融资160亿港元创始人宣布“0薪酬”：MiniMax能改写大模型资本竞赛的终局吗？

晨阳 2026-07-13 11:27

晨阳 2026/07/13 11:27

邦小白快读

全文速览

本文核心介绍了国内头部大模型企业的最新融资动态、两种商业化路径对比和行业整体现状，核心干货信息如下：

1. 最新事件方面，7月9日、10日头部大模型企业智谱、MiniMax先后在港股完成融资，分别募资314亿港元、160亿港元，MiniMax本次融资吸引百余家机构参与，实现7倍认购覆盖。MiniMax创始人闫俊杰宣布从即日起至实现AGI不再领薪酬，同时拿出总股本4%做团队激励、1%设立基金支持开源社区。

2. 行业现状方面，当前大模型赛道处于烧钱竞赛阶段，头部企业融资需求旺盛，行业存在B端服务、C端原生应用两种主流路径，两种路径的盈利表现差异明显，B端路径粘性更高、盈利能力更强，整个赛道融资已经从轮次驱动转向持续输血模式。

本文对比了大模型赛道两种主流商业化路径的发展结果，对AI领域品牌商的经营策略有较高参考价值，核心干货如下：

1. 定价与竞争方面，B端客户对大模型产品的价格敏感度低，换模型的迁移成本远高于涨价幅度，智谱API涨价83%后调用量反而增长400%，再次涨价后依旧保持增长，验证了产品不可替代性带来的定价权；C端用户对价格敏感度极高，MiniMax模型涨价引发用户不满，被迫降价回原水平。

2. 研发与消费趋势方面，当前头部品牌都将超八成以上融资投入算力基建和模型研发，技术迭代是核心竞争力；B端大模型服务当前已经跑出可行商业模式，毛利率比C端路线高16个百分点，更受资本市场认可，C端路线目前仍需要解决盈利不稳定的问题。

3. 资本环境方面，当前大模型赛道迎来融资热潮，头部品牌上市后仍可获得大额资本支持，只要技术和商业化路线清晰，就能获得持续资金输入。

本文梳理了2026年国内大模型赛道的最新变化，能给AI领域创业者和卖家提供方向参考，核心干货如下：

1. 机会提示方面，当前大模型行业融资热情高涨，一级二级市场都有充足的资金供给，上市不再是融资终点，反而成为新一轮大规模融资的起点，B端大模型服务路线已经得到市场验证，客户粘性高、毛利率高，是当前更稳妥的创业方向；海外C端市场也有增长空间，MiniMax超70%收入来自国际市场。

2. 风险提示方面，大模型行业烧钱速度极快，收入增长的同时亏损规模也在同步扩大，融资不是锦上添花而是生存必需，C端路线在算力成本持续攀升的背景下，盈利难度越来越大，MiniMax毛利率仅25.4%，亏损规模逐年扩大。

3. 可借鉴经验方面，创始人可以通过零薪酬、股权激励的方式，将个人利益与公司长期发展绑定，向市场和团队传递信心，稳定公司声誉和预期。

本文披露了大模型行业的发展现状，对需要落地智能化转型、寻求AI相关商业机会的工厂有较多启示，核心干货如下：

1. 数字化转型启示方面，当前国内B端大模型服务已经发展成熟，智谱已经服务超12000家机构客户，覆盖4500万开发者，国内Top10互联网公司有9家使用其大模型，工厂推进智能化改造，可以优先选择这类成熟的B端大模型服务商，技术迭代和服务稳定性更有保障。

2. 商业机会方面，当前大模型行业持续高投入研发，新产品新模型落地速度快，给下游工厂带来了更多赋能机会，工厂可以借助大模型能力优化生产流程、升级产品设计，提升自身竞争力；同时C端AI应用的海外市场增长较快，做出海相关产品的工厂可以绑定头部C端大模型企业挖掘相关机会。

3. 发展方向启示，大模型行业的核心竞争力是技术迭代，工厂布局AI相关业务也需要持续投入研发，跟上技术迭代节奏才能保持竞争力。

本文梳理了大模型行业的最新发展趋势和核心痛点，能给AI相关服务商提供业务方向参考，核心干货如下：

1. 行业发展趋势方面，当前大模型赛道进入融资爆发期，头部企业的融资模式已经从传统的轮次驱动转向持续输血，上市成为新一轮大规模资本运作的起点，行业整体将资金集中投向算力基建和模型研发，技术迭代速度越来越快，需要持续的资本和服务支撑。

2. 客户核心痛点方面，所有大模型企业都面临巨额算力成本压力，亏损规模随收入增长同步扩大，对持续融资、降本增效相关服务有强烈需求；C端大模型企业的成本压力更大，毛利率更低，对商业化变现服务的需求更高。

3. 业务机会方面，B端大模型服务当前已经验证了商业可行性，客户扩张速度快，服务商可以围绕B端大模型的落地拓展配套服务；同时开源社区得到行业头部企业的支持，MiniMax拿出1%股份设立专项基金支持开源，围绕开源生态的配套服务也有较大发展空间。

本文披露了大模型企业的融资和发展需求，对服务AI企业的平台商有较高参考价值，核心干货如下：

1. 大模型企业对平台的需求方面，大模型行业技术迭代快、烧钱速度快，上市不代表融资结束，反而会产生更大规模的持续再融资需求，传统的融资节奏已经不适应大模型行业，平台需要调整规则适配大模型企业持续融资的需求，提供更便捷的再融资渠道。

2. 平台招商方向方面，当前大模型赛道融资热度高，头部企业增长速度快，有强烈的上市和再融资需求，智谱已经在推进A+H双重上市，是优质的招商标的，平台可以针对性吸引头部大模型企业落地，平台自身也能获得增长动力。

3. 风险规避方面，当前大模型行业整体还未实现稳定盈利，大部分头部企业都处于亏损状态，不同路线的盈利不确定性差异很大，C端路线的不确定性远高于B端路线，平台需要加强对入驻大模型企业的风险管控，重点关注其技术能力和商业化进展，规避行业波动带来的风险。

本文呈现了国内大模型赛道最新的产业动向和商业模式对比，为产业研究提供了丰富的一手资料，核心干货如下：

1. 产业新动向方面，2026年国内大模型赛道迎来全面融资潮，不仅二级市场头部企业密集完成大额再融资，一级市场也出现单笔超500亿元的私募融资，行业融资模式发生根本变化，从轮次驱动转向持续输血，上市成为新一轮资本运作的起点而非终点。头部企业全部将融资主要投向算力基建和核心模型研发，技术竞赛持续升级。

2. 产业新问题方面，大模型行业的发展逻辑是收入增长与资本消耗同步加速，头部企业的亏损规模随营收增长快速扩大，融资已经从发展支持变为生存必需，如何在技术迭代的同时跑通盈利模式，是整个行业面临的核心问题。

3. 商业模式对比研究方面，当前行业形成了两种清晰的主流商业模式，B端大模型API服务模式客户粘性高、毛利率更高，已经验证了商业可行性；C端AI原生应用模式靠C端收入供养基座研发，当前面临成本高、用户价格敏感的问题，盈利能力仍待市场验证。

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 highlights the latest financing developments, a comparison of two commercialization paths, and the current industry landscape of China's leading large language model (LLM) companies. Key takeaways are as follows:

1. Latest developments: On July 9 and 10, two leading Chinese LLM firms, Zhipu AI and MiniMax, completed new financing on the Hong Kong Stock Exchange, raising HK$31.4 billion and HK$16 billion respectively. MiniMax's financing round attracted more than 100 institutional investors and was 7 times oversubscribed. MiniMax founder Yan Junjie announced he will forgo salary until the company achieves artificial general intelligence (AGI), while setting aside 4% of total equity for team incentives and 1% to establish a fund supporting the open-source community.

2. Current industry status: The LLM sector is currently in a capital-intensive race, with top players facing strong financing demand. The industry has two dominant commercial paths with sharply different profitability profiles. The B2B path delivers higher customer stickiness and stronger profitability, and sector financing has shifted from a round-by-round model to a "continuous capital injection" model.

This article compares the performance of the two dominant commercialization paths in the LLM sector, offering valuable insights for brand operators in the AI industry. Key takeaways are as follows:

1. Pricing and competition: B2B clients have low price sensitivity for LLM products, and the switching cost of changing models far outweighs the impact of price hikes. After Zhipu AI raised its API prices by 83%, API call volume grew 400%, and continued to grow after a second price increase, confirming that product irreplaceability grants firms pricing power. In contrast, C-end users are extremely price-sensitive: a price hike by MiniMax triggered widespread user dissatisfaction, forcing the company to roll prices back to their original level.

2. R&D and consumer trends: Leading brands currently allocate over 80% of financing to computing infrastructure and model R&D, with technical iteration as the core competitive advantage. The B2B LLM service path has already proven its commercial viability, with gross margins 16 percentage points higher than the C-end path, making it more favored by capital markets. The C-end path still faces challenges of unstable profitability.

3. Capital environment: The LLM sector is currently experiencing a financing boom. Leading brands can still secure large-scale capital support after going public, and as long as their technology and commercialization roadmap are clear, they can access continuous capital injection.

This article outlines the latest 2026 developments in China's LLM sector, providing directional guidance for AI entrepreneurs and founders. Key takeaways are as follows:

1. Opportunity outlook: The LLM industry is currently seeing a surge in financing activity, with abundant capital available in both primary and secondary markets. Going public is no longer the end point of financing, but rather the starting point of a new round of large-scale fundraising. The B2B LLM service path has been market-validated, with high customer stickiness and high gross margins, making it a more reliable entrepreneurial direction today. The overseas C-end market also offers growth room: over 70% of MiniMax's revenue comes from international markets.

2. Risk warnings: The LLM industry burns capital at an extremely fast pace, with loss scales expanding in lockstep with revenue growth. Financing is not a value-add, but a requirement for survival. Against the backdrop of continuously rising computing costs, the C-end path faces increasing profitability challenges: MiniMax posted a gross margin of just 25.4%, with losses expanding year over year.

3. Actionable takeaways: Founders can align personal interests with long-term company development and signal confidence to the market and their team through forgoing salary and issuing equity incentives, stabilizing corporate reputation and market expectations.

This article outlines the current development status of the LLM industry, offering key insights for factories pursuing intelligent transformation and seeking AI-related business opportunities. Key takeaways are as follows:

1. Insights for digital transformation: China's B2B LLM service market is now mature. Zhipu AI already serves more than 12,000 institutional clients and reaches 4.5 million developers, with 9 of China's top 10 internet companies using its models. Factories pursuing intelligent upgrading should prioritize partnering with established B2B LLM service providers like Zhipu, which offer more reliable technical iteration and service stability.

2. Business opportunities: Continuous high R&D investment in the LLM industry is driving faster rollout of new products and models, creating more empowerment opportunities for downstream factories. Factories can leverage LLM capabilities to optimize production processes, upgrade product design, and improve their overall competitiveness. Meanwhile, the overseas market for C-end AI applications is growing rapidly, and factories producing export-oriented products can partner with leading C-end LLM companies to unlock new opportunities.

3. Directional guidance: Technical iteration is the core competitive advantage of the LLM industry. For factories building AI-related businesses, this means sustained R&D investment is required to keep up with iteration pace and maintain competitiveness.

This article outlines the latest development trends and core pain points of the LLM industry, providing directional guidance for AI-related service providers. Key takeaways are as follows:

1. Industry development trends: The LLM sector is currently experiencing a financing boom, and leading firms have shifted from traditional round-by-round financing to a continuous capital injection model, where going public serves as the starting point for a new round of large-scale capital operations. The industry as a whole is concentrating capital on computing infrastructure and model R&D, driving faster technical iteration that requires sustained capital and service support.

2. Core client pain points: All LLM companies face enormous pressure from high computing costs, with loss scales expanding alongside revenue growth, creating strong demand for services related to continuous financing, cost reduction and efficiency improvement. C-end LLM firms face even greater cost pressure and lower gross margins, leading to higher demand for commercial monetization services.

3. Business opportunities: The B2B LLM service path has already proven its commercial viability and is rapidly expanding its client base, so service providers can develop supporting solutions for B2B LLM implementation. The open-source community also has backing from industry leaders: MiniMax has set up a special fund with 1% of its equity to support open-source development, creating significant room for growth in supporting services around the open-source ecosystem.

This article outlines the financing and development needs of LLM companies, offering valuable insights for platforms serving AI firms. Key takeaways are as follows:

1. Demand for platform services: LLM companies face fast technical iteration and high capital burn, so going public does not end their financing needs—it instead creates demand for larger-scale continuous follow-on financing. Traditional financing rhythms no longer fit the LLM industry, so platforms need to adjust their rules to adapt to LLM companies' need for continuous financing, and provide more convenient access to follow-on fundraising.

2. Investment targeting: The LLM sector is currently seeing high financing enthusiasm, with leading firms growing rapidly and having strong demand for IPO and follow-on financing. Zhipu AI is already pursuing a dual A+H listing, making it a high-quality listing candidate. Platforms can proactively attract leading LLM companies to list on their exchanges to drive their own growth.

3. Risk mitigation: The LLM industry as a whole has not yet achieved stable profitability, with most leading firms still operating at a loss. Profitability uncertainty varies sharply between commercial paths, with the C-end path carrying far higher uncertainty than the B2B path. Platforms need to strengthen risk management for listed LLM companies, prioritize monitoring their technical capabilities and commercialization progress, and mitigate risks from industry volatility.

This article presents the latest industrial developments and a business model comparison of China's LLM sector, providing rich first-hand data for industry research. Key takeaways are as follows:

1. New industrial developments: In 2026, China's LLM sector is seeing a broad financing boom, with leading listed firms completing large-scale follow-on financing in the secondary market, and single private financing deals exceeding RMB 50 billion in the primary market. The industry's financing model has fundamentally shifted from round-by-round funding to continuous capital injection, with IPO serving as the starting point rather than the end point of capital operations. All leading firms allocate most of their financing to computing infrastructure and core model R&D, intensifying the ongoing technology race.

2. New industrial challenges: The growth logic of the LLM industry means revenue growth and capital consumption accelerate in tandem, with leading firms' loss scales expanding rapidly alongside growing revenue. Financing has shifted from a driver of growth to a requirement for survival, and the core industry-wide challenge remains how to deliver profitable growth while maintaining continuous technical iteration.

3. Comparative business model research: The industry has formed two clear dominant business models. The B2B LLM API service model delivers higher customer stickiness and higher gross margins, and has already proven its commercial viability. The C-end native AI application model funds base model R&D through C-end revenue, and currently faces challenges of high costs and high user price sensitivity, with its profitability yet to be validated by the market.

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.

作者丨晨阳

编辑丨小龙

7月10日早间，港股大模型公司MiniMax宣布完成新一轮160亿港元融资。

本次融资采用“配股+零息可转债”的组合结构：公司以每股268港元配售3560万股新A类股份，较前一交易日收市价297.4港元折让约9.89%，预计募资约95.41亿港元；同步发行65亿港元于2027年到期的零息有担保可转换债券，初步转换价为每股335港元，较收市价溢价约12.64%。

两笔交易合计募资逾160亿港元（约20亿美元）。

据披露，本次交易吸引了多家顶级国际主权基金、长线基金、一线中资机构和顶级多策略基金参与，覆盖亚太、欧洲和美国市场，其中长线及主权基金达20余家。

交易共吸引百余家机构参与，实现7倍认购覆盖，最初发行规模约18亿美元，在机构需求推动下最终扩大至20亿美元以上。

现有股东、基石投资人及国际长线资金仍对MiniMax中长期基本面和行业位置持续关注。

同日，MiniMax创始人兼CEO闫俊杰发布内部全员信，宣布自即日起至公司实现AGI之日不再领取任何薪酬，未来四年将拿出个人名下相当于公司总股本4%的股份用于激励团队成员，另拿出1%的股份设立专项基金支持开源社区发展。

“大模型双雄”的资本博弈

就在MiniMax融资公告的前一日——7月9日，另一家港股大模型公司智谱刚刚宣布启动约314亿港元的配售融资。

智谱以每股1588港元配售最多1978万股新H股，较前一交易日收盘价1825港元折价约12.99%，预计募资总额约314.11亿港元。

这一规模创下2026年港股科技企业单次配售募资规模新高，也成为国内大模型赛道企业上市后最大一笔股权融资。

7月9日配售公告发布后，智谱股价上涨11.34%，市值重回9059.53亿港元。

MiniMax 7月10日融资公告发出后，股价全天走弱，盘中最大跌幅超14%，截至收盘股价大跌9.68%，总市值842.42亿港元。

智谱与MiniMax虽同属大模型头部企业，但二者的业务方向从一开始就分道扬镳。

智谱押注B端，收入来自企业级大模型、智能体、开放平台API服务，截至2026年5月已服务超12000家机构客户，覆盖4500万开发者。中国前十大互联网公司中，9家在使用智谱GLM大模型。这种“向老板伸手”的模式，构建了高黏性的企业客户网络。

MiniMax则走了一条截然不同的路——主攻C端AI原生应用，旗下拥有MiniMax Agent、Hailuo AI、Talkie、星野等多款APP，累计超3亿全球C端用户。它的商业模式是“靠C端虚拟恋人赚快钱，去供养昂贵的基座模型研发”。这条路在算力成本持续攀升的2026年，正变得越来越难走。

两家公司核心业务方向的不同，最直观地体现在定价权上——这是资本市场给出截然不同估值的分水岭。

今年一季度，智谱将API调用价格上调83%。按常理，涨价应压制需求，但智谱的调用量同期增长了400%。6月GLM-5.2发布后，智谱直接取消低价档位，综合API价格再涨一成。涨完价，模型在Code Arena全球开发者盲测中排到第一。

MaaS平台年化经常性收入达到17亿元，12个月增长60倍。B端客户已将GLM嵌入业务流程，换模型的迁移成本远高于多付的钱——涨价反向验证了产品的不可替代性。

MiniMax则遭遇了截然相反的境遇。6月1日发布旗舰模型M3当天，公司高开跳水收跌15%。M3定价约为前代M2.7的两倍，引发用户强烈不满，一周后MiniMax被迫宣布永久降价50%，回落至接近前代水平。

两家公司2025年营收规模大体相当——智谱7.24亿元，MiniMax约5.7亿元——但毛利率拉开了16个百分点的差距：智谱41%，MiniMax仅25.4%。

这16个百分点，是两种商业路径的真实成本差异。

智谱更贴近国内基础模型、企业服务、开发者平台和国产AI基础设施的叙事，在当前市场风格下更容易获得估值溢价。而MiniMax的优势在多模态、C端产品和海外增长，但市场仍在观察它能否把这些优势转化为稳定的商业收入。

资金去向：

算力基建与模型研发的双轮驱动

两家公司本轮融资的资金投向高度一致，均指向AI基础设施与模型研发这一核心方向。

MiniMax公告显示，募集资金净额的80%将用于继续加强AI基础设施及模型研究与开发，约10%用于加快Harness产品的全球商业化及开发，约10%用于营运资金及一般企业用途。

据消息人士透露，MiniMax正研发一款2.7万亿参数的大语言模型，内部暂称M3 Pro，最快或于今年三季度发布，并计划开源，该模型规模远超其现有旗舰M3的4280亿参数。

智谱的314亿港元募资则更为集中——全部资金将在2027年底前使用完毕，三大投向清晰：核心研发与算力基建、商业化扩张与产业并购、补充运营资金与优化资本结构。

公司此前披露，截至2026年6月30日，IPO所得款项净额已动用逾93%。

与此同时，智谱正在推进A股科创板上市，拟募资150亿元，搭建A+H双重上市架构。

两家公司资金消耗速度之快、融资规模之大，折射出大模型赛道“烧钱”的残酷现实。

MiniMax2025年总收入7903.8万美元，同比增长158.9%，超过70%收入来自国际市场，但经调整净亏损为2.5亿美元。从2023年营收2451万元到2025年5.56亿元，营收增长迅猛，但归母净亏损从19.07亿元扩大至131.6亿元。

大模型公司的商业模型决定了——收入增长与资本消耗同步加速，融资不是锦上添花，而是生存必需。

一封全员信：零薪酬背后的创始人信号

就在MiniMax宣布160亿港元融资的同日，创始人兼CEO闫俊杰向全员发布了一封内部信。

信中只有短短几段话，但信息密度极高。

闫俊杰宣布：从即日起，直到公司实现AGI的那一天，他将不再从公司领取任何薪酬。未来四年，他将拿出个人名下相当于公司总股本4%的股份，用于激励长期与公司并肩作战、共同创造价值的团队成员；同时拿出1%的股份设立专项基金，持续支持相关开源社区的发展。

闫俊杰在信中写道：“市场会有波动，外界会有杂音，但前进的方向不会改变。身处行业一线的我们，比任何人都更清楚技术演进的真实速度，也更清楚我们正在创造和积累的长期价值。”他称此举是“作为创始人作出的长期承诺”。

全员信以一句话收尾：“We will keep going until we get there.Intelligence with Everyone.”

“零薪酬”在商业史上并非孤例，但每一次出现都伴随着特定的信号——往往意味着创始人正在用个人利益为公司的长期信念做一次公开的、不可撤回的背书。

在科技界，最知名的“零薪酬”案例当属马斯克。马斯克在特斯拉常年不领基本工资。2025年，特斯拉在监管文件中公布马斯克薪酬总额高达1580亿美元，但这一天文数字背后，马斯克实际到手金额为零——因为特斯拉当年的市值指标、经营业绩等多项考核全部未达标，马斯克无缘任何绩效奖励。此前一笔数百亿规模的临时奖励也已被公司全额作废。

马斯克的“零薪酬”本质上是“绩效对赌”——薪酬与公司市值、经营目标深度绑定，达不成目标则分文不取。这是一种将创始人利益与公司命运完全锁定的制度设计。

而闫俊杰的“零薪酬”能否稳住市场信心，答案不在信里，在MiniMax未来的模型能力、商业化进展和盈利路径上。

行业变局：大模型融资潮的全面爆发

智谱与MiniMax的隔日融资并非孤立事件，而是2026年国产大模型赛道融资热潮的集中缩影。

据证券时报统计，本次智谱港股配售募资规模仅次于比亚迪、宁德时代，成为今年港股科技企业单次配售募资规模新高。

商汤在今年4月完成32.47亿港元新股配售；科大讯飞落地40亿元定向增发。

一级市场同样风起云涌——深度求索DeepSeek完成首轮510亿元融资，创下国内AI企业单笔私募融资纪录；月之暗面（Kimi）在6月传出正在寻求20亿美元融资；阶跃星辰接近完成近25亿美元融资。

大模型赛道的融资节奏正在从“轮次驱动”转向“持续输血”模式。

上市不再意味着融资的终点，反而成为新一轮更大规模资本运作的起点。

这种“上市即再融资”的节奏，在传统行业中极为罕见，却正在成为AI大模型公司的常态。

更深层的产业逻辑在于：大模型的技术竞赛没有中场休息。训练、推理、算力锁定和全球化投入都需要持续资本支持。

头部企业必须在技术迭代与商业落地之间保持双线推进，而这两条战线都需要巨额资金的持续灌溉。