[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fhAdM6cX5z8BSZBIyqemzeOpsrCZIZoaKjxa6uQAsFFA":3},{"code":4,"msg":5,"data":6},200,"操作成功",{"id":7,"title":8,"content":9,"digest":10,"source":10,"coverPath":11,"thumbsCoverPath":12,"isTop":13,"isShow":14,"baseClick":13,"clickCount":15,"createTime":16,"typeId":17,"isNewest":18,"newsInfoTypeRespVo":19,"voiceUrl":22,"voiceSize":23,"taskId":24,"releaseTime":25,"titleEn":26,"contentEn":27,"voiceUrlEn":28,"taskIdEn":29,"voiceSizeEn":30},1279,"快手的“AI商业内循环”跑通了 ","\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">每当提到那些划时代的新技术时，大多数人脑海中，总会浮现出摧枯拉朽般的颠覆景象：蒸汽机一响，工业革命就迅速推倒农业文明；互联网诞生，信息时代便席卷全球，颠覆传统生活方式。\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">但实际上，真正能扎根壮大、改变世界的颠覆性技术，鲜少是靠纯粹的“替代”一蹴而就，往往是在提升效率、降低成本的过程中，与传统模式共生共长，最终悄然重塑格局。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">今年中报期的谷歌与Meta就是最好的例证，在纯AI产品商业化有限的情况下，依旧双双大幅beat了市场预期，依靠的正是AI加持下，Advantage+、Reel、Pmax等广告业务突飞猛进。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">聚焦到国内市场，当前的AI发展与应用主逻辑亦是如此——就在昨日，中国视频大模型领先者快手发布了2025年中报：营业收入录得350.46亿，同比增长13.1%，经调整利润56.18亿，同比增长21.6%，营收利润双双大幅beat了预期。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002Fef1688bf97ed4580bf82218e5bafa460\u002F3c6fb6dfbdf043cfae0abf44e293477a.webp\" width=\"746\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">并且在未来预期层面，快手也展现出对AI发展及整体业务发展相对乐观的信号，业绩会中CFO金秉透露，可灵AI 2025年全年收入预计比今年初目标翻倍，这使快手坚定了在可灵AI上进行长期投入的决心，预计可灵AI相关的2025年Capex上的投入较年初预算也将翻倍。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">同时，快手宣布派发上市以来首次特别股息，每股0.46港元，合计共约20亿港元，在回馈股东的同时，也向资本市场表明对新周期充满自信。财报业绩会后，资本市场也给予快手这份超预期中报积极反馈，快手-W（HK1024）今日盘中拉升接近5%。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">快手这份优异的中报，为我们提供了观察AI如何深度改造成熟商业生态的先进样本：依托这份财报，我们将从供需两侧，以及内容和电商两个维度，“2X2”四个方面，详细探讨下AI如何重塑既有短视频商业生态。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">01 需求破壁：AI撬开快手增长天花板\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">根据中国互联网发展状况调查报告显示，去年下半年至今年年初，受限于上网设备达到增长的天花板，实际网络视频的整体用户使用率，在过去18个月内呈现了高位回落的态势，从97.7%下降至96.7%。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F365bc8bda5414ab9a534d0622da18c7b\u002F90c13213dfb34b1d8064903385e85499.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp class=\"ql-align-center\">\u003Cspan class=\"ql-lineHeight-1-75\" style=\"color: rgb(187, 187, 187);\">图：过去两年网络视频用户规模及使用率，来源：中国互联网络信息中心\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">尽管产业大势依然式微，快手却呈现出一副“特立独行”的图景，中报显示：\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">今年二季度，快手的DAU和MAU分别达到了4.09亿及7.15亿，分别同比增长了3.4%及3.3%，平均DAU再创新高并连续四个季度超过4亿。平均使用时长达到了126.8分钟，用户总使用时长同比增长了7.5%。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002Fa92051a092614d458ea51b6d0040d385\u002F5058578fffc048648cd907f1c4d1ce78.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">快手各项核心运营数据的稳步增长，核心逻辑即在于通过AI技术提升推荐机制，实现了内容供给和用户需求更进一步的匹配：\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">在用户量稳定的业态下，内容供给扩容与需求的增长关系并不是线性的，尤其是短视频内容平台，本质是供需非对称商业模式。比如爱看搞笑视频的用户，即便有再多的科普类供给，也不会导致其需求增长。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">诚然，与实物商品供给类似的是，现如今内容供给的数量绝不是阻碍需求的核心，核心在于能否通过供需匹配的优化，提高用户粘性从而带动整体使用时长增长，提升总流量。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">而快手通过AI技术的应用，找到了提升内容侧供需匹配的有效方法，比如已在快手应用双端（APP、极速版）上线应用的OneRec端到端推荐大模型。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">传统的推荐模型，大致都是一套“多阶段流水线”架构，其核心的问题在于，每一层模型只管自己那摊事，很难从整体出发考虑用户的真实兴趣。并且多层模型会形成掣肘，一边要拉高点击率，一边还要兼顾冷门和多样性，最终导致一致性和效率持续恶化。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">OneRec 本质上是把推荐问题从多阶段、分层筛选，彻底切换成了端到端的大模型生成。过去，推荐系统像装配线，一层层地筛选内容；而现在OneRec 通过大模型，直接生成出一串用户最可能感兴趣的视频内容列表。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">总体而言，OneRec端到端大模型解决了两个核心问题：\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px;\">一是成本和效率。\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">成本侧通过架构革新，训练和推理的算力利用率（MFU） 分别飙升至 23.7%和 28.8%，运营成本（OPEX）仅为传统方案的10.6%。效率侧大语言模型在顶级芯片H100 上的算力利用率普遍能达到 40% 以上。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px;\">二是推荐效果。\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">OneRec 在内容理解上引入多模态语义分词器。不仅只看视频的标签、标题或者简单的视觉特征，而是把每个视频拆解成更细致的信息片段——画面内容、文字、配音、音乐，甚至用户的互动。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">比如一条美食短视频，分词器不仅会捕捉画面里的食材、烹饪动作、厨房环境，还能分析配音里的生活化语气，用户评论和弹幕，都会变成内容理解的一部分，在进行更细分更立体的理解用户行为后，自然而然能明确用户的喜好，从而改善推理的逻辑。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">我们可以直接来看效果，快手针对OneRec上线后进行了一周 5% 流量 AB测试，类似于医药行业的头对头实验，在点赞、关注、评论等所有交互指标上均获正向收益。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F88f1d0d5e752470c984d2c6e3a0b6c83\u002Fd4333544b4ec432491b528a355ee04a4.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">OneRec在快手本地生活服务场景同样表现惊艳：AB对比实验表明该方案推动GMV暴涨 21.01%、订单量提升 17.89%、购买用户数增长 18.58%，其中新客获取效率更实现 23.02%的显著提升。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">用户能够看到内容，和用户能够看到自己想看的内容，两字之差，万字之别。自然能够帮助快手提升用户粘性，抬升总流量。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">再来聊一聊商品需求侧，即快手通过AI技术正在逐步重构内容电商的决策链路。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">传统货架电商的商业链条维持的核心是搜索，即人找货；内容电商的商业链条维持的核心是推荐算法，即货找人。而快手依托大模型的能力，在进一步完善进化推荐算法的基础上，不仅实现了更精准的货找人，还实现了服务AI化，完善了消费链路。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">比如同样买一件衣服，在快手的APP上使用基于AIGC技术AI试衣功能，可以更直观地向用户展示上身效果，缩短传统种草-下单-到货试衣-退换货的复杂步骤，笔者亲测，效果非常不错。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F080502cb736b4f0e87b6861e2d8bb772\u002Fb7ad413425a144b3b0dc627191f44474.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp class=\"ql-align-center\">\u003Cspan class=\"ql-lineHeight-1-75\" style=\"color: rgb(187, 187, 187);\">图：快手APP试衣功能图示，来源：快手APP\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">得益于商品需求侧AI赋能的优化，消费决策链路显著减短，快手电商二季度活跃买家数量达到了1.34亿，电商活跃用户的复购次数也不断提高。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">同时快手也将AI技术运用在了人货匹配和售前售后体验上，快手基于大模型的能力能够准确预测用户购买行为和购买偏好，自动识别强关联类目和兴趣标签，让算法更理解用户潜在需求。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">快手售前售后服务的智能客服聪慧度也要远高于传统自动客服，可以解决很多商家日常面对的高频问题。比如今年618翠绿珠宝的案例，就依托于快手AI助手，售前售后回答了用户大量高频、专业化的产品信息，解决了60%-70%的常规问题，大大减轻了人工客服的压力。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">商品侧通过AI技术得到放大，也直观地体现在了财报中，二季度快手整体的电商GMV录得3589亿元，同比增长17.6%。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">总结而言，快手之所以能够在内容平台触底的瓶颈期，依旧能够取得不俗的增长的核心原因之一，就是通过AI技术实现了内容层面的良性循环，并实现了用户商品需求的精准触达，从而突破了需求侧预期内的天花板。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">02 供给革新：AI重塑短视频商业生态\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">还是延循着我们前文的逻辑，如果从商业视角出发，供给侧也分为商品供给和内容供给，供给优化最直观的账面体现在于，快手中报期披露，今年二季度新入驻快手的商家数量同比增长50%，商城场域的日均动销商家同比增长30%。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">在现如今平台分化割据的大背景下，快手依旧能够攫取到供给侧新增量，除了前文中提到快手是目前为数不多能够抬升总流量的平台外，更重要的是快手通过技术赋能，不断纾解电商商家两个核心痛点：高昂的营销成本和经营成本。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">电商商家想要搞营销，首先要考虑的就是计算ROI，但是传统的投流工具最大的问题在于，整个营销链路每一步都无法精准测算，最终可能导致投后ROI效果远离预期。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">第一个难点便是，营销素材的成本无法有效控制，电商营销节奏快，成本高，如果营销素材跟不上流量趋势，投放的效果就差强人意。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">但快手应用AI技术的出现解决了这一难题，比如今年磁力大会中提到的磁力开创平台，商家可以一键剪辑爆火视频同款、智能生成高质量脚本、制作数字人视频，以更低成本、更快速度获得更高品质的投放素材，有效解决了成本和效能之间的矛盾。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">再比如女娲数字人，支持虚拟人24小时开播，拥有矩阵运营能力，直播营销真正实现了“门槛低，效率高”。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F8c9a45ca01a74029b8153ca767bf10fa\u002F1b6ee46bf4ca46978949bb1b8f7194e8.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">根据年报季快手CEO在电话会中透露，快手内部测算，AI大模型预计可以把客户的短视频营销素材制作成本降低60-70%甚至更高。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">第二个难点是推广流量的转化率普遍不高，推广广告和自然流量很难交互匹配。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">而快手依托于AI能力推出了UAX（Universal Auto X，全自动投放）解决方案，其拥有智能基建、智能调控、智能创意、智能定向四大模块核心能力，可以有效辅助客户、代理商增强跑量能力、稳定跑量成本、提升投放效率。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">根据快手商业化算法负责人江鹏披露，营销推荐大模型的思维链决策学习能够让模型像人类一样逐步拆解和归纳复杂问题，推理出用户的购买意图、商品的营销点，更精准的人货匹配带来商品订单量提升10%，新品消耗提升20%。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">UAX全自动投放解决方案完成了由规则决策到模型化决策的升级，今年二季度，UAX的营销消耗占外循环总消耗的比例提升至65%左右。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">优化了ROI，解决了投放转化率的堵点，商家投流的倾向自然就会提升，也会带动预算的倾斜，今年上半年，快手整体的线上营销收入录得198亿元，同比增长12.8%便是最好的印证和结果。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">第三个难点是，从投放到经营落地的传统链条很长，商家需要大量的基建工作才能维护好投流带来的业务转化，对于部分中小商家而言，很难有足够的经营经验和人工成本去匹配投流转化的实际业务。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">快手从技术和政策两个维度解决了这个问题，一方面前文中提到的从素材制作、广告投放、到开播讲品、智能客服都有相对应的低成本AI工具提供给商家，帮助商家承接投流带动的业务增长。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">另一方面，快手拿出了真金白银的补贴，面向新手商家，快手电商上线了新商冷启计划，重点针对首播到90天内的新商定向扶持与成长提拉。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">鞋靴主播范小洛老板娘，今年2月底开始经营快手，截至5月，账号累计GMV已超1000万。范小洛在入驻之初，保证日均直播时长 5 - 7 小时，借助首播 30 天专属扶持、首播 90 天优质新商持续提拉实现增长。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">同时，利用快手跨场域流量资源，通过爆款切片、短视频种草实现场域间流量互通，超3 成成交来自短视频引流。全站推广也帮助范小洛实现了更快地成长，使投放 ROI 稳定在 30 左右。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">综上所述，快手利用AI技术，解决了传统商户可能面临的三大问题，自然而然会带动供给侧繁荣的生态，叠加需求侧优化匹配的双向作用，最终促成了GMV的高速增长。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">03 王牌“可灵”：快手价值蜕变的高阶叙事\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">行文至此，我们大致理清了快手中报期所透露出AI改变供给、需求、内容、商品的四个维度的逻辑，也便不难理解快手的增长，其实与meta、谷歌二季度财报展现的逻辑别无二致。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">但快手的AI叙事远不止于此，在全球AI竞赛叙事的大背景下，其王牌AI应用——“可灵AI”的价值，其实要远高于目前市场的普遍认知，这也是快手价值进一步蜕变的重要变量。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">首先，可灵AI是中国参与全球AI应用巅峰角逐的关键力量。中国需要有标志性的AI应用来对标全球AI前沿应用，而可灵AI的能力出众。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">目前多模态领域的AI应用，强敌环伺，国内有可灵AI、即梦、海螺等等，海外有老牌多模态产品Runway、Sora、Stability AI。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">所有的技术应用发展伊始，都是小部分技术爱好和科技尝鲜者率先成为意见领袖，但随着产业不断深入发展，更多专业人士可以从更多的切面和细节中探寻真正有价值的技术创新。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">在今年的WAIC上，可灵AI的能力得到了专业人士的广泛认可，全球知名创意素材平台Freepik的CEO Abela表示，“我们接入了超过10个不同的视频生成大模型，用户选择使用可灵AI生成的视频数量，超过其他模型加起来的总和。”\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">显然，专业人士有力印证了可灵AI在全球范围内的技术领先性和用户认可度，确立了其作为中国AI应用的代表地位。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">其次，可灵AI正持续向市场证明其作为先进生产力的巨大价值。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">AI技术的生命力在于能否真正转化为推动产业升级和社会进步的引擎。我们可以看到，可灵AI正通过广泛的商业化落地场景，生动诠释着这一点，比如前文中提到的广告素材制作，再比如电影工业化落地的场景：\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">6月25日，全球首部AI单元故事集《新世界加载中》于北京举行首映礼，该作品由快手可灵AI与星芒短剧联合出品、异类Outliers团队制作，总时长达180分钟，该剧也是目前业内已知题材涉猎最广、叙事体量最大的AIGC剧集作品，全球累计播放量达到2亿。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F4ff8eada9bed41e0995ce80260d144da\u002F7631916303ef4b09bb7bc31d06590818.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">甚至游戏工业领域可灵AI的落地场景也不少，热门手游《逆水寒》与可灵AI达成了深度合作，将业内领先的AI视频生成能力内置于游戏中，丰富了游戏的社交玩法。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">数据显示，截至今年7月，可灵AI已累计生产超2亿个视频和4亿张图片，服务超2万家企业客户。程一笑也在电话会上提到，“未来我们希望随着模型综合性能的提升，让可灵AI逐步具备在游戏制作和专业影视制作等工业级场景的深度应用能力，拓展更多产业用户。”\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">最后，可灵AI通过良好的商业化表现，不仅提升了估值，更塑造了中国AI应用资本开支良性循环的范本。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">今年二季度，可灵AI营业收入超过2.5亿元，现如今，鲜少有AI应用能够自信披露商业化收入进程，可灵AI自信交出商业化的答卷，不仅能够证明其实际产品力，同时也向市场证明AI应用商业化的潜力和价值。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">根据中金研报数据显示，“可灵AI”系列模型在全球AI视频工具市场的访问量份额已高达30.7%，稳居行业第一。按照业绩会中CFO金秉透露，可灵AI 2025年全年收入预计比今年初目标翻倍。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">快手依托于可灵AI优异的商业化表现，也为中国AI领域资本开支升级开了个头，CFO金秉透露，可灵AI相关Capex投入也较年初预算翻倍，并且考虑到可灵AI推理算力层面已实现毛利率打正且保持稳定，因此追加算力投入对集团整体利润影响可控，资本开支迈入良性循环。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">我们曾经多次分析过目前中美AI发展角力的关键点就是资本开支，美国大厂寄希望于通过大幅度增加AICapex不断建立和巩固属于自己的业务高墙，从而提高了AI的行业的进入门槛。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">现如今我们头部大厂因为受限于AI资本回报率较低，没有足够的信心投入更多资本开支来弥补技术差距。而快手可灵AI通过产品硬实力，抬高ROI，不仅为自身资本投入建立了良性循环，也为中国AI应用发展打了个样。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">如果说AI赋能即有业务带动业绩增长是快手中短期潜在的预期差，那么可灵AI的业绩表现，无疑为快手长周期发展打上了一针强心剂。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">总结而言，透过快手这份中报，我们清晰地看到：AI并非仅仅是遥远未来的图景，在成熟的平台上，AI技术已经能够深入渗透到连接用户和内容，赋能商家和供给的每一个环节，成为驱动核心业务增长的关键引擎。\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(187, 187, 187);\">【新闻来源】蓝鲸新闻  文｜锦缎  \u003C\u002Fspan>\u003Ca href=\"https:\u002F\u002Fwww.sohu.com\u002Fa\u002F927609704_250147\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(187, 187, 187);\">https:\u002F\u002Fwww.sohu.com\u002Fa\u002F927609704_250147\u003C\u002Fa>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(187, 187, 187);\">（本网转发此文章，旨在为读者提供更多的信息资讯，所涉内容不构成投资、消费建议。文章事实如有疑问，请与有关方核实，文章观点非本网观点，仅供读者参考。）\u003C\u002Fspan>\u003C\u002Fp>","","https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F3507f905ff2f43899a0758dcf38bf548\u002FAI领域.jpg","https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002Fthumbs\u002F3507f905ff2f43899a0758dcf38bf548\u002FAI领域.jpg",0,1,220,"2025-08-26 21:10",2,false,{"id":17,"name":20,"enName":21},"芯位视野","Xinwei Vision","https:\u002F\u002Fxinwei-dev-test.oss-cn-shenzhen.aliyuncs.com\u002Fintelligent\u002Faudio%3Added88dc-5c24-4c60-bd41-76d8e0d735ce%3A0.wav?Expires=1756218698&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=2xlmOeRu6mzti3L412J7WKPfop8%3D",31569426,"dded88dc-5c24-4c60-bd41-76d8e0d735ce","2025-08-26 20:57","Kuaishou's \"AI Commercial Inner Cycle\" has been successfully implemented.","\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">Whenever people talk about groundbreaking new technologies, most people's minds are always filled with the image of a revolutionary impact: the steam engine sounds, and the Industrial Revolution quickly overthrows the agricultural civilization; the birth of the Internet brings the information age to sweep the globe, overturning traditional lifestyles.\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">In fact, truly rooted and powerful disruptive technologies that change the world rarely come from pure \"replacement\" in one go. They often coexist and grow with traditional models during the process of improving efficiency and reducing costs, eventually quietly reshaping the landscape.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Google and Meta's mid-year reports this year are the best examples. Despite limited commercialization of pure AI products, they both greatly beat market expectations, relying on the rapid progress of advertising businesses such as Advantage+, Reel, and Pmax under AI empowerment.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Focusing on the domestic market, the current development and application logic of AI is also similar - just yesterday, China's leading video large model company Kuaishou released its mid-year report for 2025: revenue reached 35.046 billion, an increase of 13.1% year-on-year, adjusted profit was 5.618 billion, an increase of 21.6% year-on-year, both revenue and profit significantly beat expectations.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002Fef1688bf97ed4580bf82218e5bafa460\u002F3c6fb6dfbdf043cfae0abf44e293477a.webp\" width=\"746\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Moreover, in terms of future expectations, Kuaishou also showed relatively optimistic signals regarding AI development and overall business development. During the earnings call, CFO Jin Bing revealed that the revenue of XINWEI AI is expected to double compared to the target at the beginning of this year, which has strengthened Kuaishou's determination to make long-term investments in XINWEI AI. The Capex investment related to XINWEI AI in 2025 is also expected to double compared to the initial budget.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">At the same time, Kuaishou announced the distribution of its first special dividend since listing, 0.46 Hong Kong dollars per share, totaling approximately 2 billion Hong Kong dollars. While returning value to shareholders, it also signaled confidence in the new cycle to the capital market. After the earnings call, the capital market gave a positive response to Kuaishou's unexpectedly strong mid-year report, with Kuaishou-W (HK1024) rising nearly 5% during the day.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Kuaishou's excellent mid-year report provides us with an advanced sample of how AI can deeply reform an established commercial ecosystem: based on this report, we will explore in detail how AI reshapes the existing short video commercial ecosystem from four aspects - supply and demand sides, and content and e-commerce dimensions.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">01 Demand Breakthrough: AI Opens the Growth Ceiling of Kuaishou\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">According to the survey report on the development status of China's Internet, from the second half of last year to the beginning of this year, due to the growth ceiling of online devices, the overall user usage rate of online videos showed a high-level decline in the past 18 months, decreasing from 97.7% to 96.7%.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F365bc8bda5414ab9a534d0622da18c7b\u002F90c13213dfb34b1d8064903385e85499.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp class=\"ql-align-center\">\u003Cspan class=\"ql-lineHeight-1-75\" style=\"color: rgb(187, 187, 187);\">Figure: The scale and usage rate of online video users over the past two years, source: China Internet Network Information Center\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Although the industry trend is still weak, Kuaishou shows a unique picture. The mid-year report shows:\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">In the second quarter of this year, Kuaishou's DAU and MAU reached 409 million and 715 million respectively, increasing by 3.4% and 3.3% year-on-year. The average DAU set a new record and has exceeded 400 million for four consecutive quarters. The average usage time reached 126.8 minutes, and the total user usage time increased by 7.5% year-on-year.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002Fa92051a092614d458ea51b6d0040d385\u002F5058578fffc048648cd907f1c4d1ce78.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The steady growth of Kuaishou's core operational data is fundamentally due to the improvement of the recommendation mechanism through AI technology, achieving further matching between content supply and user demand:\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Under a stable user base, the relationship between content supply expansion and demand growth is not linear, especially for short video content platforms, which essentially operate on an asymmetric supply and demand business model. For example, users who like funny videos will not see their demand increase even if there are more science and technology content supplies.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Indeed, similar to physical product supply, the quantity of content supply is no longer the core obstacle to demand. The core lies in whether it can improve user stickiness through optimizing supply-demand matching, thereby driving an increase in total usage time and traffic.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Kuaishou found an effective way to improve the supply-demand matching on the content side through the application of AI technology, such as the OneRec end-to-end recommendation large model already launched on both ends of the Kuaishou app (APP and极速版).\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Traditional recommendation models generally adopt a \"multi-stage pipeline\" architecture, and the core problem lies in each layer only focusing on its own tasks, making it difficult to consider the user's true interests from an overall perspective. Moreover, multiple layers of models create bottlenecks, one side needs to increase click-through rates, while the other needs to ensure cold content and diversity, ultimately leading to continuous deterioration in consistency and efficiency.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">OneRec essentially transforms the recommendation problem from multi-stage, layered screening into an end-to-end large model generation. In the past, the recommendation system was like an assembly line, screening content layer by layer; now, OneRec directly generates a list of videos that users are most likely interested in through a large model.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Overall, the OneRec end-to-end large model solves two core issues:\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px;\">First, cost and efficiency.\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">On the cost side, through architectural innovation, the computing power utilization (MFU) for training and inference has surged to 23.7% and 28.8% respectively, and the operating expenses (OPEX) are only 10.6% of traditional solutions. On the efficiency side, large language models can achieve a computing power utilization of over 40% on top-tier chips H100.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px;\">Second, recommendation effectiveness.\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">OneRec introduces a multimodal semantic tokenizer in content understanding. It doesn't just look at the tags, titles, or simple visual features of videos, but breaks down each video into more detailed information segments—visual content, text, voiceover, music, and even user interactions.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">For example, a food short video, the tokenizer not only captures ingredients, cooking actions, and kitchen environments in the visuals, but also analyzes the casual tone in the voiceover, and user comments and弹幕 will become part of the content understanding. After a more detailed and three-dimensional understanding of user behavior, it naturally clarifies user preferences, thus improving the reasoning logic.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">We can directly see the results. Kuaishou conducted a 5% traffic AB test after the launch of OneRec for a week, similar to a head-to-head experiment in the pharmaceutical industry. All interaction indicators such as likes, follows, and comments showed positive gains.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F88f1d0d5e752470c984d2c6e3a0b6c83\u002Fd4333544b4ec432491b528a355ee04a4.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">OneRec also performed impressively in Kuaishou's local life service scenarios: AB comparison experiments show that this solution boosted GMV by 21.01%, order volume increased by 17.89%, and the number of purchasing users grew by 18.58%. Among them, the efficiency of acquiring new customers improved by 23.02% significantly.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Users seeing content and users seeing what they want to see differ by just two characters, but the difference is vast. Naturally, it helps Kuaishou enhance user stickiness and increase total traffic.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Let's talk about the goods demand side next, that is, Kuaishou is gradually reengineering the decision-making process of content commerce through AI technology.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The core of the traditional shelf e-commerce business chain is search, i.e., people find goods; the core of the content e-commerce business chain is recommendation algorithms, i.e., goods find people. Based on the capabilities of large models, Kuaishou has not only achieved more accurate goods finding people, but also realized service AIization, completing the consumption chain.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">For example, when buying a piece of clothing, using the AI fitting function based on AIGC technology on the Kuaishou APP can more intuitively show the wearing effect, shortening the complex steps of traditional grass-rooting - ordering - receiving - trying on - returning. I personally tested it, and the effect was very good.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F080502cb736b4f0e87b6861e2d8bb772\u002Fb7ad413425a144b3b0dc627191f44474.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp class=\"ql-align-center\">\u003Cspan class=\"ql-lineHeight-1-75\" style=\"color: rgb(187, 187, 187);\">Figure: Kuaishou APP Fitting Function Diagram, Source: Kuaishou APP\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Thanks to the optimization of AI empowerment on the goods demand side, the consumption decision-making chain has been significantly shortened, and the number of active buyers on Kuaishou e-commerce in the second quarter reached 134 million, and the repurchase frequency of active e-commerce users has continued to increase.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">At the same time, Kuaishou has also applied AI technology to person-goods matching and pre-sale and after-sales experience. Based on the capabilities of large models, Kuaishou can accurately predict user purchase behavior and preferences, automatically identify strongly related categories and interest tags, allowing algorithms to better understand users' potential needs.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The intelligence of Kuaishou's pre-sale and after-sales customer service is far superior to traditional automatic customer service, and can solve many frequent problems faced by merchants daily. For example, in the case of翠绿珠宝 during this year's 618, it relied on Kuaishou AI assistant to answer a large number of high-frequency, professional product information from users, solving 60%-70% of routine issues, greatly reducing the pressure on manual customer service.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The amplification of goods through AI technology is also directly reflected in the financial report. In the second quarter, Kuaishou's overall e-commerce GMV reached 358.9 billion yuan, an increase of 17.6% year-on-year.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">In summary, one of the core reasons why Kuaishou can still achieve remarkable growth during the bottleneck period of content platforms is that it has achieved a virtuous cycle in the content aspect through AI technology and achieved precise touchpoints of user goods demands, thereby breaking through the ceiling of the demand side within expectations.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">02 Supply Innovation: AI Reconstructs the Short Video Commercial Ecosystem\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Still following our previous logic, if we look from a business perspective, the supply side is also divided into goods supply and content supply. The most direct accounting manifestation of supply optimization is that the number of new merchants joining Kuaishou in the second quarter this year increased by 50%, and the daily turnover merchants in the mall field increased by 30% year-on-year.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">In today's context of platform segmentation, Kuaishou can still capture new supply-side increments, not only because Kuaishou is currently one of the few platforms that can increase total traffic, but more importantly, Kuaishou has continuously alleviated two core pain points for e-commerce merchants: high marketing costs and operating costs.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">E-commerce merchants want to do marketing, the first thing they need to consider is calculating ROI. However, the biggest problem with traditional traffic tools is that every step of the marketing chain cannot be accurately calculated, ultimately leading to post-marketing ROI effects being far from expectations.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The first challenge is that the cost of marketing materials cannot be effectively controlled. E-commerce marketing is fast-paced and expensive, and if the marketing materials fail to keep up with the traffic trends, the effectiveness of the campaign will be poor.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">But the emergence of AI technology on Kuaishou has solved this problem. For example, the Maglev Creation Platform mentioned at this year's Maglev Conference allows merchants to instantly edit viral videos, smartly generate high-quality scripts, and produce digital human videos, obtaining higher quality promotional materials at a lower cost and faster speed, effectively resolving the contradiction between cost and efficiency.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Another example is Nvwa Digital Human, which supports 24-hour live broadcasts with matrix operation capabilities, truly achieving \"low barriers, high efficiency\" in live marketing.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F8c9a45ca01a74029b8153ca767bf10fa\u002F1b6ee46bf4ca46978949bb1b8f7194e8.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">According to the CEO of Kuaishou during the call, internal calculations show that AI large models are expected to reduce the cost of short video marketing materials by 60-70% or even more.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The second challenge is that the conversion rate of promoted traffic is generally not high, and promoted ads and natural traffic are difficult to interact and match.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Kuaishou has launched the UAX (Universal Auto X, fully automated placement) solution based on AI capabilities. It has four core capabilities: intelligent infrastructure, intelligent control, intelligent creativity, and intelligent targeting, which can effectively assist clients and agents to enhance their capacity to run traffic, stabilize running costs, and improve placement efficiency.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">According to Jiang Peng, the head of Kuaishou's commercial algorithm, the thought chain decision-making learning of the marketing recommendation large model enables the model to gradually decompose and summarize complex problems like humans, infer user purchase intent and product marketing points, and achieve more accurate person-goods matching, resulting in a 10% increase in product orders and a 20% increase in new product consumption.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The UAX fully automated placement solution upgraded from rule-based decision-making to model-based decision-making. In the second quarter of this year, the proportion of UAX marketing spending in the external circulation total spending increased to around 65%.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Optimizing ROI, solving the bottleneck of placement conversion rate, the tendency of merchants to invest in traffic naturally increases, which will also drive budget allocation. This first half of the year, Kuaishou's overall online marketing revenue reached 19.8 billion yuan, an increase of 12.8% year-on-year, is the best proof and result.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The third challenge is that the traditional chain from placement to business implementation is long, and merchants need a lot of infrastructure work to maintain the business conversion brought by traffic. For some small and medium-sized merchants, it is difficult to have enough management experience and labor costs to match the actual business of traffic conversion.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Kuaishou has solved this problem from both technical and policy perspectives. On one hand, as mentioned earlier, from material production, advertising placement, to live broadcasting, intelligent customer service, there are corresponding low-cost AI tools provided to merchants to help them take advantage of traffic-driven business growth.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">On the other hand, Kuaishou has offered real financial subsidies, launching the New Merchant Cold Start Plan for new merchants, focusing on targeted support and growth enhancement for new merchants within 90 days of their first broadcast.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Fan Xiaoluo's wife, a shoe and boot host, started operating Kuaishou in February this year. By May, her account's cumulative GMV had exceeded 10 million. Fan Xiaoluo ensured a daily live broadcast duration of 5-7 hours upon entry, leveraging the 30-day exclusive support for the first broadcast and the 90-day quality new merchant continuous growth to achieve growth.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">At the same time, by utilizing Kuaishou's cross-field traffic resources, through popular clip slicing and short video seeding, inter-field traffic sharing was achieved, with over 30% of transactions coming from short video referrals. Full-site promotion also helped Fan Xiaoluo achieve faster growth, keeping the placement ROI stable around 30.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">In summary, Kuaishou used AI technology to solve three major issues that traditional merchants may face, naturally driving the prosperity of the supply side ecosystem. Combined with the dual effects of demand-side optimization and matching, it ultimately led to the rapid growth of GMV.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cstrong class=\"ql-lineHeight-1-75\" style=\"font-size: 18px; color: rgb(255, 153, 0);\">03 Powerhouse \"XINWEI\": Kuaishou's High-Level Narrative of Value Transformation\u003C\u002Fstrong>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">By this point, we have roughly clarified the four dimensions of AI changing supply, demand, content, and goods revealed in Kuaishou's mid-year report, and it is not hard to understand that Kuaishou's growth is actually no different from the logic shown in Meta and Google's second-quarter financial reports.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">However, Kuaishou's AI narrative goes beyond this. In the context of the global AI competition narrative, its flagship AI application - \"XINWEI AI\" - has a value that is much higher than the current market's general perception, which is also an important variable for Kuaishou's further value transformation.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">First, XINWEI AI is a key force for China to participate in the global AI application competition. China needs a landmark AI application to match global AI frontier applications, and XINWEI AI's capabilities are outstanding.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Currently, in the multimodal field, there are many strong competitors. Domestically, there are XINWEI AI, Ji Meng, Hai Luo, etc., and internationally, there are old-established multimodal products like Runway, Sora, and Stability AI.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">At the beginning of all technological applications, it is usually a small group of technology enthusiasts and tech early adopters who become opinion leaders first. But as the industry continues to develop, more professionals can explore truly valuable technological innovations from more angles and details.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">At this year's WAIC, XINWEI AI's capabilities were widely recognized by professionals. Abela, CEO of the global well-known creative material platform Freepik, said, \"We have integrated more than 10 different video generation large models. The number of videos generated by XINWEI AI chosen by users exceeds the sum of other models.\"\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Clearly, professionals have effectively verified XINWEI AI's technological leadership and user recognition globally, establishing its position as a representative of Chinese AI applications.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Second, XINWEI AI is continuously proving its tremendous value as an advanced productivity.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">The vitality of AI technology lies in whether it can truly transform into an engine that drives industrial upgrading and social progress. We can see that XINWEI AI is vividly illustrating this through extensive commercial application scenarios, such as the advertising material production mentioned earlier, and the scenario of movie industrialization:\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">On June 25th, the world's first AI unit story collection \"Loading the New World\" held its premiere in Beijing. The work was jointly produced by Kuaishou XINWEI AI and Starlight Short Drama, and made by the Outliers team. The total duration was 180 minutes, and this series is the largest known in the industry in terms of subject coverage and narrative volume of AIGC series works, with a global cumulative play count reaching 200 million.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cimg alt=\"undefined\" src=\"https:\u002F\u002Fimage.51xinwei.com\u002F2025\u002F08\u002F4ff8eada9bed41e0995ce80260d144da\u002F7631916303ef4b09bb7bc31d06590818.webp\" width=\"undefined\" height=\"undefined\" style=\"display: block; margin: auto;\">\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Even in the game industry, XINWEI AI has many practical application scenarios. The popular mobile game \"Nihility\" has reached a deep cooperation with XINWEI AI, integrating the industry-leading AI video generation capability into the game, enriching the social gameplay of the game.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Data shows that by July this year, XINWEI AI has produced more than 200 million videos and 400 million images, serving over 20,000 enterprise customers. Cheng Yixiao also mentioned in the call that, \"We hope that as the comprehensive performance of the model improves, XINWEI AI will gradually possess the ability to deeply apply in industrial scenarios such as game production and professional film and television production, expanding more industrial users.\"\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Finally, XINWEI AI, through its good commercial performance, not only enhances its valuation but also shapes a model of a healthy capital expenditure cycle for Chinese AI applications.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">In the second quarter of this year, XINWEI AI's revenue exceeded 250 million yuan. Now, few AI applications are confident enough to disclose their commercial income process. XINWEI AI confidently submitted its commercial results, which not only proves its actual product strength but also demonstrates the potential and value of AI application commercialization to the market.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">According to data from CICC research reports, the access share of the \"XINWEI AI\" series model in the global AI video tool market has reached 30.7%, ranking first in the industry. According to CFO Jin Bing's disclosure at the earnings call, XINWEI AI's revenue for 2025 is expected to double compared to the target at the beginning of this year.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Through XINWEI AI's excellent commercial performance, Kuaishou has also opened the door for capital expenditure upgrades in China's AI field. CFO Jin Bing revealed that the Capex investment related to XINWEI AI has also doubled compared to the initial budget. Considering that XINWEI AI has achieved a positive gross margin in the inference computing power level and maintained stability, the additional computing power investment has a controllable impact on the group's overall profit, and capital expenditures have entered a healthy cycle.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">We have analyzed multiple times that the key point of the current Sino-US AI development competition is capital expenditure. American big companies hope to build and consolidate their own business high walls by significantly increasing AICapex, thus raising the entry barrier for the AI industry.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">Nowadays, our top companies are constrained by low AI capital return rates and lack sufficient confidence to invest more capital expenditure to compensate for the technical gap. However, XINWEI AI has raised ROI through its product strength, not only establishing a virtuous cycle for its own capital investment but also setting a benchmark for the development of Chinese AI applications.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">If AI empowerment of existing businesses driving performance growth is a potential discrepancy in Kuaishou's short-term expectations, then XINWEI AI's performance is undoubtedly a strong stimulant for Kuaishou's long-term development.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cspan style=\"font-size: 18px;\" class=\"ql-lineHeight-1-75\">In summary, through Kuaishou's mid-year report, we clearly see that AI is not merely a distant future scenario. In mature platforms, AI technology has already penetrated every link of connecting users and content, empowering merchants and supply, becoming a key engine driving core business growth.\u003C\u002Fspan>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp>\u003Cbr>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(187, 187, 187);\">[News Source] Blue Whale News | Wen | Jinduan \u003C\u002Fspan>\u003Ca href=\"https:\u002F\u002Fwww.sohu.com\u002Fa\u002F927609704_250147\" rel=\"noopener noreferrer\" target=\"_blank\" style=\"color: rgb(187, 187, 187);\">https:\u002F\u002Fwww.sohu.com\u002Fa\u002F927609704_250147\u003C\u002Fa>\u003C\u002Fp>\u003Cp class=\"ql-align-justify\">\u003Cspan style=\"color: rgb(187, 187, 187);\">（This article is reprinted by the website to provide readers with more information and news. The content does not constitute investment or consumer advice. If there are any questions about the facts of the article, please verify with the relevant parties. The views expressed in the article are not the views of the website and are for reference only.）\u003C\u002Fspan>\u003C\u002Fp>","https:\u002F\u002Fxinwei-dev-test.oss-cn-shenzhen.aliyuncs.com\u002Fintelligent\u002Faudio%3A4bc0883b-62b2-4a3c-8da2-2c2970303fa6%3A0.wav?Expires=1774838485&OSSAccessKeyId=LTAI5tNvY2RkKjZw4LLWsrPK&Signature=OAl1zwGw5YX%2FDmBZL%2F2eoaqUix4%3D","4bc0883b-62b2-4a3c-8da2-2c2970303fa6",18139558]