• 亿邦公众号
    • APP下载

单日暴跌40% 「AGI第一股」要回答的不只是股价问题

作者:市值榜团队 文章来源:市值榜

2026-07-13 14:05

作者:市值榜团队 文章来源:市值榜

2026-07-13 14:05

邦小白快读

EN
全文速览

本文围绕被称为“港股AGI第一股”的云知声股价暴跌事件,梳理了事件来龙去脉,拆解了AI行业当前面临的核心问题,干货如下:

1. 事件核心脉络:云知声2025年6月登陆港股,上市初期借AGI概念和筹码稀缺,最高市值达620亿港元,解禁后股价持续下挫,2026年7月市值仅55亿港元,较高点蒸发超90%。解禁抛压只是导火索,本质是资本市场对AI企业的估值逻辑已经从看概念转向看商业兑现。

2. 普通投资者参考:投资AI相关标的不能只看AGI标签和技术故事,要重点考察企业的商业模式、盈利能力和自我造血能力,警惕只讲故事没有业绩兑现的标的,避开估值泡沫风险。

本文通过云知声的案例,为AI赛道品牌商总结了行业现状和发展参考,干货如下:

1. 行业估值与消费趋势变化:当前资本市场对AI品牌的估值逻辑已经脱离早期的技术概念炒作,不再仅凭AGI标签和筹码稀缺性给高估值,长期价值必须回归商业模式和盈利能力的硬考核,市场更关注技术能力能否转化为实际收入利润。

2. 产品研发方向:面向垂直产业的AI产品不需要盲目堆砌参数,要紧扣客户真实需求,侧重提升智能密度,用较小参数、较低成本满足行业对部署成本、推理效率的要求,依托长期行业积累打造差异化竞争力。

3. 商业模式升级:要尽早摆脱项目制定制交付的旧模式,逐步转向标准化产品、Token计费、订阅制的轻模式,提升收入可持续性和毛利率,同时要调整组织架构适配新的商业模式。

本文为AI赛道创业者、卖家梳理了当前行业变化、机会与风险,干货如下:

1. 行业发展阶段变化:大模型发展已经完成热身赛,在2026年正式进入真实商业化竞争阶段,市场已经从追捧技术参数、榜单成绩转向关注技术到商业变现的落地能力,靠概念讲故事已经无法获得市场认可。

2. 明确的市场机会:垂直产业AI落地是高潜力的增长方向,借助大模型能力可以把原来高成本的定制化项目转化为标准化可复制产品,大幅压缩交付周期、降低成本,比如云知声将保险理赔审核交付周期从3个月缩短到1周,成本降低80%,变现效率提升明显。

3. 核心风险提示:要警惕项目制交付的陷阱,该模式回款周期长、没有持续复购收入、难以形成规模效应,会导致企业财务脆弱,同时要注意组织架构必须适配新商业模式,避免旧架构拖慢新业务布局错过机会。

本文对AGI产业发展的分析,给传统工厂推进数字化转型、布局AI相关业务带来多方面启示,干货如下:

1. 产品开发需求方向:工厂布局AI相关产品或推进自身数字化转型,不需要盲目追求万亿参数大模型,要侧重“智能密度”,在满足场景需求的前提下控制成本,适配产业端对部署成本、推理效率的要求,提升产品性价比。

2. 商业机会挖掘方向:AI切入垂直产业场景是明确的发展趋势,工厂可以结合自身所在垂直领域的行业经验和Know-how,对接场景需求开发AI落地产品,像医疗、轨交、物联网等领域都有大量未被满足的落地需求,长期积累的行业经验能形成差异化竞争力。

3. 转型核心启示:工厂推进AI和数字化转型,不能只停留在技术升级层面,还要同步调整业务模式和组织架构,逐步从定制化项目交付转向标准化可复制产品,打造持续稳定的收入,避免旧的业务模式和组织架构拖累转型进度。

本文梳理了当前AI大模型服务商面临的行业趋势、客户痛点和可行解决方案,干货如下:

1. 行业发展趋势:大模型竞争已经进入商业化深水区,当前行业形成两类发展路径,一类是做通用基础模型拼规模和调用量,一类是走“产业深潜”路线扎根垂直场景,对于多数服务商来说,依托行业经验走垂直路线更容易形成差异化优势。

2. 客户核心痛点:产业端客户对AI落地的需求已经不单是基础智能能力,还要求模型懂行业知识,满足数据安全、合规、可靠性要求,同时普遍对部署成本和推理效率比较敏感,原有定制化项目交付模式成本高、周期长,供需双方都承受较大压力。

3. 可行解决方案:走垂直路线的服务商可以采用“强基模+深应用”的模式,打造小参数高智能密度的模型适配场景需求,逐步将能力模块化、标准化,降低交付成本,从定制项目转向可持续的订阅收入,同时要同步调整组织架构适配新的商业模式。

本文以云知声事件为样本,为AI相关平台梳理了市场需求、运营方向和风险点,干货如下:

1. AI企业对平台的核心需求:处于竞争阶段的AI企业,在研发、算力、人才方面都需要持续的资金投入,上市后有持续融资的需求,平台可以在规则框架内适配AI企业合理的融资需求,支持企业储备资金应对行业竞争。

2. 招商与运营方向:平台引入AI企业时,不能仅被AGI概念、技术标签吸引,要建立更理性的考察标准,重点关注企业的商业化落地能力、商业模式质量,优先引入已经实现路径验证的企业,提升平台整体标的质量。

3. 风险规避要点:平台要引导市场关注AI企业的真实经营情况,警惕频繁融资但资金大量闲置、商业化进展不及预期的企业,提示相关风险,推动市场建立更成熟的AI企业估值逻辑,促进行业良性发展。

本文以云知声股价暴跌事件为研究样本,呈现了当前AGI产业发展的新动向、新问题,对产业研究有较高参考价值,干货如下:

1. 产业发展新动向:AGI产业已经完成早期的概念热炒和技术热身,正式进入商业化竞争阶段,资本市场的估值逻辑发生根本性变化,从看重技术叙事、筹码稀缺性转向看重商业模式质量和盈利能力,云知声市值较高点蒸发超90%就是这一变化的典型案例。

2. 产业现存核心问题:当前走产业深潜路线的AGI企业普遍面临多重矛盾:长期项目制交付没有规模效应,客户留存率低,收入可持续性差,应收账款高、回款慢导致财务脆弱,企业长期亏损缺乏自我造血能力,同时组织架构升级滞后,跟不上技术和商业模式的迭代速度。

3. 商业模式研究方向:当前行业已经形成两类清晰的AGI商业化路径,分别是通用基础模型路线的AGI商业价值=智能上界×Token消耗规模,和产业深潜路线的AI行业价值=智能密度×Token价值,后者从项目制向标准化订阅制转型的创新方向,值得深入研究。

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 unpacks the sharp stock price drop of Unisound, known as the "first AGI stock" on the Hong Kong stock market, outlines the full context of the incident, and breaks down core challenges facing the current AI industry. Key takeaways are as follows:

1. Timeline of the incident: Unisound listed on the Hong Kong stock exchange in June 2025. Riding on the AGI hype and limited tradable shares, its market capitalization peaked at HK$62 billion. After the lock-up period expired, the stock price declined steadily, and by July 2026 its market cap stood at only HK$5.5 billion, representing a more than 90% drop from the peak. Lock-up selling pressure was only the trigger; the underlying cause is a shift in how capital markets value AI companies, from concept-driven performance to business reality.

2. Guidance for retail investors: When investing in AI-related assets, investors should not rely solely on AGI labels and technology narratives. Instead, they should prioritize evaluating a company's business model, profitability and ability to generate self-sustaining cash flow, avoid firms that only pitch stories without delivering actual financial results, and hedge against valuation bubble risks.

Using the Unisound case, this article summarizes the current state of the AI industry and provides key takeaways for AI brands, as follows:

1. Shifts in valuation and market demand: Capital markets have moved past early-stage hype around technology concepts for AI brands, and no longer assign high valuations based solely on AGI labels and limited share supply. Long-term value must now pass rigorous tests of business model and profitability, with the market focusing more on whether technical capabilities can translate into actual revenue and profit.

2. Product R&D priorities: AI products for vertical industries do not need to blindly chase larger model parameters. Instead, they should align with real customer needs, prioritize higher "intelligence density", and use smaller models with lower costs to meet industry requirements for deployment cost and inference efficiency. Companies can build differentiated competitiveness through long-term industry expertise.

3. Business model transformation: Companies should move away from the traditional custom project-based delivery model as early as possible, and gradually shift to a lighter model based on standardized products, token-based billing and subscription services. This improves revenue sustainability and gross margins, and requires corresponding organizational restructuring to support the new business model.

This article sorts out current industry changes, opportunities and risks for AI entrepreneurs and sellers, with key takeaways as follows:

1. Shift in industry development stage: Large model development has completed its early "warm-up" phase, and officially entered the stage of real commercial competition in 2026. The market has shifted its focus from technical parameters and benchmark rankings to the ability to turn technology into monetizable business value, and concept-based pitches no longer win market recognition.

2. Clear market opportunities: AI deployment in vertical industries is a high-growth potential direction. Leveraging large model capabilities, high-cost custom projects can be transformed into standardized, replicable products that drastically cut delivery cycles and lower costs. For example, Unisound reduced the delivery cycle for insurance claim review from 3 months to 1 week, cut costs by 80%, and significantly improved monetization efficiency.

3. Key risk warnings: Companies should beware of the pitfalls of project-based delivery. This model features long payment cycles, no sustained recurring revenue and difficulty achieving economies of scale, leaving companies financially fragile. In addition, organizational structures must be adjusted to align with new business models, to avoid outdated structures slowing new business expansion and causing missed opportunities.

This article's analysis of AGI industry development offers multiple insights for traditional factories advancing digital transformation and deploying AI-related businesses, as follows:

1. Product development direction: When developing AI products or advancing in-house digital transformation, factories do not need to blindly pursue trillion-parameter large models. Instead, they should prioritize "intelligence density", control costs while meeting scenario-specific requirements, adapt to industry demands for deployment cost and inference efficiency, and improve product cost-effectiveness.

2. Business opportunity exploration: AI penetration into vertical industry scenarios is a clear industry trend. Factories can leverage their own vertical sector experience and industry know-how to develop AI solutions that match on-the-ground demand. Sectors including healthcare, rail transit and the Internet of Things still have a large volume of unmet deployment needs, and long-term accumulated industry expertise can help build differentiated competitiveness.

3. Core takeaways for transformation: When advancing AI and digital transformation, factories should not stop at only upgrading technology. They also need to adjust business models and organizational structures in parallel, gradually shifting from custom project delivery to standardized, replicable products to build stable, sustained revenue, and avoid having outdated business models and organizational structures hold back transformation progress.

This article sorts out current industry trends, customer pain points and viable solutions for large model AI service providers, with key insights as follows:

1. Current industry trends: Large model competition has entered the deep end of commercialization, and the industry has formed two distinct development paths: one focuses on general foundational models, competing on scale and query volume; the other pursues an "industry deep dive" route, rooting operations in vertical scenarios. For most service providers, building on industry experience to pursue the vertical route makes it easier to establish differentiated advantages.

2. Core customer pain points: Industry clients' demand for AI deployment goes far beyond basic intelligence capabilities today. They also require models that understand industry knowledge, meet data security, compliance and reliability requirements, and are generally sensitive to deployment costs and inference efficiency. The traditional custom project delivery model has high costs and long cycles, putting heavy pressure on both suppliers and clients.

3. Viable solutions: Service providers pursuing the vertical route can adopt a "strong base model + deep application" framework, building small-parameter models with high intelligence density to fit scenario-specific needs. They should gradually modularize and standardize capabilities to cut delivery costs, shift from custom projects to sustainable subscription revenue, and adjust organizational structures in parallel to align with the new business model.

Using the Unisound incident as a case study, this article sorts out market demand, operational priorities and risk points for AI-related platforms, as follows:

1. Core demands from AI companies: AI companies in the competitive phase require continuous capital input for R&D, computing power and talent, and have ongoing financing needs after listing. Within regulatory frameworks, platforms can accommodate reasonable financing demands from AI companies to support them in reserving capital for industry competition.

2. Priorities for business development and operation: When onboarding AI companies, platforms should not be attracted only by AGI concepts and technology labels. They need to establish more rational evaluation criteria, focusing on companies' commercialization capability and business model quality, and prioritize onboarding companies that have already validated their go-to-market paths to improve the overall quality of platform-listed assets.

3. Risk mitigation guidance: Platforms should guide the market to focus on AI companies' real operating performance, watch for firms that raise capital frequently but hold large idle funds while falling short of commercialization milestones, disclose relevant risks, help the market establish a more mature valuation framework for AI companies, and promote healthy industry development.

Using Unisound's stock price collapse as a research sample, this article presents new trends and emerging problems in current AGI industry development, offering high reference value for industrial research. Key insights are as follows:

1. New industry trends: The AGI industry has completed early-stage concept hype and technical preparation, and officially entered the phase of commercial competition. The valuation logic of capital markets has undergone a fundamental shift, from prioritizing technology narratives and limited share supply to focusing on business model quality and profitability. Unisound's 90%+ market cap drop from its peak is a典型 example of this shift.

2. Core existing industry problems: AGI companies pursuing the industry deep dive route generally face multiple structural contradictions: long-term project-based delivery fails to generate economies of scale, leading to low customer retention, poor revenue sustainability, high accounts receivable and slow payment cycles that leave firms financially fragile. Most companies operate with sustained losses and lack the ability to generate self-sustaining profits, while organizational upgrading lags behind the iteration speed of technology and business models.

3. Directions for business model research: The industry has now formed two clear AGI commercialization paths. For the general foundational model route, business value is calculated as "intelligence upper bound × token consumption scale"; for the industry deep dive route, industry value is defined as "intelligence density × token value". The latter's transformation direction from project-based delivery to standardized subscription services is an innovative track worthy of in-depth research.

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.

■ 当筹码稀缺的故事讲完

■ 实现AI产业化的路径

■ 商业模式质量再检验

作者|齐笑

编辑|赵元

解禁当日,股价重挫41.19%的云知声,牵动着很多投资者的神经。

云知声股价走势的意义不止于其自身,它更像一面镜子,照出了AI企业在资本市场上不愿意面对却不得不面对的问题:当筹码稀缺的故事结束,技术叙事还需要接受商业兑现的检验,“AGI”标签还能否支撑长期估值?

2025年6月底,云知声登陆港股,被市场称为“港股AGI第一股”。上市初期,在AI概念和稀缺筹码的推动下,公司股价快速上涨,最高触及879港元,对应市值一度超过620亿港元。

2025年9月,云知声的股价持续下行,再加上解禁的影响,云知声股价最低跌至61.2港元。2026年7月9日,收盘价为73.85港元,总市值约55亿港元,较高点蒸发超过90%。

解禁带来的抛压只是表象,背后是资本市场对AI企业估值逻辑的深层变化。大模型热潮下,技术故事可以撑起上市初期的想象空间,但长期价值终须回归商业模式和盈利能力的硬考核。

而云知声面临的压力,其实早有迹象。

01

解禁只是导火索

上市半年后,云知声便开启了密集融资。

2026年1月16日,云知声首次配售,以每股252港元向不少于六名承配人配售78万股新H股,较前一交易日收市价折让约16%,净募资约1.92亿港元。

随后,公司于2月、5月分别进行了第二次和第三次配售,配售价格分别为310港元和228港元,折让幅度扩大至17.7%和19.89%,募资净额分别为3.07亿港元、3.8亿港元。而一般情况下,发行价不得较基准价折让20%或以上。

三次配售累计募资超过8.8亿港元,约为IPO募资净额2.06亿港元的4倍以上。

对于处于大模型竞争阶段的AI企业而言,研发投入、算力建设和人才投入都需要持续资金支持,融资无可厚非。

云知声创始人黄伟此前曾表示,2023年至2025年是大模型发展的“热身赛”,2026年才进入真正竞争阶段。

截至2025年底,公司现金及现金等价物约3.4亿元。一般香港主板上市公司上市后,原则上6个月内不得进行新股发行。云知声的首次配售是在上市6个半月左右,持续且急切地融资背后显然有为后续竞争储备粮草的考虑。

从资金使用方向来看,三次配售的资金,指向了大模型竞赛中最核心的几个环节,包括底层算力基础设施、前沿技术研发、新市场与新场景的开拓。

但频繁融资还是引发了外界对于其资金使用效率、资源配置能力的质疑。

公告显示,截至2026年4月30日,1月配售所募约1.92亿港元有55.7%闲置。2月配售所募约3.07亿港元中,已使用约0.07亿港元,超过97%的资金未动。

对此,新浪证券的文章提出疑问:是战略过于前瞻,需要不断储备资金应对未来,还是当前业务进展不及预期,导致前次资金无法按计划投向?

如果说融资节奏引发了市场疑问,那么U2大模型的发布,则进一步放大了市场对于“技术与商业之间距离”的关注。

6月8日,云知声发布U2大模型,不盲目堆砌参数,不比拼输出长度,更强调面向真实任务的连续执行能力。U2在国内外权威评测中也取得了亮眼的成绩,有媒体将U2的发布解读为云知声的DeepSeek时刻。

但这样的“利好”并没有对冲掉当日指数的下跌,公司股价下跌近11%,并跌破发行价205港元。

过去,AI企业可以凭借模型参数和榜单成绩获得追捧,但随着大模型竞争进入深水区,市场开始把目光更多投向技术与商业之间的那段距离。

U2在设计理念上确实回应了商业化关切,但从理念到业绩之间仍隔着产品化、获客和规模化落地的漫长链条。

对于云知声而言,U2证明的是技术能力、产品能力,但市场需要看到的是这一能力如何转化为实实在在的收入和利润。

02

云知声如何通向AGI?

当前大模型企业大致形成了两类发展路径。一类更强调基础模型能力,通过提升模型智能上限,扩大模型调用规模;另一类则试图将AI能力嵌入具体产业,通过解决真实业务问题实现商业价值。

智谱更强调基础模型能力的平台化价值,其核心逻辑是提升模型能力和应用覆盖范围,通过更大的Token使用规模释放价值,其商业逻辑的公式是:AGI商业价值 = 智能上界??Token消耗规模。这是一场以“通用性”换取“规模效应”的实验。

云知声则选择了后一种“产业深潜”路线,试图通过“强基模+深应用”,在医疗、轨交等具体场景里挖掘价值。

这也是云知声提出“AI公司的行业价值=智能密度??Token价值”的原因。在产业侧,客户不仅要求模型懂行,更受限于本地算力资源,一般对部署成本和推理效率也更敏感。因此,云知声强调“智能密度”,即用较小参数、较低成本达到足够高的智能水平,以满足专业场景的高性价比需求。

这一逻辑在U2模型上得到了体现,相比一味追求万亿参数,U2以不到3000亿参数的规模,试图在特定场景下实现更优的效能比。

理论上,在这个过程中,通用基座不断吸收垂直领域的知识,变得越来越专业,垂直应用也会因为通用基座能力的提升,变得越来越强,形成正循环。

以医疗为例,医疗AI需要进入医生和医院的工作流程。真正能够落地的医疗AI,需要理解专业医学知识,处理私域数据,同时满足医疗行业对于安全、合规和可靠性的要求。这意味着,通用模型提供的是基础智能能力,而行业应用仍需要结合场景经验和行业Know-how。

过去,云知声长期围绕语音识别、自然语言处理、人机交互等领域进行技术积累,并持续布局医疗、物联网等垂直场景。相比于后来进入大模型领域的企业,云知声希望依靠长期行业经验形成差异化。

仍以医疗领域为例,云知声已经服务超过400家三甲医院和超过500家企业客户,并积累了大量真实业务数据。

但这份“深厚积累”在商业化层面却呈现出了另一面:项目制交付。不同医院的业务流程、系统环境差异巨大,AI企业不得不投入大量人力进行定制化开发。项目越多,团队越忙,但没有规模效应。

云知声寄希望于大模型打破这一死结。

大模型时代,企业可以通过统一模型底座和智能体框架,将部分能力模块化、标准化,从而降低交付成本,提高复制效率。

黄伟曾举例,公司保险理赔审核系统过去需要较长周期的定制开发,而基于大模型能力后,客户经过简单配置即可上线,交付周期从3个月缩短至1周,成本降低约80%。

如果这一模式能够成立,云知声希望实现的不只是AI项目增长,而是商业模式的变化:从依靠项目交付获得收入,转向通过标准化产品和智能能力持续创造价值。

其实,不同AGI路线最终都需要回答同一个问题:技术能力如何转化为商业回报。基础模型路线需要证明模型规模能够带来持续调用价值,而产业智能路线则需要证明行业积累能够转化为可复制的产品和收入。

这一点也是市场真正关注的问题。

03

“龙虾”引起的焦虑背后

从2025年的经营数据来看,云知声距离“可复制的产品和收入”还有相当距离。

2025年,云知声收入12.11亿元,同比增长29%。需要单独拿出来讲的是,2025年,云知声的大模型相关收入同比增长超10倍。

但云知声的整体收入主要来源于面向政企客户的AI解决方案项目(如智慧轨交、智慧座舱的语音全链路方案、医疗信息化等),多属于典型的项目制系统集成业务。

比如,智慧生活板块中“解决方案收入”占比约69.8%,这类业务一般按合同履约进度或验收时点确认收入,具有明显的阶段性或一次性特征。

项目制的特点是“一锤子买卖”或“阶段性交付”,难以沉淀为可持续的订阅收入。

客户留存率的数据也可以佐证。以核心的智慧医疗业务为例,客户留存率从2022年的70.4%下滑至2024年的53.3%。53.3%的留存率意味着,近半数客户在项目周期结束后,并未转化为长期用户。

进一步看,这种旧模式带来了财务脆弱。

一方面,政企项目回款周期极长。2025年底,云知声应收账款达到7.8亿元,占资产总额的一半以上,2022年—2024年,应收账款周转天数在250天—290天,2025年周转速度有明显的提升,但回款时间仍接近200天,这意味着,公司做生意,要先替客户垫资半年。

另一方面,公司一直在亏损,2022—2025年,亏损额分别为3.7亿元、3.8亿元、4.5亿元、3.3亿元。

不管是从现金流的角度来看(2024年和2025年经营活动净额分别为-3.2亿元和-2.1亿元)还是从盈利的角度来看,云知声都没有自我造血的能力。

云知声在年报中明确提及“正积极探索以API调用、Token计费等模式构建经常性收入体系”,可以说还处在由重转轻的过程中。

API调用、Token计费等模式的典型是代表Agent(智能体)形态的“龙虾。

抓住“龙虾”机会的一个案例是智谱。

智谱的业务中也有相当比例是项目制落地方式,定制化特点突出。得益于更早的平台化布局以及龙虾的爆火,智谱的业务变得更轻、更标准化,商业化方面也不再是线性增长。截至2026年3月,智谱MaaS平台ARR约17亿元,过去一年增长约60倍。同时,API业务毛利率从3.3%提升至18.9%,业绩又成为了股价的催化剂。

龙虾也是黄伟最近的一次危机感。

当龙虾爆火时,黄伟很焦虑,因为发现团队反应迟缓。“后来我反思了一下,不是他们不想动,是现有的组织架构是按原来的业务模式设计的,没有多余的人和脑子来想这个事。”

黄伟这番话,道出了云知声当下的深层矛盾:技术已经迭代到了2.0,但支撑技术的组织架构依然停留在1.0。

而组织架构往往要适配于其业务形态与盈利模式。对于云知声而言,“龙虾”带来的焦虑不应只是一次组织架构的检讨,更应成为商业基因重塑的契机。毕竟,在AGI的长跑中,技术决定了起跑的速度,但商业模式的质量,才能决定跑多远。

注:文/市值榜团队,文章来源:市值榜(公众号ID:shizhibang2021),本文为作者独立观点,不代表亿邦动力立场。

【版权提示】亿邦动力倡导尊重与保护知识产权。未经许可,任何人不得复制、转载、或以其他方式使用本网站的内容。如发现本站文章存在版权问题,烦请提供版权疑问、身份证明、版权证明、联系方式等发邮件至run@ebrun.com,我们将及时沟通与处理。

已点赞

+1

{{ praise_text }}

好文章,值得被更多人看到

FAQ回顾

云知声的AGI发展路线是什么样的?

云知声选择“产业深潜”的AGI发展路线,提出“AI公司的行业价值=智能密度×Token价值”公式,采用“强基模+深应用”模式,聚焦医疗、轨交等垂直场景,以较小参数、较低成本达到高智能水平,满足专业场景高性价比需求。

云知声当前面临的主要商业化困境是什么?

云知声核心收入多来自项目制AI解决方案,缺乏规模效应,2024年智慧医疗业务客户留存率降至53.3%,应收账款占资产总额超一半,2022到2025年持续亏损,暂不具备自我造血能力,尚未形成稳定的经常性收入体系。

目前AI大模型企业主要有哪两类发展路径?

当前大模型企业主要分为两类发展路径,一类侧重基础模型能力提升,通过扩大模型调用规模获取规模效应,代表企业为智谱;另一类走产业深潜路线,将AI能力嵌入具体产业场景创造商业价值,代表企业为云知声。

广告

下一篇

跨境电商资源对接

《全球跨境电商服务商资源手册》第二期手册现已发布,触达更多资源,了解跨境服务商信息,欢迎点击链接下载。

了解详情

未成年人 黑名单人员 失德艺人……这些人都不能在拼多多直播了

拼多多将持续加大直播间排查和清理力度,并依据平台协议及规则采取相应处置措施,包括关闭直播、回收直播权限,以及平台认为必要的其他处理措施。

电商平台 拼多多 未来零售

2026-07-13 11:26:31