初创公司尽调
尽调报告 AI / application software Series A+ 2026-07-09

Tripo AI

快速增长的中国 3D 生成式 AI 平台,产品纵深和用户规模都已成形,但在据报 $1.5B 的估值之下,公开经济性仍然偏薄。

Tripo 在 AI 3D 创作的品类领先和可信采用值得跟踪,但据报道 $1.5B 估值仍跑在公开收入披露之前,还要打真实的中国风险折扣。

封面要素

最近一轮融资 01
$200M Series A+ / A++ (Jun 2026) [CV001]
估值 02
1500 USD M [CV004]
用户 03
10M+ registered / claimed [CU018, CV005]
企业客户 04
90000 studio clients / claimed [CU018, CV005]

公司概况

Tripo AI 是中国 AI 创业公司 VAST 的旗舰产品,也是当下能见度最高的 AI 原生快速 3D 资产生成平台之一。公司由 Simon Song 于 2023 年创立;他曾在 SenseTime 任职,也是 MiniMax 联合创始人。Tripo 已从文字 / 图像生成 3D,扩展到 API、工作流、商业、媒体以及接近世界模型的产品界面。公开报道把 VAST 与 2026 年 3 月 Alibaba 领投的一轮融资,以及规模大得多的 2026 年 6 月 A+ / A++ 融资联系起来;后者让公司进入独角兽区间,Forbes 引用一位知情人士称估值约 $1.5B。公开采用信号很强,包括 10M 用户和 90K 工作室客户,但这些口径背后的经济性仍未公开。

官网
tripo3d.ai
成立时间
2023-01-01
创始人
Simon Song
创立地点
Beijing, China
总部
Beijing, China
产品
Tripo3D.ai 平台覆盖文字生成 3D 和图像生成 3D 资产;同时提供开发者 API、面向游戏、电商和媒体制作的工作流工具,以及 Project Eden——公司在 2026 年 6 月融资同时推出的世界模型研究计划。
客户
全球游戏开发者、XR 和媒体团队、电商及目录运营方、工业或创意设计师,以及个人创作者;中国以外使用量可观,但公开收入构成仍未披露。
商业模式
免费增值自助订阅,付费后获得商业使用权;更高并发的创作者或工作室层级;另有开发者 API 和企业工作流变现。
阶段
Series A+
融资情况
公开报道支持 Alibaba 领投的 2026 年 3 月 $50M 融资,以及约 $200M 的 2026 年 6 月 Series A+ / A++ 融资;后者把 VAST / Tripo 推入独角兽区间,Forbes 引用知情人士称估值约 $1.5B。
[CO004, CO006, CO008, CO010, CO011, CO012, CU018, CV001]

执行摘要

主要优势

  • Tripo 的产品宽度真实,覆盖创作者 UI、付费工作流工具、API 和伙伴集成,不只是单一 demo 功能。
  • 对一家年轻 3D AI 公司来说,公开采用信号异常强:到 mid-2026,公司声称已有 10M 用户和 90K studio 客户。
  • Runway 等私有 creative-AI 可比公司表明,只要稀缺性和动能真实,高倍数可以成立。

主要风险

  • 公开来源仍未披露 ARR、留存、毛利率或付费转化,$1.5B 估值很难承保。
  • 与中国相关的出口管制、算力获取和生成式 AI 合规风险,即使需求保持强劲,也可能压缩估值支撑。
  • 独立评测仍把 Tripo 描述为快速工作流加速器,可能还需要清理,而非通用的生产级替代品。

未决问题

  • 按套餐或 cohort 验证的 ARR、确认收入、毛利率和免费到付费转化。
  • 企业 logo 质量、ACV 区间、续约行为和具名客户集中度。
  • 股权结构表机制、优先权条款,以及可能让 headline valuation 偏离普通股价值的任何结构。

目录

Chapter 01

01公司概览

1.1 身份、结构、产品与阶段

更准确地说,Tripo AI 是更大 VAST/Holymolly 公司栈里的产品品牌,而不是一家单一且披露清楚的运营公司。官方条款把 Holymolly Ltd 列为 tripo3d.ai 背后的签约实体,并给出香港办公室地址;South China Morning Post 则称,公司以 Vast 名义注册于 Cayman Islands,并于 2023 年在北京设立。CB Insights 的口径又不同,列出 Shanghai Pilot Free Trade Zone 总部。合在一起看,公开记录支持一个务实尽调结论:Tripo AI 是面向客户的品牌,VAST 是融资报道里的母体身份;集团大概率采用多司法辖区结构,这在与中国有关的风投支持公司里很常见。 产品主张本身在年轻 AI 公司里相当具体。官网、定价页、API 和多篇产品博客都把 Tripo 定位成端到端 3D 创作工作流:用文字提示、图像、多视角输入和草图,在数秒内生成更接近生产可用的资产。官方材料反复强调的不只是生成,还包括分割、绑定、动画、格式转换、低多边形优化、导出到标准 3D 格式等下游资产处理能力。这一点重要,因为 Tripo 推销的不是新奇图像生成器,而是为创作者、开发者和工作室提供的工作流替代或加速层,交付的是实际网格输出。 截至报告日,这家公司已不再是早期概念验证。Forbes、SCMP/Yahoo 和公司材料都把 Tripo 放在产品市场契合之后、高增长私营公司的类别里:公司已越过消费者演示阶段,同时销售订阅和项目制企业服务,并开始把自己从 3D 资产生成器重新叙述为更广的空间智能与世界模型平台。最需要警惕的是,成熟叙事在运营层面更可见,在机构层面仍不够可见:公开产品界面很丰富,但公司栈和治理界面滞后。[CO001, CO002, CO003, CO004, CO005, CO006]

快照 KPI 表
指标数值 / 状态日期 / 版本期置信度尽调缺口
成立20232023核实各法律实体的准确注册和上线日期
母公司 / 运营名称Tripo AI 产品品牌,条款主体为 VAST / Holymolly Ltd2025-2026从已签署文件梳理产品品牌、母公司和签约主体
公司足迹开曼注册、北京研发运营、香港办公地址、上海数据库记录2025-2026核对注册实体、董事会席位和受益所有权
最新披露融资$200M Series A+/A++,发生在 2026 年 3 月 $50M Series A 之后2026-06取得已签署融资文件和股权表
最新估值锚知情人士报道的投后估值约 $1.5B2026-06用已签署融资文件验证,而不是依赖二手报道
用户规模披露公开来源同时出现 6.5M、10M 和 20M 用户数字2025-2026要求提供经审计 MAU、付费用户和企业账户定义
具名企业客户Tencent、NetEase、Sony、Microsoft、Pop Mart 等被公开引用2025-2026确认哪些是付费生产客户,哪些是试点或合作伙伴
公开定价官网提供 Free、Pro、Max 和 Team 自助订阅2026-07-09需要企业定价表和 API 定价附件
变现模式月度订阅,加项目制企业服务和开发者 / API 分发2025-2026要求按自助、API 和企业服务拆分收入结构
财务披露深度收入、ARR、利润率、烧钱速度、跑道和员工数,在高声誉公开来源中仍披露不足2026-07-09判断基本面之前需要管理层材料包

本表刻意区分稳健的公开锚点和尚未解决的公司画像字段;冲突的规模数字按冲突展示,而不是被调和成估算。

[CO002, CO003, CO004, CO005, CO006, CO007]
FO002: 公司快照逻辑

Tripo 的公开逻辑把多司法辖区公司架构、3D 创作工作台、生态插件、企业客户和世界模型野心串在一起。

[CO001, CO002, CO012, CO020, CO021, CO022]
FO003: 快照 KPI

速读公司卡片区分坚实公开锚点和仍过于不透明、无法干净做投资判断的领域。

KPI 卡片有意混合数字锚点和状态标记,让读者区分扎实事实与未解决的尽调领域。

[CO004, CO024, CO030, CO031, CO034, CO041]

1.2 创始人、领导层与治理

创始人记录是本案最清晰的正面信号之一。多家独立来源都确认 Simon Song 是创始人兼 CEO;他的创始人与市场匹配既来自领域审美,也来自此前运营 AI 公司的经验。Forbes 和 SCMP 称,Song 曾就读于 Johns Hopkins,任职于 SenseTime,随后联合创办 MiniMax,并在 2022 年离开去创建 VAST。这些细节重要,因为 Tripo 位于生成式 AI、游戏、动画和工具的交叉点,而 Song 自己的公开叙事也明确锚定在玩家身份:他想让 3D 创作更快。 看不见的部分几乎同样重要。公开材料高度围绕创始人展开:Song 是融资报道里的发声者、被点名的技术叙事者,也是显眼的人才磁铁。但已抓取资料没有给出扎实的公开董事会名单、具名 CFO,或清晰独立的治理层。即便公司面向用户的文件在产品权利和用户义务上写得很细,也没有实质披露控制权、委员会结构或更宽的高管梯队。这对一家 2023 年创立的私营公司并不少见,但当估值进入独角兽区间,它就是实打实的尽调缺口。 因此,治理判断是混合,而非负面。创始人集中度可能加快产品速度,也帮助 Tripo 下重注:发布开源模型、投入世界模型、扩展工作流插件。代价也很清楚:关键人物依赖明显,外部很难看见谁在内部挑战创始人、谁负责财务纪律,以及香港、Cayman、北京和可能与上海有关的注册主体之间如何治理。[CO008, CO009, CO010, CO011, CO042]

领导层与创始人表
人员 / 职务公开支持的背景当前公开角色创始人-市场匹配 / 依赖度披露限制
Simon Song / 创始人兼 CEOJohns Hopkins 校友;曾在 SenseTime 任职;MiniMax 联合创始人核心创始人、CEO,以及产品和融资对外发声者在 AI、游戏和生成式工具上创始人-市场匹配强,但关键人集中度高更广执行团队和董事会结构未清晰公开
财务 / 治理班底已抓取公开资料中未找到具名 CFO、董事会委员会和独立董事已审阅材料未披露尚不清楚治理成熟度是否匹配独角兽估值跃迁需要董事会名单、章程文件和组织架构图
研究组织SCMP 提及北京研究中心;arXiv/GitHub 可见开源模型工作技术能力通过模型发布间接可见暗示创始人早期能招到可信研究人才需要研究、基础设施和商业职能负责人名单
商业化领导层SCMP 访谈称彼时只有两名销售,增长主要来自自然传播当时动作更偏服务驱动,而非大规模 BD 驱动暗示产品驱动分发强,但商业执行也集中需要当前销售、CS 和企业客户组织细节

本表只部分列举公开可见的领导层和治理表面;缺少公开董事会与高管名单,本身就是一个尽调信号。

[CO008, CO009, CO010, CO011, CO029, CO042]

1.3 融资、规模与商业化

公开融资动能真实,而且压缩得不同寻常。Forbes 报道称,2026 年 3 月 Alibaba 和 Hengxu 领投了 $50 million Series A;约三个月后,INCE Capital、Genesis Capital 和 Primavera 等组成的财团又参与了约 $200 million 的 2026 年 6 月 Series A+/A++ 轮。AI Weekly 也提到 $200 million 融资,并把一家 China Life 关联基金加入轮次叙事。这些来源共同支撑核心结论:即使确切投后估值仍非官方、且依赖不同来源,VAST/Tripo 也在很短时间内从有前景的产品公司变成中国 AI 独角兽。 商业牵引也方向性强,但具体规模数字必须谨慎看待。SCMP/Yahoo 引述 6.5 million 用户,其中超过 85% 在中国以外,并列出 Tencent、NetEase、Sony、Microsoft 和 Pop Mart 等企业用户。Forbes 2025 和 2026 年文章分别提到超过 3 million 专业用户、20 million 全球用户;AI Weekly 则称有 10 million 个人用户和 90,000 家工作室客户。这些口径彼此差距太大,不能当作单一经审计 KPI 承销,但方向一致:Tripo 不是只有几个试点 logo 的小众玩具。它已经形成可观的全球分发,尤其在游戏、创意以及接近工业设计的工作流里。 变现模型比财务披露更清楚。官方定价显示,公司提供从免费到团队层级的自助订阅;Forbes 称,NetEase、Sony 等企业按项目收费,终端用户按月订阅,价格大致从 $20 到 $140。消费者 / 专业消费者订阅、项目制企业变现、开发者 API 组合在一起,契合公司既做创作软件、也做 3D 基础设施的定位。仍然缺的是每个投资人最想先看的数字:经审计收入、ARR、毛利率和现金消耗。[CO024, CO025, CO026, CO027, CO028, CO029]

利益相关方 / 投资者图谱
利益相关方角色 / 关系重要性公开信号尽调问题
Alibaba Group2026 年 3 月 Series A 领投方;后续财团名单中也被点名参与带来中国互联网云分发和验证Forbes 和 Auganix 都把 Alibaba 与 3 月轮融资联系起来确认持股比例、商业合作和云承诺
INCE Capital2026 年 6 月 Series A+/A++ 领投方独角兽跃迁的定价投资人Forbes、AI Weekly 和 InforCapital 都引用 INCE 参与要求提供估值备忘录、治理条款和清算优先权
Genesis Capital2026 年 6 月财团参与方支撑中国机构 VC 需求深度Forbes 和 InforCapital 点名确认持股、按比例跟投权和董事会观察员身份
Primavera Capital Group2026 年 6 月财团参与方显示游戏垂直投资人之外还有更大资本支持Forbes 和 InforCapital 点名澄清参与是战略性质还是纯财务性质
China Life 关联方 / China Life Science & Technology Innovation Fund视来源不同,2026 年融资中为联合领投或参与方增加准机构资本和政策贴近度AI Weekly 和 InforCapital 提到 China Life 关联参与方核实准确基金名称,并检查是否有国家资本背景治理条款
NetEase具名企业客户游戏生产客户证明是投资论点核心Forbes 和 SCMP/Yahoo 都提到 NetEase确认合同规模、续约节奏和生产部署深度
Sony具名客户和官方硬件 / 显示合作伙伴支撑企业可信度和空间计算叙事Sony 合作伙伴页面以及 Forbes/AI Weekly 提到 Sony区分 PR 合作价值和收入贡献
Stability AI开源和生态合作方提高研究可信度和西方生态相邻性TripoSR GitHub 和 AccessNewswire 提到合作澄清技术和商业关系的深度

这是一张部分公开利益相关方图谱,覆盖会实质影响公司概览尽调视角的具名资本提供方、客户和生态伙伴。

[CO024, CO028, CO030, CO031, CO032, CO033]
FO001: 公司里程碑时间线

公开时间线显示,Tripo 2023 创立,2025 产品打磨,2026 融资快速升级并转向世界模型野心。

部分日期从抓取到的公开记录中只能支撑到月份或年份精度。

[CO004, CO005, CO024, CO025, CO026, CO030]

1.4 里程碑、技术信号与不利因素

里程碑记录显示,平台扩张速度异常快。官方和半官方来源把公司从 2023 年创立,追踪到 2025 年产品强化,再到 2026 年资本加速。到 2025 年末,Tripo 已经越过原始提示词到模型生成,进入纹理引擎、分割、绑定、格式转换、ComfyUI 节点、Blender 工作流,以及 Sony Spatial Reality Display 合作等企业合作叙事。研究界面也有实质内容:TripoSR、TripoSG 和 GitHub / 开源足迹,让公司相对许多只有产品、缺少公开研究产物的 AI 应用公司,拥有更强技术信号。 同时,证据集还不足以让人盲目接受每一句管线就绪声明。本章最有用的不利来源是独立 Medium 拆解,它提出了一个公平观点:3D AI 演示在导出前可以很漂亮,但真实生产价值取决于 mesh 是否干净、UV 是否可迁移、绑定是否稳健,以及在真实下游工具里的清理时间。这个批评也呼应 Tripo 自己反复强调的分割、绑定、低多边形优化和导出处理——只有当下游摩擦真实存在时,公司才会重压这些功能。换句话说,不利情形并不是产品造假;而是真实生产就绪度是一条光谱,不是二元标签。 因此,战略弧线可信,但尚未充分去风险。VAST 显然想超越单一资产生成应用:Project Eden 和 W1.0 表明公司有意向世界模型与空间仿真上爬;Sony 以及工作流插件合作则显示出真正的商业化落地想象力。尽调的开放问题是,产品和资本形成的速度,是否已经由企业控制、治理成熟度和独立测量的生产性能跟上。[CO020, CO021, CO022, CO023, CO035, CO036]

里程碑表
日期事件类型金额 / 估值 / 状态参与方含义
2023VAST / Tripo 成立创立公司成立Simon Song后续所有融资和产品里程碑的创立锚点
2023SCMP 出现北京设点与开曼注册叙事治理多司法辖区结构VAST / Tripo显示中国运营足迹与离岸融资结构并存
2024Musk 转发和全球创作者兴趣之后,早期用户增长突破被提及规模用户认知加速全球创作者暗示自然采用和创作者驱动分发
2025-09Forbes 将 Tripo 描述为拥有 3M+ 专业用户的 3D 基础模型公司规模>3M 用户 / >40k 合作伙伴Forbes、Simon Song独立报道开始把 Tripo 描述为不只是 demo 工具
2025-09AccessNewswire 发布 Tripo 3.0 上线消息产品20B 参数 / 300% 细节提升说法Tripo / Stability 生态公开产品叙事转向管线就绪
2025-12SCMP/Yahoo 报道 6.5M 用户和超过 85% 中国以外用户占比规模6.5M 用户SCMP / Simon Song证实强劲国际采用
2026-03Series A 融资宣布融资$50M多个来源引用 Alibaba、Hengxu、Baidu Ventures资本支持模型和开发者平台扩张
2026-03H3.1、P1.0 和 W1.0 模型家族叙事公开产品新模型架构家族Tripo显示向原生 3D 生成和世界模型推进
2026-06Series A+ / A++ 融资宣布融资约 $200M / 报道估值约 $1.5BINCE、Genesis、Primavera、China Life 关联基金等Tripo 达到独角兽规模,并获得招聘和 R&D 资金
2026-06Project Eden 世界模型计划随融资报道公开产品空间 / 世界模型扩张VAST显示野心从资产生成延伸到模拟环境
2025-2026Sony 商业合作和 WAIC / ChinaJoy demo合作显示 + 生成生态Sony、VAST强化企业和空间计算商业化叙事
2026-05独立拆解提醒,导出资产仍需真实管线测试反向工作流警示独立评测者提醒投资人,生产就绪主张尚未完全得到独立基准验证

这条时间线是公司概览章节的单一记录,保留了看多里程碑和警示信号,而不是把它们压平成单一进展叙事。

[CO004, CO005, CO024, CO025, CO026, CO030]

1.5 展品

Chapter 02

02市场分析

2.1 市场边界与规模测算视角

分析 Tripo 的第一条纪律,是把市场定义窄到有用。Tripo 不是面向整个中国 AI 市场销售,甚至也不是面向整个全球创作者软件市场。它所在的是 AI 原生 3D 资产创作工作流层:文字 / 图像 / 草图生成 3D、纹理 / 材质生成、网格清理与重拓扑、绑定、导出、API 驱动接入,以及从创作者到引擎的工作流工具。因此,它的真实市场边界更接近「把创意意图转换成生产可用 3D 资产的软件和服务」,而不是泛泛的生成式 AI 支出。 多个公开规模测算视角都有用,但各自回答的问题不同。Axis 和 Grand View 给出宽口径中国 AI 市场收入区间,Research and Markets 描述具体 3D 资产类别结构,China Daily/IDC 提供沉浸式计算支出视角,Digital in Asia 则用中国电商和游戏提供消费者需求背景。不能把这些数字硬塞进一个假的 TAM/SAM/SOM 栈。更合适的做法是建立分层边界:最上层是巨大的中国数字化和 AI 预算;下面是更小的沉浸式、游戏 / 电商内容层;再往下才是 3D 资产生成工作流可变现的窄楔子。 这种分层很重要,因为它避免了创业公司常见错误:用一张巨大 AI 市场幻灯片支撑估值。Tripo 的机会真实存在,但它不是完整的 $62 billion IDC 式中国 AI 市场,也不是完整的 $2.93 trillion 中国电商市场。可用机会只占这些预算的一小部分:这些预算确实需要反复创作 3D 资产、快速迭代、标准导出,并具备足够下游信任,让团队敢把 AI 生成资产放进真实生产系统。[CM001, CM002, CM003, CM004, CM005, CM006]

市场定义表
分部 / 类别纳入支出排除支出买方 / 付款方契合度
AI 原生 3D 资产生成软件文本 / 图像 / 草图转 3D、网格生成、纹理、重拓扑、绑定、导出、工作流插件通用 LLM 席位、2D 图像生成、非 3D 创作者工具创作者、工作室、工程团队、设计团队与 Tripo 产品功能直接对齐的核心市场
游戏与动画管线原型道具、环境资产、角色、引擎就绪网格、API 自动化完整游戏预算、发行支出、广告 UA美术总监、技术美术、制作负责人高度契合,因为资产吞吐和引擎兼容性很重要
XR / AR / 空间计算实时 3D 资产、低多边形优化、交互内容、显示管线仅头显硬件销售、一般 AR 广告支出XR 构建者、空间内容团队、硬件伙伴需要交互式 3D 和低摩擦迭代的场景契合度高
电商产品可视化3D 商品页、WebAR 资产、目录转化、数字原型整体零售 GMV、支付收入、物流支出商品运营、增长、数字商务、AR 体验负责人有吸引力,因为转化和退货率经济性可衡量
室内 / 工业设计概念模型、房间规划器、产品可视化、数字样机完整 CAD/PLM 栈替换、制造资本开支独立设计师、产品团队、工业设计负责人有用切入点,但对精度和信任要求更高
教育 / 爱好 / 3D 打印学生项目、创客资产、可打印模型、创作者实验全机构 LMS 或 edtech 预算个人、教师、爱好者漏斗顶部好,但客单价较低,企业锁定较弱
现状替代方案手工 Blender/Maya/CAD、摄影测量、服务工作室、素材库N/A现有美术或设计预算关键在于 Tripo 必须替代劳动力和工作流摩擦,而不只是增加新奇感

本市场围绕把创意意图转化为可用 3D 资产的工作流层定义,而不是围绕总体 AI 或总体数字经济支出。

[CM001, CM002, CM003, CM004, CM018, CM021]
TAM / SAM / SOM 或市场规模测算视角表
发布方 / 视角年份地域数值CAGR / 增长方法 / 边界置信度局限
Axis / Fortune Business Insights 保守 AI 收入口径2024-2025中国$21.6B 2024;$28.2B 2025 预测CAGR 32.5% 至 2032 年统计较窄的 AI 收入,而不是更宽的 AI 赋能经济活动对 Tripo 本身过宽,但可作宏观 AI 上限
Axis / Grand View AI 收入口径2025中国$31.6Bn/a更宽的 AI 软件和服务收入口径仍远宽于 AI 3D 创作本身
Axis / IDC 广义 AI 活动口径2025中国$62Bn/a覆盖基础设施和 AI 赋能活动在内的宽口径 AI 相关市场不能与窄口径软件 TAM 直接相比,否则会高估 Tripo 机会
Grand View Research 中国生成式 AI 展望数据手册2024-2030中国2024 年软件占比 63.9%预测至 2030 年中国生成式 AI 内部的细分结构份额数据有助于搭建市场结构,但无法单独拆出 3D
Research and Markets 生成式 AI 3D 资产视角2020-2035全球和国家拆分覆盖软件 / 硬件 / 服务及终端用户类别历史 + 预测生成式 3D 资产的具体类别框架已抓取预览比精确总量值更能说明范围和分层
China Daily 转引 IDC 的 AR/VR 支出视角2022-2026中国2026 年达到 $13.1B43.8% CAGR沉浸式技术支出,包含游戏和 VR 偏重用例沉浸式支出只是一个相邻市场,不是 Tripo 的完整市场
Digital in Asia 电商视角2025-2026中国2025 年电商 $2.93T;2026 年直播电商 8.16T 元电商年增长 6.4%数字商业和直播电商支出池大多数商业支出并不需要定制 3D 资产
Digital in Asia 游戏视角2025中国游戏收入 350.8B 元 / $49.8B同比 7.7%用消费者游戏收入代理内容预算深度游戏收入是下游需求,不是直接工具支出

这些是分层的规模测算视角,不是可相加的 TAM/SAM/SOM 模块;表格有意保留口径差异, 而不是强行制造虚假的可比性。

[CM005, CM006, CM007, CM008, CM009, CM010]
FM001: 市场规模测算视角

Tripo 的相关支出池从巨大的中国数字经济大盘,收窄到更小的沉浸式和特定工作流 3D 创作切入点。

这是邻近性阶梯,不是干净相加的 TAM/SAM/SOM 模型;每一层都是更宽或更窄的支出池,Tripo 只能部分触达,而非全部拿下。

[CM011, CM014, CM015, CM016, CM026, CM033]
FM002: 市场估计区间

公开中国 AI 市场估计差异很大,取决于边界是保守软件收入,还是更宽泛的 AI 相关活动。

差距反映真实的边界差异,而不是测量误差;因此这一区间应理解为相互冲突的市场定义,不是简单置信区间。

[CM005, CM006, CM036]

2.2 买方、用户与付费方

买方地图是 Tripo 最有吸引力的结构特征之一。产品可以从个人、设计团队或技术组织切入,不必等待单一大型采购方。官方定价用 credits 和订阅支持个人创作者和小团队;官方 API 页面和工作流页面则明确拆出开发者 / API 产品线,服务用量更高、集成更重的用户。这让 Tripo 能覆盖很不同的待完成任务:原型化一个道具,为 WebAR 生成家具资产,快速生成房间布局,把资产接入 Unity 或 Unreal,或为工作室管线生成定制 3D 元素。 实际细分比确切收入拆分更清楚。游戏工作室和 XR 团队看重低多边形网格、绑定、导出纪律和引擎桥接。电商和零售团队看重 GLB/USDZ 部署、商品页转化提升、AR 摆放和退货率下降。室内与工业设计师更看重快速迭代和标准文件导出,而不是大规模 API 自动化。独立创作者和爱好者看重速度、credit 包和浏览器易用性。每个细分的付费方都不同:创作者用信用卡自助购买,工作室动用美术或制作预算,企业 API 买方需要工程或产品基础设施批准,商业团队往往归在商品、增长或数字体验预算之下。 这种多样性利好需求,却不利于讲一个简单故事。公司可以展示大量用户,却尚未证明深度企业变现;也可以展示强企业 logo,却尚未证明跨细分有标准化、可重复预算。Tripo 的市场在资产吞吐反复发生、文件导出重要、节省人力显而易见的地方最强——尤其是游戏管线、XR / 空间内容和商业可视化。偶发概念图或一次性新奇资产已经足够的地方则更弱,因为这些买方可以继续做免费用户,或低成本在工具之间切换。[CM018, CM019, CM020, CM021, CM022, CM023]

细分市场 / 买方图谱
细分市场买方用户付款方 / 预算负责人工作流采用触发因素
独立 / 中端游戏工作室技术美术或制作负责人美术师、技术美术、关卡设计师美术 / 制作预算制作原型资产并导出到 Unity / Unreal人工做资产太慢,跟不上快速迭代
AAA 或大型长线运营工作室管线工程或内容工具团队大型内部内容团队平台 / 工具 / 中央工程预算API 驱动生成,加内部导入管线需要可重复的资产吞吐和工具杠杆
XR / 空间计算团队产品或空间内容负责人3D 创作者、引擎开发者XR 产品 / 创新预算面向实时场景的低多边形或交互式资产需要比人工团队更快的交互式 3D 创作
电商商家和零售商商品运营或数字体验经理目录、内容和增长团队数字商业 / 商品运营预算3D 商品页和 WebAR 资产需要用更好的可视化提高转化、降低退货
独立室内 / 产品设计师独立专业人士或工作室所有者设计师个人 / 项目预算基于浏览器创建房间和资产,并支持标准导出需要不背沉重 CAD 负担,也能交付给客户看的概念方案
工业设计 / 机器人仿真团队设计负责人或创新经理设计师、仿真团队设计 / 研发预算高细节概念迭代和仿真资产需要更快的概念循环,但仍要求几何可靠
开发者平台客户CTO、PM 或平台工程负责人开发者和管线工程师工程 / 产品基础设施预算按 API 合约程序化生成资产需要可扩展的自动化资产生成,而不是人工工作室席位
创作者 / 爱好者 / 教育个人创作者或教师个人终端用户个人卡或课堂预算基于 Web 的生成和导出需要低门槛上手 3D 创作

同一产品家族横跨自助、团队和 API 动线,因此单看用户数看不出变现质量; 不同细分市场的买方和付款方差异很大。

[CM018, CM019, CM020, CM021, CM022, CM023]
FM003: 买方 / 细分市场地图

Tripo 最好的买方细分,是那些有经常性 3D 产出需求、预算归属清晰且工作流痛点强的群体。

[CM020, CM021, CM022, CM027, CM028, CM029]
FM004: 采用漏斗或价值链地图

多数市场会从广泛的手工 3D 痛点,逐步收窄到足够信任 AI 生成资产、愿意为工作流集成付费的群体。

阶段值是用于展示漏斗收窄的序数索引,不是实测转化率。抓取到的来源没有披露 AI 3D 采用的完整市场漏斗。

[CM023, CM024, CM029, CM030, CM034, CM037]

2.3 增长驱动因素与采用约束

最强增长驱动来自 Tripo 想替代的难看基线。传统 3D 工作流很慢,重度依赖专业经验,还被建模、UV、贴图、重拓扑、绑定和导出切得很碎。Tripo 每个公开工作流页面都在抓同一个市场事实:团队愿意为更短的从概念到资产周期、更少的一版几何美术工时,以及更容易进入引擎、查看器或 AR 店面而付费。当游戏管线、XR 内容、数字商业和工业可视化都要求比人工团队更高的资产量时,这种需求拉力会变得更强。 第二个驱动是工作流集成。即使市场预算很大,AI 点解决方案的输出进不了真实下游系统,也很难转化。Tripo 围绕 ComfyUI、Blender、Unity、Unreal、REST APIs、标准导出和 DCC 桥接的官方信息之所以重要,正是因为工作流深度常常决定一个 pilot 能否变成预算项。用市场语言说,互操作性降低切换成本,也让无法忍受孤儿资产或每次交接都手工返工的团队更容易服务。 最大约束同样清楚。第一,输出质量仍需要人工验证:官方页面强调分割、低多边形和导出处理,正是因为下游摩擦仍真实存在;第一章的不利来源说明了这点为何重要。第二,公开市场规模估算会随着边界是窄口径软件收入还是宽口径 AI 关联经济活动而大幅波动,这削弱了巨型 TAM 幻灯片的用处。第三,中国生成式 AI 监管和标识规则,会给面向中国公众用户或跨境数据流的服务带来反复合规工作。因此,市场有吸引力,但只奖励那些把原始生成能力、工作流可靠性、合规纪律和足够产品深度组合起来、能跨过专业信任门槛的供应商。[CM021, CM022, CM023, CM024, CM029, CM030]

增长驱动因素与约束表
驱动因素 / 约束方向时点含义尽调追问
人工 3D 工作流的成本和速度痛点驱动当前一稿生成、重拓扑和导出自动化能讲清 ROI向管理层索取客户实测省时或清理时间基准
中国游戏市场规模和内容深度驱动当前支撑道具、环境、角色和迭代工作流的重复需求要求披露游戏和娱乐账户的分部收入或用户结构
中国电商和直播电商规模驱动当前至中期为 3D 可视化和 AR 商品陈列打开庞大的相邻预算池索取真实商家的转化率 / 退货率案例,不要示意性样例
AR/VR / 空间计算扩张驱动当前至中期支撑交互式 3D、显示设备合作伙伴和实时资产需求索取 XR 客户集中度和平台伙伴路线图
标准导出、插件与 API 解耦驱动当前降低从试用切到生产使用的成本检查付费客户群对 API、DCC 桥接和导出格式的真实使用
中国庞大的 AI / 云 / 5G 基础设施底座驱动当前相比数字生态更薄的市场,更容易做分发和算力密集实验确认模型推理经济性是否受益于国内云厂商关系
输出质量与清理风险约束当前减慢从演示热情转向规模化付费部署的速度拿真实引擎、电商和设计管线跑基准测试
市场定义分散约束当前如果把宽口径 AI 数字直接当 TAM,估值材料很容易讲过头要求管理层按用例展示可服务市场,而不只是自上而下的 AI 图表
生成式 AI 的数据、安全和算法规则约束当前面向中国公众用户或处理可编程接口时,会增加合规工作审查数据来源、日志、标识和备案义务的合规归属
AI 生成内容标识规则约束当前至中期生成内容被下载、发布或再分发时,产品和导出义务会上升询问 Tripo 如何在 API 和导出流程里处理元数据、水印和标识

核心市场问题不是需求是否存在;而是 Tripo 能否把需求转成重复付费工作流, 同时越过质量和合规门槛。

[CM011, CM012, CM015, CM016, CM017, CM021]

2.4 中国语境与 SAM 含义

中国对 Tripo 有两重意义:一是基础设施和政策背景,二是部分但不完整的需求上限。中国在 AI、云、5G、游戏、电商和数字支付上的规模,为 3D 内容工具提供了肥沃土壤。但第一章证据显示,超过 85% 的用户在中国以外;纯中国 TAM 叙事会低估 Tripo 需求已经有多全球化。更好的框架是:中国生态给 Tripo 便宜实验、人才密度和产业政策支持,而商业机会在全球范围内存在,前提是当地有重复 3D 内容创作需求。 这引出 SAM 含义。可触达市场不是单一地理区域,也不是一个采购渠道。Tripo 可服务的市场,是中国内外同时需要 3D 资产吞吐、又能容忍 AI 生成资产进入真实生产工作流的团队集合。游戏、XR、电商可视化、室内 / 产品设计,以及部分工业仿真用途都符合条件,只是置信度和付费意愿不同。公司的创作者基础可能是全球的,但更高置信度的变现口袋,大概率集中在手工 3D 工作痛点频繁、可测量且昂贵的场景。 未解决的尽调问题在于,没有公开来源把这个可服务市场干净切出来。投资人可以用相邻支出池和工作流经济学去夹逼估算,但仍需要管理层展示用户细分构成、各用例转化率、企业 ACV,以及公司到底在哪里真正替代了既有人力或软件支出。在此之前,市场故事最好被视为大且上行——但只有一部分已经转化为可证明的可服务收入基础。[CM014, CM015, CM016, CM017, CM026, CM027]

2.5 展品

Chapter 03

03竞争格局

3.1 竞争版图与买方选择集

Tripo 同时在不止一条赛道竞争,所以简单同行组会误导。买方如果想要生产可用的 3D 网格、导出、低多边形优化以及从浏览器到管线的工作流,相关替代品是 Meshy 这类直接 AI-3D 产品,以及围绕 Stability 发布版搭建的开放模型栈。买方如果主要想要营销活动内容、分镜动态或创意概念生成,Adobe Firefly、Luma 和 OpenAI/Azure 即便不交付同一套 3D 网格工作流,也能拿走预算。对很多团队来说,默认替代品仍然不是另一家创业公司,而是手工 Blender/Maya/Substance 工作、外包工作室、摄影测量,或由 API 和插件拼起来的内部工具链。 因此,公开材料指向一个分层竞争地图。Tripo 和 Meshy 是最清晰的自助、web 优先、AI 原生 3D 工作台替代品:两者都营销速度、民主化、导出和更宽的工作流工具,而不只是实验室演示。Stability 和 NVIDIA 重要,是因为它们可以把高质量 3D 生成推到开放层或平台放大的层面,从而削弱纯算法护城河。Adobe 重要,是因为它已经掌握大量创意预算,并把多个合作伙伴模型、商业安全定位和 Content Credentials 打包在一起。Luma 和 OpenAI 重要,是因为买方注意力正漂向多模态和世界模型叙事;即使它们不能逐资产替代 Tripo,也能重定向企业实验预算和投资人框架。 战略含义是,Tripo 不能只问今天谁的网格最好来评估竞争。它必须同时赢三场比较:相对 Meshy 的直接打包 3D 工作流,相对 Stability 式开放生态的模型 / 平台经济性,以及相对创意套件既有平台的信任加分发。这让市场比狭义创业公司对创业公司比较更危险;但也意味着 Tripo 不需要在所有地方击败 Adobe 或 OpenAI,它只需要在生产就绪 3D 资产生成确实是待完成任务的场景里赢得清楚。[CP001, CP002, CP003, CP004, CP008, CP011]

竞争对手画像表
竞争对手类别规模 / 融资信号目标细分市场差异化局限
Tripo AIAI 原生 3D 工作台公开披露 1000 万+ 用户 / 9 万+ 企业或工作室客户;前章提及近期 $200M 融资创作者、游戏 / XR 团队、电商、设计、API 买方端到端 3D 原生工作流、创作者品牌、空间智能叙事留存和企业付费深度的公开竞争证据仍偏薄
MeshyAI 原生 3D 工作台自称 $15M ARR、30% MoM 增长、600 万用户、85% 毛利率创作者、游戏开发者、3D 打印、团队、企业定价透明、插件 / API 叙事强、企业控制、2026 年快速发版主张主要来自公司自有材料;独立验证有限
Stability AI开放 / API 3D 模型平台企业平台叠加开源仓库和开发者 API开发者、游戏、VR、零售、设计团队单图生成 3D 快、API / 社区分发、开放生态杠杆不像 Tripo 或 Meshy 那样包装成打磨完整的创作者工作台
NVIDIA + SPAR3D平台放大器 / 生态入场者CES 发布叠加 RTX AI PC 分发与 NVIDIA 合作叙事开发者、产品设计师、环境搭建者硬件绑定分发与实时点云编辑叙事留存证据未显示它是独立面向终端用户的 3D SaaS 目的地
Luma相邻多模态创意平台获 HUMAIN、a16z、Amazon、AMD Ventures、NVIDIA、Amplify、Matrix 支持工作室、代理商、企业营销、开发者Ray3.x 电影级控制、HDR/EXR、API、创意智能体平台重点在视频和多模态创作,而非可导出的 3D 网格
Adobe Firefly + Substance 3D创意套件在位替代品庞大装机基础和捆绑的 Creative Cloud 分发设计师、营销人员、创意团队、企业买方商用安全叙事、Content Credentials、合作伙伴模型中心、Substance 毗邻生态专门的 3D 网格生成和导出工作流不如 Tripo/Meshy 明确
OpenAI / Azure Sora 2相邻平台替代品OpenAI 模型经 Azure 部署,带企业 API 和审核层开发者、企业应用构建者、创意团队文本 / 图像 / 视频转视频、重混、音频、按秒计费、Azure 信任层留存证据未显示专门的 3D 网格工作流或资产库产品
人工工具 / 内部自建现状替代品既有人力预算和在位 DCC 技术栈工作室、设计师、内部管线团队控制力最高、工具熟悉、不依赖模型供应商迭代最慢,对非专家的普及效果最弱

买方需要在真实管线中使用可导出的 3D 资产时,直接竞争最强;更宽的创意平台 更多是在相邻预算和信任上竞争,而不是争夺完全相同的网格工作流。

[CP001, CP004, CP006, CP008, CP011, CP013]
FP001: 竞争定位图

Tripo 在 3D 工作流专精度上处于高位,但 Adobe、Azure/OpenAI 和 NVIDIA 相关生态的分发或信任杠杆强于纯 3D 初创公司。

坐标轴是基于保留的产品、定价、分发和信任信号证据综合出的序数判断,不是经审计的市场份额或收入数据。

[CP003, CP011, CP018, CP021, CP022, CP023]

3.2 直接 3D 原生同行

Meshy 是最重要的直接公开基准,因为它的包装与 Tripo 的重叠度高于任何更宽的多模态平台。Meshy 营销文字 / 图像生成 3D、后处理、导出、插件、REST API 访问、团队工作区、企业控制和安全姿态;定价页足够透明,让买方容易比较;2026 年博客 / 新闻节奏显示,它在工作流、打印、智能体式创作和合作节点工具上快速出货。如果 Tripo 想主张自己独特地生产就绪,Meshy 是成熟客户或投资人会第一个拉出来的反例。 Stability 是另一类威胁。它的公开 3D 产品没有包装成同样精致的创作者工作台,但 Stable Fast 3D 和 Stable Point Aware 3D 缩小了核心生成任务上的技术差距,并借 API、GitHub 和社区许可渠道分发。对 Tripo 最不利的证据很直接:Stability 自己的 Stable Fast 3D 代码库称,该模型基于 TripoSR,并加入 UV 展开和材质技术。这意味着 Tripo 的研究领先可以向外扩散到生态工具,而不是被锁在公司付费产品界面里。NVIDIA 又把这个威胁放大,因为 SPAR3D 有了绑定 RTX AI PCs 和 CES 式关注的硬件分发故事。 结果是一场双线直接同行竞争。Tripo 必须在打包产品和商业转化层面击败 Meshy,同时证明自己的端到端工作流广度,能持续领先于开放模型和 API 生态低成本重组出的能力。因此,竞争尽调不应只看表面模型质量,还要看转化、留存、团队工作区采用,以及当相似重建能力在别处广泛可用后,付费客户是否还会留下。[CP004, CP005, CP006, CP007, CP013, CP014]

功能 / 能力矩阵
采购标准Tripo AIMeshyStability AILumaAdobe FireflyOpenAI / Azure Sora 2
文本或图像转 3D 网格None有限 / 不明None
3D 导出后处理None有限 / 不明None
API 访问有限 / 间接
插件 / 工作流桥接深度
团队工作区 / 管理控制有限
多模态视频生成有限有限
商用安全 / 信任叙事
开源或社区可扩展性有限有限有限有限

强 / 中 / 有限 / 无,是基于留存公开资料得出的证据型判断,不是经审计的基准分数。 Adobe 和 Sora 在信任 / 分发上得分高,但留存证据不支持它们拥有专门网格工作流。

[CP004, CP010, CP011, CP013, CP014, CP015]
FP002: 功能广度 / 能力地图

专业 3D 厂商在网格工作流上领先;Adobe、Luma 和 OpenAI 则在邻近多模态和企业平台能力上领先。

“高 / 中 / 低 / 无”标签概括本章保留的公开产品叙事和包装证据;它们不是经审计的同等能力评分。

[CP010, CP011, CP014, CP015, CP019, CP024]

3.3 既有平台与邻近替代品

Adobe、Luma 和 OpenAI 争夺的是同一决策中的另一块,虽然不同但仍重要。它们目前都不像 Tripo 或 Meshy 那样,被呈现为专用 AI 原生 3D 网格工作站。但当买方更看重营销活动制作、多模态输出、商业安全或广义创意套件便利性,而不是专门 3D 工作流时,三者都能赢得预算。Adobe Firefly 尤其相关,因为它把 Adobe 自有模型与来自 Google、OpenAI、Luma 和 Runway 的合作伙伴模型组合在一个商业安全界面里,加入 Content Credentials,并贴近 Substance 3D 和 Creative Cloud。即便 Firefly 不是最好的单用途网格生成器,这也是一个强采购故事。 Luma 最适合被理解为邻近替代品,而不是直接网格对手。它 2026 年的定位强调创意智能体、多模态生成、HDR/EXR 输出、帧级控制、1080p 视频,以及面向工作室和代理商的 API 访问。当交付物是视频或品牌叙事,而不是可导出的资产时,这些能力可能把创意团队从专用 3D 工具拉走。OpenAI 和 Azure Sora 2 延续同一模式:文本 / 图像 / 视频转视频、重混、音频、按秒计费,以及 Responsible AI 护栏。即便独立 Sora 应用和 API 停止,也说明大平台可以多快地重新包装或重新部署能力,而不保留稳定产品边界。 因此,这些邻近平台的重要性,不在于一对一替换 Tripo 网格,而在于挤压注意力、实验预算和信任。如果企业买方已经使用 Adobe 或 Azure,Tripo 必须说明为什么一个独立 3D 原生工具值得单独占一个席位、API 预算项或采购周期。只要 Tripo 在网格工作流、引擎就绪导出和资产吞吐上明显更强,这一点可以承受;但如果买方把 3D 生成视为更大多模态套件里的一个勾选项,就会很危险。[CP008, CP009, CP010, CP011, CP012, CP017]

定价 / 包装对比
公司价格 / 单位 / 合同模式包含能力折扣或未知项含义
Tripo AI免费;Pro 约 $19.9/月 / 年付折合 $13.93/月;Max 约 $89/月 / 年付折合 $53.94/月;Team 按席位定价积分、并发、网格质量升级、批量生成、私有模型、团队工作区实际企业定价和 API 经济性未公开清晰的自助升级梯度有助于创作者转化,但也暴露直接比价风险
Meshy免费 $0;Pro $20/mo;Studio $60/mo;Enterprise 定制积分、更快生成、API 访问、私有所有权、并发、企业控制促销折扣会改变入门价格;企业定价未披露两家公司都发布面向消费者的档位,买方会立刻拿 Tripo 和 Meshy 对比
Luma基于积分的套餐,叠加团队和企业承诺;Ray3.14 base 约 4 credits/sec视频 / 图像模型、Luma Agents 用量、团队管理、SSO、分析积分折算成美元会随套餐和输出模式变化买方按媒体输出而非网格资产做预算时,竞争范围会扩大
Adobe Firefly每日免费生成次数,加带生成积分的付费套餐图像、视频、音频、矢量、编辑、合作伙伴模型、商用安全工作流留存证据中的公开套餐页偏高层;具体企业折扣未知捆绑积分和在位信任会让单独采购专门工具更难自证
Stability AIAPI / 平台和社区许可分发;留存页面未见简单消费级档位Stable Fast 3D、SPAR3D、API 访问、开放 / 社区生态所审公开来源未保留 3D 用量标价主要靠开发者采用和商品化压力竞争,而不是清晰席位定价
OpenAI / Azure Sora 2通过异步 API 模型按秒计费文本 / 图像 / 视频转视频、重混、音频、Azure 安全护栏和内容审核已留存的公开资料没有给出与 Tripo 面向终端用户套餐的简单对比主要关系到相邻预算争夺和平台整合

直接打包竞品的定价透明度明显高于平台或 API 替代品;这本身就是 Tripo 和 Meshy 做自助获客的竞争优势。

[CP002, CP005, CP009, CP019, CP024, CP027]

3.4 护城河耐久性与竞争风险

Tripo 最强的公开护城河,并不是别人生成不了 3D 物体。公开记录已经推翻了这一点。相反,它的护城河在这些地方最强:工作流包装、创作者 UX、导出纪律、团队功能和带有领域感的 3D 身份,合在一起形成一个比从模型、代码库和通用创意套件拼栈更容易采用的产品。这是真实价值,尤其适合需要反复资产吞吐,而不是偶发 AI 实验的游戏、XR、电商和设计团队。 但护城河裂缝可见。Meshy 证明另一家创业公司可以用透明定价、企业控制和大规模宣称,包装出高度相似的故事。Stability 证明源于 Tripo 的研究思路可以扩散到开放或半开放生态。Adobe、Luma、OpenAI 和 Microsoft 证明,围绕工作流的信任、安全控制、品牌认知和打包客户入口,可能和生成器本身一样重要。这些力量指向一种可能的多栖市场结构:买方测试多家供应商,把手工 DCC 工具作为后备,并避免深度依赖单一供应商,除非 API、工作区、资产库或协作功能创造出真正锁定。 因此,实际尽调问题很简单:Tripo 哪些切换成本已经能穿越模型商品化?如果付费收入仍主要来自追逐最新模型输出的自助创作者,耐久性就弱。若收入越来越多嵌入 API 驱动资产管线、共享工作区、企业治理和可重复垂直工作流,耐久性就更好。在管理层用队列、留存和赢单 / 输单数据证明这次转变前,公司应被视为竞争前景不错,但防线尚未牢固。[CP016, CP025, CP028, CP033, CP034, CP035]

护城河耐久度 / 竞争风险清单
护城河主张威胁严重性缓释因素 / 当前信号尽调要求
端到端 3D 工作流打包Meshy 已经提供类似的 Web、导出、插件和 API 打包Tripo 仍在主打更完整的一体化 3D 工作站定位按创作者和工作室客群复盘相对 Meshy 的赢单 / 输单数据
专有模型领先Stability 公开基于 TripoSR 构建,说明研究成果已扩散到生态工具Tripo 仍可靠工作流深度和集成化产品落地拉开差异询问付费使用中有多大比例依赖开放替代品没有的功能
空间智能 / 世界模型叙事Luma、OpenAI 和 NVIDIA 也在讲物理世界或世界模型愿景Tripo 有 3D 原生产品验证,不只是叙事定位验证客户购买 Tripo 是为了当前工作流,还是为了未来世界模型故事
企业采购信任Adobe 和 Azure 自带更强的捆绑信任、审核能力和采购关系当专业 3D 输出比套件标准化更重要时,Tripo 仍有赢面索取大型账户的安全、合规和续约证据
自助式创作者获客多栖使用和低切换成本让创作者会追逐最新模型质量公开定价和高频发布仍能让 Tripo 留在候选清单衡量竞品重大发布后付费转化和流失变化
API 与工作区锁定如果客户只用生成端点,切换仍然容易共享工作区、资产历史和管理员功能可以加深嵌入拆分 API / 工作区客群与轻度席位买家的收入
手工 DCC 回退许多团队可以保留 Blender / Substance / 手工流程,只在有机会时用 AI显著节省时间、导出可用度高,仍可带来重复使用跑通管线基准,说明 Tripo 在哪里减少真实美术工时,而不只是演示时间

严重性衡量的是耐久度风险,而不是短期份额流失。共同模式是基础生成商品化,价值迁移到工作流、信任和嵌入式使用上。

[CP016, CP025, CP028, CP033, CP034, CP035]
FP003: 护城河 / 就绪度 KPI

公开竞争快照显示直接产品定位强,但护城河耐久度取决于转化、工作流嵌入,以及 Tripo 能否跑赢商品化压力。

KPI 标签刻意混合具名对手和定性状态信号,用来区分可见强项与尚未解答的耐久性问题。

[CP016, CP030, CP031, CP033, CP038, CP039]

3.5 展品

Chapter 04

04财务

4.1 收入模型与定价界面

Tripo 的公开变现架构,比它的公开财务披露具体得多。公司清楚销售三类经济产品:自助 Studio 订阅、按使用量计费的 API credits,以及更高席位数的团队或企业计划。定价页展示了经典 SaaS 阶梯——Free、Pro、Max、Team;API 文档则展示了独立的先付后用 credit 账本。这种拆分很重要,因为它意味着 Tripo 不依赖单一变现动作。它可以靠循环订阅变现爱好者和专业人士,靠用量和预付 credits 变现开发者,并靠团队治理、并发、工作区和管理功能变现组织。 公开界面也显示 Tripo 实际在为什么收费。定价页不只是限制模型访问,还限制私有模型、商业使用、并发、批量导出、专用处理、共享工作区、存储深度和编辑历史。API 文档再加一层:复杂度更高的工作流消耗更多 credits,SDK 和任务端点暴露精确的已消耗 credit 与余额计量数据。换句话说,Tripo 已经搭出一个商业控制系统,不只按席位数,也能按真实工作流强度计量价值。 但这不能证明收入质量。便宜入门价格会扩大漏斗,也意味着如果免费用户占主导,或专业用户没有转化为 API 或团队支出,表面用户规模可能掩盖弱变现。公开材料证明 Tripo 知道自己想怎么收费;还没有证明这些收费能否高效转化为高毛利、可重复的企业收入。[CI001, CI002, CI003, CI004, CI005, CI006]

收入流表
收入流机制计量单位当前价值 / 状态质量尽调要求
Studio 订阅面向创作者和专业用户的经常性付费计划月度席位 / 账户Free、Pro、Max、Team 已公开列出中;定价可见,但转化未知按计划拆分付费用户、月付与年付占比、流失率
API 积分预付使用积分,由生成和后处理任务消耗积分与任务消耗公开文档称先付后用,大用量可签定制合同中高;计量收入模型可见,但使用集中度未知提供 API 收入占比、头部客户集中度、每名活跃开发者月均支出
团队 / 企业席位按席位或托管团队计划,包含共享工作区和管理员控制席位合同 / 工作区Team 层级已公开;定制企业经济性未公开中;收入质量可能好于创作者层级,但尚未披露展示企业 ACV、部署规模、续约率和服务附加率
商业权利和隐私增购相比公开 CC BY 的免费输出,付费计划解锁私有模型和商用权访问权 / 计划层级定价阶梯可见中;基于权利的增购逻辑成立,但真实付费意愿未知量化隐私或商用权要求推动转化的频率
工作流功能增购更高计划用并发、批量导出、专用处理、编辑历史、存储和共享资产设门槛功能包Pro、Max、Team 中公开可见中;变现的是工作流摩擦,而不只是新鲜感展示付费客群中高级工作流功能的附加率或使用率

Tripo 展示了多条变现杠杆,但公开证据没有披露各项收入占比。

[CI001, CI002, CI003, CI004, CI005, CI013]
定价 / 变现表
产品 / 计划价格 / 单位 / 合同包含能力折扣 / 未知项来源
Free Studio$0 / month200 积分,1 个并发任务,公开模型,下载受限免费转付费转化未披露Tripo 定价页
Pro Studio$19.9 / 月,或年付折算 $13.933,000 积分,10 个并发任务,私有模型,商用权,智能网格实际折扣和留存未知Tripo 定价页和 Costbench
Max Studio$89.9 / 月,或年付折算 $53.9425,000 积分,100 个并发任务,专用处理,无限重试重度用户与工作室的客户结构未披露Tripo 定价页和 Costbench
Team每席位 $109.9 / 月,或年付折算每席位 $5445,000 积分,200 个并发任务,共享工作区,集中计费Team 之外的企业定制条款未公开Tripo 定价页和 Costbench
API 积分账本$1 = 100 积分;大用量合同通过销售定制按积分计费,按任务和功能加收费用企业批量折扣未公开Tripo OpenAPI 文档
合同 / 账单条款通过 Stripe 以 USD 计费;除条款说明外不退款;取消后当前周期结束不再续费商户收款、税费、付款执行有滞纳金和税务处理,但企业开票条款未公开Tripo 条款

作为 AI 3D 初创公司,Tripo 的标价透明度异常高,但实际成交价和企业合同结构仍不透明。

[CI002, CI003, CI004, CI005, CI006, CI007]
FI001: 收入模型桥接图

Tripo 同时向席位和用量收费;付费价值越来越绑在工作流强度、隐私、并发和协作上。

[CI001, CI003, CI004, CI005, CI008, CI013]
FI003: 财务估算区间

公开付费席位价格区间足够宽,可以支撑向上销售;但绝对水平仍不高,转化率和企业客户组合比对外披露的用户数更重要。

该区间只展示公开标价。它不是每用户收入、实际合同价值,也不能代理企业 ACV。

[CI002, CI003, CI004, CI032, CI033]

4.2 牵引、轮次与资本部署

公开牵引和融资信号很多,但也很乱。独立和公司周边来源指向快速扩大的产品足迹——数百万用户、数万开发者或工作室客户,以及逐渐增强的企业叙事——但这些数字在媒体之间严重冲突。融资金额也一样。2026 年 3 月报道集中在 $50 million Series A 和新模型发布。2026 年 6 月报道集中在近 $200 million 的 Series A+ 和 A++ 融资,以及世界模型雄心。2026 年 7 月中文报道又描述了超过 RMB1 billion 的 A3 战略轮。数据聚合商随后发布了仍然不同的总额,包括约 $397 million 累计融资和最近一笔 $147 million 融资。 正确解读不是某个来源显然为真、其他都捏造。更可能的解释是,Tripo 在相近时间关闭了几笔分批融资,各自有不同本地命名惯例、战略投资人组合和发布时间。这仍支撑一个核心结论:2026 年公司获得新资本的能力异常强。资金用途表述在保留记录中也方向一致。管理层强调模型研究、算法迭代、数据积累、基础设施、人才,以及全球产品或生态扩张,而不是承诺短期经营杠杆。 从财务上看,这是一把双刃剑。它提高了近期资本充足度,让 Tripo 不太可能马上放慢产品开发。但它也意味着当前投资人仍在资助建设扩张阶段,而不是收获披露充分的收入引擎。因此,今天的公开故事是支撑继续扩张的资本充足性——还不是成熟、高效软件业务的证明。[CI015, CI016, CI017, CI018, CI019, CI020]

资本充足性表
资本项目公开证据置信度重要性尽调要求
2026 年 3 月融资Auganix 报道,公司在发布新模型的同时完成 $50M A 轮融资显示 2026 年夏季前的资本支持和增长节奏确认轮次日期、投资方,以及是否已完全计入后续总额
2026 年 6 月融资Forbes 和 AIThority 报道约 $200M 的 A+ / A++ 轮融资中高显示投资人需求强,近期现金大概率得到补强提供准确融资总额、交割日期和收款法律实体
2026 年 7 月融资Cyzone 报道超过 RMB1B 的 A3 轮战略融资暗示又一笔重大现金注入,或前序分批融资延续澄清 A3 是增量现金,还是前序轮次改名后的子集
资金用途多个来源反复提到研发、算法、数据、基础设施、人才和全球生态扩张说明管理层仍把建设扩张放在近期盈利之前拆分研究、产品、算力、销售和运营预算分配
在手现金决定 2026 年连续融资后的真实现金跑道提供最近一次交割后的期末现金余额
现金跑道(月)需要判断融资依赖度和下一轮时间点提供当前招聘和算力计划下的基准情景现金跑道
法律交易对手 / 收款实体Holymolly Ltd 出现在条款中;香港注册处镜像显示 2023 年成立且仍存续影响合同可执行性、税务和融资结构识别哪些实体确认收入、持有 IP,并接收各轮融资款

公开来源有力证明公司能反复获得资本,但没有给出干净、单一来源的现金桥。

[CI009, CI012, CI021, CI022, CI023, CI024]
FI004: 资本强度 / 现金流图

2026 年融资叙事看起来把资金投向研发、数据、基础设施和全球扩张,而不是短期盈利冲刺。

[CI021, CI022, CI023, CI024, CI026, CI035]

4.3 单位经济与披露缺口

用公开信息承销 Tripo,最难的不是理解产品可能如何赚钱,而是理解它已经做得多好。保留来源没有披露收入、ARR、毛利率、现金消耗、现金余额、客户集中度、净留存、企业 ACV 或算力承诺。这迫使分析师从收入机制而不是业绩证据出发。好消息是机制清楚:credits 可以把使用量映射到计费,团队计划可以变现协作和治理,私有 / 商业权利可以把免费用户推向付费层。坏消息是,看得懂机制不等于证明机制有效。 公开标价足够低,仅凭用户数说明不了多少。大量免费或低支出创作者,可能与不高的实际收入并存。如果存在更高质量收入,它大概率位于循环团队合同、共享工作区、API 集成,以及管线反复消耗 credits 的客户中。因此,单位经济问题本质上是构成问题。没有细分转化和使用强度,投资人无法区分炒作式规模和变现规模。 还有一个不利运营角度。独立拆解的怀疑和竞争定价压力提示,Tripo 仍必须证明生产就绪输出足够好,可以支撑持久预算,而不是反复实验。如果导出清理或工作流摩擦仍高,哪怕漏斗顶端增长很强,变现也可能卡住。因此,公开证据支持可信的计费设计,但还不能公开证明 Tripo 已经从令人兴奋的采用,跨入高质量软件经济性。[CI027, CI028, CI029, CI030, CI031, CI032]

单位经济模型表
指标数值 / null置信度重要性尽调要求
标价下限$19.9 Pro / 月确立转化后专业席位的最低变现水平按付费客群和账单节奏展示实际 ARPU
公开团队席位最高价每席位 $109.9 / 月指示打包自助式团队产品的变现上限提供实际企业席位数和谈判折扣
API 基础兑换率$1 = 100 积分把使用量接到美元计费上,可支撑利润率分析提供每名付费开发者平均消耗积分和有效美元支出
毛利率决定重算力增长是有吸引力,还是会毁掉资本效率按产品线拆分推理、存储、支持和支付处理成本
CAC / 回本期需要判断免费漏斗增长能否高效转成收入提供创作者与企业 CAC、付费渠道和回本周期
净收入留存判断 API / 工作区嵌入和企业耐久度的关键指标按开发者 / API 账户和团队合同报告 NRR
烧钱 / 月度现金消耗把融资额换算成现金跑道时必须用到提供当前月度烧钱和融资后招聘计划
收入集中度如果少数 API 客户或合作伙伴贡献支出,就很重要提供前 10 大客户收入占比和销售管线集中度

公开证据足以画出价格曲面,但不足以衡量真实的软件经济性。

[CI006, CI007, CI027, CI028, CI029, CI030]
公开财务缺口表
缺失的私有指标对投资测算的影响具体尽调路径
收入 / ARR无法把用户和开发者数量换算成实际业务规模索取按产品线拆分的月度经常性收入、季度收入和 ARR
扣除推理成本后的毛利率无法判断更高使用量带来经济杠杆,还是毛利率压缩复盘包含 GPU、存储、带宽、支付和支持成本的 COGS 瀑布
付费转化和免费客群质量大免费漏斗未必能高效变现索取按计划和用例拆分的免费转付费转化率
企业 ACV 和合同期限Team 和 API 经济性可能显著好于自助式业务,但幅度未知索取头部企业合同规模、标准期限长度和续约率
净收入留存 / 扩张尚无证据证明客户会随着工作流成熟而加深支出索取 API、Team 和创作者客群的 NRR
烧钱和现金余额尽管融资新闻金额很大,仍无法估算现金跑道索取当前现金、月度烧钱和未来 12 个月运营计划
云 / GPU 承诺如果推理或训练承诺已锁定,资本需求可能激增复盘云合同、预留算力承诺和模型训练预算
收入集中度少数合作伙伴或工作室可能贡献过大的收入占比索取前 10 大客户收入占比和销售管线集中度

这些缺口决定了只是欣赏收入设计,还是能给这门生意做投资测算。

[CI017, CI027, CI037, CI038, CI039]
FI002: 单位经济性桥接图

公开证据能看出钱从哪里进来、算力密集成本从哪里出去,但桥两端的规模都没有公开。

[CI006, CI007, CI027, CI028, CI029, CI037]

4.4 财务结论

Tripo 的公开财务图景,架构最强,证明最弱。架构是合理的:低摩擦获客、与使用挂钩的 API 变现、分层向上销售到商业权利和工作流功能,以及看起来足够大的资本基础,能让产品开发继续推进。这正是 AI 基础设施邻近应用公司该有的设置,因为模型创新和工作流包装都很重要。 但证明缺口大到投资人不应过度自信。公司没有公开展示决定其定价界面是否有经济吸引力的核心数字:付费转化、收入集中度、实际 ARPU 或 ARPA、扣除推理和支持后的毛利率、企业留存和现金消耗。竞争平台也在发布易获取的定价,仅凭标价很难推断定价权。 实际结论是,Tripo 看起来可融资,但还不能完全承销。公开信息支持看多资本获取能力,也支持对变现设计作出中等正面判断。它不支持对效率、收入质量或达到自我维持经济性的时间作出精确判断。在管理层打开构成、毛利率和现金消耗账本之前,投资逻辑仍取决于相信快速扩张和未来强转化,而不是已展示的公开财务表现。[CI013, CI026, CI027, CI035, CI036, CI037]

4.5 展品

Chapter 05

05产品与技术

5.1 产品界面与工作流覆盖

Tripo 当前产品界面明显宽于简单文字生成 3D 演示。官方首页和功能页展示了一个托管工作流,从文字提示、图像和多视角参考开始,再延伸到贴图、分割、四边面重网格、绑定和导出等后处理模块。定价页进一步说明,这套产品想跨不同用户类型运行:轻量免费层用于实验,专业消费者和工作室层提供更高并发,面向团队的包装则加入共享工作区和集中管理。这种广度重要,因为 Tripo 试图把 3D 管线的多个步骤压缩到一个订阅界面里,而不是以单一狭窄模型端点竞争。 输出叙事也比许多通用生成式 AI 产品更懂工作流。Tripo 官方材料和教程讨论 GLB、FBX、OBJ、STL,以及面向 Roblox 的导出;JavaScript 指南明确建议网页部署使用 GLB,并提到 USDZ 在某些集成工作流中可作为替代路径。实践中,当用户想为游戏、网页场景、3D 打印或交互原型快速拿到第一版资产,并愿意把它们送入下游清理或引擎检查时,Tripo 看起来最强。因此,产品最好被框定为从创作者到生产的加速层,而不是每个 DCC 或引擎原生任务的替代品。[CE001, CE002, CE003, CE004, CE005, CE006]

产品模块 / 资产矩阵
模块 / 资产主要用户当前公开状态差异化信号主要尽调缺口
核心文本 / 图像生成创作者、工作室、开发者Web 界面和定价层级已 GA托管式 3D 秒级生成,并面向广泛创作者打包需要按资产类别审计质量分布
多视角 / 部件生成准专业用户和工作室用户公开包含在 Pro 层级说明产品正从一次性提示词演示走向可控工作流需要准确成功率和积分经济性
AI 贴图 + 风格化美术师和产品视觉团队公开功能页和工作流指南让 Tripo 从只做网格走向更完整的资产精修需要证明贴图可跨引擎迁移
自动绑定 + 动画角色团队和游戏创作者公开功能页加 SDK 支持将用途拓宽到角色或动画工作流需要困难非对称角色或布料密集角色的基准结果
分割 + 四边面重拓扑 + 清理技术美术和下游编辑者公开功能页和教程直接处理常见 AI-3D 清理瓶颈重度强调也说明原始输出仍经常需要辅助
API / SDK / 插件开发者和工作流集成方开发者门户、Python SDK、Blender、ComfyUI、MCP、JS 指南嵌入面比多数只服务创作者的工具更深企业管控和 SLA 披露仍不足

各行只总结抓取到的公开页面可见模块;不能证明每项功能在 Web 应用、API 和插件中的成熟度相同。

[CE001, CE003, CE005, CE006, CE007, CE008]
工作流 / 用例表
用户任务当前工作流痛点Tripo 方案可衡量收益 / 公开信号限制或注意事项
Web 创作者概念设计手工建模做首版资产很慢浏览器工作流中的文本转 3D 或图像转 3D秒级生成和免费入门层级降低试验成本质量有波动;首次导出仍可能需要清理
游戏资产准备网格、UV 和缩放问题可能让引擎出错游戏就绪清单、重拓扑、绑定和 FBX / GLB 工作流官方文档对 Unity / Unreal 约束写得异常明确发布清单本身就说明真实下游失效风险
开发者嵌入从零搭建定制 3D 生成很复杂开发者门户、Python SDK、认证 API、异步任务独立开发指南和官方 SDK 指向低摩擦集成正常运行时间或管理员保障的公开证据很少
节点式 AI 工作流在工具之间来回切换会割裂创作流ComfyUI 节点和官方合作伙伴文档原生合作伙伴节点支持,代码库更新活跃仍依赖托管 API 和密钥管理
基于 Blender 的编辑AI 输出常常需要在 DCC 侧清理官方 Blender 插件、分割、重拓扑,以及 MCP / Cursor 工作流拉短生成到人工修正的距离重度清理仍要在 Blender 里做,并不会凭空消失
Web / AR / 3D 打印部署格式不匹配和比例错误拖慢部署以 GLB 为先的 Web 指南、Roblox 比例指南、MakerWorld 打印合作伙伴案例已有文档说明如何交接到 Babylon.js、Three.js、Roblox 和 MakerWorld核心文档对 USDZ 和部分移动 AR 路径的确认仍不够清楚

收益是方向性的、基于工作流判断;除 MakerWorld 案例研究外,公开来源很少给出转化率或节省时间的硬指标。

[CE012, CE013, CE014, CE015, CE018, CE020]
FE001: 产品架构图

公开证据把 Tripo 描述成分层托管栈:从创作者包装起步,穿过模型和后处理层,最后由插件或 API 分发。

[CE001, CE003, CE005, CE008, CE009, CE010]
FE002: 客户工作流 / 操作流程

公开工作流从简单提示词或图像输入开始,进入生成和可选清理,再导出或嵌入下游工具。

[CE001, CE005, CE009, CE010, CE012, CE015]

5.2 开发者界面与集成生态

开发者界面是公开记录里最清楚的差异化之一。Tripo 不只是暴露一个营销网站;它有开发者门户、需要密钥的 API、官方 Python SDK、ComfyUI 合作文档、官方 ComfyUI 和 Blender 代码库,以及 MCP/Cursor 和 Babylon.js 或 Three.js 部署工作流指南。这套组合释放出的产品策略信号是:公司想嵌入创作者和开发者栈,而不是强迫所有使用都穿过单一浏览器 UI。OpenAPI 任务端点要求认证,SDK 是异步的,插件生态覆盖 Blender、ComfyUI、web 运行时,甚至面向 Roblox 的导出教程——这些都指向托管、以 API 为中心的运营模型。 生态证据不只是轶事。GitHub 元数据显示,TripoSR 牵引异常强,TripoSG 代码库、ComfyUI 节点、Blender 扩展和 Python SDK 也有较小但有意义的牵引。与 MakerWorld/Bambu Lab 的合作案例也重要,因为它表明 Tripo 可以嵌入别人的工作流或市场,而不只是作为独立创作者应用使用。主要限制是,公开界面里的企业治理文档仍偏薄:功能文档很多,但 SLA、SSO、管理控制或可用性承诺的可见证据很少,难以帮助技术买方承销关键任务部署。[CE017, CE018, CE019, CE020, CE021, CE022]

技术 / 运营架构表
层级 / 组件在技术栈中的作用关键依赖主要技术风险
托管式生成 UI承接提示词、图像和创作者任务的入口界面云端推理和 Tripo 账户系统打磨过的预览可能掩盖输出质量波动
TripoSG 图像转 3D 基础模型基于图像条件输入生成高保真 3D大规模 rectified-flow 训练、精选数据、GPU 推理公开研究有多少能直接落到付费生产端点,仍不清楚
TripoSR 重建模型快速单图重建,也提供开放技术可信度A100 级基准测试环境和 CUDA 兼容栈开源基准测试速度不等于真实负载下的产品延迟
后处理层分割、重网格、贴图、绑定、低多边形、格式转换工作流编排和面向具体模型的编辑工具过度依赖修补,可能说明首轮输出还不够干净
API 和 SDK 层用程序创建任务、下载结果、自动化流程并嵌入应用鉴权端点、异步任务处理、客户端库公开文档没有展示企业级 SLA 或管理保障
插件 / 生态适配器Blender、ComfyUI、JS 运行时、Roblox、合作伙伴嵌入、MCP/Cursor第三方工具兼容性和插件持续维护每个适配器都会多出一个可能失效或跑偏的集成点

这张架构图根据公开论文、仓库、SDK 文档、插件教程和 API 行为推断;Tripo 没有发布统一的官方系统图。

[CE017, CE018, CE019, CE021, CE022, CE023]
FE003: 关键依赖图

Tripo 的可用性靠托管推理、开放研究、插件维护和外部运行目的地共同喂给同一个交付层。

[CE017, CE018, CE021, CE023, CE024, CE026]

5.3 模型架构与公开路线图信号

Tripo 最强技术可信度信号来自研究层。TripoSG 论文和代码库展示的是真实模型架构,而不是模糊产品营销:大规模 rectified-flow transformer,结合 SDF、normal 和 eikonal losses 的 hybrid VAE supervision,以及 2 million 个样本的 image-to-SDF 数据管线。代码库还记录了 1.5B 参数版本和涂鸦条件变体,说明公司愿意向研究和开发者社区暴露至少一部分模型谱系。TripoSR 从重建侧补全了这个故事:它是领先单图模型,声称在 A100 硬件上推理不到 0.5 秒,并以 MIT license 发布,由 Stability AI 共同开发。 2026 年路线图叙事越过了这些研究产物。公开新闻和赞助报道把 P1.0 和 H3.1 描述为更整体的空间架构、更面向生产,目标是降低许多 3D 生成器面临的手工编辑负担。这不能证明完整产品已经在所有地方交付每一项声称能力,但它确实展示了一条连贯战略方向:Tripo 想从「快速生成一个网格」走向「生成更干净、更能进管线的资产,并借插件、API 和创作者工作区分发」。剩余尽调任务,是把这些研究和营销标签干净映射到实际生产端点和变现界面。[CE028, CE029, CE030, CE031, CE035, CE036]

路线图 / 发布 / 开发阶段表
日期 / 阶段功能或里程碑状态含义来源
2024-03TripoSR 开源发布已观察到快速重建模型给 Tripo 带来早期技术可信度和开发者触达GitHub 仓库 + API 元数据
2025-03TripoSG 论文和仓库发布已观察到明确披露架构,显示其图像转 3D 基础模型野心arXiv + GitHub 仓库
2025-03 至 2025-04TripoSG 1.5B 发布及涂鸦条件变体已观察到释放迭代速度和可控性实验信号TripoSG 仓库
2026-03GDC 2026 前后的 P1.0 和 H3.1 生产级叙事第三方报道叙事从纯生成转向引擎就绪的原生 3D 扩散TechTimes + Stanford Daily
2026-06-30ComfyUI 插件更新:Mesh Segmentation v2.0 和直接 URL 导入已观察到表明到 2026 年中,生态工具仍在积极维护ComfyUI-Tripo GitHub 仓库
2026-07-01Python SDK 已推送,并覆盖当前 API 面已观察到开发者自动化仍是技术栈中持续维护的一环GitHub API 元数据

这些日期只是公开发布信号,不是管理层批准的 GA 承诺;因此路线图解读只能视为方向性判断,不具合同约束力。

[CE022, CE029, CE030, CE032, CE035, CE036]

5.4 质量控制、成熟度与技术限制

负面证据并不是说 Tripo 弱,而是在说最后一公里仍然关键。Tripo 自己的游戏就绪、清洁网格和补洞指南反复强调生成之后还要处理拓扑、UV、比例、重拓扑和修复。独立评测把同一点说得更尖锐:Tech On Play 把 Tripo 看成工作流加速器,而不是完整替代品;Medium 拆解则提醒,预览质量可能掩盖缺陷,只有导出到真实工具链后才暴露。这些信号重要,因为核心尽调问题不是 Tripo 能不能做出好看的资产,而是这些资产交给 Blender、Unity、Unreal、WebGL 或打印流程后,能否用可接受的清理时间跑完。 落到实践里,这项技术最强的场景是创意构思、第一版资产生成,以及速度比首轮几何精度更重要的近生产工作流。弱点则出现在用户需要确定性绑定、稳定的复杂有机网格,或清晰企业部署控制的时候。因此,产品成熟度在覆盖广度上扎实,模型深度也越来越可信;但在区分「出色 demo」和「默认生产标准」的关键环节上,表现仍然参差。严肃投资人应要求用真实资产做基准导出,而不是只看产品营销或视觉效果很强的 demo。[CE015, CE016, CE037, CE038, CE039, CE040]

信任 / 质量 / 合规表
控制项 / 质量指标当前公开状态范围缺口 / 含义
API 鉴权可见OpenAPI 任务端点拒绝匿名访问说明有基础密钥门禁,但看不到企业身份控制
公开使用权可见免费计划明确将公开模型标注为 CC BY 4.0还需要更完整的矩阵,说明付费计划输出权利和参考数据处理
游戏就绪验证可见官方清单覆盖多边形、UV、纹理、比例和导入检查清单有用,但不能替代基准测试通过率
清理 / 修复工具可见Tripo 宣传重新拓扑、分割、洁净网格和补洞工作流说明配套工具存在,也意味着原始输出仍会频繁出问题,足以影响使用
企业部署控制公开能见度弱公开来源未显示清晰的 SLA、SSO 或管理员治理界面可能拖慢企业采用,或迫使上线前做定制尽调

这张表有意把已经存在的控制和缺失项放在一起;投资判断要回答的是,Tripo 的信任面是否足够支撑生产部署,而不是它是否没有任何控制。

[CE004, CE015, CE017, CE041, CE042, CE045]
FE004: 产品成熟度 / 能力图

公开证据显示,面向创作者的能力覆盖很广;但企业控制和稳定达到生产级完美输出,确定性明显低得多。

[CE015, CE018, CE023, CE032, CE034, CE037]

5.5 佐证材料

Chapter 06

06客户

6.1 客户分层与工作流组合

Tripo 还很年轻,但客户图谱已少见地分散。官方材料没有把产品锁定在单一买方,而是反复面向创作者、独立游戏开发者、工作室、产品团队、建筑师、零售商,以及影视或媒体用户。这种多样性重要,因为它给平台带来多个入口:个人创作者可以从免费计划开始,游戏团队可以用它快速原型,电商运营可以把它接入商品目录工作流,平台伙伴则可以通过 API 或工作区界面嵌入。时尚 AR、目录扩展、影视级预演和游戏绑定的垂直页面,都指向同一个结论:只要 3D 资产吞吐是反复出现的问题,Tripo 的卖点就是工作流加速器。 广度的另一面,是不同用户分层的变现能力并不相同。免费层降低了使用门槛,创作者和爱好者采用大概率充足;但企业级付费意愿会集中在 3D 资产具有运营重要性、而不只是装饰性的场景。因此,公开用例页面在游戏、商业可视化、平台嵌入、建筑或工业设计,以及 XR 或空间显示上最有说服力。拿来证明深度企业粘性时,它们的说服力较弱,因为公开记录更多说明谁能使用 Tripo,而不是哪些分层会续约、扩张,并长期贡献可观付费。[CU001, CU002, CU003, CU004, CU005, CU006]

客户分层表
分层买方 / 用户 / 付款方主要用例公开规模信号收入或战略价值缺口
个人创作者 / 爱好者用户和付款方往往是同一个人快速概念设计、资产生成、实验2025–2026 年来源显示用户达数百万庞大的漏斗顶部和社区增长引擎付费转化率未知
游戏开发者 / 工作室美术、技术美术、制作人、独立创始人道具原型、已绑定角色、引擎就绪资产官方游戏开发内容,加上 NetEase 证据和调研口径可能是最容易变现的重复资产创建客群之一需要 ACV、部署深度和重复使用数据
电商 / 零售团队商品、产品、增长或工程团队商品可视化、AR 试穿 / 试用、大 SKU 量 3D 目录多个官方商业垂直页面和产品原型指南可能带来高容量 API 或批量工作流需求结果指标和活跃客户名称都偏少
建筑 / 工业设计设计专家、建筑师、项目团队草图转 3D、客户迭代、设计预览Dewan 客座文章和建筑指南精度足够时,工作流价值较高缺少具名付费客户和精度基准测试
平台 / API 集成方开发团队和平台运营方把 3D 生成嵌入目录、工具或工作空间ComfyUI 文档、SDK、面向 PIM 的 API 页面、Replit 引用对可扩展分发和更有黏性的技术采用很重要公开使用量和合同质量未披露
XR / 空间 / 媒体用户硬件、媒体和空间体验团队空间显示、预可视化、数字孪生、互动内容Sony Spatial Reality 和媒体制作页面战略上证明 Tripo 不局限于玩具级创作者场景生产规模和支出仍不清楚

分层标签结合了官方用例页面和第三方报道;它们说明 Tripo 能触达谁,而不是经审计的各客群收入结构。

[CU001, CU002, CU004, CU005, CU006, CU007]
FU001: 客户旅程图

Tripo 的公开旅程从免费创作者试验开始,经由导出和插件加深;只有商业需求或工作流关键需求出现后,才会扩张。

[CU002, CU003, CU009, CU024, CU030, CU031]

6.2 采用趋势与地域分布

即便把口径噪音打折,公开采用曲线仍然强。多个 2025 和 2026 年来源显示,Tripo 已远不只是小众开发者玩具:用户数从 2025 年夏末约 300 万,升至 2025 年末 650 万,再到某个 2026 年来源称 1000 万个人用户;另有来源还提到 35,000 名活跃开发者、40,000 家工作室或企业伙伴、90,000 名开发者,或 90,000 个工作室客户。这些数字明显在衡量不同人群,不能合并成一个干净 KPI;但它们足以支撑一个稳健的方向性结论:Tripo 以少见速度建立了全球分发。 地域格局同样值得注意。Yahoo 转发的 SCMP 报道称,超过 85% 用户在中国以外,欧洲和美国是最大市场;Forbes 又把日本和韩国加入集中市场清单。这一点重要,因为它说明 Tripo 的增长不只是一个中国 AI 故事。与此同时,全球足迹还没有告诉投资人付费收入实际在哪里。没有董事会级别的定义表和区域收入拆分之前,公开采用故事很亮眼,但只有部分达到投资级质量。[CU013, CU014, CU017, CU018, CU019, CU020]

客户增长 / 采用轨迹表
指标数值日期 / 来源可信度含义缺失分母
用户基数8 月基线为 3M 用户2025 / Yahoo-SCMP 参考点确立 2025 年末加速前的拐点前规模需要用户定义
用户基数6.5M 用户2025 年末 / Yahoo-SCMP公开规模在几个月内翻倍以上需要厘清 MAU 与注册用户
个人用户10M 个人用户2026 / AI Weekly最新顶线增长口径又明显变大需要经审计的 2026 年用户定义
开发者35K 活跃开发者2025-09 / ACCESS开发者客群本身已大到值得重视需要明确「活跃」如何计算
工作室 / 合作伙伴40K 家工作室和企业合作伙伴2025-09 / Forbes2025 年 logo 与工作室触达已经很广需要付费与引用 logo 的拆分
开发者90K 开发者2026-03 / Stanford Daily技术用户社区似乎已经显著放大需要与工作室客户的重叠情况
工作室客户90K 工作室客户2026 / AI Weekly若定义成立,最新商业足迹口径很大需要精确定义工作室客户
已生成模型100M+ 个生成的 3D 模型2026-03 / Stanford Daily即使变现不清楚,使用强度看起来很高需要独立用户和重复使用背景
收入信号beta 后一个月月收入增长 2.5x,三个月增长 5x2025 年末 / Yahoo-SCMP说明不只是用户增长,也有变现牵引力需要实际收入基数和可持续性
地域85%+ 用户在中国以外;欧洲和美国领先2025 年末 / Yahoo-SCMP确认全球足迹,也确认本土市场之外的需求需要按地区拆分的付费账户结构

这张表有意保留彼此冲突的分子,因为不同来源衡量的人群不同;把它们强行统一成一个标准 KPI,会夸大精度。

[CU013, CU014, CU017, CU018, CU019, CU020]
FU002: 采用 / 部署漏斗

公开证据显示,漏斗顶端试验很宽,但到可见企业或平台部署前会大幅收窄。

数值是序数索引,表示可能收窄的阶段,不是实测转化率;公开来源没有披露真实客户漏斗。

[CU003, CU016, CU030, CU031, CU032, CU033]

6.3 具名客户证明与开发者信号

具名 logo 证明存在,但质量不均。Sony、NetEase、MakerWorld 或 Bambu Lab、Replit 是最强的公开样本,因为它们要么出现在多个独立来源中,要么有官方工作流页面解释合作如何运作。NetEase 尤其有用,因为 KR Asia 明确称 Tripo 用在 Where Winds Meet 中,把证据从泛泛罗列 logo 推向有记录的下游使用。Sony 的公开合作伙伴页面也重要,因为它把 Tripo 连接到一个具体工作流环境——Spatial Reality Display——以及零售、教育和数字孪生等行业场景。 开发者信号从另一个角度强化了同一故事。ComfyUI 官方伙伴文档、Tripo SDK,以及 TripoSR 和 ComfyUI-Tripo 周边的 GitHub 足迹显示,技术用户不只是旁观产品,而是在把它接入工具链。这很关键,因为 3D-AI 公司可以有病毒式创作者采用,却不一定变得运营上重要。Tripo 的公开开发者与伙伴界面更有力地说明,客户基础中至少有一部分正在真实工作流里使用系统。剩下的缺口是商业具体性:公开来源仍未清晰区分付费生产客户、试点、合作伙伴和参考用户。[CU011, CU012, CU015, CU019, CU021, CU022]

具名客户证据表
客户 / 合作伙伴分层部署或用例生产 / 试点结果 / 信号限制
Sony Spatial Reality DivisionXR / 空间显示围绕裸眼 3D 显示、内容生成和互动体验的官方合作带有生产意图的合作,但尚未披露为付费续约合同强品牌背书,也有具体工作流语境商业条款、规模和经常性收入未公开
NetEase游戏工作室 / 发行商据称用于 Where Winds Meet,且多次出现在客户名单中最强的公开生产使用信号证据不再只是泛泛 logo 引用,而是落到具名游戏部署收入、合同范围和续约质量未披露
MakerWorld / Bambu Lab创作者平台 / 3D 打印生态图像转模型和灯笼工作流嵌入给创作者使用运营型合作伙伴嵌入案例研究称原型制作快 90%,社区使用范围也更广案例研究由公司撰写,缺少独立 KPI 验证
Replit开发者平台ACCESS 和 Stanford Daily 将其列为合作伙伴或平台协作者合作 / 集成可见度支持 Tripo 能嵌入开发者工作流的判断具体产品使用和商业状态仍模糊
Tencent / Microsoft / Pop Mart / HTC 群组企业 logo 集群公开来源反复将它们列为重要用户或合作伙伴混合引用集合显示其在科技、游戏和消费品牌中的品类级企业能见度公开记录没有清楚区分活跃付费部署和引用 logo

这里只部分列举最强的公开 logo 证据;入选行要么在多个来源中反复出现,要么有官方案例研究式工作流描述。

[CU011, CU012, CU015, CU018, CU019, CU021]
FU003: 客户验证矩阵

Tripo 能同时给出具名 logo 和具体工作流时,公开客户验证最强;只有 logo、没有合同或部署细节时,验证最弱。

[CU012, CU022, CU024, CU025, CU026, CU036]

6.4 留存、扩张与客户质量风险

最大投资判断问题不是缺少采用,而是客户质量不清楚。公开来源大量谈到创作者、logo 和增长,却很少披露 NRR、流失率、合同期限、按席位或 API 量扩张的情况,或数百万用户中到底有多少会变成付费订阅者。价格阶梯暗示了一条可行的落地扩张路径:从免费创作者,到付费 Pro,再到团队或 API 使用;但实际转化数学缺失。独立评测和定价摘要强化了反向证据:免费层足够慷慨,容易支撑大量试用;只有商业或私有使用变得重要时,付费权利才真正必要;更大的席位制计划价格不低,不是每个试用都会扩成严肃企业部署。 因此,投资人应把客户故事视为有前景但尚未完全证实。获客动作看起来是产品驱动且国际化分布;如果转化健康,这很高效,反之风险也高。具名 logo 有帮助,但 logo 密度本身不等于耐久收入。正确的下一步尽调,是从亮眼的公开触达转向有 cohort 支撑的收入质量:转化、续约、使用强度,以及按支出区间划分的 logo 状态。看不到这些之前,Tripo 的客户章节支持正面的需求判断,但变现判断只能给中等置信度。[CU003, CU016, CU030, CU031, CU032, CU033]

留存 / 重复使用 / 满意度表
指标数值 / null分层可信度尽调追问
NRRnull企业 / 团队按计划、API 与工作区客群索取 NRR
GRR / logo 流失null企业 / 团队索取续约、流失和降档历史
合同期限null企业 / 平台索取平均期限、续约时点和试点转生产转化
免费转付费null创作者 / 自助式索取免费账户到付费计划、再到首次商业导出的转化漏斗
公开评价情绪混合:原型制作评价强,最终生产确定性较弱独立评测者把创作者满意度与生产团队满意度分开看
重复使用信号只能从收入增长、模型数量和开发者 / 社区活跃度做方向性判断跨客群需要真实队列留存,而不是使用轶事

null 是有意保留的,因为公开来源没有披露留存指标;这种缺失本身就是重要尽调发现。

[CU017, CU031, CU032, CU033, CU034, CU035]
扩张与集中风险表
扩张驱动集中风险影响尽调路径
Free → Pro → Max / Team 套餐庞大免费客群可能永远不转化能做出亮眼用户数,但变现可能比表面采用度更薄按计划索取漏斗和 ARPA
创作者 → 工作室的工作流深度高级用户可能采用,但组织范围席位不一定扩张收入质量可能落后于技术参与度追问席位增长队列和工作室扩张模式
API / 平台嵌入少数平台交易可能承载过大的战略权重一旦某个工作流变化,合作伙伴集中可能带来渠道风险审查头部平台账户和管线依赖
国际用户集中据报道大多数用户在中国以外,但区域收入结构不透明付费需求可能比全球使用广度暗示的更窄索取按地域拆分的收入结构,以及本地合规 / 支持模式
知名 Logo 营销密度Logo 可能混有试点客户、合作伙伴和参考用户投资人可能只凭品牌名高估客户质量要求每个具名 Logo 提供状态、支出区间和续约日期
直销覆盖偏薄产品驱动带来的效率有吸引力,但可能服务不好大型企业采购流程重的客户里,进一步渗透可能变慢要求企业销售人数、转化率和实施支持数据

核心客户风险不在需求不足,而在漏斗顶部很宽,内部转化和扩张质量却看不清。

[CU016, CU030, CU031, CU032, CU036, CU038]
FU004: 扩张循环与客户质量关口

用户从创作者兴奋转向商业权利、工作流嵌入和团队采用,且没有不可接受的清理或采购摩擦,扩张才会耐久。

[CU030, CU031, CU032, CU033, CU034, CU035]

6.5 佐证材料

Chapter 07

07风险

7.1 监管与法律压力

Tripo 处在当前 AI 版图中合规压力最重的区域之一:与中国有关联的生成式模型业务,拥有全球用户、可下载输出,训练数据来源也不清晰。公开记录已经显示,公司要穿过的规则密度很高。中国监管环境现在覆盖生成式 AI 安全要求、标识执法、网络安全法修订,以及多部门 AI 智能体指引。与此同时,Tripo 自身条款和隐私政策显示,服务处在正式法律框架下,触及企业客户数据,并涉及国际数据传输。这意味着合规负担不是理论问题,而是直接落在客户和产品表面上。 更难的法律问题是 IP。美国版权争论围绕训练数据合理使用、市场损害和输出责任持续变化,但远未落定。近期分析称,训练在某些情形下可能构成合理使用;但盗版或未经授权的来源,或与权利人竞争的输出,仍可能带来责任。由于 Tripo 没有公布其 3D 模型背后数据的清晰来源图,投资人不能把训练数据合法性视为已解决。对一家输出可能与商业 3D 资产重叠的公司来说,这种不确定性不是脚注,而是核心投资判断问题。[CR001, CR002, CR003, CR004, CR005, CR006]

监管 / 法律风险登记表
规则 / 案件 / 问题司法辖区状态发生概率严重程度缓释措施剩余敞口尽调路径
训练数据版权和来源全球 / 以美国为中心的法律外溢公开记录尚未解决记录数据集来源、许可和输出审查政策高:来源未公开披露,判例法仍在变化要求训练数据来源图谱和版权敞口法律备忘录
生成式 AI 安全和评估要求中国已生效且在收紧维持本地合规审查和安全评估流程中高:规则仍在演进审查合规责任归属、备案和内部审计节奏
AI 生成内容标识执法中国2026 年已可见实际执法建设显性 / 隐性标识、元数据和发布控制中高:不合规可能引发监管行动测试产品导出中的标识和审计日志
网络安全 / 跨境数据义务中国 + 全球隐私制度生效中中高数据映射、驻留选项和传输保护高:国际传输和企业客户数据都在范围内要求 DPAs、传输控制和数据本地化设计
客户输出责任全球现实法律风险中高用户政策、审查流程和禁用场景执行中:即使训练合法,客户仍可能生成侵权或不合规输出审查条款执行和内容审核控制
实体 / 合同架构清晰度跨境公司结构仅部分公开明确合同和治理披露中:法律实体和司法辖区选择会影响救济和合规处理要求法律实体图和适用法律选择理由

各行按投资判断中的严重程度排序,而不是按时间排序;关键在于法律不确定性会如何随模型规模和企业使用叠加。

[CR001, CR002, CR003, CR004, CR006, CR007]

7.2 地缘政治、算力与依赖风险

Tripo 的算力风险并不抽象。美国贸易和出口管制来源明确显示,先进计算限制正直接瞄准放慢中国获得顶级芯片及配套 AI 应用的速度。指引已经延伸到总部位于受限地区、但在其他地方运营的实体;美国官员也持续用军民融合和执法视角看待与中国有关联的交易对手。对一家耗算力的生成式 3D 公司来说,实际后果很清楚:获得最佳芯片更难或更慢,会传导到模型质量、推理成本、产品延迟和发布节奏。 依赖故事不止硬件。Tripo 还依赖云交付、插件生态和下游工作流信任。与此同时,巨型平台公司正在整合多模态模型目录、定价弹性和开发者工具。Microsoft Foundry 和 Azure OpenAI 展示了既有巨头如何把前沿模型、可观测性和吞吐定价打包,从而让单点方案更难防守。OpenAI 的 Sora 时代模型阵容也从另一方向强化同一风险:Tripo 不只与其他 3D 专家公司竞争,还要与更大的模型平台竞争;后者可能在客户提出专门 3D 供应商需求前,就先拿走预算份额。[CR010, CR011, CR012, CR013, CR029, CR030]

合作伙伴 / 依赖风险登记表
依赖项交易对手 / 系统作用集中度失效场景严重程度缓释措施剩余敞口
先进 GPU 获取受出口管制的芯片和云生态基础模型训练和推理新的出口限制拖慢升级,或推高单次生成成本分散供应商、优化模型、预购算力、制定应急方案
中国合规姿态CAC / MIIT / 跨境监管机构规则制定、标识、备案、执法产品变更或执法行动中断分发或导出流程设立专门合规责任人,并在设计中内置标识中高
大型多模态平台OpenAI / Microsoft / Azure 生态预算竞争、功能打包、企业采购杠杆客户选择更大的产品目录和平台定价,而不是专家工具更深嵌入 3D 专属工作流和质量优势
创意套件既有巨头Adobe / NVIDIA 相邻视觉模型栈分发、品牌、集成工作流能力中高既有巨头补齐功能差距,压塌独立工具的付费意愿在需要专业 3D 工作流深度的场景胜出中高
插件和合作伙伴生态Blender / ComfyUI / 合作伙伴嵌入技术分发和客户工作流集成适配器漂移或伙伴变化打断高粘性工作流采用主动维护和官方支持
具名战略客户和合作伙伴Sony、NetEase、MakerWorld、Replit 等证明、分发、市场信号一两个 Logo 的重要性可能高于公司承认的程度中高扩大生产客户基础,并厘清状态组合中高

最重要的依赖是算力;最显眼的市场依赖,是大型平台塑造客户预期和价格天花板。

[CR010, CR011, CR012, CR013, CR029, CR030]
FR002: 风险传导图

多数下行路径从监管、算力或质量问题开始,传导到延迟、客户信任、转化,最终压缩估值。

[CR007, CR010, CR018, CR023, CR033, CR034]
FR003: 依赖图

Tripo 同时依赖监管者、算力获取、生态系统,以及少数信号很强的合作伙伴或客户。

[CR010, CR011, CR029, CR030, CR031, CR032]

7.3 产品质量与商业模式风险

主要运营风险在于,Tripo 可能足以让用户惊艳,但资产进入真实生产工作流后,不一定足以把用户留住。独立评测反复把产品描述为快速、有用,但仍更适合作为工作流加速器,而不是技术美术或既有资产管线的完全可信替代。质量警示在复杂设计、绑定、UV 清洁度和最终导出可靠性上最严重。这并非 Tripo 独有,而是整个品类的问题;但当投资人为差异化执行付费时,品类共性问题仍然重要。 商业风险会放大质量风险。Tripo 的免费增值结构和商业权利门槛,构成了一个合理的创作者漏斗;但也可能形成陷阱:数百万用户制造大量噪音,只有很小一部分转化为耐久收入。如果平台被嵌入工作流,更大的付费层可以扩大变现;如果产品仍然需要大量清理,它们也会制造价格摩擦。公开采用指标仍然足够嘈杂,投资人还无法判断这组权衡哪一边占优。因此,客户质量尽调而不是虚荣增长,仍是档案中的关键商业风险。[CR018, CR019, CR020, CR021, CR022, CR023]

运营 / 质量 / 安全风险登记表
失效模式发生概率严重程度缓释成熟度剩余敞口未解决缺口
导出资产在高难生产场景仍需要大量清理需要复杂角色和真实管线的第三方基准通过率
免费增值漏斗顶部无法转化为稳定付费客群中高中低需要分群转化、流失和 ARPA 数据
全球数据处理跑在合规控制前面中低需要企业级数据驻留、审计和响应控制
客户质量指标噪声太大,难以支撑清晰投资判断中高需要标准 KPI 定义和收入质量仪表盘
销售 / 支持覆盖偏薄,消化不了企业部署复杂度中低中高需要实施、支持和 CS 组织细节
公开评价热情掩盖生产使用不满需要续约、客户访谈和部署后满意度数据

运营风险高度耦合:转化弱叠加质量弱,会同时拖累增长和利润率。

[CR018, CR019, CR020, CR021, CR022, CR023]
FR001: 风险热力图

最高严重度风险集中在算力获取、法律不确定性和变现质量,而不是某个简单产品 bug。

[CR017, CR023, CR027, CR029, CR033, CR040]

7.4 团队、执行与止损标准

Tripo 的执行风险与人员风险紧密相连。公开报道把 Simon Song 放在创始人、运营者和对外叙事者的中心位置,而更广的接班梯队仍不清楚。这种集中可能帮助速度,但如果创始人分心、离开,或无法招募并留住前沿 3D 平台所需的研究和基础设施负责人,也会增加脆弱性。更广泛的 AI 市场已经人才紧缺;热门创业公司在景气期或许还能应对,一旦产品失手或政策挤压,就会很痛。 对投资人来说,关键问题不是这些风险是否存在——它们显然存在——而是哪些风险最可能先打破投资论点。信号最强的触发器很直接:可见的监管行动、芯片获取明显恶化、无法把头条规模转化为付费质量、生产质量基准恶化,或领导层流失且没有可见梯队。Tripo 确实有一些可见缓释因素——开放研究信号、活跃发版、伙伴界面和创作者驱动分发——但还不足以把残余风险压低。它仍是高风险、高上行的运营画像,需要扎实的私下尽调之后,才能有信心支撑价格。[CR025, CR026, CR027, CR028, CR041, CR042]

人员 / 执行风险登记表
角色 / 职能依赖或缺口发生概率严重程度缓释措施尽调路径
创始人 / CEO(Simon Song)创始人是核心技术叙事者和战略门面中高拓宽领导层梯队,并让被授权的运营负责人对外可见要求继任计划和具名领导层名单
研究 / 基础设施人才全球 AI 人才短缺推高招聘和留任压力留任激励、使命吸引力和研究品牌信号要求流失率、录用邀约接受率和关键招聘管线数据
销售 / 客户成功公开证据显示 GTM 人员配置很精简中高在企业客户推进前先补实施和 CS 能力要求人数和企业支持模型
合规 / 安全负责人公开资料看不到强企业级控制界面中高设置专门隐私、法律和合规负责人要求安全 / 合规组织图和汇报线
范围纪律垂直场景押注太多,会拆散执行力优先投入商业化最高的工作流向管理层索取资源分配和路线图评分模型

人员风险不只在创始人集中度;还在于规模跑过流程之前,公司能否补齐合规、企业支持和基础设施岗位。

[CR025, CR026, CR027, CR028]
缓释与放弃标准表
风险可监控触发信号阈值 / 事件行动含义
算力 / 出口管制风险规则变化后延迟或成本恶化产品发布明显变慢、推理成本上升,或模型路线图下调重新评估利润率假设,并暂停支持溢价估值
监管 / 标识风险正式监管问询或可见不合规事件任何与标识、隐私或合成内容处理相关的 CAC 式执法事件升级法律尽调,压力测试中国敞口,并下调信心
客户质量风险免费转付费弱,或续约数据低转化显著低于管理层叙事,或 NRR 不透明下调增长质量判断,并推动更低进入价格
产品质量风险核心资产类别第三方基准失败复杂角色或引擎导入测试显示清理负担重把 Tripo 视为概念工具,而非默认生产标准
人员风险创始人离开或高级人才流失激增Simon Song 或多名关键模型 / 基础设施负责人流失在继任和执行连续性得到证明前,暂停投资信念
指标不透明风险管理层无法对齐用户 / 开发者 / 客户分母没有标准 KPI 定义,或董事会层面指标不一致假设商业化质量更弱,并扩大下行情景

这些放弃标准服务于投资纪律,而不是公司运营;每一条都把风险信号映射到投资应对。

[CR023, CR025, CR033, CR041, CR042]

7.5 佐证材料

Chapter 08

08估值

8.1 当前价格锚与商业化表面

公开记录已经足以验证,Tripo 是一个真实融资故事,而不是传闻。2026 年 6 月的官方发布、SaaS 行业报道、AI Weekly 和 Forbes 指向同一幅大体图景:VAST 完成约 $200 million 的 A+ 和 A++ 融资,借此推出世界模型路线图,并已经被讨论为独角兽。Forbes 是最干净的可得估值锚,因为它把该轮融资与包括 INCE Capital 在内的财团联系起来,并称一位知情人士给出的公司估值约为 $1.5 billion。确切投后条款仍有一些模糊,但足以把 $1.5 billion 作为尽调使用的工作市场价格。 商业化表面也比纯 demo 叙事更真实。Tripo 定价页显示出明确的免费增值漏斗,然后清楚抬升到付费商业权利、更高并发、批量导出和私有模型。API 页面和官方垂直案例研究又显示出第二层变现:开发者使用、伙伴嵌入和工作室工作流。问题不是缺少可见变现表面,而是公开档案从未告诉投资人,1000 万用户或 90,000 个工作室客户中到底有多少在付费、花多少钱,或产品是否能产生软件式利润率。[CV001, CV002, CV003, CV004, CV005, CV006]

建议摘要表
维度当前观点重要性信心
建议跟踪公司具备战略吸引力,但公开材料仍太薄,不足以在当前估值下给出买入判断。
估值立场偏高如果 Tripo 已经达到约 $100M 以上 ARR 且留存强,这个价格可以成立;但公开证据尚未证明。
主要支撑产品驱动商业化已可见定价、API、企业用例和具名 Logo 显示,这不是研究演示,而是真实商业界面。
主要担忧分母缺失ARR、留存、毛利率和股权结构条款都未见于公开记录。
最大外部风险中国折价出口管制、生成式 AI 合规和更薄的退出流动性,可能在需求转弱前先压缩倍数。
上调触发点经营经济性披露要进入可投资区间,需要 ARR、留存、毛利率和付费企业客户质量的硬证据。

这个判断刻意对价格敏感,并不意味着 Tripo 是弱公司。

[CV010, CV011, CV013, CV039, CV040, CV041]
正向观点 / 反向观点表
维度正向论点反向论点改变观点的证据
采用10M 用户、90K 工作室客户和具名 Logo,意味着真实的产品市场拉力。这些数字没有披露多少用户在付费,也没有披露贡献了多少收入。按套餐划分的付费账户分群和分客群 ARR
商业化免费增值、API 和企业工作流打包,拼出多条收入界面。收入界面很宽,仍可能掩盖弱转化或重折扣。实际 ACV 区间、ARPA 和转化漏斗
品类溢价创意 AI 和世界模型相邻资产,仍可能拿到私募市场溢价倍数。增长质量或流动性支撑变弱时,溢价倍数会很快压缩。经验证的增长质量,以及持久退出需求证据
竞争位置在速度、3D 专业化和集成很重要的细分市场,Tripo 有动能。Adobe、Roblox 生态、Unity 工具和 AI 新进入者,会随时间削弱稀缺性。相对替代方案的输赢数据和续约证明
中国语境2026 年,境内资本已激进重估中国 AI 胜出者。中国注册地也带来出口管制、合规和地缘政治折价。可信的算力和合规应急方案

反向观点主要指向估值支撑和证据质量,而不是质疑 Tripo 是否做出了有用产品。

[CV004, CV005, CV010, CV011, CV013, CV029]
FV001: 建议逻辑

决策路径从真实融资标尺和可见产品动能出发,经过分母不透明和中国相关折价,落到「跟踪」立场。

[CV004, CV005, CV010, CV013, CV033, CV039]

8.2 可比公司语境与反推收入约束

评估 Tripo 估值最诚实的方式,是先看市场语境,再反推出公司需要的收入分母。Multiples.vc 和 Acquiry 都认为,2026 年软件倍数仍然分化:AI 原生资产仍可拿到真实溢价,但买方非常看重增长质量、留存和盈利能力。公开创意与 3D 邻近可比公司让这个判断落地。按市值口径,Adobe 当前收入倍数约 3.6x,Roblox 和 Unity 更接近 7x 到 8x。这些同行并不完美,但为创意软件、创作者生态和 3D 工作流基础设施提供了有用的公开市场边界。 最关键的前沿 AI 私募可比对象是 Runway。TechCrunch 报道了该公司 2025 年 $300 million 年化收入目标,以及 2026 年 $5.3 billion 估值。这不是干净的同口径倍数,因为分母是目标而不是审计数字;但它确实说明,稀缺创意 AI 只要靠近世界模型,投资人有时愿意支付十几倍后段的倍数。从这个视角看,Tripo 的 $1.5 billion 并不荒唐。不过,它要求背后有相当可观的隐含收入基础:20x 时约 $75 million,15x 时约 $100 million,12x 时约 $125 million,10x 时约 $150 million。没有披露,这个分母仍然只是猜测。[CV014, CV015, CV016, CV017, CV018, CV019]

乐观 / 基准 / 悲观情景表
情景可支撑倍数区间支撑 $1.5B 所需 ARR 或收入必须成立的前提会打破情景的因素
乐观15x-20x$75M-$100MTripo 已经具备可支撑高端估值的 ARR 规模、强留存、高毛利软件经济性,以及 3D 生成工作流里的真实稀缺性。隐藏的收入分母被证实更小,或者世界模型叙事无法变现。
基准情景10x-12x$125M-$150M采用是真实的,但投资人仍会给私营公司不透明度和中国风险打折。企业客户占比、利润率或留存明显弱于叙事暗示。
悲观情景8x-10x$150M-$187.5M市场更像给一款竞争激烈的应用软件定价,并叠加显著监管和流动性折价。免费转付费或计算成本恶化到足以削弱高效增长。

公开收入未披露,用反推阈值比假装精确的 DCF 更站得住。

[CV015, CV023, CV024, CV025, CV026, CV027]
可比估值表
参考对象类型当前或公开状态为什么有用局限
Adobe上市可比公司July 2026 市值 ~$88.5B,TTM 收入 ~$24.45B展示规模化创意软件在公开市场中的披露和变现水平。体量大得多、业务更分散,也不是前沿 AI 初创公司。
Roblox上市可比公司July 2026 市值 ~$40.38B,TTM 收入 ~$5.29B为创作者经济、UGC 和 3D 生态提供参照。平台经济性与 Tripo 的软件栈差异很大。
Unity上市可比公司July 2026 市值 ~$13.4B,TTM 收入 ~$1.92B可参照引擎邻近的 3D 工作流基础设施。Unity 是成熟上市平台,变现和成本结构不同。
Runway私营创意 AI 可比公司Feb 2026 估值 $5.3B;此前讨论过 ~$300M 年化收入目标高端创意 AI 中最好的私营可比对象,并带有世界模型邻近性。收入分母不是经审计的公开披露。
Moonshot / MiniMax / Zhipu中国 AI 情绪可比对象2026 年报道中,Moonshot 私募估值 $20B,同业公开估值 $13B-$55.9B说明中国 AI 赢家在 2026 年可能大幅重估。前沿模型实验室和上市股票,与私营 3D 应用软件结构不同。
2026 软件倍数基准分析师市场视角AI 原生 SaaS 常被引用在约 8x-15x ARR,高增长公司更高有助于把 Tripo 隐藏的分母转换为可支撑的倍数区间。基准区间无法弥补 Tripo 缺失披露。

这组可比对象用于设定合理边界条件,不是为了机械算出精确公允价值。

[CV014, CV015, CV017, CV018, CV019, CV020]
FV002: 估值敏感性

固定 $1.5B 估值对应的 ARR 门槛差异很大,取决于最终哪一个倍数站得住。

数值是用 $1.5B 除以各倍数倒推出的门槛,只作为情景工具,不是 Tripo 披露指标。

[CV024, CV025, CV026, CV027, CV028]
FV003: 估值 / 回报区间

一旦投资人纳入信息不透明、质量和涉华折价,可支撑的倍数区间会迅速收窄。

区间描述的是可支撑收入倍数带,不是 Tripo 股权价值的点估计。

[CV015, CV023, CV036, CV041]

8.3 中国 AI 语境,以及为什么溢价仍要打折

确有为溢价买单的牛市逻辑。2026 年,中国 AI 资本市场愿意给类别龙头打出极高估值:Moonshot 以 $20 billion 估值融资,CNBC 和 TechCrunch 描述的 Zhipu 与 MiniMax 公开市场价值,也会随着模型发布改变情绪而大幅移动。Tripo 也有可信的增长可选性。Forbes、Yahoo/SCMP、ACCESS 和官方新闻稿合在一起,描绘的是一家公司正从快速用户增长,走向更大的世界模型、内容和生态野心。换句话说,溢价叙事不是虚构。Tripo 是少数拥有可见消费者漏斗、创作者心智和企业 logo 的中国应用 AI 公司之一。 但同一叙事需要纪律性打折。Moonshot、Zhipu 和 MiniMax 更接近前沿模型融资故事,而不是私有 3D 应用栈。Tripo 还暴露在中国特有的监管和出口管制风险下,可能冲击算力成本、全球企业胃口和投资人退出假设。CNBC 的 AI 退出报道给出第二个折扣:资本充裕,但流动性仍然稀薄且不均衡。最后,独立评测仍把 Tripo 描述为常常受益于清理的工作流加速器,而不是通用生产替代品。这些事实不会杀死估值逻辑,但会让投资人更难在没有不透明折扣的情况下,支付干净的前沿 AI 倍数。[CV029, CV030, CV031, CV032, CV033, CV034]

投资逻辑破裂与终止触发项表
触发项阈值或事件对投资逻辑的传导行动含义
ARR 分母不及预期经验证的 ARR 明显低于 ~$75M-$100M当前高端估值叙事会立刻失去最清晰的支撑区间。下调至回避 / 放弃,除非价格重置。
付费转化偏弱庞大用户基数难以转成付费队列或企业 ACV10M 用户会变成虚荣指标,而不是估值支撑。出资前要求完整队列数据。
毛利率或推理成本偏弱高计算成本让利润率明显低于软件级水平收入质量弱,AI 原生溢价随之压缩。要求提供利润率桥接,并审查供应商集中度。
中国监管或出口管制冲击算力获取收紧,或跨境商业化变得更难即使客户需求未崩,估值倍数也可能收缩。提高折现率,或暂停投资判断。
结构化轮次压力优先股堆叠或附带条款撑高名义价格报道估值不再是可靠的公允价值锚。按普通股等价口径重算回报。

这些触发项可监控,设计目的在于机械地打破投资逻辑,而不是被情绪左右。

[CV024, CV025, CV028, CV033, CV036, CV042]
FV004: 投资关键指标

投委会式记分卡,覆盖市场动能、产品验证、经济性可见度、风险和估值支撑。

分数是为投委会框架服务的分析判断,不是标准化外部评级。分数越高越好。

[CV005, CV010, CV013, CV033, CV037, CV038]

8.4 建议与最终尽调问题

本章结论应该谨慎,但不应轻易否定。Tripo 有足够的产品广度、采用证明和可比市场支撑,可以留在投资人的主动跟踪名单上;当前价格也不像某些纯炒作 AI 轮次那样明显不理性。然而,缺失分母的问题具有决定性。投资人仍看不到 ARR、付费 logo 质量、cohort 留存、毛利率,或估值背后的真实清算优先权结构。这意味着市场价格也许说得过去,但仅凭公开证据,公允价值仍未充分证明。 因此,可支撑的建议是跟踪,而不是买入。估值是偏紧,而非明显昂贵,因为真实动能存在,Runway 式私募可比对象也说明,市场有时会为稀缺创意 AI 支付上沿倍数。置信度应保持中等,风险保持高位,因为下行路径很容易描述:如果 Tripo 的 ARR 仍远低于约 $100 million,如果免费到付费转化较弱,或如果中国因素引发的算力和合规冲击加剧,估值压缩风险会很大。接下来的尽调任务很明确:核实分母,检查留存和利润率质量,并弄清当前估值是由经济性支撑,还是由轮次结构撑住。[CV039, CV040, CV041, CV042, CV043, CV044]

最终尽调请求表
主题缺失证据为什么重要负责人或尽调路径
收入分母ARR、月经常性收入、确认收入与过去十二个月收入桥接这是判断 $1.5B 究竟合理、偏高还是昂贵的唯一门槛变量。管理层资料室和董事会 KPI 包
留存与扩张NRR、GRR、队列留存、企业续约,以及席位或用量扩张曲线高倍数靠持续扩张支撑,不只是客户 logo 堆积。财务和 GTM 尽调
利润率质量按产品拆分的毛利率、每个生成资产的推理成本,以及云或 GPU 集中度收入成本一重,AI 原生倍数会很快压缩。工程与财务尽调
股权结构表与轮次结构优先股堆叠、清算优先权、反稀释棘轮、员工老股条款和稀释历史名义价格可能不等于普通股价值。法律顾问与融资备忘录审阅
客户质量具名付费客户、ACV 区间、企业客户收入占比和免费转付费只有底层收入质量足够强,10M 用户才有投资价值。销售运营、产品分析和客户访谈核验

这些请求正好补齐当前阻碍更高确信度建议的缺口。

[CV013, CV016, CV024, CV028, CV040, CV043]

免责声明

本报告仅供信息参考,基于截至 2026-07-09 的公开来源,不构成投资建议。Tripo AI / VAST 是私营公司,许多对投资判断至关重要的指标尚未经审计或未披露, 因此所有财务结论都应独立验证。

证据索引

结论
编号陈述可信度来源
CO001 Tripo AI is the flagship customer-facing product brand within a broader VAST / Holymolly corporate stack rather than a plainly single-named operating company. SO001, SO005, SO013
CO002 The official tripo3d.ai terms identify Holymolly Ltd as the service provider behind the Tripo websites and services. SO005
CO003 The same terms page lists a Hong Kong office address for notices to the company. SO005
CO004 VAST / Tripo AI was founded in 2023. SO007, SO013, SO015
CO005 SCMP-republished reporting says Tripo AI was registered as Vast in the Cayman Islands and set up in Beijing shortly after Google released DreamFusion. SO015
CO006 The same SCMP-republished reporting says Tripo AI has a research centre in Beijing. SO015
CO007 CB Insights instead lists Tripo AI as based in Shanghai, creating a live headquarters conflict in the public record. SO016
CO008 Simon Song is the founder and CEO publicly associated with VAST and Tripo AI across official and independent sources. SO013, SO015
CO009 Forbes reports that Simon Song studied economics and international studies at Johns Hopkins University. SO013
CO010 Forbes and SCMP report that Simon Song worked at SenseTime before founding VAST. SO013, SO015
CO011 Forbes and SCMP report that Simon Song co-founded MiniMax and left in 2022 before starting VAST. SO013, SO015
CO012 Official Tripo materials say the platform can generate 3D assets from text prompts, images, sketches, and multi-view inputs in seconds. SO001, SO008, SO010
CO013 Public descriptions consistently place Tripo use cases in gaming, animation, advertising, 3D printing, industrial design, XR, and digital commerce. SO009, SO014, SO019
CO014 The official pricing page shows Free, Pro, Max, and Team subscription tiers. SO002
CO015 The free tier provides 200 monthly credits, one concurrent task, and public models under CC BY 4.0 terms. SO002
CO016 The Pro tier offers 3,000 monthly credits, multi-view-to-3D, parts generation, Smart Mesh, private models, and commercial use. SO002
CO017 The Max and Team tiers scale to 25,000-45,000 monthly credits and 100-200 concurrent tasks for heavier production usage. SO002
CO018 The privacy policy says Tripo collects user-generated content, device and usage data, and may transfer data internationally. SO004
CO019 The terms say outputs may be inaccurate, incomplete, or not representative of real-world people, places, or facts and should not be treated as definitive truth. SO005
CO020 Official 2025-2026 product updates add Ultra HD Texture, texture upscaling, 100+ mocap-quality biped animations, lock-frame export, automatic cleanup, and format conversion. SO006
CO021 Official workflow pages show Tripo integrating into ComfyUI, Blender-centered workflows, and broader export toolchains rather than remaining a closed web toy. SO008, SO010
CO022 TripoSR is a public single-image 3D reconstruction model with an arXiv paper and GitHub repository associated with VAST AI Research. SO022, SO024
CO023 TripoSG is presented in arXiv as a high-fidelity 3D shape-synthesis model built on large-scale rectified flow methods. SO023
CO024 Forbes reported in September 2025 that Tripo had more than three million professional users worldwide and more than 40,000 studios and corporate partners. SO014
CO025 ACCESS Newswire said in September 2025 that Tripo served more than three million global creators, 35,000 active developers, and 700+ enterprise customers. SO020
CO026 SCMP-republished reporting says Tripo AI's user base more than doubled from three million in August to 6.5 million by the time of the interview. SO015
CO027 More than 85% of Tripo AI's users are reported to be outside China, with Europe and the United States as the biggest markets. SO015
CO028 SCMP-republished reporting names Tencent, NetEase, Sony, Microsoft, and Pop Mart among Tripo AI's prominent enterprise users. SO015
CO029 SCMP-republished reporting says Tripo AI had only two salespeople at that point and relied primarily on word-of-mouth and organic growth. SO015
CO030 Forbes reported that VAST raised about $200 million in June 2026 Series A+ and Series A++ financing from investors including INCE Capital, Genesis Capital, and Primavera Capital Group. SO013
CO031 The same Forbes report says the June 2026 financing implied at least a $1 billion valuation and that one informed source put VAST at about $1.5 billion. SO013
CO032 Forbes says VAST raised $50 million in a March 2026 Series A led by Alibaba and Hengxu Capital. SO013
CO033 Auganix says the March 2026 funding was backed by Alibaba and Baidu Ventures and was earmarked for large-scale 3D foundation models and the global developer platform. SO019
CO034 AI Weekly says the June 2026 round was led by Ince Capital with a China Life-affiliated fund and that Tripo had 10 million individual users and 90,000 studio clients. SO018
CO035 The official Sony collaboration page says Sony and VAST are working together on glasses-free 3D display, AI-generated content, and interactive 3D experiences. SO007
CO036 The September 2025 ACCESS Newswire release says Tripo 3.0 carried more than 20 billion parameters, 300% enhanced detail accuracy, and an ecosystem of 18+ open-source projects. SO020
CO037 Forbes reported that VAST charges enterprises such as NetEase and Sony on a project basis while charging end users monthly subscription fees from roughly $20 to $140. SO013
CO038 The independent Medium teardown concludes that Tripo should be treated as a workflow accelerator or concepting layer rather than an outright replacement for production 3D pipelines. SO021
CO039 The same teardown warns that exported assets can expose mesh, UV, material, or rigging problems that are not obvious in a web preview. SO021
CO040 InforCapital adds claims about a 50 million-asset dataset, 40,000+ enterprise clients, $12M ARR, and a later July 2026 round, but the source should be treated as low-confidence supplementary data rather than a canonical ledger. SO017
CO041 Public scale disclosures conflict materially, ranging from 6.5 million creators to 10 million individual users and 20 million global users, so exact current scale is directionally large but not independently audited. SO013, SO015, SO018, SO019
CO042 Public materials do not disclose a robust board roster, current executive bench, or clear control-rights picture commensurate with a freshly reported unicorn valuation.
CO043 Current revenue, ARR, burn, runway, and exact headcount remain under-disclosed in high-reputation public sources.
CO044 Project Eden and the W1.0 world-model initiative show VAST trying to expand from 3D asset generation toward simulated spatial environments and world models. SO013, SO019
CO045 Official and independent sources consistently support that Tripo is more mature than a toy demo, but independently benchmarked proof of production readiness remains thinner than the company's own marketing language. SO011, SO012, SO014, SO021
CM001 Tripo’s relevant market is AI-native 3D asset-creation and workflow software, not the whole AI economy. SM001, SM003, SM015
CM002 Included spend covers text/image/sketch-to-3D generation, texturing, mesh cleanup, rigging, export, workflow plugins, and API-based asset delivery. SM001, SM003, SM006, SM007
CM003 Excluded spend includes generic LLM subscriptions, pure 2D image generation, hardware-only AR/VR sales, and general cloud infrastructure that does not directly buy 3D workflow output. SM013, SM015, SM016, SM018
CM004 Tripo-adjacent demand pools include gaming, animation, XR/spatial computing, e-commerce visualization, industrial design, education, and 3D printing. SM010, SM012, SM015
CM005 Axis Intelligence says China’s conservative AI market revenue was about $28–31 billion in 2025 and could reach about $202 billion by 2032 at a 32.5% CAGR. SM013
CM006 Axis also highlights that public China AI market baselines vary widely, from roughly $21.6 billion to $62 billion, depending on whether the perimeter is narrow revenue or broad AI-related activity. SM013
CM007 Grand View’s China generative AI horizon databook says software represented 63.9% of market revenue in 2024. SM014
CM008 The same Grand View material frames AR/VR development and regulation as major growth-shaping forces inside the China generative AI market. SM014
CM009 Research and Markets defines the generative AI for 3D assets category across software, hardware, and services, and across end-user groups including media, gaming, architecture, e-commerce, and education. SM015
CM010 Research and Markets identifies text-to-3D acceleration, real-time 3D asset iteration, synthetic data generation, and cloud-based 3D production pipelines as major trends in the category. SM015
CM011 China Daily, citing IDC, says China’s AR/VR market should grow at a 43.8% CAGR from 2022 to 2026 and reach $13.1 billion in spending by 2026. SM016
CM012 IDC’s China AR/VR lens says VR accounts for 70% of investment and gaming is the leading VR use scenario at roughly one-third of use. SM016
CM013 Consumer-focused goods are projected to make up nearly 40% of China’s AR/VR market by 2026. SM016
CM014 Digital in Asia says China’s digital economy reached about 54 trillion yuan (~$7.5 trillion) in 2023. SM018
CM015 The same market overview says China’s e-commerce market is expected to reach about $2.93 trillion by 2025 and live commerce could reach 8.16 trillion yuan by 2026. SM018
CM016 Digital in Asia says China’s gaming market reached 350.8 billion yuan ($49.8 billion) in 2025. SM018
CM017 China’s 5G, cloud, and digital-infrastructure scale lowers distribution friction for rich 3D and AI content relative to thinner ecosystems. SM016, SM018, SM021
CM018 Official Tripo commerce materials claim that interactive 3D product assets can reduce buyer uncertainty, improve engagement, and lower return rates relative to static photos. SM004, SM008
CM019 Official Tripo commerce and design pages emphasize GLB, USDZ, FBX, OBJ, STL, and related export formats for storefronts, AR, and downstream rendering tools. SM008, SM009
CM020 Official Tripo design pages say browser-based studios help independent designers cut early drafting and concepting time. SM009
CM021 Official Tripo gaming pages claim AI 3D can compress manual game-asset creation cycles from weeks to seconds or minutes. SM006
CM022 The same gaming materials emphasize segmentation, auto-rigging, and smart low-poly retopology as core requirements for real-time game deployment. SM006
CM023 Official API workflow materials say the Tripo API is a separate product line from Tripo Studio and is intended for bulk, programmable enterprise integration. SM003, SM007
CM024 Official workflow materials say DCC bridges, REST APIs, and standard exports reduce friction for moving generated assets into Unity, Unreal, and other downstream systems. SM007, SM024
CM025 Forbes reported that Tripo’s API powers many startups offering AI 3D while Tripo Studio serves studios, game developers, and XR companies. SM010
CM026 SCMP-republished reporting says more than 85% of Tripo users are outside China, meaning global demand matters more than a China-only TAM story. SM011
CM027 The practical buyer map spans creators, studios, enterprise developers, commerce teams, independent designers, and industrial or XR builders. SM002, SM003, SM004, SM006, SM009, SM010
CM028 Budget ownership varies materially by segment: creators self-serve on cards, studios buy from art or production budgets, enterprise API customers buy from engineering or product infrastructure budgets, and commerce teams buy from merchandising or growth budgets. SM002, SM003, SM007, SM008, SM009
CM029 The strongest gaming adoption trigger is faster prototype-to-engine cycles plus lower manual retopology and rigging labor. SM006, SM007, SM010
CM030 The strongest e-commerce adoption trigger is the promise of better conversion and lower returns through spatial product visualization and AR-ready assets. SM004, SM008, SM018
CM031 Industrial design and XR uses are attractive because they need rapid concept iteration, but they impose tighter tolerance for geometry, scale, and export errors than hobbyist creation. SM010, SM012, SM023
CM032 Manual Blender/Maya/CAD work, photogrammetry, service studios, and stock-asset libraries remain the status-quo substitutes that Tripo must displace. SM006, SM008, SM009
CM033 Large Chinese AI, gaming, e-commerce, and digital infrastructure pools make the macro demand backdrop supportive for 3D automation tools. SM013, SM016, SM018
CM034 Standard export formats, DCC bridges, and API infrastructure reduce switching cost from experimentation to production. SM007, SM008, SM009, SM024
CM035 No fetched public source isolates a clean China or global SAM specifically for AI-native 3D asset generation, so the market story has to rely on layered adjacent lenses.
CM036 Market-size estimates vary materially by perimeter, so a single giant AI TAM slide would overstate the direct market Tripo can monetize. SM013, SM014
CM037 China’s 2023 generative-AI measures require lawful data, user-log protections, provider duties, and algorithm filing or security assessment in certain cases. SM019
CM038 China’s 2025 synthetic-content labeling rules require explicit and implicit labels, metadata, and propagation handling when generated content is downloaded or distributed. SM020
CM039 China’s May 2026 AI-agent policy direction emphasizes application-driven growth, standards, and safety/controllability rather than pure speed. SM021
CM040 The market is attractive, but vendors will win on workflow depth, interoperability, and compliance readiness rather than on demo quality alone. SM007, SM019, SM020, SM021
CP001 Tripo presents itself as an AI-native, all-in-one 3D creation workbench rather than as a single-purpose model endpoint. SP001, SP004
CP002 Tripo publicly separates Studio pricing from an API product and exposes a ladder from Free to Team tiers. SP002, SP003
CP003 Tripo's buyer alternatives split into direct AI-3D peers, broader multimodal creative platforms, and status-quo manual or internal workflows. SP001, SP005, SP009, SP013, SP024
CP004 Meshy markets text/image-to-3D creation, export-ready workflows, API access, plugins, and enterprise administration on its official surfaces. SP005
CP005 Meshy's public pricing page lists Free, Pro, Studio, and Enterprise plans with explicit monthly price points for Pro and Studio. SP006
CP006 Meshy claims it surpassed $15 million ARR, maintained 30% month-over-month growth in 2025, reached 6 million users, and operates at 85% gross margin. SP007
CP007 Meshy's 2026 blog shows rapid shipping across 3D agents, workspaces, ComfyUI workflow, game-development, and 3D-printing features. SP008
CP008 Luma positions itself as a creative-agents platform spanning video, image, audio, and text with a mission around operating in the physical world. SP009
CP009 Luma's pricing is credits-based, includes team and enterprise plans with SSO, and exposes Ray3.14 cost per second rather than a simple seat-only model. SP010
CP010 Luma's Ray3.2 and Ray3.14 releases emphasize frame-level creative control, HDR or EXR output, 1080p delivery, and API access for professional video workflows. SP011, SP012
CP011 Adobe Firefly spans image, video, audio, and vector generation and includes partner models from Google, OpenAI, Luma, Runway, and others inside one product surface. SP013
CP012 Adobe pairs Firefly with commercial-safety messaging, Content Credentials, and adjacency to the broader Creative Cloud and Substance 3D stack. SP013, SP015
CP013 Stability AI's platform markets 3D and 4D creative production to enterprises and developers rather than only to hobbyist users. SP016
CP014 Stability says Stable Fast 3D can turn a single image into a UV-unwrapped 3D asset with materials in about 0.5 seconds for gaming, VR, retail, architecture, and design use cases. SP017
CP015 Stability says SPAR3D provides real-time point-cloud editing and complete 3D structure generation from a single image in under a second, with NVIDIA RTX AI PC positioning. SP018, SP021
CP016 Stability's official Stable Fast 3D GitHub repository explicitly says the model is based on TripoSR while adding UV unwrapping, illumination disentanglement, and material-parameter prediction. SP019
CP017 OpenAI says the deployed version of Sora still has limitations including unrealistic physics and struggles with complex long-duration actions. SP023
CP018 OpenAI says the Sora web and app experiences were discontinued on April 26, 2026 and the Sora API will be discontinued on September 24, 2026. SP022
CP019 Microsoft Foundry positions Sora 2 as a text-to-video, image-to-video, and video-to-video model with audio, remix, asynchronous jobs, Responsible AI protections, and per-second billing. SP024
CP020 NVIDIA matters less as a standalone SaaS rival than as a distribution amplifier for 3D models optimized around RTX hardware and partner ecosystems. SP018, SP021
CP021 CB Insights places Tripo in a 3D asset creation landscape that includes other peers such as Kaedim, Kinetix, and Spuree, showing the market is broader than a Tripo-versus-Meshy framing. SP025
CP022 Dedicated 3D-native rivals are materially closer substitutes for Tripo's core workflow than Luma, Adobe, or Sora are today. SP001, SP005, SP017, SP024
CP023 Adobe, Luma, and OpenAI compete more through distribution, trust, and adjacent creative workflow than through dedicated 3D mesh parity in the retained evidence. SP009, SP013, SP024
CP024 Tripo's pricing and home pages emphasize 3D-specific workflow features including mesh quality, low-poly handling, animation, batch generation, and model downloads. SP001, SP002
CP025 Meshy's official site highlights enterprise controls such as SOC 2 Type II, ISO 27001, GDPR, SSO via SAML, shared workspaces, and centralized billing. SP005
CP026 Luma and Adobe can win budgets where the deliverable is campaign video, branded content, or multimodal creative output rather than an exportable 3D asset. SP009, SP011, SP013
CP027 Adobe's free-plus-paid generative-credit model and commercial-safety story can reduce procurement friction relative to pure-play AI 3D startups. SP013, SP014
CP028 Stability's community-license and GitHub distribution increase the risk that baseline single-image 3D reconstruction becomes commoditized faster than closed pure plays expect. SP017, SP018, SP019, SP020
CP029 Tripo still differentiates publicly through 3D-native workflow breadth and a spatial-intelligence identity focused on asset creation rather than on generic video generation. SP001, SP004
CP030 Tripo and Meshy are the clearest publicly visible packaged substitutes for self-serve AI 3D generation with web app, transparent pricing, and export workflow. SP001, SP002, SP005, SP006
CP031 OpenAI Sora is a weaker direct substitute for exportable 3D meshes but a stronger competitor for investor attention and multimodal world-model narrative. SP023, SP024
CP032 NVIDIA's role is best modeled as ecosystem leverage rather than as a conventional app-level seat competitor. SP018, SP021
CP033 Switching costs are moderate for casual creators but higher for teams that embed APIs, workspaces, admin controls, or shared asset history into production workflows. SP002, SP003, SP005, SP010
CP034 Meshy's claimed ARR, users, and rapid 2026 shipping are the clearest public adverse signals that Tripo is not alone in scaling AI 3D demand. SP007, SP008
CP035 Stability building on TripoSR is direct adverse evidence that Tripo's research lead may diffuse into rival products rather than remain proprietary. SP019
CP036 Adobe, Azure/OpenAI, and Luma hold trust, safety, or incumbent-distribution advantages that can matter more than raw generator quality in enterprise buying. SP012, SP013, SP024
CP037 The near-term market structure is likely multi-homing, with teams testing several AI providers while keeping manual DCC tools as fallback. SP001, SP005, SP013, SP024
CP038 Tripo's moat appears stronger in workflow packaging and embedded 3D product depth than in irreproducible core reconstruction science alone. SP001, SP002, SP019
CP039 The biggest competitive risk is commoditization of baseline generation while surrounding workflow, trust, and customer access consolidate around larger ecosystems. SP013, SP019, SP024
CP040 Critical unresolved diligence items are Tripo's free-to-paid conversion versus Meshy, enterprise retention against platform incumbents, and the share of revenue already protected by API or workspace embed. SP005, SP013, SP024
CP041 Luma's January 2026 Business Wire release says the company is backed by HUMAIN, Andreessen Horowitz, Amazon, AMD Ventures, NVIDIA, Amplify Partners, and Matrix Partners. SP012
CP042 Tripo and Meshy both market democratized, production-ready AI 3D workflows, which means messaging overlap is already high and product proof matters more than category narrative. SP001, SP005, SP007
CI001 Tripo publicly monetizes through Studio subscriptions, API credits, and team-oriented paid packaging rather than through a single undifferentiated plan. SI002, SI004, SI005
CI002 Tripo's pricing page publicly lists Free, Pro, Max, and Team plans. SI002, SI020
CI003 The Pro plan publicly includes 3,000 monthly credits, 10 concurrent tasks, private models, commercial use, and mesh-quality upgrades. SI002, SI020
CI004 Max and Team tiers publicly gate higher concurrency, bulk export, dedicated processing, shared workspaces, centralized billing, and larger credit pools. SI002, SI020
CI005 Tripo's API is a pay-before-you-go product, and volume API pricing or custom contracts require contacting sales. SI004, SI005
CI006 Tripo's API docs say the base exchange rate is $1.00 for 100 credits. SI004, SI021
CI007 The public API pricing docs show task-level monetization such as 20-50 credits for image-to-model paths, 25 credits for rigging, 10 per animate for retargeting, and 5-credit conversion surcharges. SI004, SI021
CI008 The official Python SDK documents account balance, frozen amount, and task-level consumed_credit fields, indicating Tripo has usage-metering instrumentation even if public revenue is undisclosed. SI007
CI009 Tripo's terms identify Holymolly Ltd as the company and service counterparty behind the Tripo platform. SI006
CI010 Tripo's terms say fees are billed in U.S. dollars, are generally non-cancelable and non-refundable, use Stripe as payment processor, and can accrue 5% per day in late fees on overdue amounts. SI006
CI011 Tripo's terms say users may cancel during the current subscription period, after which the subscription will not renew and no further fees will be charged beyond the current term. SI006
CI012 A Hong Kong company-registry mirror lists Holymolly Limited as incorporated on 20 July 2023 with CR No. 3300305 and live status. SI008
CI013 Tripo's public monetization architecture mixes recurring seat revenue, usage-linked API revenue, and higher-value team or enterprise packaging. SI002, SI004, SI005, SI006
CI014 Tripo monetizes workflow friction as much as raw generation output by gating privacy, commercial use, concurrency, storage, batch export, and collaboration features. SI002, SI020
CI015 AccessNewswire said in September 2025 that Tripo had more than 3 million global 3D creators, 35,000 active developers, 700+ enterprise customers, and 40 million generated models. SI009
CI016 Forbes said in September 2025 that Tripo had more than 3 million professional users and more than 40,000 studios and corporate partners. SI011
CI017 Yahoo Tech / SCMP reported that Tripo had 6.5 million users and that more than 85% of them were outside China. SI012
CI018 AI Weekly reported 10 million individual users and 90,000 studio clients as VAST raised $200 million. SI014
CI019 InforCapital reports 6.5 million creators, 90,000+ developers, 40,000+ enterprise clients, and roughly $400 million raised. SI018
CI020 Public user, developer, and enterprise-client figures conflict enough that revenue per user cannot be responsibly inferred from the retained public record. SI009, SI011, SI012, SI014, SI018
CI021 Auganix reported a $50 million Series A in March 2026 alongside more than 6.5 million creators, 90,000 developers, and nearly 100 million generated assets. SI013
CI022 Forbes reported about $200 million in fresh funding from more than a dozen investors including INCE Capital, Genesis Capital, and Primavera Capital, pushing Tripo to unicorn status. SI010
CI023 AIThority reported nearly $200 million in Series A+ and A++ financing and said the capital would expand research teams, core algorithm development, data and infrastructure systems, and global product and ecosystem presence. SI015
CI024 Cyzone reported that VAST completed an A3 strategic round above RMB1 billion in July 2026 and would use the funds for algorithm iteration, data, talent, and global commercialization or ecosystem buildout. SI016
CI025 CB Insights publicly lists about $397.14 million total raised and a $147.14 million last raise 10 days earlier, which does not cleanly align with the other retained public financing narratives. SI017
CI026 The most plausible reading of the 2026 funding record is that Tripo closed several closely spaced tranches whose naming differs across English and Chinese sources. SI013, SI015, SI016, SI017
CI027 None of the retained public sources disclose Tripo's revenue, ARR, gross margin, burn, cash balance, net revenue retention, enterprise ACVs, or CAC. SI010, SI015, SI017, SI018
CI028 Public list prices are low enough that large user counts alone do not imply strong revenue without high conversion, heavy usage, or an enterprise-heavy mix. SI002, SI020
CI029 Usage-based API credits create a path to higher monetization intensity than self-serve subscriptions alone, especially if developers automate repeated high-credit workflows. SI004, SI007, SI021
CI030 The existence of get_balance and consumed_credit fields suggests Tripo can internally reconcile usage and billing at task level even though external revenue reporting is absent. SI007
CI031 Stripe-based payments, USD billing, payment enforcement rights, and service suspension for nonpayment can reduce collection risk but also show dependence on third-party payment rails. SI006
CI032 Costbench independently corroborates Tripo's main paid tier prices and warns that the product can carry hidden costs beyond list price and auto-renewing contracts. SI020, SI002
CI033 Meshy's public $20 Pro pricing indicates that Tripo's self-serve pricing already sits in a narrow competitive band rather than enjoying obvious standalone pricing power. SI022, SI002
CI034 Luma and Adobe expose pricing or credit structures across broader creative workflows, which raises bundling and substitution pressure even when they are not direct 3D mesh equivalents. SI023, SI024
CI035 Across June and July 2026 financing coverage, Tripo consistently describes planned use of funds in terms of R&D, algorithms, data, infrastructure, talent, and global ecosystem expansion rather than near-term profitability. SI015, SI016
CI036 That capital deployment pattern implies a company still in model-platform buildout mode, not yet one publicly optimizing for near-term operating leverage. SI015, SI016
CI037 Tripo's public financial verdict is constrained not by lack of monetization design but by lack of disclosed revenue quality, margin, and efficiency metrics. SI002, SI004, SI015, SI017
CI038 Near-term capital adequacy appears strong because repeated 2026 financings likely added substantial cash, but actual runway remains unobservable without burn and cash-balance disclosure. SI013, SI015, SI016
CI039 The most important private diligence requests are paid conversion, API or Team revenue mix, enterprise ACV, retention, gross margin after inference, cloud commitments, burn, and revenue concentration. SI002, SI004, SI015
CI040 Independent teardown skepticism about production-readiness is an adverse sign because monetization only strengthens if free experimentation converts into trusted, repeat paid workflows. SI019, SI002
CI041 The Tripo billing platform and third-party API guide reinforce that developer onboarding is relatively accessible, which helps acquisition but does not by itself prove sticky monetization. SI021, SI025, SI003
CI042 The public narrative around Tripo is dominated by user growth and fundraising headlines rather than by disclosed unit economics, which increases underwriting uncertainty. SI010, SI014, SI015, SI017
CE001 Tripo positions itself as an AI 3D generator that turns text prompts and images into 3D models in seconds. SE001, SE004
CE002 The homepage explicitly markets export of generated assets in GLB, STL, and OBJ formats. SE001
CE003 The official pricing page shows four live self-serve packages: Free, Pro, Max, and Team. SE002
CE004 The Free plan includes 200 monthly credits, one concurrent task, public CC BY 4.0 models, and limited downloads. SE002
CE005 The Pro plan adds 3,000 monthly credits plus multi-view to 3D, batch generation, Smart Mesh, private models, and commercial use. SE002
CE006 The Max and Team plans raise concurrency to 100 and 200 tasks respectively and add shared workspace or centralized administration features. SE002
CE007 Tripo markets its core generator around accurate, fast, clean-mesh output aimed at both beginners and professional creators. SE004
CE008 The public feature set includes auto rigging for humans, animals, and stylized characters. SE005
CE009 The segmentation feature is marketed as pipeline-ready editing that preserves solid topology and logical grouping across parts. SE006
CE010 The quad remesher feature promises automatic AI retopology into clean quad topology for animation-ready models in seconds. SE007
CE011 The texturing module emphasizes consistent global styles and pixel-level control for high-quality textured assets. SE008
CE012 Tripo’s official format guide says GLB is best for web and mobile, FBX fits animation and game-development workflows, and OBJ fits simpler static or print-oriented files. SE009
CE013 Tripo’s JavaScript integration guide recommends GLB for web deployment and lists OBJ, FBX, STL, and USDZ as alternative export paths. SE018
CE014 Tripo publishes an official Roblox export workflow focused on scale correction for downstream engine import. SE011
CE015 Tripo’s own game-ready checklist says useful outputs still need validation on polygon count, topology, UVs, textures, and real-world scale inside Unity or Unreal. SE010
CE016 The same game-ready guide says Algorithm 3.0 can generate models in roughly 10 seconds. SE010
CE017 Tripo’s public OpenAPI task endpoint rejects anonymous access with a 401 authentication failure, confirming that the production API is key-gated. SE036
CE018 The official Python SDK exposes text-to-3D, image-to-3D, and multi-view-to-3D generation through an asynchronous client. SE022
CE019 The Python SDK also exposes one-shot animation, external model import, segmentation, mesh completion, smart lowpoly, conversion, and stylization. SE022
CE020 An independent Apidog guide describes Tripo as a REST-style developer service for gaming, e-commerce, VR, and architecture use cases. SE031
CE021 ComfyUI’s official partner-node docs say the Tripo API is natively integrated and currently supports model generation plus rig-model operations inside ComfyUI. SE023
CE022 The official ComfyUI-Tripo GitHub repository notes a 2026-06-30 update adding Mesh Segmentation v2.0 and direct URL import. SE024
CE023 Tripo’s Blender plugin tutorial says creators can generate models from text or images directly inside Blender in minutes. SE013
CE024 The Blender-plus-Cursor tutorial shows Tripo supporting an MCP-server workflow that connects Blender, API keys, and Cursor for AI-assisted 3D work. SE017
CE025 Tripo’s JavaScript tutorial documents Babylon.js and Three.js import paths, with GLB described as the preferred web format. SE018
CE026 Tripo’s MakerWorld case study says MakerWorld and Bambu Lab embedded Tripo image-to-3D features rather than merely linking out to the app. SE016
CE027 The same case study claims 90% faster prototyping plus higher engagement and a broader user base after integrating Tripo image-to-model features. SE016
CE028 The TripoSG paper describes a large-scale rectified-flow transformer, hybrid supervised VAE losses, and a 2 million sample image-to-SDF data pipeline. SE019
CE029 The TripoSG repository says the model targets high-fidelity image-to-3D generation, released a 1.5B-parameter version, and added a scribble-conditioned variant for fast prototyping. SE020
CE030 The TripoSR repository says the model reconstructs 3D objects from a single image in under 0.5 seconds on an NVIDIA A100 GPU and can run a default single-image path in about 6GB VRAM. SE021
CE031 TripoSR is presented as a collaborative open-source release with Stability AI under the MIT license. SE021, SE026
CE032 GitHub API metadata shows TripoSR at 6,704 stars and 857 forks as of the run date, indicating unusually strong open-source reach for a 3D model repo. SE026
CE033 GitHub API metadata shows TripoSG at 1,718 stars and 187 forks as of the run date. SE027
CE034 GitHub API metadata shows smaller but real workflow-tooling traction, with the ComfyUI-Tripo repo at 340 stars and the Python SDK and Blender plugin at 52 and 51 stars. SE028, SE029, SE030
CE035 TechTimes reports that Tripo P1.0 uses a unified probabilistic spatial framework to generate assets directly in native 3D space rather than stepwise reconstruction. SE032
CE036 A Stanford Daily article says Tripo H3.1 and P1.0 improved geometric precision, generation speed, and production-ready quality while sitting inside a broader plugin, workspace, and API ecosystem. SE035
CE037 Tech On Play says Tripo works best as a workflow accelerator rather than a full replacement for manual modeling, with models averaging under 30 seconds to generate. SE033
CE038 The same Tech On Play review estimates topology accuracy around 80–90% for base meshes and 75–85% for intricate designs. SE033
CE039 The Medium teardown warns that an attractive preview can still fail on export through mesh issues, UVs, texture portability, rigging behavior, or poor part naming. SE037
CE040 The same teardown recommends benchmarking Tripo on real assets and treating it as a first-pass accelerator rather than the center of a production pipeline. SE037
CE041 Tripo’s own clean-mesh guide says mesh cleanup remains a crucial downstream step even when assets originate from AI generation. SE014
CE042 Tripo’s hole-fixing guide frames hole repair and edge smoothing as common post-processing work for game, print, and animation assets. SE015
CE043 MakerStack says Tripo’s image-to-3D path generally performs better than text-to-3D because the model has more visual context and can export GLB, FBX, OBJ, USD, and STL. SE034
CE044 MakerStack gives Tripo a 7.2/10 overall review score, supporting a view of broad capability with remaining rough edges. SE034
CE045 The fetched public developer surface does not visibly publish uptime SLAs, enterprise SSO controls, or admin-governance guarantees.
CE046 Because Tripo markets retopology, segmentation, hole fixing, and game-ready validation so heavily, the company’s own product narrative implies raw generations still often need cleanup before production use. SE006, SE007, SE010, SE014, SE015
CE047 No fetched public source provides a rigorous independent benchmark set focused specifically on complex organic characters, cloth-heavy rigs, or full production animation pass rates.
CE048 USDZ appears in Tripo’s web integration guidance, but the fetched official product pages do not clearly prove that USDZ is a first-class current export option across the main product surface. SE001, SE003, SE018
CU001 Tripo publicly targets game developers, product designers, marketing creatives, architects, and digital artists rather than a single buyer archetype. SU004
CU002 The pricing ladder effectively segments creators from professionals, power users, studios, and seat-based teams. SU001, SU022
CU003 Free-plan outputs remain public and non-commercial, while paid tiers unlock commercial rights, higher concurrency, and team administration. SU001, SU019, SU021, SU022
CU004 Official Tripo material positions e-commerce teams as users who need better product visualization and faster AI-driven product prototyping. SU003
CU005 A Tripo-hosted article featuring Dewan Architects & Engineers frames architects and design professionals as a real workflow audience for the platform. SU007
CU006 Tripo’s solo-game-dev example shows a usable fit for individual Unreal creators trying to build a playable prototype quickly. SU005
CU007 The rigging-for-games page explicitly addresses indie game teams that need production-ready animated assets rather than only static concept models. SU009
CU008 Tripo’s e-commerce vertical pages frame high-volume fashion AR try-on as a pipeline problem spanning 1,000 or more SKUs. SU010
CU009 The e-commerce API page positions Tripo as a programmatic catalog-scaling tool that plugs into product-information-management systems and concurrent asset pipelines. SU011
CU010 Tripo also markets film-ready and pre-visualization workflows, indicating customer ambition beyond games and commerce into media production. SU012
CU011 The MakerWorld case study says Bambu Lab’s MakerWorld integrated Tripo image-to-model and Make My Lantern features into its creator workflow. SU008
CU012 The official Sony collaboration page says Sony Spatial Reality Display and VAST formed a business partnership around 3D display, content generation, retail, education, and digital-twin use cases. SU002
CU013 Yahoo-syndicated SCMP reporting says Tripo’s user base grew from 3 million in August to 6.5 million by the time of the article. SU013
CU014 The same reporting says more than 85% of users are outside China, with Europe and the United States as the biggest markets. SU013
CU015 Yahoo-syndicated SCMP reporting names Tencent, NetEase, Sony, Microsoft, and Pop Mart among Tripo’s prominent enterprise users. SU013
CU016 The same article says Tripo had only two salespeople and relied heavily on organic growth and word of mouth. SU013
CU017 After the beta launch, Tripo Studio’s monthly income reportedly grew 2.5 times within a month and fivefold within three months. SU013
CU018 AI Weekly says Tripo had 10 million individual users and 90,000 studio clients by June 2026, including Sony and NetEase among those client names. SU014
CU019 ACCESS Newswire said in September 2025 that Tripo served more than 3 million global creators, 35,000 active developers, and 700 or more enterprise customers including Bambu Lab, Tencent, NetEase, Replit, HTC, and Fal. SU015
CU020 Forbes reported in September 2025 that Tripo had more than 3 million professional users worldwide and more than 40,000 studios and corporate partners concentrated in Europe, the US, Japan, and Korea. SU016
CU021 The Stanford Daily article says the platform serves more than 6.5 million creators, 90,000 developers, and more than 100 million generated 3D models, while working with Replit, Sony’s Spatial Reality Division, and NetEase. SU018
CU022 KR Asia reports that Tripo was used in NetEase’s game Where Winds Meet, making NetEase one of the strongest public signs of real downstream production usage. SU017
CU023 Digital Media Net described a growing creator and developer ecosystem around Tripo at GDC 2026. SU023
CU024 ComfyUI’s official partner-node documentation shows Tripo is natively integrated for model-generation workflows, indicating a real technical-user cohort beyond consumer creators. SU028
CU025 GitHub API metadata shows TripoSR at 6,704 stars, ComfyUI-Tripo at 340 stars, and the Python SDK at 52 stars as of the run date. SU025, SU026, SU027
CU026 The strongest named public proofs are Sony, NetEase, MakerWorld/Bambu Lab, and Replit because they recur across more than one source or have an official workflow description attached to them. SU002, SU008, SU015, SU017, SU018
CU027 Public customer numerators conflict across sources: 700+ enterprise customers, 35,000 active developers, 40,000 studios and corporate partners, 90,000 developers, and 90,000 studio clients are all used for different audiences or periods. SU014, SU015, SU016, SU018
CU028 Despite denominator drift, every public adoption source points in the same direction: Tripo is materially beyond toy scale in both creator reach and enterprise visibility. SU013, SU014, SU015, SU016, SU018
CU029 The customer base is global rather than China-only, with some of the clearest public evidence pointing to Europe, the United States, Japan, and Korea as important user regions. SU013, SU016
CU030 Tripo’s go-to-market motion appears strongly product-led because public sources emphasize organic growth, creator distribution, and self-serve plans more than a large direct-sales organization. SU001, SU013, SU021, SU022
CU031 The free plan structure makes it easy for many users to remain non-paying until they need commercial rights, privacy, or materially higher generation volume. SU001, SU019, SU021
CU032 CostBench’s plan breakdown suggests Max and Team tiers are priced for power users and seat-based teams, which can narrow the set of customers willing to expand broadly on price alone. SU021, SU022
CU033 Tech On Play says free users get enough monthly credits for roughly 10 standard models, making the product easy to trial but not necessarily sticky enough for deep production without paying. SU024
CU034 MakerStack says serious work begins at the paid tiers where private models and commercial rights become available. SU019
CU035 The Medium teardown argues that Tripo is easier to justify for concepting and first-pass asset generation than as a proven final-production standard. SU020
CU036 Named logos in the public record appear to mix paying customers, workflow partners, and reference users rather than one clearly disclosed paid-production cohort. SU002, SU013, SU015, SU018, SU020
CU037 Public retention metrics such as NRR, GRR, churn, renewal rates, and contract length are not disclosed in the fetched source set.
CU038 Public evidence does not reveal how many of the millions of users convert into paying subscriptions, active API accounts, or recurring enterprise contracts.
CU039 The strongest publicly documented verticals are gaming, e-commerce or retail, creator tooling, architecture or industrial design, and XR or spatial display. SU002, SU003, SU005, SU007, SU009, SU010, SU011, SU012
CU040 Official use-case pages show credible expansion paths from self-serve creation into large-SKU retail pipelines, film previsualization, and studio workflows. SU003, SU010, SU011, SU012
CU041 The MakerWorld case study claims 90% faster prototyping after integrating Tripo’s image-to-model functions. SU008
CU042 Stanford Daily says Tripo’s subscriptions, creator applications, and developer APIs allow studios, platforms, and independent developers to integrate generated 3D content into production pipelines. SU018
CU043 ACCESS describes Tripo Studio as an end-to-end AI-driven 3D pipeline and workspace, supporting the idea that enterprise or platform customers can adopt more than a one-off generation tool. SU015
CU044 Forbes and Yahoo together imply that Tripo’s higher-value customer base is internationally distributed rather than centered on mainland China alone. SU013, SU016
CR001 Tripo’s terms say Holymolly Ltd and its affiliates provide the service under a legally binding user agreement. SR001
CR002 The terms prohibit using Tripo services or outputs in ways that violate applicable laws or regulations. SR001
CR003 Tripo’s privacy policy says some personal information may be processed on behalf of enterprise customers rather than solely for direct consumer use. SR002
CR004 The privacy policy explicitly lists international data transfers as part of the service’s risk surface. SR002
CR005 The free plan keeps models public under CC BY 4.0 while paid plans are needed for private models and commercial use. SR003
CR006 China Briefing says China’s generative-AI draft security requirements cover training data, model protection, overall security protocols, and security assessments. SR010
CR007 SCIO reported in April 2026 that Chinese regulators punished online platforms for failing to comply with AI-generated-content labeling rules. SR011
CR008 NPC Observer says China’s Cybersecurity Law amendment effective January 1, 2026 tightens compliance and raises penalties. SR012
CR009 An official State Council article says CAC, NDRC, and MIIT jointly issued 2026 guidelines to regulate and standardize AI-agent development. SR013
CR010 BIS guidance says a license is required to export advanced computing items to entities headquartered in Country Group D:5 or Macau, even when those entities are located elsewhere. SR004, SR005
CR011 Trade.gov says China’s military-civil-fusion strategy makes it difficult to identify China-based counterparties with links to military end users, creating compliance risk. SR006
CR012 The GAO says Commerce implemented 2022 and 2023 advanced-semiconductor rules but still had to address compliance challenges. SR007
CR013 The CRS says U.S. actions have sought to restrict PRC access to advanced chips and related computing and AI applications. SR008
CR014 The U.S. Copyright Office says AI policy review includes the use of copyrighted materials in AI training and is still being reported in multiple parts. SR009
CR015 The ABA analysis says June 2025 court rulings found AI training can be fair use while keeping businesses liable for infringing outputs they create and publish. SR028
CR016 IPWatchdog says 2025 decisions drew a line between transformative training and pirated or unauthorized source acquisition, keeping legal uncertainty high. SR029
CR017 Tripo’s public legal and product surfaces do not disclose a clear provenance map for the 3D assets, references, or training datasets behind its models.
CR018 The Medium teardown says Tripo previews can look finished even when exported assets are not production-ready for real pipelines. SR020
CR019 The same teardown says Tripo should be treated as a workflow accelerator first, not a replacement for a technical artist or production pipeline. SR020
CR020 Tech On Play says Tripo averages under 30 seconds per model but still performs best as a workflow accelerator rather than a full replacement for manual modeling. SR021
CR021 Tech On Play says topology accuracy falls to roughly 75–85% on intricate designs, underscoring residual production risk on hard assets. SR021
CR022 MakerStack’s 7.2/10 review score supports a view that the product is useful but still imperfect for serious production teams. SR019
CR023 Because commercial rights and private models unlock only at paid tiers, Tripo’s large free cohort may generate activity without proportional revenue. SR003, SR019
CR024 Team and Max pricing indicate that broad enterprise expansion can become materially more expensive than casual creator usage. SR003
CR025 Yahoo-syndicated SCMP reporting says Tripo had only two salespeople and relied mainly on organic growth and word of mouth. SR022
CR026 Forbes says Simon Song previously worked at SenseTime and co-founded MiniMax before leaving to launch Tripo, reinforcing key-person dependence on a founder with scarce AI-builder credibility. SR023
CR027 The Financial Times says the AI boom faces a looming talent shortage and a race to build enough human capability, which raises retention risk for smaller frontier AI teams. SR017
CR028 Founder-centric and founder-quoted public coverage suggests Tripo’s external risk surface remains heavily concentrated around Simon Song as technical storyteller and strategic face. SR022, SR023, SR024
CR029 OpenAI’s model catalog now includes Sora 2 and a wide frontier multimodal stack that can compete for the same customer budgets Tripo depends on. SR014
CR030 Microsoft Foundry aggregates models from Microsoft, OpenAI, DeepSeek, Hugging Face, Meta, and others with evaluation and deployment tooling in one catalog. SR015
CR031 Azure OpenAI offers both pay-as-you-go pricing and provisioned-throughput units, giving a major platform player more flexibility to subsidize or standardize enterprise workloads. SR016
CR032 NVIDIA’s visual-design-model surface and Adobe Firefly’s official presence show that well-capitalized incumbents continue to circle adjacent creative-generation territory. SR030, SR031
CR033 Taken together, U.S. export-control guidance, compliance challenges, and advanced-semiconductor restrictions create a non-trivial risk that Chinese AI companies face slower access to top-end compute. SR004, SR006, SR007, SR008
CR034 Chinese labeling enforcement means risk extends beyond model development into the distribution of generated content itself. SR011, SR013
CR035 Tripo’s global user base plus explicit international data transfers imply recurring cross-border privacy and localization work rather than a one-time compliance project. SR002, SR022
CR036 AI Weekly’s claim of 10 million users and 90,000 studio clients still does not reveal paying conversion, revenue quality, or retention. SR024
CR037 ACCESS Newswire’s 700+ enterprise-customer claim in 2025 conflicts with later public numerators such as 90,000 studio clients or 90,000 developers, making customer-quality metrics hard to underwrite. SR024, SR025
CR038 TechTimes says many AI-generated 3D assets still require manual editing before professional pipeline use, so Tripo’s quality advantage can erode if rivals close the gap. SR026
CR039 Digital Media Net says P1.0 was introduced against a backdrop where developers still struggle to use generated 3D assets directly in production, confirming the category-wide race on usability. SR027
CR040 The combination of unresolved training-data provenance, evolving fair-use doctrine, and output-liability risk creates an ongoing IP overhang for Tripo and similar model builders. SR009, SR028, SR029
CR041 Visible mitigations today include post-processing tools, plugin depth, open research signaling, and a product-led distribution engine rather than a pure demo product. SR020, SR021, SR025, SR026, SR027
CR042 Residual exposure remains high because compliance, compute, customer-quality, and product-usability risks can all transmit into margin, growth, and valuation at the same time. SR006, SR020, SR022, SR024, SR026
CV001 The strongest public financing fact is that Tripo AI completed Series A+ and Series A++ financing totaling nearly $200 million in June 2026. SV003, SV004
CV002 The official press release says the new capital is earmarked for AI 3D and world-model research, core algorithms, data and infrastructure systems, and global ecosystem expansion. SV003
CV003 Forbes reported that Tripo moved from a March 2026 round led by Alibaba and Hengxu into a roughly $200 million June 2026 syndicate that included INCE Capital, Genesis Capital, and Primavera. SV001
CV004 The same Forbes reporting says VAST became a unicorn and that one informed source put the company at about a $1.5 billion valuation. SV001
CV005 AI Weekly says Tripo reached 10 million individual users and 90,000 studio clients by June 2026. SV002
CV006 AI Weekly and Phemex-style follow-on coverage both name Sony and NetEase as enterprise customers or clients, which is meaningful proof that adoption is not limited to hobbyists. SV002, SV007
CV007 Tripo’s pricing page shows a genuine freemium top-of-funnel: the free plan costs $0, includes 200 monthly credits, and caps users at one concurrent task. SV005
CV008 The Pro plan is priced at about $19.90 monthly and adds 3,000 monthly credits, multi-view generation, batch export, and commercial-use rights. SV005
CV009 The Max plan is priced at about $89 monthly, includes 25,000 monthly credits and up to 100 concurrent tasks, and is clearly aimed at heavier professional or team usage. SV005
CV010 Commercial use and private models are gated behind paid plans, implying that management is intentionally using rights and workflow scale as monetization levers. SV005
CV011 Tripo exposes a developer-facing API surface in addition to the creator UI, which means monetization can come from direct subscriptions and embedded developer usage rather than from a single web-app tier. SV006, SV008
CV012 Official customer-facing materials span Sony collaboration, scalable API use cases, media production, and MakerWorld integration, supporting a multi-vertical revenue narrative across gaming, commerce, media, and partner ecosystems. SV007, SV008, SV009, SV010
CV013 Despite headline scale claims, Tripo does not publicly disclose ARR, recognized revenue, NRR, gross margin, churn, or cash burn in the fetched materials. SV001, SV002, SV003, SV005
CV014 Multiples.vc says July 2026 public software valuations are highly segmented and that design-and-engineering and AI software categories still command premium revenue multiples versus weaker SaaS cohorts. SV026
CV015 Acquiry’s 2026 benchmark argues that non-AI SaaS often clears around 4x to 7x ARR while AI-native SaaS more often clears around 8x to 15x ARR, with the fastest growers able to stretch toward 10x to 20x. SV027
CV016 The same Acquiry benchmark says NRR, gross margin, CAC payback, and profitability are central multiple drivers, which matters because Tripo does not disclose any of them. SV027
CV017 Adobe’s July 2026 public market cap of about $88.5 billion against roughly $24.45 billion of TTM revenue implies a market-cap-to-revenue ratio of about 3.6x. SV020, SV021
CV018 Roblox’s July 2026 public market cap of about $40.38 billion against roughly $5.29 billion of TTM revenue implies a market-cap-to-revenue ratio of about 7.6x. SV022, SV023
CV019 Unity’s July 2026 public market cap of about $13.4 billion against roughly $1.92 billion of TTM revenue implies a market-cap-to-revenue ratio of about 7.0x. SV024, SV025
CV020 Public creative and 3D workflow comps therefore trade at far lower revenue ratios than frontier private AI marks unless growth, scarcity, or optionality justify a substantial premium. SV017, SV018, SV019, SV026, SV027
CV021 Runway said in April 2025 that it hoped to reach $300 million of annualized revenue that year. SV028
CV022 TechCrunch later reported that Runway raised a 2026 Series E at a $5.3 billion valuation, showing that premium creative-AI private comps can sit well above ordinary software multiples. SV029
CV023 If Runway’s $5.3 billion valuation is compared with its earlier $300 million annualized-revenue target, the resulting ratio is roughly 17.7x, which is a plausible upper-band private comp for a scarce frontier creative-AI asset. SV028, SV029
CV024 At a $1.5 billion valuation, Tripo would need about $75 million of ARR or revenue to clear a 20x multiple. SV001, SV027
CV025 At the same $1.5 billion valuation, Tripo would need about $100 million of ARR or revenue to clear a 15x multiple. SV001, SV027
CV026 At a 12x multiple, the implied denominator rises to about $125 million of ARR or revenue. SV001, SV027
CV027 At a 10x multiple, the implied denominator rises to about $150 million of ARR or revenue. SV001, SV027
CV028 At an 8x multiple, the implied denominator rises to about $187.5 million of ARR or revenue, and the public record does not show whether Tripo is anywhere near that level. SV001, SV027
CV029 TechCrunch says Moonshot AI reached a $20 billion valuation in May 2026 and had ARR above $200 million in April, confirming that Chinese AI leaders could still command extreme private-market premia in 2026. SV030
CV030 CNBC reported in January 2026 that Moonshot was being valued around $4.8 billion while public-market values for Zhipu and MiniMax sat around $13 billion and $15.2 billion respectively. SV031
CV031 TechCrunch later described Zhipu and MiniMax public-market values of roughly $55.9 billion and $33 billion after rallies tied to new model releases, underscoring how quickly China AI comparables can re-rate. SV030
CV032 Those China AI comps are useful sentiment markers but are still structurally different from Tripo because they are frontier LLM labs or public-market stories rather than private 3D application-software companies. SV030, SV031, SV033
CV033 BIS, GAO, and China Briefing materials all support the view that export controls, advanced-computing restrictions, and generative-AI compliance rules remain live valuation haircuts for a China-based model company. SV016, SV017, SV018
CV034 CNBC’s 2025 AI funding report says funding has greatly outpaced exits and that lower-value bolt-on acquisitions still dominate, which weakens private-market liquidity support for aggressive entry prices. SV032
CV035 Independent teardown coverage says Tripo’s meshes still need workflow testing and cleanup before buyers build around them as production-safe defaults. SV014, SV015
CV036 That combination of quality caveats, geopolitical overhang, and still-thin exit markets justifies a real haircut to any straight AI-native revenue multiple. SV014, SV015, SV016, SV017, SV018, SV032
CV037 Forbes and Yahoo growth coverage show that Tripo’s user base expanded sharply from 2025 into 2026 and that management is already positioning the platform for larger consumer and global ecosystem ambitions. SV011, SV012
CV038 ACCESS Newswire and the official June press release reinforce that management is expanding the product stack into richer generation, infrastructure, and world-model narratives rather than standing still on a single text-to-3D feature. SV003, SV013
CV039 The right valuation frame is therefore price-sensitive rather than admiration-sensitive: Tripo looks strategically interesting, but the public file is not rich enough to underwrite the current mark with confidence. SV001, SV002, SV003, SV027, SV032
CV040 A buy recommendation would require evidence that the revenue denominator, retention profile, and cost structure are already strong enough to support a premium multiple without leaning on narrative alone. SV013, SV015, SV016, SV027
CV041 A stretched valuation stance is more supportable than an outright expensive call because Tripo has real adoption, visible monetization surfaces, and an adjacent private comp in Runway that also prices far above ordinary SaaS. SV002, SV005, SV028, SV029
CV042 Risk should still be rated high because a disclosure miss on ARR or paid conversion, or a China-specific regulatory or compute shock, can compress multiples quickly even if product demand remains healthy. SV013, SV016, SV017, SV018, SV032
CV043 The cleanest upgrade trigger would be verified evidence that Tripo is already above roughly $100 million of ARR with strong retention, healthy gross margin, and meaningful paying enterprise concentration. SV024, SV025, SV026, SV027
CV044 The clearest thesis-break trigger would be disclosed ARR materially below about $75 million to $100 million, weak free-to-paid conversion, or financing terms that show the headline valuation is being propped up by structure rather than economics. SV005, SV024, SV025, SV032
CV045 Project Eden gives Tripo a world-model option value that can justify investor attention, but the public record still treats that initiative as a research roadmap rather than a monetized product line. SV003, SV029
CV046 Adobe’s 10-K illustrates what mature creative-software disclosure looks like, which sharpens the contrast with Tripo’s absence of filing-grade information on revenue recognition, risk factors, and customer concentration. SV019
来源
编号出版方标题引文
SO001 Tripo AI AI 3D Model Generator from Text & Images | Tripo 3D
SO002 Tripo AI Tripo Studio Pricing | Generate High-Quality 3D Models with AI Start Free. Upgrade to Unlock Ultra Mesh, Smart Low-Poly, High-Quality Textures, Animation, and More Premium Features.
SO003 Tripo Developers Tripo Developers — AI 3D Generation API
SO004 Tripo AI Privacy Policy Your information may be transferred to and processed in countries outside your residence.
SO005 Holymolly Ltd / Tripo AI Terms of User Agreement Welcome and thank you for your interest in Holymolly Ltd ... and our website at <www.tripo3d.ai>.
SO006 Tripo AI Blog Introducing Tripo's Latest Update: Ultra HD Textures, Smarter Rigging & Unlimited Creativity Over 100+ New Mocap-Quality Biped Animations
SO007 Tripo AI Blog Sony × VAST: Opening the Future of 3D Creation Sony and VAST Officially Announce 3D Business Collaboration to Expand the 3D Content Ecosystem.
SO008 Tripo AI Blog How E-commerce Users Can Use Tripo AI for 3D Modeling and AI Product Prototyping Compatible seamless exports to major 3D platforms such as Blender, Unity, and Unreal Engine using OBJ, FBX, GLB formats.
SO009 Tripo AI Blog Exploring Tripo AI: Powerful 3D Workflows for Diverse Creator Types
SO010 Tripo AI Blog Tripo X ComfyUI: Official API Node Available Tripo has officially integrated into ComfyUI's API Nodes.
SO011 Tripo AI Explore How I Evaluate AI 3D Models for Real-World Production Success
SO012 Tripo AI Explore AI 3D Model Generator: Why I Evaluate the Silhouette First
SO013 Forbes Under 30 Alum’s 3D Model Startup Hits Unicorn Status Amid AI Frenzy VAST ... has become a unicorn startup after raising about $200 million in fresh funding from over a dozen investors including INCE Capital, Genesis Capital and Primavera Capital Group.
SO014 Forbes Tripo Grows 3D AI Business, Eyes Consumer Future With 3D TikTok Tripo now counts more than three million professional users across the world ... and has signed more than 40,000 studios and corporate partners.
SO015 Yahoo Tech / South China Morning Post China-founded Tripo AI updates 3D content creation platform as users more than double More than 85 per cent of Tripo AI's users are based outside China, with Europe and the United States as its biggest markets.
SO016 CB Insights Tripo AI - Products, Competitors, Financials, Employees, Headquarters Locations It was founded in 2023 and is based in Shanghai, China.
SO017 InforCapital Tripo AI - Deep Tech Startup, $400M Raised | InforCapital Founded in March 2023 by Simon Song ... The platform serves 6.5 million creators, 90,000+ developers, and 40,000+ enterprise clients.
SO018 AI Weekly Tripo AI Hits 10M Users as Vast Raises $200M The platform reports 10 million individual users and 90,000 studio clients including Sony and NetEase.
SO019 Auganix Tripo AI Raises $50M and Unveils New 3D Generation Models Its platform serves more than 6.5 million creators and 90,000 developers worldwide, with nearly 100 million 3D assets generated to date.
SO020 ACCESS Newswire Tripo, the Frontrunner of 3D AI Boom, Supercharges New Era in Content Creation with 3.0 Upgrade Tripo 3.0 ... contains more than 20 billion parameters-a 20x increase over previous versions-and 300% enhanced detail accuracy.
SO021 Medium Tripo AI Technical Teardown 2026: Test the Mesh Before You Build Around It I would treat it as a workflow accelerator first, not a replacement for a technical artist or production 3D pipeline.
SO022 arXiv TripoSR: Fast 3D Object Reconstruction from a Single Image
SO023 arXiv TripoSG: High-Fidelity 3D Shape Synthesis using Large-Scale Rectified Flow Models
SO024 GitHub GitHub - VAST-AI-Research/TripoSR: TripoSR: Fast 3D Object Reconstruction from a Single Image TripoSR: Fast 3D Object Reconstruction from a Single Image
SO025 The Stanford Daily Tripo AI is revolutionizing AI generated 3D models
SM001 Tripo AI AI 3D Model Generator from Text & Images | Tripo 3D
SM002 Tripo AI Tripo Studio Pricing | Generate High-Quality 3D Models with AI
SM003 Tripo Developers Tripo Developers — AI 3D Generation API
SM004 Tripo AI Blog How E-commerce Users Can Use Tripo AI for 3D Modeling and AI Product Prototyping
SM005 Tripo AI Blog Exploring Tripo AI: Powerful 3D Workflows for Diverse Creator Types
SM006 Tripo AI Rapid Prototyping Game Environments With AI | Tripo AI
SM007 Tripo AI A Comprehensive Guide to Connecting AI 3D Workspaces to Game Engines via REST API for Automated Import | Tripo AI
SM008 Tripo AI [Guide] 2D Image to 3D Furniture Conversion | Tripo AI
SM009 Tripo AI [Guide] Evaluate 3D Room Planner AI Tools | Tripo AI
SM010 Forbes Tripo Grows 3D AI Business, Eyes Consumer Future With 3D TikTok
SM011 Yahoo Tech / South China Morning Post China-founded Tripo AI updates 3D content creation platform as users more than double
SM012 Auganix Tripo AI Raises $50M and Unveils New 3D Generation Models
SM013 Axis Intelligence Research China AI Statistics 2026: Market Size, Investment & Global Competitive Position China’s AI market reached approximately $28–31 billion in 2025 revenue, on a trajectory toward $200 billion by 2032 at a 32.5% CAGR.
SM014 Grand View Research / Web Archive snapshot China Generative AI Market Size & Outlook, 2030 Software was the largest segment with a revenue share of 63.9% in 2024.
SM015 Research and Markets Generative AI for Three-Dimensional (3D) Assets Market Report 2026
SM016 China Daily citing IDC Growth a big reality for AR, VR domestic market Spending on AR and VR in China is predicted to hit $13.1 billion by 2026.
SM017 Marketing to China Virtual Reality Industry (VR) in China Explained 2026
SM018 Digital in Asia What is the State of China's Digital Economy in 2026? A Comprehensive Market Overview China’s gaming market posted 350.8 billion yuan ($49.8 billion) in 2025.
SM019 Cyberspace Administration of China 生成式人工智能服务管理暂行办法 提供具有舆论属性或者社会动员能力的生成式人工智能服务的,应当按照国家有关规定开展安全评估,并按照《互联网信息服务算法推荐管理规定》履行算法备案。
SM020 Cyberspace Administration of China 人工智能生成合成内容标识办法 人工智能生成合成内容标识包括显式标识和隐式标识。
SM021 State Council Information Office / Xinhua China unveils guidelines to regulate, boost innovative development of AI agents The guidelines identify 19 typical application scenarios spanning scientific research, industrial development, consumption boost, public well-being and social governance.
SM022 ACCESS Newswire Tripo, the Frontrunner of 3D AI Boom, Supercharges New Era in Content Creation with 3.0 Upgrade
SM023 arXiv TripoSG: High-Fidelity 3D Shape Synthesis using Large-Scale Rectified Flow Models
SM024 arXiv TripoSR: Fast 3D Object Reconstruction from a Single Image
SM025 Tripo AI Blog Sony × VAST: Opening the Future of 3D Creation
SP001 Tripo AI AI 3D Model Generator from Text & Images | Tripo 3D
SP002 Tripo AI Tripo Studio Pricing | Generate High-Quality 3D Models with AI
SP003 Tripo Developers Tripo Developers — AI 3D Generation API
SP004 ACCESS Newswire Tripo, the Frontrunner of 3D AI Boom, Supercharges New Era in Content Creation with 3.0 Upgrade
SP005 Meshy Meshy AI - The
SP006 Meshy Meshy Official Pricing: Free, Pro, Studio & Enterprise Plans
SP007 PR Newswire Meshy Hits $15M ARR with 30% Month-over-Month Growth, Unveils Meshy 6 Preview for Next-Gen 3D Creation
SP008 Meshy Blog - Meshy
SP009 Luma Luma | AI Agents for Creative Work
SP010 Luma Plans & Pricing | Luma
SP011 Luma Luma Introduces Ray3.2 Model & API: Complete Creative Control for Video Generation
SP012 Business Wire Luma AI Launches Ray3.14, Eliminating Quality-Speed-Cost Tradeoff in Generative Video
SP013 Adobe Adobe Firefly
SP014 Adobe Compare plans that include generative AI | Adobe Firefly
SP015 Adobe Adobe Substance 3D
SP016 Stability AI Stability AI Developer Platform
SP017 Stability AI Introducing Stable Fast 3D: Rapid 3D Asset Generation From Single Images
SP018 Stability AI Introducing Stable Point Aware 3D: Real-Time Editing and Complete Object Structure Generation
SP019 GitHub / Stability AI GitHub - Stability-AI/stable-fast-3d: SF3D
SP020 GitHub / Stability AI GitHub - Stability-AI/stable-point-aware-3d
SP021 Analytics Vidhya NVIDIA and Stability AI Team Up to Launch SPAR3D: A New Era in 3D Generation
SP022 OpenAI Sora
SP023 OpenAI Sora is here
SP024 Microsoft Sora 2 video generation overview (preview) - Microsoft Foundry
SP025 CB Insights Tripo AI - Products, Competitors, Financials, Employees, Headquarters Locations
SI001 Tripo AI AI 3D Model Generator from Text & Images | Tripo 3D
SI002 Tripo AI Tripo Studio Pricing | Generate High-Quality 3D Models with AI
SI003 Tripo Tripo OpenAPI docs
SI004 Tripo Pricing | Tripo OpenAPI docs
SI005 Tripo Developers Tripo Developers — AI 3D Generation API
SI006 Tripo AI Terms of User Agreement
SI007 GitHub / VAST-AI-Research tripo-python-sdk/docs/API.md at master
SI008 Hong Kong Company List Holymolly Limited company registration profile
SI009 ACCESS Newswire Tripo, the Frontrunner of 3D AI Boom, Supercharges New Era in Content Creation with 3.0 Upgrade
SI010 Forbes Under 30 Alum’s 3D Model Startup Hits Unicorn Status Amid AI Frenzy
SI011 Forbes Tripo Grows 3D AI Business, Eyes Consumer Future With 3D TikTok
SI012 Yahoo Tech / South China Morning Post China-founded Tripo AI updates 3D content creation platform as users more than double
SI013 Auganix Tripo AI Raises $50M and Unveils New 3D Generation Models
SI014 AI Weekly Tripo AI Hits 10M Users as Vast Raises $200M
SI015 AIThority Tripo AI Raises Nearly $200 Million in Series A+ and Series A++ Financing to Advance AI 3D and World Model Roadmap
SI016 Cyzone 融资丨VAST完成超10亿元A3轮融资
SI017 CB Insights Tripo AI - Products, Competitors, Financials, Employees, Headquarters Locations
SI018 InforCapital Tripo AI - Deep Tech Startup, $400M Raised | InforCapital
SI019 Medium Tripo AI Technical Teardown 2026 — Test the Mesh Before You Build Around It
SI020 Costbench Tripo AI Pricing 2026: Free, Professional, Advanced & Premium Plans
SI021 Apidog How to Use Tripo 3D API: Complete Developer Guide
SI022 Meshy Meshy Official Pricing: Free, Pro, Studio & Enterprise Plans
SI023 Luma Plans & Pricing | Luma
SI024 Adobe Compare plans that include generative AI | Adobe Firefly
SI025 Tripo AI Platform of Tripo AI
SE001 Tripo AI AI 3D Model Generator from Text & Images | Tripo 3D
SE002 Tripo AI Tripo Studio Pricing | Generate High-Quality 3D Models with AI
SE003 Tripo AI Tripo Developers — AI 3D Generation API
SE004 Tripo AI AI 3D Model Generator — Create & Animate 3D Characters | Tripo AI
SE005 Tripo AI AI Auto Rigging Tool for 3D Characters & Animation - Tripo AI
SE006 Tripo AI 3D Model Segmentation AI - Tripo AI
SE007 Tripo AI AI Remesh 3D model - Clean Quad Retopology - Tripo AI
SE008 Tripo AI AI Texture Generator for 3D Models - Tripo AI
SE009 Tripo AI Which AI 3D Format Should You Use: OBJ, FBX, or GLB?
SE010 Tripo AI How to Check if Your AI 3D Model is Game-Ready
SE011 Tripo AI How to Export AI 3D Models to Roblox with Correct Scale
SE012 Tripo AI The Ultimate Guide to the Tripo-ComfyUI Plugin: From Installation to Full Usage
SE013 Tripo AI Complete Tripo-Blender Plugin Tutorial: Transform Images to 3D Models in Minutes
SE014 Tripo AI How to Create Clean Meshes in Blender & Tripo AI: The Complete Guide
SE015 Tripo AI How to Fix Holes in 3D Mesh and Smooth Edges Like a Pro
SE016 Tripo AI How Makerworld and Tripo Revolutionized 3D Printing for Creators
SE017 Tripo AI Boost Your 3D Workflow: How to Set Up Tripo in Blender and Sync with Cursor
SE018 Tripo AI Unlocking the Future of Design: How Image to 3D Model AI is Transforming Industries
SE019 arXiv TripoSG: High-Fidelity 3D Shape Synthesis using Large-Scale Rectified Flow Models
SE020 GitHub / VAST-AI-Research GitHub - VAST-AI-Research/TripoSG: TripoSG: High-Fidelity 3D Shape Synthesis using Large-Scale Rectified Flow Models
SE021 GitHub / VAST-AI-Research GitHub - VAST-AI-Research/TripoSR: TripoSR: Fast 3D Object Reconstruction from a Single Image
SE022 GitHub / VAST-AI-Research GitHub - VAST-AI-Research/tripo-python-sdk: Official Tripo3d Python SDK
SE023 ComfyUI Tripo Partner Nodes Model Generation ComfyUI Official Example - ComfyUI
SE024 GitHub / VAST-AI-Research GitHub - VAST-AI-Research/ComfyUI-Tripo: Official custom nodes for using Tripo in ComfyUI.
SE025 GitHub / VAST-AI-Research GitHub - VAST-AI-Research/tripo-3d-for-blender: Official extension for Blender
SE026 GitHub API Repository metadata for VAST-AI-Research/TripoSR
SE027 GitHub API Repository metadata for VAST-AI-Research/TripoSG
SE028 GitHub API Repository metadata for VAST-AI-Research/ComfyUI-Tripo
SE029 GitHub API Repository metadata for VAST-AI-Research/tripo-python-sdk
SE030 GitHub API Repository metadata for VAST-AI-Research/tripo-3d-for-blender
SE031 Apidog How to Use Tripo 3D API: Complete Developer Guide
SE032 TechTimes Tripo AI Debuts Production-Grade Native 3D Diffusion Model Following GDC 2026 Showcase
SE033 Tech On Play Tripo AI Review: Is 3D Model Generation Finally Good? [2026 Real Test]
SE034 MakerStack Tripo AI Review (2026): Pricing, Features & Honest Verdict
SE035 The Stanford Daily Tripo AI is revolutionizing AI generated 3D models
SE036 Tripo API OpenAPI task endpoint response
SE037 Medium Tripo AI Technical Teardown 2026: Test the Mesh Before You Build Around It
SU001 Tripo AI Tripo Studio Pricing | Generate High-Quality 3D Models with AI
SU002 Tripo AI Sony × VAST: Opening the Future of 3D Creation
SU003 Tripo AI How E-commerce Users Can Use Tripo AI for 3D Modeling and AI Product Prototyping
SU004 Tripo AI Exploring Tripo AI: Powerful 3D Workflows for Diverse Creator Types
SU005 Tripo AI Two-Week Solo Game-Demo Sprint: How AI Tools + Unreal Handle the Whole Pipeline
SU006 Tripo AI The Future of 3D AI Tools in Game Development: Insights from the AI x Game Dev Survey 2024
SU007 Tripo AI How Tripo AI Revolutionizes 3D Image Generation for Architects
SU008 Tripo AI How Makerworld and Tripo Revolutionized 3D Printing for Creators
SU009 Tripo AI Rigging AI 3D Models For Indie Game Success | Tripo AI
SU010 Tripo AI Scaling AR Virtual Try-On Pipelines Across 1,000+ Fashion SKUs
SU011 Tripo AI Implementing 3D Generation APIs for E-Commerce Catalog Scaling
SU012 Tripo AI Generate Film-Ready 3D Assets 2026 Fast with Tripo AI | Tripo AI
SU013 Yahoo Tech / SCMP syndication China-founded Tripo AI updates 3D content creation platform as users more than double
SU014 AI Weekly Tripo AI Hits 10M Users as Vast Raises $200M
SU015 ACCESS Newswire Tripo, the Frontrunner of 3D AI Boom, Supercharges New Era in Content Creation with 3.0 Upgrade
SU016 Forbes Tripo Grows 3D AI Business, Eyes Consumer Future With 3D TikTok
SU017 KR Asia Vast’s Tripo AI used in NetEase game
SU018 The Stanford Daily Tripo AI is revolutionizing AI generated 3D models
SU019 MakerStack Tripo AI Review (2026): Pricing, Features & Honest Verdict
SU020 Medium Tripo AI Technical Teardown 2026: Test the Mesh Before You Build Around It
SU021 CostBench Tripo AI Free Plan 2026: What You Get for $0
SU022 CostBench Tripo AI Pricing 2026: Free, Professional, Advanced & Premium Plans
SU023 Digital Media Net Tripo AI Debuts Production-Grade Native 3D Diffusion at GDC 2026 – Digital Media Net
SU024 Tech On Play Tripo AI Review: Is 3D Model Generation Finally Good? [2026 Real Test]
SU025 GitHub API Repository metadata for VAST-AI-Research/TripoSR
SU026 GitHub API Repository metadata for VAST-AI-Research/ComfyUI-Tripo
SU027 GitHub API Repository metadata for VAST-AI-Research/tripo-python-sdk
SU028 ComfyUI Tripo Partner Nodes Model Generation ComfyUI Official Example - ComfyUI
SU029 Apidog How to Use Tripo 3D API: Complete Developer Guide
SR001 Tripo AI Tripo AI Terms of User Agreement
SR002 Tripo AI Tripo AI Privacy Policy
SR003 Tripo AI Tripo Studio Pricing | Generate High-Quality 3D Models with AI
SR004 Bureau of Industry and Security Homepage | Bureau of Industry and Security
SR005 Bureau of Industry and Security BIS advanced computing guidance
SR006 International Trade Administration China - U.S. Export Controls
SR007 U.S. Government Accountability Office Export Controls: Commerce Implemented Advanced Semiconductor Rules and Took Steps to Address Compliance Challenges
SR008 Congressional Research Service U.S. Export Controls and China: Advanced Semiconductors
SR009 U.S. Copyright Office Copyright and Artificial Intelligence | U.S. Copyright Office
SR010 China Briefing China Releases New Draft Regulations for Generative AI
SR011 SCIO Chinese internet platforms punished for AI-generated content labeling violations
SR012 NPC Observer Cybersecurity Law - NPC Observer
SR013 The State Council of the PRC China unveils guidelines to regulate, boost innovative development of AI agents
SR014 OpenAI All models | OpenAI API
SR015 Microsoft Microsoft Foundry Models overview - Microsoft Foundry
SR016 Microsoft Azure Azure OpenAI in Foundry Models | Microsoft Azure
SR017 Financial Times How a looming talent shortage threatens the AI boom
SR018 Forbes The Chinese AI Blockade Is Coming
SR019 MakerStack Tripo AI Review (2026): Pricing, Features & Honest Verdict
SR020 Medium Tripo AI Technical Teardown 2026: Test the Mesh Before You Build Around It
SR021 Tech On Play Tripo AI Review: Is 3D Model Generation Finally Good? [2026 Real Test]
SR022 Yahoo Tech / SCMP syndication China-founded Tripo AI updates 3D content creation platform as users more than double
SR023 Forbes Tripo Grows 3D AI Business, Eyes Consumer Future With 3D TikTok
SR024 AI Weekly Tripo AI Hits 10M Users as Vast Raises $200M
SR025 ACCESS Newswire Tripo, the Frontrunner of 3D AI Boom, Supercharges New Era in Content Creation with 3.0 Upgrade
SR026 TechTimes Tripo AI Debuts Production-Grade Native 3D Diffusion Model Following GDC 2026 Showcase
SR027 Digital Media Net Tripo AI Debuts Production-Grade Native 3D Diffusion at GDC 2026 – Digital Media Net
SR028 Business Law Today from ABA What Business Lawyers Can Learn from the First AI Copyright Fair Use Rulings
SR029 IPWatchdog Copyright and AI Collide: Three Key Decisions on AI Training and Copyrighted Content from 2025
SR030 Adobe Adobe Firefly | Sign in
SR031 NVIDIA Explore Visual Design Models | Try NVIDIA NIM APIs
SV001 Forbes Under 30 Alum’s 3D Model Startup Hits Unicorn Status Amid AI Frenzy VAST ... has become a unicorn startup after raising about $200 million in fresh funding from over a dozen investors including INCE Capital, Genesis Capital and Primavera Capital Group.
SV002 AI Weekly Tripo AI Hits 10M Users as Vast Raises $200M
SV003 Yahoo Finance / GlobeNewswire Tripo AI Raises Nearly $200 Million in Series A+ and Series A++ Financing to Advance AI 3D and World Model Roadmap Tripo AI ... announced the completion of its Series A+ and Series A++ financing rounds, raising nearly $200 million in total.
SV004 The SaaS News Tripo AI Raises $200M Series A Extension Tripo AI has raised nearly $200 million across consecutive Series A+ and Series A++ financing rounds.
SV005 Tripo AI Tripo Studio Pricing | Generate High-Quality 3D Models with AI
SV006 Tripo AI Tripo Developers — AI 3D Generation API
SV007 Tripo AI Sony × VAST: Opening the Future of 3D Creation
SV008 Tripo AI Implementing 3D Generation APIs for E-Commerce Catalog Scaling
SV009 Tripo AI Generate Film-Ready 3D Assets 2026 Fast with Tripo AI | Tripo AI
SV010 Tripo AI How Makerworld and Tripo Revolutionized 3D Printing for Creators
SV011 Forbes Tripo Grows 3D AI Business, Eyes Consumer Future With 3D TikTok
SV012 Yahoo Tech / SCMP syndication China-founded Tripo AI updates 3D content creation platform as users more than double
SV013 ACCESS Newswire Tripo, the Frontrunner of 3D AI Boom, Supercharges New Era in Content Creation with 3.0 Upgrade
SV014 Medium Tripo AI Technical Teardown 2026: Test the Mesh Before You Build Around It
SV015 MakerStack Tripo AI Review (2026): Pricing, Features & Honest Verdict
SV016 Bureau of Industry and Security BIS advanced computing guidance
SV017 U.S. Government Accountability Office Export Controls: Commerce Implemented Advanced Semiconductor Rules and Took Steps to Address Compliance Challenges
SV018 China Briefing China Releases New Draft Regulations for Generative AI
SV019 Adobe ADBE 10K FY24 For the fiscal year ended November 29, 2024.
SV020 CompaniesMarketCap Adobe (ADBE) - Market capitalization As of July 2026 Adobe has a market cap of $88.50 Billion USD.
SV021 CompaniesMarketCap Adobe (ADBE) - Revenue Revenue in 2026 (TTM): $24.45 Billion USD.
SV022 CompaniesMarketCap Roblox (RBLX) - Market capitalization As of July 2026 Roblox has a market cap of $40.38 Billion USD.
SV023 CompaniesMarketCap Roblox (RBLX) - Revenue Revenue in 2026 (TTM): $5.29 Billion USD.
SV024 CompaniesMarketCap Unity Software (U) - Market capitalization As of July 2026 Unity Software has a market cap of $13.40 Billion USD.
SV025 CompaniesMarketCap Unity Software (U) - Revenue Revenue in 2026 (TTM): $1.92 Billion USD.
SV026 Multiples.vc Public Software Valuation Multiples — July 2026 Design and engineering software commands premium multiples, as companies like Autodesk and Adobe successfully integrate AI features.
SV027 Acquiry SaaS Valuation Multiples in 2026: What the Data Actually Shows 4-7x ARR multiple, non-AI SaaS (2026) ... 8-15x ARR multiple, AI-native SaaS (2026).
SV028 TechCrunch Runway, best known for its video-generating AI models, raises $308M With products like Gen-4 and its recently launched API for video models, Runway hopes to hit $300 million in annualized revenue this year.
SV029 TechCrunch AI video startup Runway raises $315M at $5.3B valuation, eyes more capable world models AI video-generation startup Runway has raised a $315 million Series E round, nearly doubling its valuation to $5.3 billion.
SV030 TechCrunch China's Moonshot AI raises $2B at $20B valuation as demand for open source AI skyrockets Moonshot AI ... has raised about $2 billion at a valuation of $20 billion.
SV031 CNBC Alibaba-backed startup Moonshot AI's valuation is up $500 million, sources say, after its rivals IPO in Hong Kong Moonshot was closing a funding round that will value it at least $500 million higher than the December round.
SV032 CNBC AI startups raised $104 billion in first half of year, but exits tell a different story The dominant exit trend right now is frequent but lower-value acquisitions and fewer IPOs with significantly higher value.
SV033 Eqvista Top 100 AI Startups by Valuation (2026) Moonshot AI $20B ... Runway $5.3B.