广告
加载中

快手亲儿子 估值1360亿

张雪 2026-05-13 09:10
张雪 2026/05/13 09:10

邦小白快读

EN
全文速览

总:本文核心介绍快手旗下视频生成大模型产品可灵的发展历程、融资规划与当前行业竞争格局,核心干货信息如下:

1. 可灵从2023年内部项目前身发展至今不到两年,当前年化经常性收入已经突破5亿美元,是全球首款开放给公众测试的文生视频大模型,上线后用户增长极快,不到半年服务用户超500万,累计生成超5100万个视频,技术一度领先行业至少一年。

2. 目前可灵计划独立融资20亿美元,估值达到200亿美元约合1360亿人民币,是当前全球估值最高的视频生成大模型独立产品,预计明年第一季度启动IPO。

3. 当前可灵面临内忧外患,先发优势已经被字节、阿里的新品反超,还存在研发投入成本高、核心人才被挖角流失的问题,后续发展还要看融资后的资源投入效果。

总:可灵的发展历程与AI视频生成赛道的变化,给品牌商带来多方面的参考信息,具体干货如下:

1. 消费趋势层面,C端用户对AI生成视频内容的接受度和需求都极高,可灵开放测试不到半年用户就突破500万,足以说明AI内容生产的大众需求已经起来,品牌可以布局AI工具降低自身营销内容生产的成本,提升内容产出效率。

2. 变现路径层面,可灵已经跑通了C端会员订阅、按生成次数计费,B端提供API服务、影视广告游戏等垂直场景解决方案的商业模式,不到两年就实现5亿美元年化收入,大部分收入来自海外市场,给品牌布局AI业务提供了清晰的变现参考。

3. 竞争层面,当前赛道技术迭代快,大厂资源投入力度大,先发优势很容易被反超,品牌布局AI新业务需要控制投入节奏,避开盲目烧钱的陷阱。

总:AI视频生成赛道的最新发展,给相关领域卖家梳理了明确的机会与风险,干货内容如下:

1. 机会层面,AI视频生成是当前一级市场热捧的增长赛道,近段时间已有生数科技、爱诗科技等玩家拿到大额融资,纷纷进入独角兽行列并筹备上市,可灵本身年化收入不到一年从1亿美元涨到5亿美元,过半收入来自海外市场,说明海内外C端、B端需求都非常旺盛,相关服务还有很大增长空间。

2. 风险层面,赛道当前普遍面临推理成本过高的问题,OpenAI的Sora就是因为日均百万美元级运营成本关停,即便是有先发优势,也很容易被字节、阿里这类资源充足的大厂反超,同时核心人才竞争激烈,中小玩家很容易出现人才流失问题。

3. 当前赛道还在洗牌阶段,卖家可以切入细分场景抓住Sora退出后的流量缺口,关注资本动向及时调整布局。

总:AI视频生成赛道的爆发,给工厂带来了新的商业机会与数字化转型启示,具体干货如下:

1. 商业机会层面,AI视频生成行业快速爆发,对算力硬件、配套存储设备、周边配套服务的需求快速增长,工厂可以围绕赛道新增需求开发对应产品,开拓新的增长曲线。

2. 生产设计端,AI视频生成技术可以帮助工厂快速生成产品展示视频、营销推广素材,大幅降低工厂设计和营销的成本,提升新品推广的效率,传统工厂可以引入这类工具优化自身运营流程。

3. 数字化转型启示,快手提前布局AI新技术抓住了先发优势,说明工厂需要尽早关注AI新技术的落地,不过也要注意平衡投入和产出,避免盲目烧钱;可灵拆分独立融资的操作,也给工厂孵化新AI业务提供了新思路,成熟的高潜力新业务可以独立融资对接外部资源发展。

总:AI视频生成赛道的最新变化,给相关AI服务商明确了行业趋势、核心痛点与发展方向,干货内容如下:

1. 行业发展趋势,AI视频生成已经从早期的技术崇拜转向成本约束阶段,当前资本关注度极高,大量资金涌入赛道,多个玩家拿到大额融资成为独角兽,技术迭代速度越来越快,目前头部格局初步形成但尚未稳定,还有较多市场机会。

2. 核心客户与行业痛点,当前行业最突出的痛点就是推理成本过高,投入产出比失衡,OpenAI的Sora就是因为难以承担高昂成本直接关停,同时行业人才竞争激烈,核心团队容易被大厂和资本挖角,中小服务商资源不足很难长期竞争。

3. 解决方案方向,可灵的发展验证了C端订阅加B端API服务的模式可以跑通,服务商可以聚焦垂直细分场景控制成本,抓住Sora退出后的流量缺口抢占市场,对接资本获取资源支撑长期研发投入。

总:可灵AI从快手体系拆分独立融资的事件,给布局AI业务的平台商带来多方面启示,具体干货如下:

1. 市场需求层面,当前C端用户对AI文生视频这类新AI应用的需求非常旺盛,可灵上线不到半年用户就突破500万,说明AI内容生成是平台用户的核心需求之一,平台需要提前布局相关业务抢占用户心智。

2. 运营管理层面,快手早期将可灵定为战略级项目,倾斜全部算力资源,后来升级为一级直接汇报的事业部,才有了快速的技术突破,后续拆分独立融资释放资本价值,给平台孵化高潜力AI业务提供了参考,成熟业务可以独立融资对接更多外部资源。

3. 风险规避层面,赛道竞争激烈,大厂可以用资源投入弥补时间差距,先发优势很容易被反超,同时高昂的研发推理成本对平台资金能力要求很高,平台布局需要平衡资源投入,还要做好核心人才的激励保留,避免核心团队流失。

总:可灵AI独立融资事件,反映了当前AI大模型赛道尤其是视频生成领域的最新产业动向,具备较高的研究价值,核心干货如下:

1. 产业新动向,国内AI视频生成赛道已经进入商业化爆发阶段,跑出了可灵这种年化收入达到5亿美元的成熟项目,资本大量涌入赛道,多个项目跻身全球独角兽行列,不少头部项目已经传出IPO计划,BAT三大互联网巨头都已经下场布局,赛道进入白热化竞争阶段。

2. 行业新问题,当前行业普遍面临推理成本过高的核心问题,技术迭代速度快,对资源投入要求极高,先发优势无法形成稳定壁垒,很容易被资源充足的大厂反超,同时行业核心人才稀缺,人才流失问题频发,这些都是行业需要解决的新问题。

3. 商业模式层面,可灵的发展验证了C端会员订阅加B端API服务加垂直场景解决方案的AI变现模式是可行的,拆分高潜力AI业务独立融资、独立IPO的模式,也为大模型业务发展提供了新的商业模式参考。

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 development trajectory, fundraising plans and current competitive landscape of Keling, Kuaishou’s text-to-video generative AI product, with key takeaways as follows:

1. Keling has developed from an internal project launched in 2023 in less than two years, and already boasts an annual recurring revenue (ARR) exceeding $500 million. As the world’s first text-to-video large model open to public testing, it has seen explosive user growth: it served over 5 million users and generated more than 51 million videos in less than half a year after launch, and once held a technical lead of at least one year over the industry.

2. Keling is currently planning an independent fundraising round targeting $2 billion, at a valuation of $20 billion (approximately 136 billion RMB), making it the highest-valued standalone video generative AI product globally. It is expected to launch its initial public offering (IPO) in the first quarter of next year.

3. Keling now faces both internal and external challenges: its first-mover advantage has already been overtaken by new products from ByteDance and Alibaba. It also grapples with high R&D costs and the loss of core talent to poaching. Its future growth will depend on how effectively it deploys new resources secured through fundraising.

Keling’s growth and shifts in the AI video generation space offer key takeaways for brands, as outlined below:

1. In terms of consumer trends: C-end user acceptance of and demand for AI-generated video content are already very high. Keling crossed 5 million users in less than half a year of public testing, confirming that mass-market demand for AI content production has arrived. Brands can adopt AI tools to cut marketing content production costs and boost output efficiency.

2. In terms of monetization paths: Keling has already validated a viable business model that combines C-end membership subscriptions and pay-per-generation pricing, plus B-end API services and custom solutions for vertical scenarios including film, advertising and gaming. It hit $500 million in ARR in less than two years, with the majority of revenue coming from overseas markets, providing a clear monetization blueprint for brands looking to build out AI-enabled businesses.

3. In terms of competition: The sector is seeing extremely rapid technological iteration, and large tech players are pouring massive resources into the space, meaning first-mover advantages can easily be erased. Brands looking to enter new AI businesses should pace their investments and avoid the trap of blind, unprofitable spending.

The latest developments in the AI video generation sector clarify key opportunities and risks for sellers in related fields, with key insights as follows:

1. Opportunities: AI video generation is a high-growth sector heavily favored by private capital markets. Recently, players including Shengshu Technology and Aishi Technology have secured large funding rounds, reached unicorn status and are preparing for IPOs. Keling’s ARR jumped from $100 million to $500 million in less than a year, with more than half of its revenue coming from overseas, indicating strong demand from both C-end and B-end clients across global markets, with substantial room for growth for related services.

2. Risks: A widespread pain point across the sector is currently extremely high inference costs. OpenAI shut down Sora due to daily operating costs reaching hundreds of thousands of dollars. Even for players with early mover advantages, large resource-rich incumbents like ByteDance and Alibaba can easily overtake the competition. At the same time, competition for core talent is cutthroat, leaving small and medium-sized players particularly vulnerable to talent drain.

3. The sector is still in a period of market consolidation. Sellers can capture market gaps created by Sora’s exit by targeting niche vertical scenarios, and adjust their strategic positioning based on capital market trends.

The boom in AI video generation has created new business opportunities and digital transformation insights for manufacturing factories, as outlined below:

1. Business opportunities: The rapid growth of the AI video generation industry has driven sharply rising demand for computing hardware, supporting storage equipment and related supporting services. Factories can develop targeted products to serve these new sector demands and open new growth curves.

2. Product design and marketing: AI video generation technology can help factories quickly generate product showcase videos and marketing materials, significantly cutting design and marketing costs while speeding up new product promotion. Traditional factories can adopt these tools to optimize their internal operational processes.

3. Digital transformation insights: Kuaishou’s early bet on AI to capture first-mover advantage demonstrates that factories need to monitor and explore new AI-enabled applications early, while also balancing investment and returns to avoid wasteful spending. Keling’s spin-off for independent fundraising also provides a new blueprint for factories to incubate new AI businesses: mature, high-potential new businesses can seek independent financing to access external resources for growth.

The latest shifts in the AI video generation sector clarify industry trends, core pain points and development directions for AI service providers, with key insights below:

1. Industry trends: AI video generation has moved past the early era of tech-focused hype and entered a phase constrained by cost efficiency. The sector attracts extremely high capital attention, with massive funding flowing in, multiple players securing large rounds to reach unicorn status, and accelerating technological iteration. A preliminary top-tier hierarchy has formed but is not yet stabilized, leaving plenty of room for new market entrants.

2. Core clients and industry pain points: The most pressing industry-wide problem is currently exorbitant inference costs, which create unbalanced input-output ratios. OpenAI shut down Sora because it could not sustain these high costs. Additionally, competition for talent is extremely fierce, and core teams are frequently poached by large tech firms and deep-pocketed investors, leaving small and medium-sized service providers with insufficient resources to compete long-term.

3. Development directions: Keling’s growth validates that a combined model of C-end subscriptions plus B-end API services is commercially viable. Service providers can control costs by focusing on niche vertical scenarios, capture market share from the gap left by Sora’s exit, and secure capital backing to support long-term R&D investment.

Keling AI’s spin-off from Kuaishou for independent financing offers multiple insights for platform companies building AI businesses, as outlined below:

1. Market demand: C-end user demand for new AI applications like text-to-video generation is already very strong. Keling crossed 5 million users in less than half a year after launch, confirming that AI content generation is a core user demand for platforms. Platforms should deploy related businesses early to capture user mindshare.

2. Operational and management insights: Kuaishou initially designated Keling as a strategic project and allocated full computing resources, later upgrading it to a first-tier division that reports directly to senior management, enabling rapid technological breakthroughs before its spin-off for independent financing to unlock capital value. This provides a blueprint for platforms incubating high-potential AI businesses: mature projects can spin off to raise independent funding and access more external resources.

3. Risk mitigation: The sector is extremely competitive, and large tech players can close the first-mover gap with massive resource investment, meaning early advantages can easily be overtaken. At the same time, high R&D and inference costs impose strict requirements on a platform’s cash position. Platforms entering the space must balance resource allocation, and put in place incentives to retain core talent to prevent key team departures.

Keling AI’s independent fundraising round reflects the latest industry developments in the large AI model space, particularly in video generation, and carries high research value, with key takeaways as follows:

1. New industry developments: China’s AI video generation sector has entered a phase of commercialization breakout, with mature projects like Keling already reaching $500 million in ARR. Massive capital is flowing into the sector, multiple projects have reached global unicorn status, many leading players have announced IPO plans, and China’s three largest internet giants (BAT) have all entered the space, pushing the sector into a phase of cutthroat competition.

2. New emerging industry problems: The core widespread industry problem remains high inference costs. Rapid technological iteration requires massive resource investment, and first-mover advantages do not create stable moats, meaning resource-rich large players can easily overtake early entrants. At the same time, the sector faces a shortage of core talent, leading to frequent talent drain. All of these are pressing new problems for the industry to resolve.

3. Business model insights: Keling’s growth validates that a combined monetization model of C-end membership subscriptions, B-end API services and vertical scenario solutions is viable for generative AI. Its model of spinning off high-potential AI businesses for independent financing and a standalone IPO also provides a new blueprint for the development of large AI model businesses.

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.

上周,可灵独立融资的消息就已经在一级市场传开了,甚至有投资人还专门在业内打听了一圈该项目的真实性。

近日消息似乎已被证实。据外媒报道,快手计划为旗下视频生成大模型业务可灵(Kling)进行独立融资,估值为200亿美元(约1360亿人民币)。随后,据晚点Latepost报道,可灵此次的融资规模为20亿美元,潜在投资方有腾讯。

若交易完成,可灵将成为目前全球估值最高的视频生成大模型独立产品。

对于可灵独立融资,我在第一时间向快手方面求证,截止发稿,未有回复。不过,一些投资人认为,“估值要的还是有点高了”。毕竟整个快手还不到300亿美元的市值。但也有投资人认为,可灵手握着今年超10亿美元的ARR预期,要这个估值也无可厚非。

此外,与融资消息一同传出的还有可灵的上市时间表,据悉其将于明年第一季度启动IPO,预计到那时的ARR将达到13亿美元。

一边是上市的确定性,一边是可观的商业化数据,这似乎完美契合了当下投资人理想项目的画像。

不过,热闹之下也暗藏隐忧,Sora倒下的身影还未散去,Seedance和Happy-horse又趁势追击,内忧外患下,可灵要杀出一条路,实现预期的商业化也没有那么容易。

发布不足两年,ARR已突破5亿美元

与字节、阿里不同,快手在视频生成领域是行动最早的。早在2023年初,快手CEO程一笑就提出要聚焦推荐和视频生成。

一个标志性事件是在2023年10月,快手重启了内部项目“噗叽”,一个将静态图片转为2秒GIF表情包的工具,这被视为可灵的技术前身。同时,该业务的负责人万鹏飞也成为了可灵早期的核心人物。

2024年2月,Sora横空出世,火遍全网,这也让万鹏飞看到了DiT新型视频生成架构的可行性,于是,从事视觉算法多年的他开始探索在快手打造“中国版Sora”。在3月初的一个内部会上,万鹏飞的想法得到了快手高级副总裁盖坤的支持,进而在“噗叽”的基础上开始打造视频生成模型。

据硅星人报道,可灵项目开始后不到一个月,就获得了程一笑的支持,将可灵项目视为公司战略级项目。甚至亲自发话:“可灵要大做”,“盖坤常说的就是,公司的卡都给你们用,公司全力支持。”

仅仅三个月后,万鹏飞带着团队就给出了结果,2024年6月6日,可灵视频生成大模型官网正式上线,同年11月,独立App正式推出,这是全球首款可供公众测试的文生视频大模型产品。

可灵AI一经推出就迅速获得了市场关注,发布一个月后,在还未开放全面内测时,其申请人数就超过了50万,视频生成数量达700万,到2024年8月27日,用户数超过160万,累计生成超1600万条视频,11月,数据有了更大的突破,服务用户已超500万,累计生成5100万个视频,超1.5亿张图片。

这一系列数据,让可灵AI在全球视频生成大模型赛道迅速站稳了脚跟,业内判断快手在视频大模型上曾一度领先了至少一年的时间。

不过当时在国内这依然还是个不怎么被看好的赛道。比如王小川称视频生成“既不代表AGI,也不代表场景”,李彦宏也在内部会上表示“Sora这种视频生成的投入周期太长,10年、20年都可能拿不到业务收益。无论多么火爆,百度都不去做。”

可事实证明,“视频模型不赚钱,是投资人的集体误判”。2025年3月,也就是在可灵AI正式上线后的第十个月,快手官方公布称,其实现年化经常性收入(ARR)超过1亿美元。这样的成绩直接超过了当时的另一明星AI编程产品Cursor(耗时约12个月)。

从商业模式上看,可灵延续了业内在C端,通过会员订阅、按生成次数计费的模式,同时在B端,主要向企业提供API服务,以及影视广告和游戏场景的解决方案。

在ARR突破1亿美元后,可灵AI趁热打铁,先后发布了可灵2.0和2.1系列模型。不仅如此,在组织架构上,快手也适时调整,把AI提上了空前高度:2025年4月30日快手成立了可灵AI事业部,快手高级副总裁盖坤担任可灵AI事业部负责人,负责LLM大模型、多模态理解大模型以及应用技术研发。调整后可灵AI将作为一级业务部门,向快手董事长兼CEO程一笑汇报。

架构上升级之后,可灵AI名副其实地成为了快手“全村的希望”,其表现也延续了之前的优势,快手在今年3月表示,可灵的年化收入将超过3亿美元,预计年收入将在2026年翻一番。其中一位人士表示,今年第一季度,可灵的营收达到了7500万美元,其中大部分来自包括北美在内的海外市场。

另据最新消息,可灵当前的年化收入(ARR)已经达到5亿美元,已比春节前翻倍。

快手养不起可灵了

从可灵AI手握的剧本来看,不管是技术还是商业化上都占据了一定的先发优势。

比如,字节发布首个类Sora模型的时间比快手晚了3个多月,2024年9月底,它才发布Pixel Dance和Seaweed,而这两款模型直到2025年5月,才整合为了豆包视频模型Seedance。

再比如,生数科技、即梦AI、海螺AI等虽然也进行了商业化,但从规模上讲比可灵小了一个数量级。

然而,这种优势却在今年初突然被打破。今年2月,字节旗下Seedance2.0 AI在国内外互联网爆火出圈,被称为“地表最强视频生成模型”,虽然可灵3.0系列模型也在同期上线,但声量却远不如Seedance2.0,业内有人保守估算,Seedance2.0春节期间月活达到4500万人,已经反超了可灵。

紧接着,3月24日,OpenAI宣布全面关停Sora,包括独立应用、API接口及ChatGPT内置视频功能,原因无他,核心直指视频生成大模型的投入与回报本质。

据《福布斯》此前报道,OpenAI在AI短视频业务上的日均成本高达1500万美元,年化支出超过50亿美元。高昂的推理成本,使得即便拥有领先效果,商业化仍难以成立,这推动行业快速从技术崇拜转向成本约束。

《华尔街日报》披露的财务面更刺眼:Sora日运营成本约100万美元、峰值推理可达1500万/天,整个生命周期in-app收入只有210万美元;用户从超100万MAU跌到不到50万,下载量三个月跌66%。

不止OpenAI,过去一年,几乎所有玩家都受制于单位视频成本。当然,核心玩家的突然离场,留下的除了反思还有流量。所以那段时间,谁能接住Sora的用户成了AI视频生成赛道的核心议题。

这时候,此前几乎没有参与竞争的阿里趁机入局。4月初,阿里淘天的HappyHorse-1.0(欢乐马)匿名空降登顶,官方随后认领,负责人是张迪,曾任快手副总裁、快手大模型团队负责人,也是可灵AI的关键人物之一。

如此一来,可灵3.0Pro被挤到第三,Seedance2.0落第二。可灵3.0被一个更小、更专一的方案反超了,地位也从华语圈第一变为了行业三强之一。

更严峻的是,视频生成赛道处于高速迭代之中,技术竞赛背后比拼的依然是真金白银的砸钱力度,字节、阿里、腾讯等互联网大厂动辄千亿投入,让这场用户争夺战变得格外激烈。

反观快手,其去年全年收入为1428亿元,相比之下,阿里和字节在资源与投入方面,显然更强势,而这两者的“后发制胜”则再次证明了技术领域用资源补时间的真理,如今可灵先发优势正在被逐渐削弱。

另一方面,可灵也面对与Deepseek同样的困境——人才流失,可灵AI在行业冒头后,早期核心团队成员也成为了同行和投资人挖角的对象,除了张迪外,2025年底,可灵基础大模型负责人周国睿也被曝离职。也有消息指出,快手可灵团队工程师的原有薪酬在市场上并不具备竞争力。

内忧外困下,可灵不得不向外界寻求一个解法。

冲击全球最高估值

“在快手体系内,可灵的资本溢价显示不出来,拆分是必然的。”一位投资人如此评价。

同时,面对字节和阿里的围追堵截,可灵AI也没有丝毫退缩。

最近的一个反向操作,或许可以解释可灵AI的野心。上个月初,可灵AI宣布推出为期三个月会员模型优惠计划,折扣从8折到9折不等,并将部分低至免费的图片模型优惠延长,活动覆盖国内Web端与App端。这与当下行业整体算力成本升高、部分厂商酝酿涨价的背景形成明显反差。

当然,最终决定能跟这两家把手腕掰到什么程度的,还取决于口袋多深。

好在,今年国内一级市场对AI视频生成大模型的追逐同样火热。2月5日,生数科技宣布完成超6亿元人民币A+轮融资,该轮融资由中关村科学城公司和星连资本领投,上市公司万兴科技、视觉中国、拓尔思进行战略投资,原有股东启明创投、北京市人工智能产业投资基金、卓源亚洲、建发新兴投资、淮海投资等投资人加码跟投。

仅一个月后,同一赛道的爱诗科技宣布完成3亿美元C轮融资,由鼎晖投资联合领投,中国儒意、三七互娱等产业资本入局,该融资刷新了亚洲AI视频赛道融资纪录。

至此,生数与爱诗双双迈入了全球独角兽行列,不久前,这两家企业也传出了即将赴港上市的消息。

而如今可灵AI开放向一级市场的融资窗口,则为这个赛道又添加了一把火。对于20亿美元的融资规模,业内虽然觉得有些困难,但也在意料之中,毕竟现在获客营销的战争远没结束,能拿钱的时候就该多拿。也有猜测称,这轮融资中也免不了有算力兑换股权的存在。

关于资方,腾讯本就是快手的大股东,根据快手最新曝光的股权,宿华持股9.8%、程一笑持股8.8%,而腾讯则持股15.7%,参与可灵AI的融资也算是名正言顺,这也被看做是腾讯入局AI视频生成大模型的一个信号,该领域也成为了BAT继大模型后的又一个战场。

不管结局如何,快手已经走出了一步,三年时间孵出一个千亿项目已然是一种胜利了,接下来,在200亿美元估值的高光下,就看可灵如何以及能否证明自己了。

注:文/张雪,文章来源:投中网(公众号ID:China-Venture),本文为作者独立观点,不代表亿邦动力立场。

文章来源:投中网

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