广告
加载中

可灵头上缺了一朵遮风挡雨的云

景行 2026-06-10 14:11
景行 2026/06/10 14:11

邦小白快读

EN
全文速览

本文核心梳理了国内头部视频大模型可灵的发展现状、增长路径与面临的困境,核心干货信息如下:

1. 可灵是国内三大顶流视频模型之一,目前正寻求Pre-IPO轮融资,计划从快手独立,核心原因是研发和算力投入过高,而快手没有云业务生态做支撑,难以长期承担高额投入。

2. 可灵走出了差异化的增长路径,不靠硬投放和技术参数吸引用户,主打简单易用+爆款模板,半年内三次引爆全球社交平台,靠用户自发传播实现低成本增长,目前主要收入来自专业创作者订阅。

3. 为应对行业价格战,可灵多次下调价格,部分图片功能已经免费开放,普通用户可以零成本体验基础功能,目前可灵仍在持续投入,试图构建独立产品的护城河。

本文梳理了可灵的品牌运营策略、价格竞争现状以及AI视频行业的消费趋势,对AI相关品牌运营有较高参考价值,干货总结如下:

1. 品牌营销层面:可灵抓住普通用户对AI产品的核心需求,不靠堆砌技术参数,靠打造爆款模板引发用户自发传播,还借助名人背书、热点事件造势,实现零成本撬动亿级传播,同时通过高额奖金、战略资源激励创作者,持续沉淀品牌核心用户。

2. 定价与竞争层面:可灵曾尝试提价转嫁算力成本,因用户反对回调,目前为应对竞争持续降价,行业当前价格战激烈,背靠云业务的竞品可以靠云业务盈利消化成本,打低价抢占市场,无云支撑的品牌需要深耕核心用户应对竞争。

3. 消费趋势层面:用户更偏好易用性产品,对AI产品新鲜感消退快,愿意为可落地的爆款功能付费,专业创作者是当前AI视频产品的核心付费群体。

本文梳理了AI视频生成赛道的竞争格局、发展机会与潜在风险,能给AI相关领域卖家提供清晰参考,干货总结如下:

1. 行业机会层面:当前AI视频生成赛道仍处于增长期,C端社交传播破圈是头部竞品尚未充分挖掘的路径,可灵已经验证了该路径的可行性,专业创作者群体的付费需求已经被充分验证,可灵近七成收入来自该群体,赛道机会主要向先发、能构建护城河的厂商倾斜。

2. 风险提示层面:当前行业价格战已经进入白热化,谷歌、阿里、字节等巨头背靠云业务,可以推出超低价产品抢占市场,AI产品普遍留存率极低,首月留存仅42%,远低于传统应用,而且需要持续投入百亿级的算力、研发成本,当前多数产品投入远高于收入,盈利拐点尚不清晰。

3. 可参考的经营策略:深耕核心创作者群体,通过资源激励拉高用户迁移成本,持续制造爆款维持用户留存,构建自身生态护城河。

AI视频生成赛道的快速发展,能给相关领域工厂带来产品设计、商业机会以及数字化转型的多方面启示,干货总结如下:

1. 产品生产设计需求层面:C端用户并不关注AI模型的技术参数,更看重易用性,能快速上手做出符合热点的内容,可灵靠简单的“一张图、一句话”+爆款模板,实现了比Runway几十种专业工具更好的破圈效果,这提示工厂设计相关AI配套产品时,要优先聚焦场景化易用性,而非堆砌功能。

2. 商业机会层面:当前头部玩家对AI大模型的投入持续增长,快手2026年新增110亿元资本开支全部投向可灵等大模型领域,赛道对算力基础设施、相关硬件配套、AI内容生产周边产品的需求持续上涨,相关工厂有较大的市场增长空间。

3. 数字化转型启示:AI已经成为内容生产领域的核心增长驱动力,提前布局AI相关能力,能帮助工厂抓住新质生产力的增长机会,依托互联网平台的社交传播经验,更容易实现新产品的C端破圈。

本文梳理了AI视频生成行业的发展趋势、核心痛点与可行解决方案,对AI相关服务商有较高参考价值,干货总结如下:

1. 行业发展趋势层面:当前AI视频生成赛道已经进入第一梯队竞争阶段,演化出两条主流发展路径,一条是背靠云业务生态,将AI视频模型作为云业务的获客入口,靠云业务整体盈利,另一条是独立发展,靠社交传播破圈,深耕专业创作者的订阅收入,目前行业已经开启白热化价格战,头部巨头纷纷降价抢占市场。

2. 行业核心痛点:AI应用普遍存在留存率低的问题,用户新鲜感消退快,独立发展的AI模型面临算力成本高、投入压力大的问题,缺少云生态支撑的独立模型,在价格战中处于劣势,争夺核心创作者的运营成本也在持续上升。

3. 可行解决方案参考:可灵探索的方案是持续打造爆款模板,靠用户自发传播降低获客成本,通过创作者激励沉淀核心用户,拉高用户迁移成本,逐步构建独立产品的生态护城河。

可灵的发展路径,能给布局AI业务的平台商带来运营、生态建设以及风险规避的多方面参考,干货总结如下:

1. AI业务对平台的核心需求:AI视频模型需要持续高额的算力、研发投入,背靠云业务生态的平台,可以通过云业务的整体盈利消化AI模型的成本,支撑长期价格竞争,缺少云生态支撑的AI项目,面临极大的生存压力,这提示布局AI业务的平台,需要提前配套完善云服务生态,才能支撑AI业务长期发展。

2. 用户运营可参考做法:可灵验证了“社交传播+爆款模板”的轻量化运营路径,获客成本远低于传统硬推广,效果更好,平台运营AI产品可以参考该路径,围绕用户对易用内容的需求,激励创作者持续产出爆款,维持用户留存。

3. 竞争风险规避:当前行业价格战激烈,平台不要盲目跟风降价,要聚焦核心用户群体,通过沉淀用户资产、拉高迁移成本构筑护城河,避免陷入持续烧钱的恶性竞争。

本文披露了国内AI视频生成赛道的最新发展动向,提出了独立AI大模型发展的新问题,对产业研究有较高的资料价值,干货总结如下:

1. 产业新动向:当前AI视频生成赛道已经形成三种成熟商业模式,分别是背靠云生态,将AI模型作为云业务获客入口,代表玩家为谷歌Veo、阿里快乐马、字节Seedance2.0;独立发展,靠社交破圈深耕专业创作者订阅,代表玩家为可灵;面向B端做专业视频编辑工具,代表玩家为Runway。目前行业头部已经开启价格战,集中度不断提升,机会向先发厂商集中。

2. 产业新问题:独立发展的AI大模型缺少云生态支撑,面临成本高、价格竞争压力大、盈利难的困境,同时AI应用普遍留存率远低于传统应用,如何维持用户长期留存是全行业共同面对的新问题。

3. 研究启示:独立AI模型可以通过社交传播、深耕核心创作者群体构建护城河,分拆独立融资是缓解高额投入压力的可行路径,为独立AI大模型的发展提供了新的探索方向。

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 current development, growth trajectory and challenges facing Keling, one of China’s leading video generative AI models. Key takeaways are as follows:

1. Keling ranks among China’s top three video AI models. It is currently seeking pre-IPO funding and planning to spin off from Kuaishou, primarily because its R&D and computing power costs are too high for Kuaishou, which lacks a cloud business ecosystem to sustain these long-term heavy expenditures.

2. Keling has carved out a differentiated growth path. Instead of attracting users via heavy advertising or technical parameter hype, it focuses on ease of use and viral templates. It has achieved three global viral hits on social media within half a year, driving low-cost growth through organic user sharing. Its primary revenue currently comes from subscriptions from professional creators.

3. To compete in the industry’s fierce price war, Keling has cut prices multiple times, and made some of its image generation features free for all users, allowing casual users to access core functions at no cost. Keling continues to invest heavily to build a moat for its independent product.

This article analyzes Keling’s brand operation strategy, current pricing competition and consumer trends in the AI video industry, offering valuable insights for AI brand operators. Key takeaways are as follows:

1. Brand marketing: Keling tapped into casual users’ core demand for easy-to-use AI products. Instead of overemphasizing technical specifications, it built viral templates to drive organic user sharing, and leveraged celebrity endorsements and trending events to generate 100 million+ impressions at zero customer acquisition cost. It also incentivizes creators with large cash prizes and strategic resources to continuously build a loyal core user base.

2. Pricing and competition: Keling previously attempted to raise prices to pass on computing costs, but rolled back the change after user pushback. It has repeatedly cut prices to stay competitive. In the current fierce industry price war, competitors backed by cloud businesses can offset AI development costs with cloud revenue to sustain low prices and seize market share. Brands without cloud support must focus on deepening engagement with core users to remain competitive.

3. Consumer trends: Users now prioritize usability over technical sophistication, and their novelty for AI products fades quickly. They are only willing to pay for practical, viral-ready features. Professional creators remain the core paying customer group for AI video products today.

This article breaks down the competitive landscape, growth opportunities and potential risks of the AI video generation track, offering clear guidance for sellers in AI-related fields. Key takeaways are as follows:

1. Industry opportunities: The AI video generation track is still in a growth phase, and C端 social viral expansion remains an under-exploited path for leading players, which Keling has already proven viable. The paying demand of professional creators is also well validated, with nearly 70% of Keling’s revenue coming from this group. Opportunities in the track are increasingly tilted toward early movers that can build sustainable competitive moats.

2. Risk warnings: The industry’s price war has intensified to a fever pitch. Tech giants including Google, Alibaba and ByteDance, all backed by cloud businesses, can launch ultra-low-priced products to grab market share. AI products also universally face extremely low user retention: their first-month retention rate stands at just 42%, far lower than that of traditional applications. The track also requires sustained 10-billion-yuan level investment in computing power and R&D, and for most players, investment far outpaces revenue, with no clear profitability inflection point in sight.

3. Actionable operating strategies: Deepen engagement with your core creator base, use resource incentives to increase user switching costs, continuously create viral content to sustain retention, and build an ecological moat for your business.

The rapid growth of the AI video generation track offers insights for related manufacturers in product design, business opportunities and digital transformation. Key takeaways are as follows:

1. Product design and demand: C端 users do not care about AI model technical parameters; they prioritize usability and the ability to quickly create trend-aligned content. Keling’s simple "one image, one prompt" plus viral template model achieved far greater mainstream traction than Runway’s dozens of professional editing tools. This suggests that when designing AI-related products, manufacturers should prioritize scenario-based usability over feature bloat.

2. Business opportunities: Leading players are continuously increasing investment in large AI models. Kuaishou has earmarked all of its RMB 11 billion in additional 2026 capital expenditure for large model development including Keling. Demand for computing infrastructure, related hardware supporting, and AI content production peripheral products continues to rise, creating significant room for market growth for relevant manufacturers.

3. Insights for digital transformation: AI has become a core growth driver for the content production industry. Early布局 of AI-related capabilities can help manufacturers capture growth opportunities from new productive forces. Leveraging the social distribution experience of internet platforms also makes it far easier for new products to break through to C端 audiences.

This article summarizes the development trends, core pain points and actionable solutions for the AI video generation industry, offering valuable reference for AI-related service providers. Key takeaways are as follows:

1. Industry development trends: The AI video generation track has entered a first-tier competition phase, and has evolved two mainstream development paths. The first is backing a cloud business ecosystem, positioning the AI video model as a customer acquisition tool for cloud services, with overall profitability coming from the cloud business. The second is independent development, breaking out via social distribution, and focusing on subscription revenue from professional creators. The industry is now in a white-hot price war, with top giants cutting prices one after another to seize market share.

2. Core industry pain points: AI applications universally struggle with low user retention as user novelty fades quickly. Independently developed AI models face extremely high computing costs and heavy investment pressure; independent models without cloud ecosystem support are at a distinct disadvantage in the price war, and the operating cost of competing for core creators is also rising continuously.

3. Reference for actionable solutions: The approach Keling has explored is to continuously develop viral templates, reducing customer acquisition costs through organic user sharing, retaining core users via creator incentives, increasing user switching costs, and gradually building an ecological moat for the independent product.

Keling’s growth trajectory offers multiple insights for platforms布局 AI businesses, covering operations, ecosystem building and risk mitigation. Key takeaways are as follows:

1. Core platform requirements for AI business: AI video models require sustained, heavy investment in computing power and R&D. Platforms with established cloud business ecosystems can offset AI model costs through overall cloud business profitability, and sustain long-term price competition. AI projects without cloud ecosystem support face enormous survival pressure. This indicates that platforms布局 AI business need to build out a complete cloud service ecosystem in advance to support long-term AI business development.

2. Reference for user operations: Keling has validated the lightweight "social viral distribution + viral templates" operation path, which delivers better results at far lower customer acquisition costs than traditional hard advertising. Platforms operating AI products can adopt this approach: focus on user demand for easy content creation, incentivize creators to continuously produce viral content, and sustain user retention.

3. Competitive risk mitigation: The current industry price war is extremely fierce. Platforms should not blindly follow competitors in cutting prices. Instead, they should focus on their core user base, build a competitive moat by retaining user assets and increasing switching costs, and avoid getting trapped in a vicious cycle of continuous cash burn.

This article discloses the latest development trends of China’s AI video generation track, and raises new questions about the development of independent large AI models, offering high value for industry research. Key takeaways are as follows:

1. New industry trends: The AI video generation track has now formed three mature business models. First: backing a cloud ecosystem, positioning the AI model as a customer acquisition channel for cloud business, represented by Google Veo, Alibaba Kuailema and ByteDance Seedance2.0. Second: independent operation, breaking out via social distribution and focusing on subscriptions from professional creators, represented by Keling. Third: offering professional video editing tools for B端 clients, represented by Runway. Top industry players have launched a price war, industry concentration is rising, and opportunities are increasingly concentrated in early market entrants.

2. New industry challenges: Independently developed large AI models lack cloud ecosystem support, and face high costs, intense price competition pressure and difficulties turning a profit. At the same time, AI applications universally have far lower user retention than traditional applications, and how to sustain long-term user retention is a new challenge facing the entire industry.

3. Research insights: Independent AI models can build competitive moats via social distribution and deepening engagement with core creator groups. Spinning off for independent financing is a viable path to ease pressure from high investment, and offers a new exploration direction for the development of independent large AI models.

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.

无云的快手最狠心?

作者|景行

编辑|古廿

据报道,可灵正在寻求Pre-IPO轮融资,投前估值为180亿美元。

从财报看,快手有意独立可灵,以拉拢外部融资,核心原因是算力与研发两个烧钱大头。

研发,一季度开支36亿元,同比增长9.8%,财报明确归因于“雇员福利开支增加及对AI的投入增加”。

销售成本端,带宽与服务器托管(对应AI训练推理的资源消耗)17.49亿元,同比增长18.4%;物业及设备折旧、无形资产摊销(对应前期AI算力基础设施折旧计入当期成本)17.75亿元,同比增长43.7%。

更大的数字在后面。快手财报会议披露,2026年集团整体资本开支将达到约260亿元人民币,较去年新增约110亿元,多出来这部分优先给可灵和大模型相关投资。

花了这么多,但是可灵能够给快手带来的收入多数以订阅费用为主。

作为国内的三大视频模型顶流之一,这种表现远不及阿里的快乐马还有字节的seedance2.0。快乐马可以给电商、阿里云带去新增量,Seedance2.0可以给字节短剧、火山云带去新增长。

以Seedance2.0为例,据报道,今年火山云MaaS收入目前已经完成去年实际营收的10倍增长。这些增长几乎都归功于Seedance 2.0的爆发。

显然,同为国内的三大视频模型顶流的第一梯队,目前可灵在缺云的快手体系内被严重低估了。

穷人发家走四方,可灵想要成长起来,从快手独立出来或许是最好的选择。

AI模型之外的可灵

相较看得见的硬成本,快手正通过一套全球社交传播、KOL创作激励、折扣促销的组合拳推动用户增长与付费转化。

换句话说,每一轮快手爆发增长的结果,靠技术自然驱动,更要靠平台不遗余力的全球推广。

分析平台Similarweb数据显示,自2024年6月上线以来,可灵在三个月里投放了大约230万元人民币,主要用于X、油管等主流平台的关键词推广,其搜索CPC费用就超过0.46美元。

多方现象则将可灵的营销重心从搜索入口引向社交造势。

其一,作为Sora之后质量最强的视频模型,可灵引发海外社交媒体的极大热情,自学可灵注册方式,跪求测试资格一度成为热点话题。

其二,马斯克在X高度评价可灵生成内容,称AI正在加速娱乐化。这番表述经过媒体引用,成为可灵出海的有力背书。

其三,就在可灵用户激增,从获客转向留客的关键时间段,2025年春节期间,可灵“万物膨胀”模板意外爆红,用户狂热地上传地标建筑,生成毛绒玩具观感的动态视频。

这令可灵看到视频模型背后的传播属性。相较投放广告,渗透短视频空间让用户自己询问制作方法,或许是更有效的增长方式。

这是Sora、Veo、Runway都未曾踏入的路径,三者都将注意力放在企业级市场。

例如,Runway在产品中放了30多种工具,有基础的自动抠像、简单裁剪,也有复杂的绿幕抠像、图像修复、动态追踪,相比可灵的“一张图、一句话”,Runway希望替代AE与PR,成为视频编辑的专业工具。

作为全球第二大短视频平台,基因优势注定快手更懂社交传播规律,懂得用户更感兴趣的不是视频模型的技术参数,而是里面有什么爆款模板,能让我快速上手。

结果是从宠物跳舞到纸手机,再到最近一轮的棒球现场,可灵在半年内连续三次引爆全球社交媒体,每次都带来下载量和收入的迅猛增长。

用传统的曝光成本逻辑审视,快手用零成本撬动了以亿计的传播价值。但背后,可灵的创作赛事与创作者激励投入,正在猛涨。

2024年首次举办的可灵创作者大赛,总奖金池是30万元;到2026年第一季度,可灵的专项赛事已经接近十场,4K影像创作大赛,仅金奖奖金就达1万美元。

在可灵AI创作者计划3.0中,可灵计划每月提供百万奖金、千万灵感值作为激励,并常态化运营。

平台认定的超级创作者则能每月拿到1314元价值的黑金会员,以及国际电影节展映机会等战略资源。

反映在财报中,快手2026年一季度销售及营销开支达到103亿元,同比增长4.4%,财报明确增长原因是推广活动开支增加。

没有云业务必须独立?

会计准则中,折扣销售应当对冲收入而非计入销售费用,以避免虚增营收嫌疑。

这一规则同样适用于处在价格战阵痛期的可灵。

可灵一度尝试提升价格,以应对水涨船高的算力成本。以5秒视频生成所需灵感值为例,在1.6版本是35灵感值,去年4月发布的可灵2.0大师版上调到100灵感值,但遭用户不满,随后在2.1版又调了回去。

在可灵2.5 Turbo中,这个价格已经降到25灵感值。

会员价格方面,可灵2.6发布时,直接推出了会员年卡6.6折活动;自2026年4月1日以来,可灵对AI产品大力降价,可灵3.0模型铂金及以上会员享受8折,部分图片功能低至免费。

对应的是,来自竞争对手的压迫感正在逐步增强。

国内市场,4月阿里快乐马以低价+刷榜态势杀入市场。以1080P视频生成为例,快乐马会员价格约为0.78元每秒,可灵约为0.84元每秒,Seedance2.0约为1元每秒。有消息称,未来字节预计推出Seedance 2.0 Lite,价格只要0.5元每秒。

对于可灵,这是一个危险信号。相较吃订阅增长的可灵,HappyHorse关注的是如何助力背后的云业务打通Token经济。

在海外市场,可灵最大对手谷歌也是类似的云生态逻辑。

2024年12月,谷歌Veo 2的API价格还是0.5美元每秒,不到一年时间,Veo 3降到0.4美元,Veo 3 Fast一路杀到0.15美元,真正的腰斩价。

直到今年4月,Veo 3.1真正将价格战甩到行业脸上——Fast版本720p只需0.1美元每秒,Lite版则只需0.05美元。

这不只是价格下降这么简单,谷歌正在重新寻找视频模型的定位。

Sora的狼狈退场向业界打了一针预防针——盯着生成质量打,纯粹的视频生成应用缺少下一个时代的入场券。当产品运营成本远超收入,看不到盈利拐点曙光时,退场是最理性的选择。

谷歌的做法是,将Veo作为谷歌云获客入口,借道Google AI Studio向付费开发者开放,用token调用计费取代会员订阅。只要谷歌云的算力、存储销售走高,Veo自身的利润情况可以忽略。

换句话说,谷歌并不想卖视频生成能力,而是为谷歌云上架一款视频产品,向企业与开发者兜售。

可灵的商业模式则是构筑一个更开放的生态,并从中筛选专业创作者。快手CEO程一笑曾在财报会议中表示,可灵接近七成的收入来源于P端订阅会员,即介于普通消费者和企业之间的专业创作者。

这也是为什么唯有可灵热衷于大规模的社交媒体破圈,当对手都在云模同步作战时,可灵没有退路,必须一条路杀到底。

骑虎难下

“这个赛道真正的机会,可能只属于那么几家先发,且能守住护城河的厂商。”有AI从业者对「市象」表示。

可灵面临的挑战也在于此。

据《智能涌现》报道,Seedance2.0单月收入已经超过10亿元。相比之下,可灵在今年整个一季度,收入水平是超过6.5亿元。

“都是边迭代边发布。”有行业观察人士表示,企业需要不断用新模型向资本市场与用户证明实力,用户预期也在变化,从专注头部产品,到谁便宜就用谁的产品。

风投机构a16z合伙人Olivia Moore曾经复盘过Sora的死因——第一天用户留存率10%,30天用户留存率只有1%,60天留存率跌至0%:

“该模型确实具有病毒式传播性,但大多数用户并未持续使用该应用。”

这契合了大部分用户对视频模型的印象,因新功能刷屏一两天,然后迅速销声匿迹。

Sequoia Capital数据显示,传统应用的首月留存率大概是63%,而AI应用只有42%,日活跃度则要更低。

快手的应对方案是,不断制造下一个爆款,用内容维护跳过新鲜感陷阱。

在公开表述中,快手高级副总裁马宏彬将可灵定义为平权的新质生产力:“AI的终极价值,是把表达权还给每一个人。”终极目的,是打破技术壁垒,实现创作平权。

而维系用户留存,离不开持续投入。

第一层是算力账。可灵2025年全年收入是10.4亿元,而快手2026年预计投向可灵等大模型,新增的资本开支额是110亿元,用百亿级别的资本投入,去支撑10亿级的产品,这是快手目前在做的事。

第二层是创作者激励。一个季度近十场专项赛事背后,快手急需扩充可灵在P端的曝光度,不仅给钱给灵感值,更给长期合作的战略资源。

这本质上是一场用户资产的投资,投资专业创作者,产出带有可灵标签的内容,进一步在C端扩散。当字节即梦与阿里快乐马下场争夺创作者时,可灵面临持续投入的压力,要么加码力度,要么看着创作者流失。

第三层是价格补贴。0.05美元每秒的Veo 3.1 Lite,甚至不是谷歌的全力。在极端情况下,谷歌可以允许视频推理亏损,只要吃掉客户的算力、存储需求,总账仍然是赚的。

没有一个云帝国在后面支撑,行业越是打破AI视频的价格认知,越是挑战独立视频模型的定位。

这迫使可灵进一步强化P端创作者的基本盘,通过更多的社交传播强化品牌认知,让专业创作者在平台沉淀更多素材设计,以拉升迁移成本,进而做成独立产品的生态护城河。

这将影响下一步市场对可灵独立的判断。一个分拆的可灵能否持续增长,取决于能积累多少用户的使用习惯,而不是能把生成价格拉到多低。

注:文/景行,文章来源:市象,本文为作者独立观点,不代表亿邦动力立场。

文章来源:市象

广告
微信
朋友圈

这么好看,分享一下?

朋友圈 分享

APP内打开

+1
+1
微信好友 朋友圈 新浪微博 QQ空间
关闭
收藏成功
发送
/140 0