广告
加载中

两条AI赛道 一场消耗战:阿里与字节耐力对决

AI新科技组 2026-07-17 11:10
AI新科技组 2026/07/17 11:10

邦小白快读

EN
全文速览

本文核心梳理了阿里与字节在AI赛道的不同布局路径与竞争现状,干货内容如下:

1. 当前AI商业化处于分岔路口,分为阿里做B端收租、字节做C端抢人两条路线,两家均靠主业盈利承担AI高额烧钱成本,最终比拼的是主业底盘韧性和现金流稳定性,2026年是两种商业模式冲刺商业闭环的第一年。

2. 阿里阿里云AI收入连续11个季度实现三位数增长,年化收入突破358亿元,AI收入占云外部收入比重首次突破30%,增速比肩微软Azure,但全年阿里自由现金流净流出466亿元,累计AI基建投入超1200亿元,目前收入以低毛利的卖算力为主,高毛利的模型服务占比很低。

3. 字节豆包月活达3.45亿,日均Token消耗180万亿,但付费转化率仅约1%,2025年字节净利润暴跌超70%,国内C端AI付费意愿低,天花板明显,不过字节手握海外TikTok流量底牌,有机会打开第二增长曲线。

本文梳理了国内头部企业AI商业化发展现状,能为品牌商把握AI相关趋势、布局营销与数字化提供参考,干货内容如下:

1. 消费端AI用户基础已经成型,豆包、阿里千问的月活均突破3亿,但国内C端用户AI付费意愿偏低,重度需求用户占比极低,付费天花板非常明显,品牌商依托C端AI产品做营销推广,应侧重挖掘免费流量的价值,不要过度押注付费渠道。

2. B端AI服务已经进入高速增长期,阿里云AI年化收入超358亿元,连续11个季度保持三位数增长,字节火山引擎占据公有云MaaS市场49.5%的份额,品牌商可对接成熟大模型能力,升级自身数字化运营与用户运营,降低自研AI的成本。

3. 当前AI行业整体处于长期烧钱阶段,后续格局仍存在变数,品牌商布局AI相关业务要预留足够的现金流缓冲,避免盲目押注投入。

本文分析了国内AI赛道的发展现状与风险机会,能帮助卖家把握AI领域的创业与合作方向,干货内容如下:

1. 当前AI商业化已经进入跑通闭环的关键阶段,B端AI服务已经步入规模商业化回报周期,阿里云等头部平台AI收入增长确定性强,卖家可对接头部平台的大模型能力,开发垂直场景的AI应用服务,对接B端客户需求。

2. C端AI已经拥有超3亿级的用户规模,但付费意愿低,付费天花板明显,做C端AI相关产品的卖家不要过高预估付费收入,可优先探索流量变现、广告变现等路径。

3. 风险与机会提示:当前AI行业整体投入巨大,头部企业都出现了现金流承压、净利润下滑的问题,卖家入局要控制前期投入成本,优先对接大厂现成能力,不要盲目投入算力基建;后续可优先绑定现金流更稳定的平台合作,降低自身经营风险。

本文分析了国内AI产业的发展动态,能给工厂推进数字化转型、挖掘商业机会提供启示,干货内容如下:

1. 当前国内AI基础设施已经成熟,阿里云已经完成全栈AI闭环,平头哥自研GPU已经进入量产阶段,大模型同时覆盖B端和C端场景,工厂想要用AI升级产品生产与设计,可低成本对接成熟大模型能力,不需要盲目投入自研,降低转型的资金压力。

2. AI产业的高速扩张带来了大量商业机会,阿里、字节近两年AI资本开支都超千亿元,大量投入流向AI算力基建,给上游硬件生产、配套零部件制造的工厂带来了稳定的订单需求,相关工厂可主动对接头部大厂的供应链需求,拓展业务规模。

3. AI转型与布局需要长期烧钱,头部大厂都需要依靠主业现金流反哺AI投入,工厂布局AI相关业务要提前评估自身现金流承受能力,优先从赋能自身生产设计切入,逐步推进,不要盲目跟风烧钱扩规模。

本文梳理了当前国内AI产业的发展趋势与现存痛点,能给AI相关服务商把握行业方向提供参考,干货内容如下:

1. 当前AI商业化发展趋势清晰,分为两种成熟路径:阿里代表的B端收租模式增长确定性高,B端客户迁移成本高,粘性强,已经进入正向规模商业化回报周期;字节代表的C端流量模式虽然用户规模大,但变现尚不成熟,付费天花板很低。

2. 当前行业的核心痛点明确:全行业都面临高额投入带来的现金流压力,阿里云年自由现金流净流出466亿元,字节2025年净利润暴跌超70%;头部大厂的AI收入结构不合理,低毛利的算力收入占比偏高,高毛利的场景化模型服务供给不足,客户对高性价比的垂直场景AI解决方案需求强烈。

3. 服务商可抓住市场缺口,深耕垂直行业的B端场景化模型开发,对接头部大厂的算力资源,为客户提供完整解决方案,获得增长空间。

本文分析了头部AI平台的发展现状与核心问题,能给各类AI平台的运营布局提供参考,干货内容如下:

1. 当前不同市场对AI平台的需求特征差异明显,B端客户看重服务稳定性与生态配套,一旦迁入平台后短期内切换成本极高,能给平台带来稳定的持续性收入;C端用户对价格和产品体验敏感度高,忠诚度低,容易被竞品抢走,做C端AI平台需要持续投入流量和产品优化,用户留存成本远高于B端。

2. 当前头部AI平台的共性风险是投入压力过大,阿里出现自由现金流净流出,字节出现净利润大幅下滑,平台运营需要平衡增长投入与现金流健康度,避免出现长期持续失血的问题。

3. 运营启示:做B端AI平台要逐步提升高毛利模型服务的占比,不能长期依赖卖算力拉动增长;做C端AI平台要尽快拉开付费版与免费版的价值差距,提升付费转化率,同时可提前挖掘海外流量空间,打开新的增长曲线。

本文梳理了当前中国AI商业化发展阶段的最新动向,对产业研究具有较高的参考价值,干货内容如下:

1. 当前AI产业出现了明确的新动向,AI商业化已经跨越初期投入阶段,进入冲刺商业闭环的关键期,国内已经形成了两种完全不同的成熟商业模式,分别是阿里代表的B端全栈AI基建收租模式,和字节代表的C端流量产品变现模式。

2. 当前产业发展暴露了多个新问题:全行业AI投入规模极高,头部企业都需要依靠主业盈利反哺AI投入,行业整体距离全面盈利还有较长距离;阿里模式存在现金流持续失血、高毛利模型服务占比偏低的问题,字节模式存在C端付费天花板低、投入产出比失衡的问题。

3. 本文给出了产业走向的判断框架,后续可通过三个核心信号判断竞争走向:分别是阿里云下一季度的自由现金流变化、豆包2026年底的付费转化率、两家企业主业对AI投入的支撑能力,最终胜出者不一定是技术最强的,但一定是现金流最稳定的。

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 maps out the distinct strategic paths and competitive landscape of AI development at Alibaba and ByteDance. Key takeaways are as follows:

1. AI commercialization is at a crossroads, with two diverging strategies: Alibaba focuses on B-end revenue generation while ByteDance prioritizes C-end user acquisition. Both companies fund their massive AI spending with profits from their core businesses, so the competition will ultimately come down to the resilience of their core operations and cash flow stability. 2026 will mark the first year both business models aim to achieve a complete profitable commercial cycle.

2. Alibaba Cloud's AI revenue has grown triple-digit for 11 consecutive quarters, with an annualized revenue exceeding RMB 35.8 billion. AI now accounts for over 30% of Alibaba Cloud's external revenue for the first time, with growth rates on par with Microsoft Azure. However, Alibaba recorded a net free cash outflow of RMB 46.6 billion for the full year, with cumulative AI infrastructure investment exceeding RMB 120 billion. Currently, most of its AI revenue comes from low-margin cloud computing power sales, while high-margin model services account for a very small share.

3. ByteDance's Doubao AI has 345 million monthly active users, with 180 trillion tokens consumed daily, but its paid conversion rate is only around 1%. ByteDance's net profit plummeted by more than 70% in 2025, reflecting low willingness to pay for AI among domestic C-end users and a clear low ceiling on C-end monetization. That said, ByteDance holds a trump card in global traffic via TikTok, giving it an opportunity to unlock a second growth curve.

This article summarizes the current status of AI commercialization among leading Chinese companies, offering insights to help brands grasp AI trends and plan their marketing and digital transformation strategies. Key takeaways are as follows:

1. The user base for consumer-facing AI is already established: both Doubao and Alibaba's Tongyi Qianwen have surpassed 300 million monthly active users. However, domestic C-end users have very low willingness to pay for AI, with only a tiny share of heavy users, creating a clear low ceiling for paid C-end AI. For brands looking to market via C-end AI products, the priority should be extracting value from free traffic rather than over-investing in paid channels.

2. B-end AI services have entered a period of rapid growth: Alibaba Cloud's annualized AI revenue exceeds RMB 35.8 billion, with triple-digit growth for 11 consecutive quarters, while ByteDance's Volcano Engine holds 49.5% of the public cloud MaaS market. Brands can leverage the mature large model capabilities of these platforms to upgrade their digital operations and user management, while cutting the costs of developing AI in-house.

3. The AI industry as a whole remains in a prolonged period of heavy investment, so its future structure remains uncertain. Brands planning AI-related initiatives should reserve sufficient cash buffer to avoid overcommitting capital blindly.

This article analyzes the current status, risks and opportunities in China's AI sector, to help sellers identify directions for AI-focused entrepreneurship and cooperation. Key takeaways are as follows:

1. AI commercialization has now entered a critical phase to achieve profitable operations, and B-end AI services have entered a cycle of scalable commercial returns. Leading platforms such as Alibaba Cloud offer strong visibility for AI revenue growth. Sellers can integrate large model capabilities from leading platforms to develop AI applications for vertical scenarios and meet B-end client demand.

2. C-end AI has already exceeded 300 million total users, but willingness to pay remains low and the monetization ceiling is clear. Sellers working on C-end AI-related products should avoid overestimating paid revenue, and prioritize exploring monetization via traffic and advertising instead.

3. Risk and opportunity note: The entire AI industry requires massive capital investment, and even leading players are facing cash flow pressure and declining net profits. New entrants should control upfront investment costs, prioritize leveraging existing capabilities from major tech players instead of investing blindly in computing infrastructure, and prioritize partnerships with platforms that have more stable cash flow to reduce operational risk.

This article analyzes developments in China's AI industry, offering insights for factories advancing digital transformation and identifying new business opportunities. Key takeaways are as follows:

1. China's AI infrastructure is now mature: Alibaba has built a complete full-stack AI closed loop, and its semiconductor subsidiary T-Head has begun mass production of self-developed GPUs, with large models covering both B-end and C-end scenarios. Factories looking to upgrade product production and design with AI can access mature large model capabilities at low cost, avoiding the capital pressure of in-house independent development.

2. The rapid expansion of the AI industry has created substantial new business opportunities: both Alibaba and ByteDance have allocated over RMB 100 billion in capital expenditure for AI in recent years, with most spending flowing into AI computing infrastructure. This has created stable order demand for upstream hardware factories and supporting component manufacturers. Relevant factories can proactively pursue supply chain partnerships with leading tech firms to expand their business scale.

3. AI transformation and initiative require sustained heavy investment, and even leading tech giants rely on cash flow from their core businesses to fund AI spending. Factories looking to pursue AI-related business should first assess their own cash flow capacity, start by leveraging AI to enhance in-house production and design, advance incrementally, and avoid blindly expanding through wasteful spending.

This article maps out current development trends and pain points in China's AI industry, providing reference for AI-focused service providers to define their strategic direction. Key takeaways are as follows:

1. AI commercialization has now formed two clear, mature development paths: The B-end rental model represented by Alibaba offers strong growth visibility; B-end clients face high switching costs, creating strong stickiness and putting this model already in a cycle of positive scalable commercial returns. The C-end traffic model represented by ByteDance boasts a large user base, but monetization remains underdeveloped with a very low monetization ceiling.

2. The core pain points facing the industry are clear: The entire sector faces cash flow pressure from massive investment, with Alibaba recording a net free cash outflow of RMB 46.6 billion and ByteDance seeing its net profit plummet over 70% in 2025. Leading players also have imbalanced AI revenue structures: low-margin computing power sales account for an outsized share of revenue, while supply of high-margin scenario-specific model services is insufficient, leaving strong unmet demand for cost-effective vertical AI solutions from clients.

3. Service providers can capture this market gap by focusing on developing scenario-specific B-end models for vertical industries, integrating computing resources from leading tech firms to deliver end-to-end solutions for clients and unlock growth.

This article analyzes the current status and core challenges facing leading AI platforms, offering reference for the operation and strategic planning of all types of AI platforms. Key takeaways are as follows:

1. Demand characteristics for AI platforms differ sharply across market segments: B-end clients prioritize service stability and ecosystem support, and face very high switching costs once they onboard a platform, generating stable recurring revenue for platforms. C-end users are highly sensitive to price and product experience, with very low loyalty and high churn risk to competitors. Operating a C-end AI platform requires sustained investment in traffic and product optimization, leading to far higher user retention costs than B-end platforms.

2. A shared risk among leading AI platforms is excessive investment pressure: Alibaba has recorded net free cash outflow, while ByteDance has seen a sharp decline in net profit. Platform operators must balance growth investment with cash flow health to avoid prolonged sustained operating losses.

3. Operational takeaways: For B-end AI platforms, operators should gradually increase the share of high-margin model services, rather than relying long-term on computing power sales to drive growth. For C-end AI platforms, operators should quickly establish a clear value gap between free and paid tiers to improve conversion rates, while proactively exploring overseas traffic opportunities to unlock a new growth curve.

This article maps out the latest developments in the current stage of AI commercialization in China, offering high-value reference for industry research. Key takeaways are as follows:

1. The industry has seen clear new developments: AI commercialization has moved beyond the initial investment phase into a critical period of working toward commercially viable closed-loop operations. Two distinct mature business models have emerged in China: the full-stack B-end AI infrastructure rental model represented by Alibaba, and the C-end traffic product monetization model represented by ByteDance.

2. Several new issues have been exposed in the current development phase: The entire industry requires extremely large-scale investment, and even leading players must rely on core business profits to fund AI spending, meaning the industry as a whole remains far from widespread profitability. The Alibaba model faces challenges of sustained cash outflow and low share of high-margin model services, while the ByteDance model faces a low C-end monetization ceiling and imbalanced return on investment.

3. This article provides a framework for assessing the industry's future trajectory. Three core signals can be used to judge the future of competition: changes in Alibaba Cloud's free cash flow in coming quarters, Doubao's paid conversion rate by the end of 2026, and the ability of both companies' core businesses to sustain AI investment. The eventual winner will not necessarily be the one with the best technology, but the one with the most stable cash flow.

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.

阿里巴巴依靠云计算现金流反哺全栈AI基建,字节跳动依托广告业务覆盖豆包大模型流量消耗。二者均以核心盈利业务承担AI高额资本开支,现金流、净利润承压显著。这场耐力赛比拼,终究是两家巨头谁的主业底盘更坚韧、更扛得住长期烧钱。

AI商业化走到了一个分岔路口。左边是阿里云,AI收入连续11个季度三位数增长,年化超358亿元,自由现金流在一年内流出466亿。右边是字节豆包,月活3.45亿、日均Token消耗180万亿,但2025年净利润(IFRS口径)暴跌超70%。一条靠B端收租,一条靠C端抢人。两条路都烧钱,问题是谁先跑通。

01

阿里云:AI账算得清,但代价比想象的大

2026年5月13日,阿里巴巴发布FY2026 Q4财报,CEO吴泳铭用了一个措辞:"AI已正式跨越初期投入阶段,进入正向的规模商业化回报周期。"

数据确实漂亮。阿里云当季收入416.26亿元,同比增长38%,外部商业化收入增速升至40%,为九个季度以来最快。其中AI相关产品收入89.71亿元,连续第11个季度实现三位数同比增长,年化收入突破358亿元。AI收入占云外部收入的比重,首次突破30%。

这个增速放在全球云市场都不逊色——比肩微软Azure,高于亚马逊AWS。Gartner数据显示,阿里云以32.8%的份额稳居中国IaaS市场第一。大模型层面,千问大模型完成从B端平台"悟空"到C端应用的全覆盖,平头哥自研GPU进入量产,全栈AI初步成闭环。

但漂亮的收入数字背面,是自由现金流的"放血"。

FY2026全年,阿里自由现金流净流出466亿元。钱花在了三个地方:AI算力基建、即时零售物流和千问APP用户的获取。过去一年,阿里投向AI+云基础设施的资本开支已累计1200亿元。阿里管理层在分析师会上给出的预期是:未来一年AI收入占比将突破50%,成为阿里云收入增长的主要引擎。

换句话说,阿里在赌一件事:只要AI收入跑得够快,自由现金流迟早能转正。

但这里有一个隐藏的风险:阿里云的AI收入中,有多少是"卖算力",有多少是"卖模型"?卖算力吃的是基础设施红利,护城河在规模而非技术。如果算力供需关系变化,或者价格战再起,这部分收入的韧性存疑。而卖模型的毛利更高,但阿里云的模型收入至今未单独披露,说明占比尚且不高。

另一个不容忽视的背景是:阿里的传统电商业务增速已放缓至个位数(中国电商CRM同比增8%),即时零售虽然增长57%,但仍在大幅亏损。这意味着阿里云的AI叙事,是阿里集团整体故事的核心支柱,这根柱子不能断。

02

字节豆包:C端流量帝国,变现还在试水温

与阿里"算账式增长"不同,字节打的是"流量式入侵"。

2026年6月,豆包大模型日均Token调用量突破180万亿,较发布时增长超1500倍。IDC数据显示,火山引擎以49.5%的份额在公有云MaaS市场独占鳌头。豆包App月活3.45亿,千问App月活也过3亿,但千问的背后是一整个阿里生态在导流,豆包是字节用算法硬推出来的。

但华丽的数据下面,有三个让投资人睡不着的问题。

问题一:收入撑不起流量。 火山引擎2025年整体营收约200亿元,仅为阿里云的五分之一。49.5%的MaaS市场份额并没有换来对等的收入规模。虎嗅7月7日的分析文章直言:"高调用量未转化为对等收入,且不排除字节内部流量内循环撑起数据的可能。"

问题二:净利润被AI吃掉了。2025年,字节净利润同比暴跌超70%。CFO李亮解释称,这是IFRS会计准则下优先股和期权成本所致,实际经营利润率仅小幅下滑。但市场不看解释,2025年字节国内外收入增速分别为20%(国内)和50%(海外),收入没少赚,净利润却从2024年的约330亿美元跌至不足100亿美元。唯一的解释就是:AI投入太大了。2026年字节计划AI资本开支约2000亿元(约280亿美元),远超2025年水平。

问题三:C端付费天花板触手可及。2026年5月4日,豆包官宣付费三档:68元/月、200元/月、500元/月。上线后月活从3.45亿降至约3.3亿,流失约610万用户。摩根士丹利测算,豆包付费订阅中性预期年收入仅4.26亿-6.84亿美元。换算成人民币约30-48亿元,这个数字放在整个字节的收入大盘中,杯水车薪。

更关键的是,豆包官方明确表示"免费版保留核心功能并持续升级"。也就是说,付费版和免费版之间的差异,不够硬。虎嗅分析指出:国内C端AI付费意愿偏低,重度需求用户占比极低,付费天花板明显,仅能作为营收补充。

03

收租的阿里,烧钱的字节

如果把两家公司的AI业务放在一起算这笔账,结论非常清楚。

阿里云的AI收入虽然只有358亿元,但增长确定性高,B端客户一旦迁移到千问生态,短期内切换成本巨大,这是典型的"收租模式"。而且阿里云已明确表态"进入商业化回报周期",意味着接下来会逐步上提毛利率,改善现金流。

字节的问题不在规模,在效率。3.45亿月活日均消耗180万亿Token,但付费用户保守估计只有300万-500万,付费转化率约1%。火山引擎整体营收仅200亿,即便未来两三年翻三倍到600亿,也依然是阿里云体量的三分之一。

而字节的AI资本开支已与阿里相当甚至反超——花更多的钱,赚更少的钱,这笔账在资本市场上很难讲圆。

但字节有一个阿里没有的底牌:海外。字节2025年海外收入增速50%,占比已超30%,快于国内。豆包海外版和TikTok生态的协同,可能在国内变现天花板到来之前,打开第二条增长曲线。至少在TikTok的流量池里,豆包还有大量免费用户等待转化。

04

三个信号,可以判断大厂AI入口战争的走向

信号一:阿里云Q1 FY2027(4-6月)财报的现金流。如果阿里云的自由现金流在未来没有明显改善,说明"商业化回报周期"的论调需要重新审视。阿里赌的是"规模换利润",但如果规模增长的代价是现金流持续失血,这个赌局的风险比管理层承认的更大。

信号二:豆包付费转化率数据。 字节至今未公开披露豆包的付费用户数,这在以"数据驱动"闻名的字节内部是罕见的。如果到2026年底付费转换率依然低于3%,豆包C端商业化的天花板将被彻底确认。届时字节只有两条路:要么接受豆包长期"不赚钱",要么加速推企业级。

信号三:谁先扛不住"撒钱模式"? 阿里自由现金流从正转负,字节净利润跌超70%。两家公司都在用主业利润喂养AI。但阿里还有云计算的基本盘在增长,字节的内容和广告业务增长已经放缓。如果2026年下半年宏观经济进一步承压,字节的广告基本盘(国内)可能率先受到冲击,从而影响其AI基础设施的持续投入能力。

说到底,阿里和字节打的不是同一场战争。阿里在打"算力-模型-应用"全栈,目标是"AI时代的亚马逊AWS";字节在打"流量-产品-付费"闭环,目标是"AI时代的抖音"。一个是基建逻辑,一个是产品逻辑。

基建逻辑的优势是,一旦基础设施铺好,后来者很难挑战。产品逻辑的劣势是,C端用户随时可能被下一个"更便宜更好用"的产品抢走。

2026年不是AI的终局,是两种商业模式同时冲刺商业闭环的第一年。活下来的,不一定是技术最强的,但一定是现金流最稳的。

注:文/AI新科技组,文章来源:科技新知(公众号ID:kejixinzhi),本文为作者独立观点,不代表亿邦动力立场。

文章来源:科技新知

广告
微信
朋友圈

FAQ回顾

阿里和字节的AI布局分别采用什么模式?

阿里采用全栈AI基建模式,依托云计算现金流支撑AI投入,AI收入连续11个季度三位数增长,年化超358亿元,走B端商业化收租路线;字节依托广告业务覆盖豆包大模型流量消耗,走C端流量产品路线,豆包月活达3.45亿,日均Token消耗180万亿。

国内AI大模型商业化普遍面临哪些核心难点?

国内AI大模型商业化普遍面临高投入压力,阿里FY2026年自由现金流净流出466亿元,字节2025年净利润暴跌超70%;同时C端付费意愿偏低,重度需求用户占比极低,付费天花板明显,仅能作为营收补充。

阿里云在中国云市场的表现怎么样?

Gartner数据显示,阿里云以32.8%的份额稳居中国IaaS市场第一,其AI相关产品收入连续11个季度三位数同比增长,年化收入突破358亿元,AI收入占云外部收入比重首次突破30%,增速比肩微软Azure、高于亚马逊AWS。

字节豆包大模型的商业化现状如何?

截至2026年6月,豆包大模型月活3.45亿,日均Token调用量突破180万亿,付费订阅中性预期年收入仅约30-48亿元,在字节整体收入中占比极低,付费转化率仅约1%,C端商业化天花板明显。

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