1. Core corporate demand: LLM companies, especially unprofitable leading players, have strong demand for IPO financing and low-cost equity financing, alongside supporting services such as business registration and capital matchmaking. The Shanghai Stock Exchange has already introduced listing standards adapted to LLM companies, attracting a large number of leading players to prepare for A-share IPOs.

2. Platform operational priorities: Platforms can position LLM as a key investment attraction track, and introduce adapted investment attraction and incubation policies tailored to the characteristics of early-stage AI firms (high valuations, temporary lack of profitability) to attract high-quality AI companies to locate, driving the development of platform-related industries.

3. Risk mitigation: The AI track is currently experiencing extremely high capital enthusiasm, with some leading players already reaching valuations of tens of billions of US dollars. Platforms need to guard against valuation bubble risks from overheated industry, strengthen qualification checks on settled enterprises, and filter for high-quality players with genuine core technological strengths.

This article documents the latest industrial and capital dynamics of China's LLM industry, providing first-hand research materials for AI industry researchers. Key insights are as follows:

1. New industrial trends: China's LLM industry has evolved from early-stage technological competition to a new phase driven by both capital and technology. Leading players are launching IPOs one after another to expand technological advantages via capital input, and the industry landscape is gradually clearing up. The two top LLM firms founded by Cantonese-origin entrepreneurs are expected to take a major share of China's LLM industry's projected 1 trillion yuan valuation market.

2. New policy changes: Chinese regulators have introduced the fifth dedicated STAR Market listing standard for the LLM industry, allowing high-quality unprofitable LLM companies to go public. This solves the early-stage financing difficulty for AI firms, will strongly drive overall industry development, and provides a new case study for research on policy support for the hard tech sector.

3. New corporate governance model: Founders of leading LLM companies use special financing structures to protect control rights, set five-year share lock-up periods to maintain strategic independence, and launch employee stock ownership plans to retain core talent. This model offers a new research direction for financing and governance of AI startups.

Disclaimer: The "Quick Summary" content is entirely generated by AI. Please exercise discretion when interpreting the information. For issues or corrections, please email run@ebrun.com .

I am a Brand Seller Factory Service Provider Marketplace Seller Researcher Read it again.

史诗级IPO。

作者/冯雨晨

报道/投资界PEdaily

IPO脚步近了。

据彭博社报道，DeepSeek已开始筹备IPO，规划在内地上市，最快可能于今年提交申请。由于相关讨论尚属保密，DeepSeek未回应置评。

这并不意外。记得6月17日，上交所宣布支持尚未形成一定收入规模的优质人工智能大模型企业在科创板发行上市。此后，智谱、MiniMax均抓紧推进科创板IPO。

同一时间，DeepSeek被曝本周开始与新的投资人展开初步接触，讨论开启新一轮融资，外媒披露投前估值约为710亿美元（约合人民币4800亿元）。

“投资人们都想给梁文锋刷卡，但找不到POS机。”不久前，一家老牌VC机构合伙人向我们透露，7月，约了梁文锋杭州见面——下一轮上牌桌的人已经在路上了。

DeepSeek融资秘密

估值比肩长鑫

“不少国字头的机构和险资都进不去。”此前一位关注交易的知情人士透露。

有了额度也不一定能进。一家知名机构合伙人记得，首轮融资期间他被通知DeepSeek有20亿额度出来，但需要一天准备好20亿出资能力证明。“时间太紧张，我们没参与进去。”

最终我们看到，国家人工智能产业投资基金成为首轮融资中的主要国资。另外的身影则是：创始人梁文锋、腾讯、宁德时代、京东、网易、IDG资本、Monolith砺思资本、正心谷资本、拾象科技.....一一浮现。

当中几家身影令人意外。

投资界留意到，DeepSeek首轮融资进程期间，Monolith砺思资本、拾象科技密集在天津新设主体，这也是首轮融资中最为年轻的两家市场化机构。

先看拾象科技。2026年5月13日，天津拾象投资合伙企业（有限合伙）成立，注册资本约为15.1亿元。资料显示，拾象科技由前红杉投资人李广密在2019年发起成立。这位活跃在AI行业的90后，曾言“Coding是通向AGI的关键”，气质与DeepSeek某种程度上相得益彰。

Monolith砺思资本这边，期间陆续设立天津砺思星灵创业投资合伙企业（有限合伙）、天津砺思明棠企业管理咨询合伙企业（有限合伙）、天津砺思星雀创业投资合伙企业（有限合伙）、天津砺思星瀚创业投资合伙企业（有限合伙）、天津砺思明钧企业管理咨询有限公司等主体。

曹曦执掌下的Monolith砺思资本，虽然年轻但气质独特。此前曹曦透露，内部并没有给投资团队预设方向，甚至连title都没有，更多是“怎样合理就怎样来”。

“未来十年最重要的方向，我认为还是新一代人工智能，我们比较侧重投资模型公司，第一期基金就有20%投到了Kimi。”2026投资界SuperLink大会上，曹曦分享道。

无论是Monolith砺思资本还是拾象科技，身后LP阵列都出现过一个共同的名字——九安医疗。

如上述拾象科技新设的主体天津拾象投资合伙企业（有限合伙）中，SX Global Flagship Fund II L.P.及九安香港有限公司分别出资了7.6亿元、7.5亿元，后者便是九安医疗的投资平台。而Monolith砺思资本多家新设主体注册地在天津，正是九安医疗大本营之地，不由也引发联想。

迎战对手

设立员工持股计划

极高的诚意，极严的规则，是梁文锋首轮中的态度和条件。

一个细节是：首轮融资中，投资方资金需注入由梁文锋管理的有限合伙企业，而非直接投向DeepSeek主体，以此保障梁文锋对公司拥有绝对控制权。同时，所有投资方股权设有五年锁定期，锁定期内不得转让所持股份。

与之对应般，纵观首轮约500亿元规模融资，若以接近4000亿元的整体投后估值计算，梁文锋对外释放股份约13%，其中梁文锋独投200亿元带头成为本轮最大投资方，那么留给外部投资方的份额剩约7%。据悉，融资完成后，DeepSeek设立了员工持股计划，按实际估值分配股份。

今时不同往日。

成立三年，DeepSeek此前从未对外融资，哪怕在大模型融资最火热的那两年。梁文锋曾传达出的讯号是，外部投资者会干预公司战略，倒逼短期商业化和上市，影响战略独立性，因此，他婉拒了一批又一批前来叩门的投资人。

置于当下来看，选择融资不是立场变了，更像是一种对新竞争秩序的承认。

中国大模型早已走过技术秀场阶段。研发纯度、成本曲线、人才竞争这些问题的难度和重要性都在提升，实现AGI不仅是要技术领先，还需要坚硬的组织和稳定的生态——而融资，不仅能给DeepSeek内部核心人才一个“手里期权在市场上值多少钱”的估值锚，还有利于算清AGI这条路上沉重的算力成本账。

这件事，对手们做得更早更快。

智谱和MiniMax在IPO前分别融资超83亿元、15亿美元。今年1月，智谱和MiniMax陆续赴港上市，两家合计募资接近100亿港元。

放眼未IPO队伍，阶跃星辰5月完成近25亿美元的新一轮融资，年内融资总额将超200亿元，IPO愈发临近。还有常常被外界拿来与DeepSeek对位的月之暗面Kimi，被曝已启动新一轮融资，投前估值上升至315亿美金。

微妙一幕是，月之暗面创始人杨植麟与梁文锋同为广东人，一个粤东一个粤西。两位“粤籍AI双杰”有望坐拥中国大模型万亿元估值版图。

科创板上市窗口

万物流变，世界从未静止。

不久前MiniMax闫俊杰在一场圆桌讨论上聊起这样一段往事：两年前，他曾问梁文锋要不要做AI Coding？梁文锋说不做——因为当时大家的共识是，全中国会写代码的人可能只有100万到200万人，这似乎不是一个足够宽广的市场。

那时，所有人都在不同程度上低估了AI的想象力，也没有料想到，C端庞大用户对生产力的追求所能给AI大模型商业化带来的巨大赋能。尤其年初OpenClaw的爆发，几乎像是一次行业分水岭：当大模型从回答问题走向执行任务，它所对应的用户边界、调用深度和商业可能，也被一并重新估值。

DeepSeek也在变。

相当长一段时间里，DeepSeek更像一家理想主义气质饱满的研究机构，留给行业最深刻的标签是模型能力。而随着今年4月下旬DeepSeekV4发布甚至更早几个月前，外界从招聘动向、技术论文等方面捕捉到，DeepSeek也开始聚焦于Agent、长文本、Coding甚至产品端、AI基建上去。最新一幕，据悉DeepSeek也正在开发AI推理芯片，并在近几个月私下加大芯片设计工程师的招聘力度。

志在AGI，但也顺势而为；开源普惠，并非不知道如何去竞争。

资本市场大门的打开，则将竞争推向到另一个新的维度。

6月17日，上交所公告，支持尚未形成一定收入规模的优质人工智能大模型企业在科创板发行上市，为此制定了《上海证券交易所发行上市审核规则适用指引第10号——人工智能大模型企业适用科创板第五套上市标准》予以发布，并自发布之日起施行。

风声落地得极快。同一天，智谱科创板IPO辅导状态变更为辅导验收，拟募资不超150亿元。更早前的5月底，MiniMax已在筹备登陆A股。

时势推着所有野心家往前走，也逼着他们不断重新回答同一个问题：准备如何继续领先，又准备为何种领先付出代价。

此刻，梁文锋很难置身事外。

注：文/冯雨晨，文章来源：投资界(公众号ID：pedaily2012)，本文为作者独立观点，不代表亿邦动力立场。

文章来源：投资界

FAQ回顾

DeepSeek目前的估值水平是多少？

DeepSeek最新推进的新一轮融资投前估值约为710亿美元（约合人民币4800亿元），其此前完成的首轮融资规模约500亿元，投后估值接近4000亿元，估值水平比肩存储芯片企业长鑫。

AI大模型企业在科创板上市有什么支持政策？

2026年6月17日，上交所发布《上海证券交易所发行上市审核规则适用指引第10号——人工智能大模型企业适用科创板第五套上市标准》，支持尚未形成一定收入规模的优质人工智能大模型企业在科创板发行上市。

DeepSeek首轮融资的投资方主要有哪些？

DeepSeek首轮融资的出资方包括国家人工智能产业投资基金，腾讯、宁德时代、京东、网易等产业资本，以及IDG资本、Monolith砺思资本、正心谷资本、拾象科技等投资机构，创始人梁文锋是本轮最大投资方。

DeepSeek为什么此前拒绝融资现在推进资本化？

此前梁文锋担心外部投资者干预公司战略，倒逼短期商业化影响战略独立性，因此拒绝融资。当前大模型行业进入综合竞争阶段，融资可提供人才期权估值锚，覆盖高昂算力成本，叠加科创板上市政策放开，因此推进融资及IPO。