Meshy
AI 3D 生成平台,拥有 12M+ 用户和约 12x ARR 增长
Meshy 是 AI 3D 生成里规模和心智领先的玩家,用户基础和增长都很突出;但约 $1.5B 估值相当于约 50x ARR,变现尚未验证,还要面对大厂、开源竞争加速,以及中国关联投资人带来的暴露。
封面要素
公司概况
Meshy 是一家扎根 Silicon Valley 的生成式 AI 公司,平台可在约一分钟内把文本提示和参考图像转成可投产的 3D 模型,并提供 AI 贴图、自动绑定与动画、3D 打印工作流和 REST API。公司由 MIT 训练背景的研究员 Ethan Hu 于 2021 年创立,注册用户已超过 1200 万,生成模型超过 1 亿;2026 年 7 月,公司以 $1.5 billion 估值融资近 $400 million,成为迄今 AI-3D 公司最大一轮融资。
- 成立时间
- 2021-01-01
- 创始人
- Ethan Hu
- 创立地点
- San Jose, California, United States
- 总部
- San Jose, California, United States
- 产品
- 文本生成 3D 与图像生成 3D 的 AI 生成能力,叠加 AI PBR 贴图、自动绑定 / 动画、重网格、Meshy 3D Agent、面向 3D 打印的 Auto Split、多格式导出(FBX、OBJ、GLB、USDZ、STL)、引擎插件,以及带企业控制项的 REST API。
- 客户
- 游戏开发者、3D 艺术家、产品和工业设计师、制造商、3D 打印创作者、电商品牌和教育用户。
- 商业模式
- 免费增值模式,叠加付费订阅层级(Pro/Premium/Studio/Ultra)、企业方案,以及按积分计费的 API 授权。
- 阶段
- Series B
- 融资情况
- 2026 年 7 月完成近 $400M Series B 轮,IDG Capital、Matrix Partners China 和 Monolith Management 领投,投后估值 $1.5B;老股东 Granite Asia、Sequoia China(HongShan)、BAI Capital 和 Source Code Capital 参投。
执行摘要
主要优势
- 品类领先的规模:注册用户超过 12M,平台已生成超过 100M 个模型
- 以 $1.5B 估值完成近 $400M Series B,是迄今最大 AI-3D 融资,ARR 同比约 12 倍增长
- 产品面铺得宽:文本 / 图像转 3D、纹理、绑定、3D 打印闭环和 API 都已覆盖,生态和企业采用信号也强
主要风险
- 约 50x ARR 估值偏高,需要多年持续执行和变现爬坡来支撑
- OpenAI、Google、NVIDIA、Adobe 切入 3D,加上开源 3D 模型快速进步,竞争可能触及生存线
- 作为美国公司,中国关联投资人基础抬高 CFIUS / 地缘政治暴露;AI 训练数据版权也有风险
未决问题
- 累计融资、股权结构表、员工数、毛利率、烧钱速度和现金跑道均未披露
- 免费转付费、留存 / NRR、单位经济(CAC/LTV)未公开
- ARR 在不同来源里口径不一致(官方 $30M,第三方标注更高),限制收入质量承销
目录
01公司概览
1.1 身份与产品边界
Meshy 更像一家扎根 Silicon Valley 的私有 AI 3D 创作平台,而不是一个狭义设计工具。官方产品页和帮助中心把公司定义在文本生成 3D、图像生成 3D、AI 贴图、动画和 API 工作流上,输出面向游戏开发、3D 打印、AR/VR、产品设计、教育和内容创作。生成式 AI 公司常把价值说得抽象,Meshy 的产品主张却相当具体:用户从文本、图像或概念出发,预览生成资产,再导出到标准 3D 工作流。公开来源支持 San Jose / Silicon Valley 是运营重心、2021 年是创立年份,但最强的官方页面更强调使命和规模,而不是注册实体或办公地址细节。因此,本章把 Meshy 视为一家定位美国的私有 Series B 公司;其准确法律实体、股权结构和员工基础仍需一手尽调。[CO001, CO002, CO003, CO004, CO024, CO026]
| 指标 | 数值 / 状态 | 日期 | 置信度 | 缺口 |
|---|---|---|---|---|
| 身份 | AI 驱动的 3D 内容创作平台 | 2026-07-24 | 高 | 尚未引用确切法律实体和注册文件 |
| 总部 / 运营所在地 | Silicon Valley / San Jose 地区 | 2026-07-24 | 中 | 注册办公地和法律住所需要一手文件 |
| 成立 | 2021 | 历史 | 中 | 部分公开叙事强调从 2023 年产品发布算起的三年运营历史 |
| 最新阶段 | Series B 轮私营公司 | 2026-07-21 | 高 | 除公开轮次之外的条款仍未披露 |
| 最新估值 | $1.5B 投后口径 | 2026-07-21 | 高 | 一份摘要使用超过 $1.38B 的口径 |
| 最新轮次 | 近 $400M Series B 轮 | 2026-07-21 | 高 | Series B 前累计融资尚未完整复原 |
| 注册用户 | 12M+ | 2026-07-21 | 中 | 公司披露,未经审计 |
| 已生成模型 | 100M+ | 2026-07-21 | 中 | 公司披露,未经审计 |
| ARR | ~$30M 公开 GDC 里程碑 | 2026-03-19 | 中 | 36Kr 后续报道 >$40M;审计收入不可得 |
| 员工数 | 2026-07-24 | 中 | 当前准确员工数缺少支撑 | |
| 定价 | 免费;Pro $20;Premium $40;Studio $60;Ultra $100;Enterprise | 2026-07-24 | 中 | Enterprise 合同定价未披露 |
私营公司 KPI 来自公开发布和官方页面;null 表示未找到可靠公开数字。
[CO001, CO003, CO004, CO009, CO010, CO017]概览逻辑把创始人能力、产品触点、自助变现、企业工作流、资本和尽调缺口串起来。
[CO006, CO024, CO026, CO027, CO034, CO037]1.2 领导层与关键人物依赖
公开领导层记录高度围绕创始人展开。Meshy 反复把 Ethan Hu 列为创始人兼 CEO,创始人与市场匹配的证据较强:Hu 被描述为受过 MIT 训练、研究计算机图形和 AI 的博士,并与 Taichi 相关联;Taichi 是用于高性能图形场景的开源 GPU 编程语言。这一背景与 Meshy 的技术难题贴合:快速生成几何、纹理、可动画化资产和生产导出格式。反向含义不是已知人事问题,而是治理透明度不足。已审阅公开来源没有披露完整高管梯队、独立董事或董事会构成。投资判断需要确认:如果创始人仍是主导技术和对外声音,Meshy 是否已有运营领导者能把企业销售、基础设施、安全和财务规模化。[CO005, CO006, CO007, CO008, CO021, CO022]
| 人物 | 职务 | 背景 | 创始人-市场匹配或职能覆盖 | 关键人依赖 |
|---|---|---|---|---|
| Ethan Hu | 创始人兼 CEO | MIT 计算机图形学与 AI 博士;Taichi 创建者 | 技术上高度匹配图形、几何、仿真和 AI 3D 生成 | 高:公开叙事和引述集中在 Hu 身上 |
| 未具名高管团队 | 已审阅来源未公开逐一列出 | 官方页面强调全球团队资历,而非具名 C-suite | 销售、财务、安全和运营的职能覆盖无法映射 | 重大:尽调应要求组织架构图和董事会材料 |
| 全球技术团队 | 公司描述的全球团队,拥有 MIT/Harvard 校友和 NVIDIA/Microsoft/Google 老兵 | 团队资历主张见于 Meshy 关于页面 | 支撑招聘叙事,但不能映射治理 | 中等:团队深度无法归因到具名领导者 |
由于公开来源未提供完整高管或董事会名单,列举并不完整。
[CO005, CO006, CO007, CO008, CO022, CO023]1.3 融资、估值与利益相关方地图
最清晰的公司阶段信号是 2026 年 7 月 Series B:Meshy 宣布以 $1.5 billion 估值融资近 $400 million,Yahoo 刊载了同一份公司分发稿,独立摘要也重复了主要融资事实。本章采用 $1.5 billion 作为基准估值,因为该披露直接且带时间戳;The SaaS News 的超过 $1.38 billion 数字只作为较低的冲突摘要,而不是全报告锚点。公开列名投资方包括 IDG Capital、Matrix Partners China、Monolith Management、Granite Asia、HongShan 或 Sequoia China、BAI Capital 和 Source Code Capital。募资用途宽泛但逻辑一致:基础模型研发、基础设施和全球企业扩张。风险在于上一轮条款、准确持股、债务、老股转让和投资者权利仍未公开;同时,对一家呈现为 Silicon Valley 公司的企业而言,已列名投资方与中国的关联度具有实质性。[CO009, CO010, CO011, CO012, CO013, CO014]
| 利益相关方 | 角色 | 控制权或经济重要性 | 尽调问题 |
|---|---|---|---|
| IDG Capital | Series B 轮具名支持方 / 领投组 | 可能是近 $400M 轮次的主要参与方 | 确认出资规模、董事会权利和信息权 |
| Matrix Partners China | Series B 轮具名支持方 / 领投组 | 最新轮次中的重要中国关联风投 | 确认基金实体、治理权利和地缘政治敞口 |
| Monolith Management | Series B 轮具名支持方 / 领投组 | 公开融资稿中的具名主要支持方 | 确认分配额度、pro rata 权利和关系历史 |
| Granite Asia | 既有 / 参投投资人 | 释放存量支持和可能跟投信心信号 | 确认上一轮进入价格和 pro rata 行权 |
| HongShan / Sequoia China | 既有 / 参投投资人 | 股权结构叙事中的知名中国风投敞口 | Sequoia China 更名后明确具体实体和治理权利 |
| BAI Capital | 既有 / 参投投资人 | 具名存量投资人,带亚洲敞口 | 确认任何经纪交易商或配售结构影响 |
| Source Code Capital | 既有 / 参投投资人 | 具名存量投资人,连接中国生态 | 确认持股和战略影响力 |
| 企业客户和合作伙伴 | 商业证明点 | Meshy 在游戏、3D 打印、消费和博物馆等细分中提及 | 区分活跃付费客户与合作伙伴、试点和营销引用 |
投资人列举基于公开 Series B 报道;没有股权结构文件,仍不完整。
[CO012, CO013, CO035, CO036, CO037, CO039]1.4 规模、KPI 与披露质量
在私有 AI 工具公司里,Meshy 的头部规模叙事异常具体,但仍未经审计。2026 年 7 月公告称注册用户超过 1200 万、已创建模型超过 1 亿,ARR 同比增长约 12x。2026 年 3 月 GDC 公告给出更精确的 $30 million ARR 里程碑;36Kr 之后报道 2026 年 4 月 ARR 超过 $40 million,并描述了激进的 AI 原生内部运营做法。这些数字方向上支持增长叙事,但不等同于已审计收入、队列留存、毛利率、净收入留存或现金消耗证据。员工数尤其未解:36Kr 提到 150 人运营理念,但其语境更像内部哲学,而不是已验证的当前员工人数。概览部分因此把缺乏支撑的封面指标列为空值或缺口项,而不是用受限数据库背后的估算数填补。[CO017, CO018, CO019, CO020, CO021, CO022]
头部 KPI 显示这是一家达到独角兽规模的私营公司,使用量说法强劲,但经审计财务细节仍缺失。
ARR 采用公开 GDC 里程碑;36Kr 后续报道了更高数字,本报告将其作为差异跟踪。
[CO011, CO017, CO018, CO019, CO021, CO022]1.5 里程碑与反向检查
里程碑记录显示,Meshy 从产品发布到规模化商业叙事推进很快。第三方资料把 Meshy-1 的首次公开发布放在 2023 年 10 月;随后,官方和公司分发来源显示其连续扩展到 Meshy 6、Meshy Labs、Meshy 3D Agent、Auto Split、可打印性、Unity 工作流,以及与 Formlabs 相关的打印工作流。反向筛查没有在已审阅来源中发现公开诉讼、制裁项目或监管执法行动,但发现了三个重要尽调提示:ARR 和估值摘要互相冲突、投资者数据库访问受限、部分产品工作流引用究竟是正式合作还是教程仍有歧义。这个区分重要,因为后续章节不能在没有源头确认的情况下,把产品营销证明转成合同客户或渠道证据。[CO030, CO031, CO032, CO033, CO036, CO038]
| 日期 | 事件 | 类型 | 金额 / 估值 / 状态 | 参与方 | 含义 |
|---|---|---|---|---|---|
| 2021 | Ethan Hu 在 San Jose 创立 | 创立 | 已成立 | Ethan Hu | 确立标准创立年份,但需要一手注册文件确认 |
| 2023-10-19 | 第三方简介提及 Meshy-1 公开发布 | 产品 | 首次公开发布 | Meshy | 标记产品时代起点,用于对齐三年增长说法 |
| 2026-03-18 | Meshy 6 发布见于公司产品叙事 | 产品 | 新模型生成版本 | Meshy | 在 GDC 前强化产品速度叙事 |
| 2026-03-19 | Meshy Labs 和 Black Box 在 GDC 2026 发布 | 产品 | $30M ARR 里程碑;10M+ 用户;100M+ 模型 | Meshy / GDC 观众 | 将故事从资产生产扩展到 AI-native 玩法 |
| 2026-05-20 | 36Kr 报道 ARR、增长、利润率和内部 AI-native 实践 | 反向 | > $40M ARR 主张;150 人团队概念 | 36Kr / Ethan Hu 评论 | 有用但未经审计,且与标准 ARR 里程碑部分不一致 |
| 2026-07-21 | Series B 轮宣布 | 融资 | 近 $400M;$1.5B 估值 | IDG、Matrix China、Monolith、既有投资人 | 定义当前阶段和估值锚 |
| 2026-07-21 | 融资稿披露新的规模指标 | 规模 | 12M+ 用户;100M+ 模型;ARR 同比增长 ~12x | Meshy | 抬高商业表现门槛,但仍是公司披露 |
| 2026-07-21 | Meshy 3D Agent 宣布向所有注册用户开放 | 产品 | 可直接打印的输出,包括 FBX、OBJ、GLB、STL | Meshy | 显示其向对话式端到端 3D 工作流推进 |
| 2026-07-21 | 面向 3D 打印发布 Auto Split | 产品 | 一键拆分可打印部件 | Meshy | 加深 3D 打印差异化和物理输出用例 |
| 2026-07-24 | 已审阅 Meshy x Formlabs 教程 | 合作 | 工作流证明,而非已确认合同 | Meshy / Formlabs 引用 | 需先尽调,才能视为正式渠道合作 |
| 2026-07-24 | 已审阅受限数据库档案 | 反向 | Crunchbase、Tracxn、PitchBook 受限或不可读 | 独立数据库 | 股权结构、员工数和既往条款仍是私有证据缺口 |
时间线结合公开发布、融资、产品、规模、合作和反向 / 披露事件;日期使用来源发布或审阅日期。
[CO009, CO010, CO017, CO018, CO019, CO020]Meshy 从 2021 年创立走到 2026 年独角兽融资,中间靠产品发布和公司披露的规模里程碑推进。
[CO030, CO031, CO032, CO033, CO038, CO041]1.6 展项
02市场分析
2.1 市场边界与替代品
Meshy 的相关市场不是整个 3D 软件行业。核心边界是借助 AI 从文本、图像、对话或 API 调用生成可编辑 3D 资产;当这些能力替代手工资产制作时,市场应包括资产生成、贴图、重网格 / 拓扑、导出和开发者集成。若艺术家仍在 Maya、Blender 或 Substance 里手工建模,传统 DCC 授权开支不应计入,尽管这些工具仍是工作流背景和替代预算。素材商店购买也不应计入,除非买方正在用生成的定制资产替代库存模型。元宇宙世界、数字孪生、3D 渲染、AR/VR 商务和 3D 打印等相邻领域重要,因为它们消耗 3D 内容;但它们应被视为需求池,而不是整体计入 Meshy 的 TAM。这个边界可以防止重复计算。[CM001, CM002, CM003, CM005, CM006, CM022]
| 类别 | 包含支出 | 排除支出 | 买方 / 付款方 | 关联度 |
|---|---|---|---|---|
| 核心 AI 3D 资产生成 | 文本 / 图像 / 对话生成 3D、贴图、拓扑、导出、API 使用 | 仅手工使用的 DCC 席位和无关 2D 图像生成 | 创作者、工作室、开发者、产品团队 | Meshy 直接收入池 |
| 传统 DCC 工具 | 与生成工作流绑定的 DCC 插件或 AI 功能 | 仅用于手工建模的基础 Maya/Blender/Substance 许可证 | 艺术家、工作室、设计部门 | 替代品和集成界面 |
| 资产商店 | 库存 3D 资产的自定义生成替代 | 未被生成替代的一次性库存模型购买 | 独立开发者、营销人员、电商团队 | 现状替代品 |
| 游戏和 VFX | AI 生成道具、环境、纹理、原型资产 | 完整游戏软件或娱乐收入 | 工作室、发行商、VFX 制作方 | 近期 SAM 锚 |
| 3D 打印 / 创客 | 生成可打印模型、修复 / 拆分、STL 导出 | 打印机硬件和材料 | 创客、爱好者、打印品牌 | 相邻采用楔子 |
| 数字孪生 / AR-VR / 电商 | 用于沉浸式可视化的 3D 内容创作 | 完整仿真平台、头显或商业 GMV | 制造商、零售商、空间计算团队 | 上行需求池 |
边界表把直接 AI 3D 生成收入与相邻需求池和替代品分开;它不是市场规模加总。
[CM001, CM002, CM003, CM005, CM006, CM022]2.2 多个市场规模口径
直接自上而下口径支持一个数十亿美元级、但仍在成形的品类:AI 3D 资产估算在 2026 年约为 $3.23 billion,到 2030 年约为 $9.4 billion;更窄的 AI 3D 资产生成和贴图预测到 2036 年达到 $12.84 billion。游戏专属 SAM 更保守,游戏中的生成式 AI 预计 2026 年为 $2.21 billion、2030 年为 $5.09 billion。更宽的元宇宙、数字孪生和 3D 渲染市场可作为天花板和需求信号,但不是立即可触达收入。尽调中的可投资视角应是一组区间:低位对应游戏优先采用,基准对应 AI 3D 资产软件,上行则要求打入数字孪生、AR/VR、电商和打印工作流。[CM007, CM008, CM009, CM010, CM011, CM012]
| 视角 | 发布方 | 年份 / 地区 | 数值 | CAGR | 方法 / 基础 | 置信度 | 局限 |
|---|---|---|---|---|---|---|---|
| 直接 AI 3D 资产 | 3D AI Studio / R&M 摘要 | 2026 全球 | $3.23B | ~31% | 综合分析师对生成式 AI 3D 资产市场的估计 | 中 | 竞争对手撰写来源;需与付费报告核对 |
| 直接 AI 3D 资产预测 | 3D AI Studio / R&M 摘要 | 2030 全球 | $9.4B | ~31% | 基于 2026 AI 3D 资产市场基数的预测 | 中 | 品类边界可能包含非 Meshy 用例 |
| AI 3D 生成与贴图 | GII / Meticulous Research | 2036 全球 | $12.84B | 20.8% | 按资产类型、AI 模型、集成方式和最终用户预测 | 中 | 周期长,模型误差扩大 |
| 游戏中的生成式 AI | The Business Research Company | 2026 全球 | $2.21B | 23.1% 2025-26 | 游戏商品 / 服务中的生成式 AI 收入 | 中 | 包含 3D 资产之外的游戏 AI |
| 游戏生成式 AI 预测 | The Business Research Company | 2030 全球 | $5.09B | 23.2% | 面向 2030 的游戏专属 AI 预测 | 中 | 仍宽于 Meshy 3D 创作 |
| 3D 渲染替代池 | Mordor Intelligence | 2026 全球 | $5.23B | 21.63% 至 2031 | 3D 渲染市场预测 | 中 | 渲染相邻,但不是生成资产收入 |
| 数字孪生相邻市场 | MarketsandMarkets | 2030 全球 | $149.81B | 47.9% 2025-30 | 数字孪生平台自上而下与自下而上预测 | 低 | 大多数支出不是 3D 资产生成 |
| 元宇宙相邻市场 | MarketsandMarkets | 2030 全球 | $1,303.4B | 48.0% 2023-30 | 广义元宇宙硬件 / 软件 / 服务市场 | 低 | 作为 TAM 过宽;仅作背景 |
所有金额均为 USD;置信度指 Meshy 可服务市场的证据置信度,不是发布方声誉。
[CM008, CM009, CM010, CM011, CM016, CM017]受约束的规模测算栈把直接 AI 3D 生成与更宽泛的邻近市场分开。
数值沿用所引发布方的单位与年份;各层边界不同,不能相加。
[CM009, CM010, CM016, CM017, CM043, CM018]以十亿美元为单位,展示 2026 年直接和近直接 AI 3D / 游戏市场估算区间。
所有数值均为 2026 年十亿美元口径;3D 渲染高位参照属于邻近 / 替代收入,不是纯 Meshy TAM。
[CM008, CM010, CM019, CM042]2.3 买方、用户与付款方分层
买方分层异常宽,因为同一个资产生成原语既可服务自助创作者,也可嵌入团队基础设施。独立游戏开发者和爱好者通常一人兼具用户、买方和付款方身份,价格、速度、导出格式和清理便利性因此决定购买。工作室、VFX 公司和发行商则把用户与付款方拆开:艺术家和技术总监使用工具,制作人、工具负责人或中央技术组控制预算和治理。产品设计、制造和电商团队来自不同预算池,更需要工作流集成、品牌控制和权利清晰度。教育和创客群体能扩大触达,但若不能转成订阅或 API 用量,近期收入可靠性较低。[CM003, CM004, CM015, CM027, CM033, CM034]
| 细分 | 用户 | 买方 | 付款方 / 预算负责人 | 工作流 | 采用触发 |
|---|---|---|---|---|---|
| 独立游戏 / 创作者 | 开发者或艺术家 | 本人或小团队负责人 | 创作者订阅 / 项目预算 | 提示词或图像生成道具 / 环境,再编辑 / 导出 | 速度和低前期成本 |
| AAA 与中型工作室 | 艺术家、技术美术、工具工程师 | 制片人或工具总监 | 工作室技术或制作预算 | API / 插件生成资产进入引擎 / DCC 管线 | 资产吞吐和管线适配 |
| VFX / 电影预演 | 概念艺术家或 VFX 通才 | VFX 主管 | 项目或工作室工具预算 | 快速迭代概念几何与纹理 | 概念周转速度 |
| 产品设计 / 制造 | 工业设计师、可视化工程师 | 设计负责人或 PLM / 数字孪生负责人 | 研发、可视化或数字化转型预算 | 生成变体、可视化产品,并接入数字孪生工作流 | 原型速度和可视化需求 |
| 电商 / 品牌 | 商品运营、创意、3D 商务团队 | 数字商务负责人 | 营销、电商或商品目录预算 | 为 3D / AR 体验生成产品视觉素材 | 转化提升和目录覆盖率 |
| 3D 打印 / 创客 | 创客或打印店经营者 | 同一用户或店主 | 创客订阅、打印店运营、合作伙伴渠道 | 生成可打印模型、修复 / 拆分并导出 STL | 打印就绪成功率和新奇需求 |
| 教育 / 爱好 | 学生、教育者、爱好者 | 教师、实验室或个人 | 教育或个人预算 | 低门槛创作和学习工作流 | 可及性和试验需求 |
细分是基于产品工作流、公开定价、文档和已报道用例推断的尽调地图,并非 Meshy 披露的收入拆分。
[CM003, CM004, CM015, CM033, CM034, CM035]不同客群的用户、买方和付费方可能是同一人,也可能是企业职能。
买方定性地图由 Meshy 工作流和公开市场用例证据推断。
[CM033, CM034, CM035, CM036, CM037, CM038]2.4 增长驱动与采用约束
增长来自真实的生产痛点:游戏、VFX、电商、数字孪生、3D 打印和 AR/VR 都需要更多 3D 资产,而稀缺专业人力很难低成本供给。Meshy 自身主张强调显著压缩时间和成本,独立 AI 3D 评论也描述了同一转变:从多天管线压到分钟级。不过,这个市场尚未去风险。专业买方仍然关注拓扑、可绑定性、纹理质量、IP 来源、工具集成和资产治理。Gartner 对 AI 周期的反向判断提醒,治理和可靠性限制一旦暴露,广泛 GenAI 热情可能降温。开源 3D 模型、Blender、素材商店和既有 DCC 套件也会限制定价权,除非 Meshy 证明自己具备可投产品质和分发优势。[CM027, CM028, CM029, CM030, CM031, CM032]
| 驱动因素 / 约束 | 方向 | 时间 | 影响 | 尽调追问 |
|---|---|---|---|---|
| 资产生成时间压缩 | 驱动因素 | 当前 | 创作从专家瓶颈转向自助工作流 | 核验客户使用前后的人工小时 |
| 游戏与 VFX 内容量 | 驱动因素 | 当前至 2030 年 | 支撑以游戏优先的 SAM 和工作室管线需求 | 索取按游戏工作室客群拆分的收入 |
| API 与插件集成 | 驱动因素 | 当前 | 可把 Meshy 从网页工具变成嵌入式基础设施 | 审查 API 使用量、正常运行时间和企业安全控制 |
| 空间计算、数字孪生、电商 3D | 驱动因素 | 2026+ | 如果资产达到生产就绪,游戏之外的上行空间会打开 | 验证游戏之外的付费试点 |
| 质量 / 拓扑 / 绑定差距 | 约束 | 当前 | 清理成本居高不下时,会限制专业采用 | 让专业美术做资产 QA |
| IP 来源与治理 | 约束 | 当前 | 权利不清时,企业买家可能推迟部署 | 审查训练数据、赔偿和内容审核条款 |
| 既有 DCC 与资产商店 | 约束 | 持续 | 限制定价权,并造成多平台并用 | 对标 Maya / Blender / Substance / 商店工作流成本 |
| 开源 3D 模型 | 约束 | 2025-2026 年起 | 草稿资产生成商品化,并压低 API 定价 | 对比质量、许可证和单位经济性与开源模型 |
| GenAI 祛魅 / 治理 | 约束 | 当前周期 | 如果可靠性不达预期,买家紧迫感可能下降 | 索取续约、扩张和生产部署证据 |
驱动因素和约束综合市场报告、官方产品证据和一条反向 Gartner 来源;时间判断为定性。
[CM027, CM028, CM029, CM030, CM031, CM032]专业用户转化取决于能否从实验走到生产部署和扩展。
示意性指数,并非实测转化;公开来源未披露 Meshy 的付费漏斗指标。
[CM015, CM029, CM030, CM031, CM032, CM041]2.5 市场规模缺口与尽调含义
市场分析最大的风险是虚假精确。公开分析报告对品类定义不一:有的包括所有 3D 资产生成式 AI,有的包括贴图,有的聚焦游戏,还有的描述相邻可视化、元宇宙或数字孪生池。Meshy 披露的规模证明了需求,但公开来源没有拆出付费客户数、转化率、企业渗透率、流失率或按垂直行业划分的收入。因此,本章保留相互矛盾的口径,而不是强行给出一个 TAM 数字。后续尽调应要求按细分市场提供队列级 ARR、API 与 Web 收入拆分、企业管线部署、注册用户付费转化、按用例划分的获客成本,以及生成资产无需大量人工返工即可通过专业生产评审的证据。注册用户规模可以与低付费转化并存,而企业 API 收入即使账户更少也可能具有实质性,因此这点尤其重要。[CM033, CM034, CM035, CM036, CM037, CM040]
2.6 展项
03竞争对手
3.1 竞争图谱与替代方案集合
Meshy 所处战场比文本生成 3D 这个说法更宽。直接同业包括 Tripo 和 Hyper3D Rodin,因为二者都承诺快速用文本或图像生成可下载 3D 资产。Luma 是相邻公司,因为其当前公开定位更偏创意智能体、图像和视频 API,而不是以网格为核心的资产生产。Kaedim 是团队工作流替代品:这些团队从草图、艺术方向、产品照片或简报出发,想要经检查的生产资产,而不是自助即时草稿。买方任务若是浏览器原生互动 3D 体验,Spline 就会参与竞争。Adobe Substance、Blender、Maya 和 ZBrush 仍是专业控制场景的既有替代品。NVIDIA Omniverse、Google DeepMind 世界模型、Hunyuan3D、TRELLIS 和 Stable Fast 3D 重要,因为它们可能重设买方对成本、控制和模型可得性的预期。这个框架也让比较回到买方中心:关键问题不是哪个模型在展示页里最好看,而是哪种替代方案能为买方特定工作流稳定交付可编辑、可授权、可下游使用的资产。[CP011, CP014, CP016, CP019, CP021, CP024]
| 竞品或替代方案 | 类别 | 规模 / 融资信号 | 目标客群 | 差异化 | 局限或观察项 |
|---|---|---|---|---|---|
| Meshy | AI 文本 / 图像转 3D 平台 | 近 $400M Series B;估值 $1.5B;12M+ 用户;100M+ 模型 | 创作者、游戏开发者、3D 美术、产品设计师 | 一分钟快速生成、大型免费增值漏斗、API、动画、3D 打印闭环 | 反向评测称,生产级几何质量比 Rodin 更粗糙 |
| Tripo AI | AI 文本 / 图像转 3D 直接竞品 | 免费层;年付折扣前 Pro 约 $19.90/月 | 游戏开发者和快速原型用户 | Smart Mesh、低多边形、批量生成、高积分额度 | Meshy 级 ARR 或用户基数证据较少 |
| Hyper3D Rodin | 高保真 AI 3D 直接竞品 | Creator 计划 $30/月,或年付折算 $24/月 | 专业美术、工作室、生产级资产创作者 | 面向生产的几何、UV、纹理、Smart Low-Poly | 实际价格更高,免费增值规模可能更小 |
| Luma AI | AI 重建 / 视频 / 创意代理 | 留存来源未显示 3D 资产专项规模 | 搭建图像 / 视频营销活动和 physical AI 工作流的创意团队 | 前沿视频 / 图像 API 和创意代理工作流 | 比 Meshy 更少以网格和游戏资产为中心 |
| Kaedim | 2D 转 3D 生产工作流 | 公开页面未核验私有定价 | 游戏、产品、电商和营销团队 | 人在回路审核和生产交付 | 小时级工作流,不是即时自助生成 |
| Spline | Web 原生交互式 3D 设计 | 年付席位价格为 $12/月和 $20/月 | 设计团队和 Web 体验构建者 | 浏览器协作编辑器、交互式导出、集成 AI | 不是专业生产级网格生成器 |
| Adobe / Substance 3D | 既有材料和纹理套件 | Adobe 产品线;公开页面显示 Substance 3D Collection | 美术和企业创意团队 | 专业材料和纹理工作流 | 补足或替代纹理环节,不是完整提示词到网格 |
| NVIDIA Omniverse | 大厂模拟和 OpenUSD 平台 | NVIDIA 生态和开发者平台 | 模拟、physical AI、企业开发者 | OpenUSD、SimReady、合成数据、场景优化 | 更像平台基础设施,不是创作者优先生成器 |
| Blender / Maya / ZBrush | 既有手工工具 | Blender 免费;Maya / ZBrush 是成熟商业工具 | 专业美术、工作室、技术美术 | 精度、控制力、管线成熟度、培训生态 | 技能门槛高,首稿更慢 |
| Hunyuan3D、TRELLIS 与 Stable Fast 3D | 开源和研究模型 | 开源仓库和模型发布 | 能自托管或改造模型的开发者和团队 | 边际成本低、可改造、重建研究迭代快 | 运营负担和产品 UX 留给采用者 |
局部图谱聚焦 2026 年公开证据、直接 AI 3D 竞品、既有工具、大厂平台,以及与 Meshy 买家最相关的开源替代方案。
[CP002, CP003, CP011, CP012, CP016, CP017]Meshy 在速度和规模上居前,但在负面评测的生产保真度评分上低于 Rodin。
序数 x=迭代速度 / 规模,y=生产保真度;评分来自公开说法和评测,不是受控基准测试。
[CP026, CP027, CP028, CP029, CP030, CP031]3.2 能力、包装与定价对比
头对头比较显示,Meshy 确有宽度优势。Meshy 把文本生成 3D、图像生成 3D、贴图、API 访问、绑定、600 多个动画片段、多格式导出,以及正在成形的打印履约闭环放在一起。创作者若想从想法一路走到资产,这个组合异常连贯。Tripo 用免费积分、smart-mesh 包装和偏游戏的批量工作流压迫 Meshy。Rodin 则在最难的专业维度压迫 Meshy:几何和拓扑质量。保留下来的 Luma 证据更指向图像 / 视频创意基础设施,而非网格生产;Spline 的强项是 Web 原生协作和互动性。定价同样重要:Blender 免费,Spline 和 Tripo 公布了低门槛付费层级;当输出质量是买方约束时,Rodin 可以拿到更高价格。因此,Meshy 必须证明速度加宽度足以抵消更便宜或更专门化的替代方案。缺少统一第三方基准时,公开功能主张本身不足以证明优势;定价和工作流主张应作为尽调筛选项,而不是最终胜出证明。[CP004, CP005, CP006, CP007, CP008, CP010]
| 采购标准 | Meshy | Tripo | Rodin / Hyper3D | Luma | Kaedim | Spline | 既有工具 / 开源 |
|---|---|---|---|---|---|---|---|
| 文本转 3D | 是;通过网页和 API 按提示词生成网格 | 是;有公开示例和 Studio 计划 | 是;支持文本 / 图像生成 | 留存来源显示并非网格优先 | 由需求简报 / 参考驱动,不是即时自助 | 是;通过 Spline AI | 开源模型可支持文本或图像提示词 |
| 图像转 3D | 是;Meshy 6 和多视角选项 | 是;付费计划支持多视角转 3D | 是;图像转 3D 和 ControlNet 选项 | 视频 / 图像导向更强 | 生产交付的核心输入类型 | 是;通过 Spline AI | Stable Fast 3D 和 Hunyuan3D 强调图像路径 |
| 速度 / 迭代 | 核心模型生成约一分钟 | 评测将其定位为速度导向 | 宣称秒级,但评测优势在保真度 | 高吞吐创意变体 | 文案显示是小时级而非秒级 | 编辑器内 AI 生成秒级完成 | Stable Fast 3D 宣称图像转 3D 只需 0.5 秒 |
| 网格 / 拓扑质量 | 最高约 600K 面的高保真;反向评测指出几何更粗糙 | Smart Mesh 和低多边形工具 | 评测中的生产级几何和四边形拓扑领先者 | 未按网格拓扑评分 | 生产审核闭环 | 聚焦交互式 Web 资产 | 手工工具控制力最高;开源模型表现不一 |
| 动画 / 绑定 | 自动绑定和 600+ 动作片段 | 付费层提供动画和 Smart Mesh | 留存定价文案中不是主打 | 视频优先的创意动效 | 生产资产交付 | Web 场景的交互和动效 | Maya / ZBrush / Blender 的手工绑定 / 雕刻成熟 |
| API / 开发者工作流 | API 需要 Pro,并披露积分 / 速率限制 | 已获取 API 页面,但公开细节有限 | 定价文案提到 API 访问 | 图像 / 视频 API 定位强 | 偏企业工作流,缺少公开 API 证据 | Pro 计划包含 API 和 webhook | 开源仓库支持直接集成模型 |
| 下游格式 / 生态 | FBX、OBJ、GLB、USDZ、STL、BLEND、3MF;Unity / Formlabs 工作流 | 付费计划支持批量导出 | 宣称支持常见 3D 格式 | 视频 / 图像生产格式 | 客户拥有资产,配套审核交付 | 多平台和代码导出 | 原生专业管线和开源管线深度 |
矩阵仅使用公开证据;没有共同第三方测试时,单元格依据定位而非基准测量。
[CP005, CP006, CP007, CP008, CP012, CP013]| 厂商 | 入门套餐 | 公开付费套餐 | 内含能力信号 | 竞争含义 |
|---|---|---|---|---|
| Meshy | 免费计划含每月积分 | Pro $20/月;Studio $60/月;更高 creator / team 层级 | 文本 / 图像转 3D、API 积分、动画、纹理和打印工作流 | 宽免费增值漏斗叠加可变现创作者层级 |
| Tripo | 免费计划含每月 200 积分 | 年付折扣前 Pro 约 $19.90/月;Max 约 $89/月 | Smart Mesh、高网格质量、批量生成、私有模型 | 可在每美元积分数和游戏资产包装上挤压 Meshy |
| Rodin / Hyper3D | 确认前可免费生成 | Creator $30/月,或年付折算 $24/月;直接购买积分 $1.50 | Smart Low-Poly、HD 纹理、烘焙法线、多边形数量选项 | 生产保真度重要时,可支撑更高价格 |
| Spline | 免费开始 | Starter $12/席位/月、Pro $20/席位/月,年付 | 协作编辑器、导出,Pro 含 AI 积分 | 竞争的是 Web 体验工作流,而非纯资产生成 |
| Blender | 免费开源软件 | 免费 | 完整建模、渲染、雕刻、UV 和生态 | 锚定买家为 AI 速度付费,而不是为核心工具付费 |
| Maya / ZBrush / Substance | 商业订阅 | 厂商专属商业计划 | 专业建模、雕刻、动画、材料 | 为 AI 生成输出设定质控基准 |
价格为截至访问日期的公开标价或页面可见计划锚点;实际企业价格和折扣不公开。
[CP004, CP012, CP017, CP023, CP029, CP030]Meshy 拥有最宽的自助式网格工作流,竞争对手则按保真度、视频、网页协作或开放模型分化。
强弱标签为定性判断,只使用已保留的公开证据。
[CP005, CP006, CP007, CP008, CP013, CP015]3.3 差异化、切换成本与工作流护城河
Meshy 最强的护城河不是某个单一功能,而是组合:大用户规模、免费增值漏斗、API 访问、快速预览生成、可下载格式、动画、面向引擎的工作流和 3D 打印相邻场景。团队一旦把提示词、资产历史、API 调用、引擎导入约定和下游工作流接到 Meshy 周围,这些能力就会形成中等切换成本。不过,切换成本并不绝对。AI 3D 买方可以多平台并用:把同一个提示词放到 Rodin、Tripo、聚合器或开放模型里跑,再留下输出最好的结果。因此,Meshy 的飞轮必须体现为肉眼可见的更好输出、更低循环成本或更顺滑的下游完成,而不是单靠品牌。Formlabs 和动画闭环有价值,因为它们把 Meshy 从第一版草稿生成推向完成环节,而竞争对手在这个环节较少拥有完全相同的触点。[CP001, CP002, CP003, CP006, CP007, CP008]
Meshy 在规模和广度上领先,但 Rodin 与开源带来最尖锐的质量和商品化风险。
1-5 序数风险 / 准备度评分来自有来源支撑的章节分析。
[CP002, CP003, CP038, CP039, CP040, CP041]3.4 竞争威胁与反向证据
反向证据已经足够清晰,可以支撑一项具体竞争风险。独立比较来源在可投产几何、干净四边面拓扑、UV 和 PBR 纹理上把 Rodin 排在 Meshy 前面,并称 Tripo 在某些工作流中的原始速度或游戏就绪拓扑更强。另一份比较批评 Meshy 的单模型锁定,这正是聚合器利用的弱点。开源模型进一步放大威胁,因为 Hunyuan3D、TRELLIS 和 Stable Fast 3D 给技术用户提供了可嵌入内部工具的低成本替代方案。大型科技公司在模拟、世界模型和物理 AI 上的工作今天更间接,但如果 OpenAI、Google、NVIDIA、Adobe 或 Autodesk 把分发转成默认 3D 生成功能,威胁可能变成生存级。尽调任务因此必须基于基准测试:用同一组提示词横向测试 Meshy、Rodin、Tripo、Spline 和开放模型,再在绑定、编辑、打印和引擎导入后给可用网格打分,而不是只看截图。[CP025, CP026, CP027, CP028, CP033, CP034]
| 护城河主张 | 威胁 | 严重性 | 证据 | 缓释措施或尽调追问 |
|---|---|---|---|---|
| 规模 / 数据飞轮 | 开源和大厂模型削弱基础生成的独特性 | 高 | Hunyuan3D、TRELLIS、Stable Fast 3D 和 Genie 类世界模型 | 衡量重复使用、付费转化和专有模型质量差异 |
| 速度和免费增值漏斗 | Tripo 和 Stability 类工具以低成本速度竞争 | 中 | Tripo 免费积分和 Stable Fast 3D 0.5 秒重建 | 按客群追踪积分经济性和创作者留存 |
| 生产生态广度 | Rodin 可拿下重视生产保真度的任务 | 高 | 独立评测将 Rodin 的几何 / 拓扑排在前列 | 用同一批提示词,让美术和引擎导入测试做基准 |
| API 和开发者采用 | 开发者可自托管开源模型,或集成 Luma 类 API | 中 | Meshy API 受计划门槛限制;开源仓库暴露模型代码 | 索取按端点拆分的 API 使用客群、延迟和流失 |
| 动画和 3D 打印延伸 | 既有工具和专业工作流守住下游控制权 | 中 | Maya、ZBrush、Blender、Substance 和 Formlabs 工作流证据 | 验证用户是否在 Meshy 内完成下游任务 |
| 品牌 / 品类领导力 | 单模型锁定批评可能把买家推向聚合器 | 中 | 3D AI Studio 反向对比建议多模型灵活性 | 评估多模型路线图和模型选择 UX |
严重性是基于已引用公开证据的尽调判断,不是实测概率。
[CP025, CP026, CP027, CP028, CP033, CP034]3.5 展项
04财务
4.1 融资规模与资本充足性
Meshy 的公开财务故事不是从详细经营报表开始,而是从一轮异常大的 2026 年 7 月 Series B 开始。公司宣布以 $1.5 billion 估值获得近 $400 million 新资本,由 IDG Capital、Matrix Partners China 和 Monolith Management 领投,老股东参投。投资判断上,这轮融资更适合作为资本充足性输入,而不是有吸引力单位经济的证明:它给 Meshy 留出资金做基础模型研发、推理基础设施和企业扩张,但没有披露账上现金、烧钱、债务、云承诺或里程碑式支出控制。早期融资记录更不完整,因为资料库来源指向更早融资,而官方来源主要强调 Series B。因此,本章只在前瞻资本分析需要时使用早期融资主张,并标出仍需私有数据才能计算真实资金跑道。[CI001, CI002, CI003, CI004, CI005, CI006]
| 事项 | 日期或状态 | 金额(USD M) | 估值(USD M) | 投资方或来源 | 财务含义 |
|---|---|---|---|---|---|
| 自举 / 未披露种子期 | 2026 年前 | Tracxn / 公开资料 | 未披露种子轮经济性;公开证据无法审计早期资本效率 | ||
| 早期阶段 / Series A 档案记录 | 2026 年资料页引用 | 50 | PitchBook / Seedtable | 可作为上一轮融资的方向性信号,但受限来源的交叉验证不完整 | |
| Series B 融资 | 2026-07 | 400 | 1500 | IDG Capital、Matrix Partners China、Monolith Management 与现有投资者 | 最大一笔已披露现金注入,也是资金续航分析锚点 |
| Series B 后公开披露融资总额 | 2026-07 | 450 | Series B 加资料页报告的早期融资 | 总额为近似值,因为早期轮次披露缺少完整一手来源 | |
| 公开申报证据 | 截至运行日 | SEC EDGAR 搜索端点 | 本次运行未取得公开财务报表或 Form D 经济条款 |
金额按百万美元四舍五入;早期阶段数值来自资料页,而非公司一手披露;null 表示未找到公开披露。
[CI001, CI002, CI003, CI004, CI006, CI038]| 输入项 | 公开证据 | 投资判断 | 风险 | 下一步尽调 |
|---|---|---|---|---|
| 账上现金 | 轮次交割后未披露 | Series B 意味着现金缓冲较大,但不是实际余额 | 中 | 银行流水、董事会现金报告和交割机制 |
| 月度消耗 | 未披露 | 研发和基础设施扩张可能很快吃掉资本 | 高 | 过去六个月现金消耗和按成本中心划分的预测 |
| 资金续航月数 | 未披露 | 没有现金和消耗,无法计算 | 高 | 基准 / 上行 / 下行续航模型 |
| 资金用途 | 研发、基础设施和全球企业扩张 | 支出重点偏增长,而非近期盈利 | 中 | 预算分配和与本轮融资绑定的里程碑 |
| 债务或项目融资义务 | 未发现公开债务义务 | 没有债务负担证据,但缺失不等于证明不存在 | 中 | 债务表、云资源承诺和供应商最低采购额 |
| 下一轮触发条件 | 未披露 | 估值上调取决于 ARR 质量和毛利率证明 | 高 | 下一轮融资或盈利所需里程碑 |
资本充足性从融资规模推断;未公开现金、消耗、债务和承诺排期是必查尽调项。
[CI006, CI027, CI028, CI040, CI045]2026 年 7 月融资主导了公开资本证据,也让早期轮次经济性相对不透明。
早期融资来自资料并已四舍五入;Series B 报道为接近 $400M。
[CI001, CI028, CI038, CI046]4.2 ARR、使用量与收入质量证据
最强的收入证据是 Meshy 在 GDC 2026 官方披露的 $30 million ARR 里程碑,以及 ARR 同比增长约 12x 的表述。该披露有力,因为它给出当前运行收入锚点;但它也未经审计且不完整。公开记录存在一处重大矛盾:36Kr Europe 标题称 ARR 超过 $300 million,而 Meshy 自己的 GDC 公告写的是 $30 million。本章把 $300 million 数字视为反向、低置信度离群值,并使用 $30 million 作为基准投资输入。使用量指标强化了增长故事——注册用户超过 1200 万、已生成模型超过 1 亿——但这些只是漏斗顶部或工作负载代理指标。它们不能揭示付费账户、免费转付费、积分消耗结构、企业 ACV、流失率或收入确认政策。[CI007, CI008, CI009, CI010, CI011, CI012]
| 指标 | 公开数值 | 证据状态 | 财务解读 | 核心尽调问题 |
|---|---|---|---|---|
| ARR 里程碑 | $30M | 公司官方口径 | 强劲收入端信号,但不是经审计收入 | 按产品、分群和合同类型拆分的 ARR 桥接 |
| 另一组 ARR 标题口径 | >$300M | 第三方冲突口径 | 按低置信度异常值处理 | 要求管理层核对公开材料并定义 ARR |
| ARR 同比增长 | ~12x | 公司官方口径 | 支撑高估值叙事 | 月度 ARR 历史和扣除流失后的扩张 |
| 注册用户 | 12M+ | 公司官方口径 | 漏斗顶部规模,不等于付费客户 | 付费用户、活跃用户和转化分群 |
| 生成模型数 | 100M+ | 公司官方口径 | 使用深度代理指标,本身不是收入 | 点数消耗、免费 / 付费使用拆分,以及按工作流划分的毛利率 |
| 前十科技公司客户 | 按市值计 10 家中的 5 家 | 公司官方口径 | 企业标杆客户信号,但没有合同金额 | 具名账户、ARR 集中度和续约状态 |
KPI 数值是公开牵引力指标,不是经审计财务报表;ARR 冲突是刻意展示的。
[CI007, CI008, CI009, CI010, CI011, CI012]Meshy 同时有 $30M ARR 里程碑、快速 ARR 增长和大规模使用量代理指标,但付费账户转化仍未披露。
$300M 项目仅用于标记冲突,并非标准 ARR 输入。
[CI010, CI011, CI012, CI047]4.3 变现机制与标价
Meshy 通过免费增值自助漏斗、付费订阅、企业方案和 API 用量变现。官方定价列出带月度积分的 Free 计划、每月 $20、$60 和 $100 的付费层级,以及定制 Enterprise 包装。文档说明积分用于生成工作流,API 页面则表示 API 使用需要 Pro 或更高层级。这支持一个合理组合:席位订阅收入、更高价值企业合同,以及基于用量的 API 或积分变现。限制在于,公开标价不是实际成交价。它几乎不说明折扣、年度合同、企业最低消费、总收入留存、付费席位扩张,或生成模型量究竟落在免费队列还是付费队列。因此,变现模型真实且可见,但收入结构和收入质量仍只能靠私有证据尽调。[CI014, CI015, CI016, CI017, CI018, CI019]
| 收入流或层级 | 标价 / 单位 | 机制 | 证据质量 | 尽调问题 |
|---|---|---|---|---|
| 免费计划 | $0,100 点数/月 | 免费增值拉新和试用使用 | 官方定价 / 文档 | 免费用户转付费席位或 API 支出的转化率 |
| Pro 计划 | $20/month | 订阅,并可使用 API | 官方定价 / 文档 | 扣除年付折扣和促销后的实际净价 |
| Studio 计划 | $60/month | 更高点数额度和工作流容量 | 官方定价 / 文档 | 按层级看席位扩张、团队使用和流失 |
| Ultra 计划 | $100/month | 公开列出的最高自助月付层级 | 官方定价 / 文档 | 实际附加率和工作负载结构 |
| 企业版 | 定制 | 议价合同、安全 / 支持和批量点数 | 官方定价页,但无价格 | ACV、折扣、合同条款、实施负担 |
| API / 点数 | Pro 层级及以上;通过生成 API 使用 | 围绕文本转 3D 和图像转 3D 的按量变现 | 官方 API 和文档 | 单次生成毛利率和 API 批量定价 |
标价只是公开套餐,不是实际收入;企业版和 API 经济性仍未公开。
[CI014, CI015, CI016, CI017, CI018, CI019]4.4 单位经济、烧钱与免费增值风险
财务反向解读是透明度不足,而不是没有牵引力。已审阅公开来源没有披露毛利率、CAC、回本周期、NRR、GRR、月度烧钱、资金跑道或付费客户数。相比低算力 SaaS,这些遗漏对 Meshy 更重要,因为 AI 3D 生成可能承担有意义的推理、模型训练和基础设施成本;免费增值产品也可能累积亮眼用户数和模型数,却没有相称的付费留存。订阅基准来源也强调,经常性收入质量取决于留存、扩张和订户行为,而 Meshy 没有公开任何一项。投资人因此应拆开三层:官方牵引力、可观察标价,以及拿不到的私有经济性。必要尽调路径是搭一座队列级收入桥,从注册用户到付费账户再到留存 ARR,并附上工作流层面的毛利率和云成本。[CI024, CI025, CI026, CI027, CI032, CI033]
| 指标 | 公开数值 | 置信度 | 重要性 | 尽调问题 |
|---|---|---|---|---|
| 毛利率 | 中(未披露) | 推理和模型服务成本决定可扩展性 | 按工作流、企业 / API 分部拆分的分群毛利率 | |
| CAC / 回本期 | 中(未披露) | 企业扩张可能掩盖昂贵的销售打法 | CAC、回本期、渠道结构和销售周期历史 | |
| NRR / GRR | 中(未披露) | ARR 质量取决于留存和扩张 | 分群留存和续约排期 | |
| 免费转付费转化率 | 中(未披露) | 12M 用户可能包含低意向免费用户 | 注册用户到付费套餐和 API 付费方的转化漏斗 | |
| 付费客户数 | 中(未披露) | 解读集中度和 ACV 必需 | 付费账户、企业标杆客户、头部客户 ARR 集中度 | |
| 收入结构 | 中(未披露) | 订阅、企业版和 API 的毛利率不同 | 按自助、企业版和 API / 点数划分的 ARR |
null 值不是零;它们表示所审阅公开来源未披露该指标。
[CI023, CI024, CI025, CI026, CI033, CI034]公开指标显示规模亮眼,但关键承销变量仍未公开。
倍数按公开估值和 ARR 的四舍五入数计算。
[CI002, CI007, CI024, CI027, CI029, CI045]4.5 估值基准与财务结论
以基准 $30 million ARR 输入计算,Meshy 的 $1.5 billion 估值约等于 50x ARR。只有当官方 12x 增长、企业采用、模型质量优势和全球扩张能转化为持久、高毛利经常性收入时,这一倍数才站得住。相对普通 SaaS 倍数基准,它偏贵;不过 AI 原生软件框架会为异常增长和可防守性给出高得多的倍数。互相冲突的 $300 million ARR 标题会导出完全不同的估值图景,但其可靠性不足以用于投资定价。因此,财务结论是「有前景,但不能仅凭公开数据下判断」:Series B 后资本充足性大幅改善,收入引擎也可见;但价值的关键决定项——留存、付费转化、毛利率、烧钱和收入结构——仍是私有信息。下一步应发起数据室请求,而不是只争论价格。[CI029, CI030, CI031, CI032, CI033, CI037]
| 基准或情景 | 倍数 / 指标 | 与 Meshy 的对比 | 解读 | 来源依据 |
|---|---|---|---|---|
| Meshy 官方 ARR 情景 | 约 50x ARR | 1.5B 估值 / 30M ARR | 除非增长、留存和毛利率质量都极出色,否则非常昂贵 | 公司 ARR 和估值披露 |
| AI 原生溢价区间 | 相对 SaaS 溢价 | 可支撑高于 SaaS 的倍数 | 叙事有支撑,但需要可防御性证据 | AI 估值框架 |
| 传统 SaaS 公开 / 私募中位数 | 个位数至低十几倍区间 | 远低于 50x | 如果 Meshy 向 SaaS 常态回归,凸显下行空间 | 2026 SaaS 基准来源 |
| 冲突的 $300M ARR 情景 | 若属实,约 5x ARR | 会让估值看起来没那么紧绷 | 因数据与官方 ARR 冲突,不纳入投资判断 | 36Kr 异常值与官方 ARR 对照 |
| 免费增值占主导的用户基数 | 转化率未知 | 12M 用户不等于付费客户 | 需要付费分群证据 | 定价和订阅基准来源 |
倍数仅作方向性参考,因为基准来源样本不同,Meshy 也没有经审计财务披露。
[CI007, CI008, CI009, CI029, CI030, CI031]按官方 ARR 口径测算,倍数明显高于普通 SaaS 基准;$300M 这一口径只作为存在争议的敏感性情景处理。
基准区间综合了引用的 2026 年分析师来源;样本和定义并不一致。
[CI008, CI030, CI031, CI032]4.6 展项
05产品与技术
5.1 产品面与工作流适配
Meshy 的产品面更应被理解为快速 3D 资产创作工作流,而不是单一模型端点。官方页面和帮助中心支持一个覆盖文本生成 3D、图像生成 3D、AI 贴图、绑定和动画、重网格控制、Meshy 3D Agent、面向 3D 打印的 Auto Split,以及 API 访问的目录。当前最强适配点是构思和原型:创作者从提示词或参考图像出发,按 Meshy 页面说法在一分钟内生成预览,细化或贴图后,再移入游戏、DCC、Web、AR 或打印工作流。尽调角度因此必须按模块拆开。文本和图像生成在公开界面上看起来成熟;Meshy Agent 和 Meshy Labs 更像探索;Auto Split 是打印工作流中有价值的补偿控制,而不是原始输出永远干净的证明。[CE001, CE002, CE003, CE004, CE011, CE024]
| 能力 | 主要用户 / 任务 | 状态 / 成熟度 | 差异化 | 尽调缺口 |
|---|---|---|---|---|
| 文本转 3D | 用自然语言描述资产的创作者 | 已上线产品和 API 能力 | 提示词到预览的流程快 | 需要第三方保真度和可重复性基准 |
| 图像转 3D | 将概念图或参考图转成 3D 的艺术家 | 已上线产品和 API 能力 | 单图路径加多图 API 变体 | 需要验证复杂视角和遮挡几何下的一致性 |
| AI 贴图 | 给现有网格贴图的游戏和内容团队 | 已上线产品和 API 能力 | PBR 贴图支持,包括金属度、粗糙度和法线贴图 | 需要用人工 Substance 流程对照测试材质准确性 |
| 动画与自动绑定 | 角色创作者和游戏原型团队 | 已上线功能和 API 能力 | 500+ 动作库和可编程动画输出 | 需要非标准角色的变形质量证据 |
| 重网格 / 智能拓扑 | 清理生成资产的技术美术 | 已上线 API 能力 | 目标面数和智能拓扑控制 | 需要生产管线中干净拓扑的证据 |
| Meshy 3D Agent | 编排多步骤 3D 创作的非技术用户 | Beta 支持入口 | 覆盖构思、概念和输出的聊天工作流 | 需要留存、任务成功率和边界条件证据 |
| Auto Split | 为大型或多色对象做准备的 3D 打印用户 | 已上线帮助和博客入口 | 部件切分加适合打印的水密封口 | 需要跨材料和打印机的切片器级验证 |
| Meshy Labs | 探索 AI 原生玩法的游戏开发者 | GDC 2026 发布 | 从资产延伸到玩法实验 | 需要技术细节和开发者准入模式 |
枚举基于官方产品、文档、帮助和 2026 发布页;成熟度指公开入口成熟度,不是非公开使用证据。
[CE002, CE003, CE004, CE005, CE007, CE024]能力评分衡量公开产品表面的成熟度,不代表私有使用指标。
1–5 序数评分,依据官方产品页、API 文档、帮助文章和独立佐证是否可得。
[CE002, CE003, CE004, CE005, CE007, CE024]5.2 架构、API 与生成管线
公开证据暴露了具体运营管线,但模型内部仍是私有黑箱。Meshy 记录了文本生成 3D、图像生成 3D、多图、重新贴图、重网格、绑定、动画、webhooks、余额和定价端点。技术模式是异步的:创建生成或后处理任务,检查输出和任务状态,用 webhooks 接收状态更新,并监控操作积分余额。从这些文档推断出的产品管线是输入 → 预览 → 细化 → 贴图 → 重网格 → 绑定或动画 → 导出。公开可见的最强技术控制包括 PBR 贴图支持、智能拓扑与面数控制,以及动画输出格式。同样重要的是未公开部分:在已审阅来源中,Meshy 没有披露模型架构、训练语料构成、基准测试方法或可重复性统计。[CE006, CE015, CE016, CE017, CE018, CE019]
| API 能力 | 端点或文档入口 | 可实现内容 | 控制信号 | 风险 / 缺口 |
|---|---|---|---|---|
| 文本转 3D | 文本转 3D API | 提示词驱动的模型任务 | 预览 / 精修任务流和 PBR 参数 | 延迟和可重复性指标未公开 |
| 图像转 3D | 图像和多图转 3D API | 参考图生成模型 | 单图和多图变体 | 缺少遮挡和视角一致性基准 |
| 重新贴图 | Retexture API | 为现有资产贴图 | 贴图操作以编程方式开放 | 材质正确性未被独立评分 |
| 重网格 | Remesh API | 改变拓扑和面数 | 智能拓扑和目标多边形数选项 | 干净拓扑仍需用户验证 |
| 绑定 / 动画 | 绑定和动画 API | 准备并动画化角色 | 绑定任务加动画输出 URL | 没有基准数据时,非人类和风格化绑定可能静默失败 |
| Webhook 与余额 | Webhooks 与余额 API | 接入异步任务状态和积分检查 | HTTPS Webhook 要求和余额端点 | 需要公开状态页和故障历史 |
API 行集聚焦与产品化最相关的端点,不覆盖每个文档页面。
[CE015, CE016, CE017, CE018, CE019, CE020]Meshy 公开流程把提示词或图像输入串到生成、精修、材质、拓扑、绑定和导出各环节。
[CE006, CE016, CE017, CE018, CE022, CE040]5.3 格式、集成与开发者界面
Meshy 的开发者和集成界面比纯消费者生成器更宽。导出证据覆盖 GLB、FBX、OBJ、STL、BLEND 和 USDZ 等常见交换格式;动画 API 则明确返回 GLB 和 FBX 动画输出,以及 USDZ 相关处理资产。官方来源还指向插件或合作路径:Unity、ComfyUI,以及一般插件帮助页面。GitHub 开发者信号通过 Meshy MCP 和 3D-agent 仓库可见;这点重要,因为智能体式创作工作流是产品的自然延伸。开放问题不是集成点是否存在,而是它们是否可靠、版本化,并在生产规模被采用。公开来源尚未提供正常运行时间、包下载量、活跃开发者数或企业 SLA 证据。[CE009, CE010, CE012, CE013, CE014, CE033]
| 入口 | 支持格式或宿主 | 工作流角色 | 证据质量 | 未决尽调问题 |
|---|---|---|---|---|
| 核心模型导出 | GLB、FBX、OBJ、STL、BLEND、USDZ 等格式 | 将生成模型带入 DCC、游戏、Web、AR 和打印工作流 | 官方帮助加独立目录确认 | 确认套餐级限制和批量导出上限 |
| 动画导出 | GLB、FBX、USDZ 相关处理后输出 | 将已绑定并已动画化的角色送入引擎 | API 文档 | 确认 Unity、Unreal 和 Godot 中动画重定向质量 |
| Unity 插件 | Unity | 为 Unity 项目生成或导入 AI 3D 模型 | 官方 2026 Meshy 博客 | 确认插件采用率、版本支持和失败模式 |
| ComfyUI 伙伴节点 | ComfyUI | 基于节点的生成和绑定工作流集成 | 官方伙伴证明博客 | 确认维护负责人和兼容性政策 |
| 插件总体 | 外部创作工作流 | 把 Meshy 输出接入创作者工具 | 帮助中心表述 | 需要完整插件列表和支持 SLA |
| 3D 打印流程 | 面向 STL / 切片器的输出和 Auto Split | 为实体制造准备资产 | 官方帮助加反向评价背景 | 验证水密性和特定打印机公差 |
格式和集成是公开兼容性信号;不能证明生产采用或质量。
[CE009, CE010, CE012, CE013, CE014, CE024]5.4 性能、质量与局限
质量案例由强烈的第一方速度主张和有意义的独立提示混合而成。Meshy 声称文本生成 3D 和图像生成 3D 工作流低于一分钟,Auto Split 结果约 40 秒,动画工作流可在数分钟内从上传走到动画角色。这些主张支撑了清晰的原型生产率命题。不过,There’s An AI For That 提示,多数生成模型按生成结果本身并非打印就绪,因为它们可能超过构建板、需要颜色分离,或包含开放网格表面。Costbench 另行报告了重试复现相同错误时与积分相关的挫败感。这些反向来源并不否定 Meshy 的效用,但会改变投资判断姿态:生成资产应被视为快速草稿,投产前仍需拓扑、可打印性和材料 QA。[CE003, CE004, CE008, CE024, CE025, CE026]
| 信号 | 公开数值 / 方向 | 来源立场 | 投资判断含义 |
|---|---|---|---|
| 文本转 3D 速度 | 官方称完整贴图模型不到一分钟 | 官方证实 | 原型速度卖点很强,但需要独立计时 |
| 图像转 3D 速度 | 官方称单张图片不到一分钟 | 官方证实 | 适合创意探索,但难例仍缺基准测试 |
| Auto Split 速度 | 拆分结果约 40 秒 | 官方证实 | 如果切片器验证成立,可用于自动化打印准备 |
| 动画速度 | 以分钟计,而不是天;500 多个动作 | 官方证实 | 相比手工绑定,角色原型更有优势 |
| 打印就绪度 | 大多数生成模型刚生成时并不适合打印 | 独立来源反向 | Auto Split 是必要补偿控制,不是源几何干净的证明 |
| 积分重试摩擦 | 重试可能复现同样错误并浪费积分 | 独立来源反向 | 规模化后,质量失败会变成经济摩擦 |
保留独立反向行,是为了避免过度解读第一方速度和质量声明。
[CE003, CE004, CE008, CE024, CE025, CE026]这张 KPI 图把流程耗时声明、库规模和控制项,与未经验证的私有基准分开。
[CE003, CE004, CE008, CE019, CE024, CE035]5.5 信任、安全、路线图与缺口
Meshy 的信任姿态方向正面,但尚未完全达到尽调可用。公开界面声称具备 SOC2 Type II、ISO 27001 和 GDPR 认证;帮助中心称支付详情由第三方网关处理,而不是由 Meshy 直接存储;另一篇支持文章则称未经同意不会把概念图上传用于训练。客户上传专有概念时,这些控制对企业和创作者采用都很重要。路线图势能也可通过 Meshy 5、Meshy 6、Meshy 3D Agent、Auto Split 和 GDC 2026 的 Meshy Labs 看到。剩余缺口很具体:在把 Meshy 作为生产级 3D 资产基础设施层来定价前,应取得认证材料和范围、服务状态与事故历史、删除和保留政策、经基准测试的输出质量,以及私有架构细节。[CE029, CE030, CE031, CE035, CE036, CE037]
| 控制 / 认证 | 公开状态 | 范围 | 缺口 |
|---|---|---|---|
| SOC2 Type II | 公开页面声称具备 | 企业级安全态势 | 需要报告期间、审计方和排除项 |
| ISO 27001 | 公开页面声称具备 | 信息安全管理 | 需要证书编号和范围 |
| GDPR | 公开页面声称具备 | 隐私和数据处理态势 | 需要 DPA 和子处理方名单 |
| 支付安全 | 第三方网关处理支付细节 | 支付数据处理 | 需要网关名称和 PCI 责任矩阵 |
| 训练数据同意 | 未经同意,概念图不会用于模型训练 | 上传的图像转 3D 输入 | 需要留存窗口和删除审计证据 |
控制项均为第一方说法,除非之后拿到认证文件。
[CE029, CE030, CE031]Meshy 的产品价值靠自有模型质量、积分经济、外部创作工具和信任控制撑住。
[CE021, CE023, CE025, CE026, CE029, CE038]5.6 展项
06客户
6.1 客户分层与买方 / 用户地图
Meshy 的客户面很宽,而不是集中在单一狭窄垂直领域。最清晰的公开细分包括游戏和媒体创作者、3D 打印和创客工作流、XR 或教育用例,以及需要更快构思的专业 3D 艺术家或设计师。官方用例语言把 Meshy 描述成进入游戏引擎、切片器、动作管线和 AR 查看器的桥,这意味着既有个人创作者采用,也有团队工作流嵌入。客户页和生产案例进一步提供了具名证据,覆盖游戏相邻工作室、裸眼 3D 显示硬件和桌面微缩模型生产。尽调中最重要的区分是:用户细分可见,但付款方分层不可见。免费创作者、付费自助订阅者、API 集成方和 Enterprise 团队都合理存在,但公开来源没有按细分分配收入或留存。这使分层可作为需求证据投资,但还不足以构成收入质量地图。[CU001, CU002, CU006, CU032, CU033, CU043]
| 细分 | 买方 / 用户 / 付款方 | 主要用例 | 规模或牵引信号 | 收入或战略价值 | 尽调缺口 |
|---|---|---|---|---|---|
| 游戏工作室和独立开发者 | 艺术家、制作人、技术美术;工作室或自助付费方 | 基础网格、角色、道具、可进游戏引擎的资产 | 客户页和案例文章提到 Jupiter、37 Interactive 以及游戏工作流 | 游戏资产需求反复发生且嵌入工作流,战略价值高 | 未公开付费席位数、ACV 或游戏工作室留存 |
| 3D 打印爱好者和 TTRPG 创作者 | 创客、TTRPG 创作者、打印服务用户;自助或 API 付费方 | 可打印微缩模型、手办、钥匙扣、个性化物件 | Thorns Tavern 和 Form Now 证据显示 AI 到打印工作流 | 从数字模型通向实体履约的战略桥梁 | 打印适配质量和履约经济性未披露 |
| 专业 3D 艺术家和设计师 | 艺术家、产品设计师、工业设计师;个人或团队付费方 | 概念迭代、贴图、重网格、绑定,导出到 DCC 工具 | 官方用例和插件支持 Blender、Unity、Unreal 工作流 | 工具和积分付费可变现高频专业用户 | 评论质疑其面向高级用户的功能深度 |
| 教育和 XR 创作者 | 教师、学生、教育工作者、AR/VR 创作者;学校或个人付费方 | 学习资产、兼容 Roblox 的课堂 / 游戏设计、空间内容 | 帮助中心提供教育计划,官方用例包含 XR 与教育 | 低摩擦采用可培育未来创作者和机构用户 | 未公开学校数量、续费率或教育收入 |
| 企业 / API 集成方 | 产品团队、游戏平台运营方、制造商;企业付费方 | 程序化模型生成、批量工作流、嵌入式定制产品 | API 文档定义 Pro-plus 权限和企业速率限制 | 潜在 ACV 最高,工作流锁定最强 | 企业账户数、NRR 和合同期限未披露 |
细分基于公开用例、客户故事、帮助中心和文档来源;不估算 收入结构。
[CU001, CU008, CU017, CU021, CU031, CU032]Meshy 可从免费试用切入,再扩到付费积分、API 集成、插件、企业版限额和实体制造。
[CU018, CU019, CU020, CU021, CU022, CU023]6.2 具名客户与量化工作流结果
最强客户证明不是客户标识数量,而是具名引用和量化工作流结果同时出现。Meshy 自己的客户页列出 Stratton Studios、Thorns Tavern 和 Jupiter;HackerNoon 生产文章又加入 37 Interactive Entertainment,并给出更细的生产机制。Jupiter 是最干净的量化案例,因为 Meshy 页面和生产文章都指向从一周压到两小时的工作流变化,报道称生产时间减少 98%。Thorns Tavern 的战略意义不同:Meshy API 被描述为嵌入一个面向消费者的定制微缩模型管线,把建模时间从一到两周压到分钟级,并把单模型成本削减 80%。这些案例支持真实工作流价值,但仍是供应商可见引用,而非独立审计的续约或收入承诺。[CU002, CU003, CU004, CU005, CU006, CU007]
| 客户 | 细分 | 用例 | 生产 / 试点 | 结果 | 来源 |
|---|---|---|---|---|---|
| Stratton Studios | 游戏 / 创意工作室 | 3D 创意探索和资产构思 | 公开证言;生产深度未经独立审计 | 数周建模压缩成数小时探索 | SU001; SU022 |
| Jupiter | 裸眼 3D 显示硬件 / 移动视觉内容 | 基础网格生成和客户展示内容工作流 | 生产案例,带量化工作流指标 | 从一周降至两小时;据称减少 98% | SU001; SU020 |
| 37 Interactive Entertainment | 游戏发行商 / 角色生产 | 用于角色生产的分部件图像转 3D 工作流 | 生产文章披露客户 | 高模雕刻工作量减少 30% 到 40% | SU020; SU022 |
| Thorns Tavern | TTRPG 微缩模型和 3D 打印产品 | Meshy API 嵌入定制微缩模型生成 | 流水线已运行;消费端平台被描述为内测 / 上线前 | 建模从 1–2 周压到几分钟;成本降低 80% | SU001; SU020 |
| Formlabs Form Now 用户 | 制造 / 按需打印分发 | 从提示词或照片生成专业 SLA/SLS 打印件 | RAPID + TCT 2026 报道的伙伴集成 | 最快两天交付制造件 | SU021; SU030 |
| Enterprise API 用户 | 程序化工作流集成方 | 更高体量 API 使用和定制排队任务 | 官方 API / 定价页面,客户未具名 | 企业速率限制默认 100 RPS 和 50 个排队任务 | SU004; SU008 |
枚举是公开具名或具名类别客户验证的样本,并非完整客户清单。
[CU002, CU003, CU004, CU005, CU006, CU007]公开案例显示时间或工作量大幅下降,但每项指标都来自客户故事,不是经审计的分群数据。
百分比是披露的降幅;分钟数和天数作为单独成效条,呈现合作伙伴流程速度。
[CU005, CU007, CU009, CU010, CU026, CU027]6.3 采用轨迹与转化可见度
在私有 AI 工具公司里,Meshy 的漏斗顶部牵引力异常可见:2026 年 7 月融资公告称注册用户超过 1200 万、已创建模型超过 1 亿;2026 年 3 月 GDC 信息则提到用户超过 1000 万、模型 1 亿。这些数字有意义,因为它们显示全球认知度和工作负载规模,并得到独立转载和目录覆盖佐证。但它们不能证明变现质量。注册用户可包含沉睡免费账户,生成模型可包含低价值试验;已审阅公开来源没有披露付费订阅者、活跃用户、队列留存、企业账户数或免费转付费转化。投资判断因此应把 Meshy 视为具备强采用证据,但从使用量到持久收入的转化桥仍是私有信息。[CU011, CU012, CU013, CU014, CU015, CU016]
| 指标或漏斗步骤 | 数值 | 日期 / 新鲜度 | 来源属性 | 含义 | 缺失分母 |
|---|---|---|---|---|---|
| 注册用户 | 超过 1200 万 | 2026 年 7 月 | Yahoo Finance 转发的公司公告 | 强大的漏斗顶部知名度和账户基础 | 未披露活跃用户和付费用户 |
| 生成 3D 模型 | 超过 1 亿 | 2026 年 7 月 | Yahoo Finance 转发的公司公告 | 大规模工作负载和重复试验信号 | 未披露付费、留存或生产模型占比 |
| 先前用户里程碑 | 超过 1000 万用户 | 2026 年 3 月 | 官方 GDC 公告 | 显示 7 月公告前用户数至少增加 200 万 | 注册口径未披露 |
| 免费计划 | 每月 100 积分 | 当前定价页 | 官方定价 | 降低采用摩擦,并带来更多试验 | 免费转付费转化未披露 |
| Pro 计划 | 每月 1,000 积分和 API 访问 | 当前定价 / API 页面 | 官方定价和 API 页面 | 形成自助变现台阶 | Pro 订阅人数未披露 |
| Enterprise API | 100 RPS,排队任务可定制,默认 50 个 | 当前 API 页面 | 官方 API 页面 | 支持更高体量工作流集成 | 企业客户数和 ACV 未披露 |
数值来自公开披露;缺失分母是有意保留的尽调问题,不代表零。
[CU011, CU012, CU013, CU014, CU015, CU016]公开证据在注册和模型生成规模上最强;到了付费转化和留存,就变得不透明。
由于活跃用户和付费转化数量未披露,漏斗混合使用规模指标和套餐上限。
[CU011, CU012, CU017, CU018, CU019, CU021]6.4 满意度、留存与反向信号
公开满意度记录呈混合状态,而且这对尽调重要。独立目录通常把 Meshy 描述为易用且适合快速 3D 生成;TopAI.tools 在一个小样本评论基数上报告了较高推荐占比。与此同时,SaaSHub 记录了反向主题:产品可能无法提供高级用户要求的深度,性能或可扩展性也可能让要求较高的设计环境担忧。Meshy 自身帮助内容也通过发布中空 3D 打印模型和可打印性检查文章,承认了实际生产问题。这些本身不破坏投资命题;每个生成式 3D 工作流都需要清理。但它们说明,客户成功、质量控制、积分经济和付费转化证据应是尽调优先项,而不能从原始用户数直接假设。公开留存指标缺席。[CU034, CU035, CU036, CU037, CU038, CU039]
| 信号 | 数值或证据 | 受影响细分 | 置信度 | 尽调要求 |
|---|---|---|---|---|
| 目录推荐度 | TopAI.tools 报告 13 条评论的推荐率为 92.3% | 泛创作者群体 | 中 | 获取原始评论集、时间分布和已验证用户状态 |
| 目录用户数 | Toolify 的产品资料显示 38.8K 用户 | 目录受众,不是 Meshy 账户 | 低 | 不要当作公司用户数;需与官方注册数对账 |
| 功能深度担忧 | SaaSHub 称技术类评论提到高级用户功能深度有限 | 专业艺术家和企业团队 | 中 | 按高级生产要求测试输出质量 |
| 性能 / 可扩展性担忧 | SaaSHub 提到高要求环境中的性能和可扩展性担忧 | 企业和高用量团队 | 中 | 索取 SLA、队列延迟、失败率和企业支持指标 |
| 免费层署名 | 免费输出按 CC BY 4.0 授权,并非自有资产 | 免费计划上的商业创作者 | 高 | 衡量署名 / 所有权是否推动转化或造成流失 |
| 打印适配限制 | 帮助中心发布空心模型和可打印性修复流程 | 3D 打印用户 | 中 | 审计切片器通过率、修复频率以及退款 / 重做率 |
评论证据只具方向性,不能替代私有队列留存、续费或支持工单数据。
[CU023, CU024, CU034, CU035, CU036, CU037]公开评价信号存在,但留存和付费转化 KPI 仍未披露。
零值表示对应 KPI 未公开披露,不代表实际经营表现。
[CU038, CU039, CU044, CU045]6.5 合作、分发与扩张路径
Meshy 的扩张路径比浏览器工具更宽。Formlabs Form Now 报告是最具体的分发证明,因为它把 Meshy 创作与专业实体制造连接起来,并报告两天交付成品零件的承诺。同一报告还列名 xTool、Snapmaker、Flashforge 和 MakerWorld / Bambu Lab,称其是围绕 AI 生成实体输出的生态伙伴或集成。在数字生产侧,官方文档和帮助材料支持 Blender、Unity、Unreal、Roblox 和 Godot 的插件或工作流。这创造了多条先落地再扩张的向量:创作者可以先免费起步,为积分和 API 访问升级,把资产集成到生产工具里,并可能因更高 API 限额和留存需求转到 Enterprise。剩余风险是集中度不透明:合作伙伴依赖、渠道收入分成和客户集中度均未公开量化。[CU017, CU018, CU019, CU020, CU021, CU022]
| 扩张驱动因素 | 证据 | 对客户或渠道的影响 | 集中风险 | 尽调路径 |
|---|---|---|---|---|
| API 嵌入 | Thorns Tavern 和文档显示 API 驱动工作流 | 让 Meshy 从工具变成产品基础设施 | 嵌入式 API 用户收入占比未知 | 索取 API 客户数、使用集中度和总留存 |
| 企业级额度 | 企业 API 提供更高 RPS 和可定制排队任务 | 支持规模化团队和更大合同 | 企业 ACV 和续费条款未披露 | 复核企业合同队列和支持义务 |
| Formlabs Form Now | 独立报道把 Meshy 与专业制造连接起来 | 将 Meshy 延伸到实物履约 | 伙伴经济性和对 Formlabs 的依赖未披露 | 获取合作协议和抽成经济性 |
| 打印机生态集成 | 据报道包括 xTool、Snapmaker、Flashforge、MakerWorld/Bambu Lab | 拓宽进入创客生态的渠道 | 对硬件伙伴和兼容路线图存在依赖 | 梳理活跃集成、收入分成和技术归属 |
| DCC 和引擎插件 | 文档 / 帮助覆盖 Blender、Unity、Unreal、Roblox、Godot | 嵌入既有工作流,降低切换成本 | 插件使用量和维护负担未披露 | 索取插件 MAU、崩溃率和支持版本承诺 |
扩张向量公开可见;集中度和经济性仍只能靠私有证据确认。
[CU020, CU021, CU025, CU026, CU027, CU028]6.6 展项
07风险
7.1 按严重程度排序的风险视图
Meshy 的反向案例不是单点失败,而是估值、竞争、法律来源、地缘政治审查和执行负荷之间的相互作用。公司有可信规模信号,包括公开声称生成资产超过 1 亿,以及第三方报道注册用户超过 1200 万;但这些采用指标本身不能证明企业转化、毛利率或持久留存。2026 年 7 月融资抬高了门槛:按报道 $1.5 billion 估值,即使用约 $30 million ARR 的基准尽调假设,进入倍数也约为 50x ARR。因此,最高优先级尽调是把已观察事实与剩余暴露拆开:验证 ARR 质量、客户集中度、GPU 经济性,以及 Meshy 关于隐私、训练数据和输出所有权的缓释主张是否能对目标企业用例形成合同约束。[CR003, CR004, CR005, CR006, CR007, CR046]
| 类别 | 风险 | 证据基础 | 可能性 | 影响 | 时间窗口 | 缓释成熟度 | 剩余敞口 | 尽调要求 |
|---|---|---|---|---|---|---|---|---|
| IP / 版权 | 训练数据来源声明公开度不够,无法排除受版权保护的 3D、图片或扫描数据暴露 | AI 诉讼追踪和 Anthropic 先例显示,训练数据责任仍是现实风险 | 中 | 高 | 0–24 个月 | 中 | 重大 | 复核数据集来源、许可证、退出记录和赔偿上限 |
| IP / 输出所有权 | 生成的 3D 资产可能继承用户提示词或上传参考图的问题 | Meshy 条款要求用户拥有输入权利,并区分付费 / 免费输出权利 | 中 | 高 | 当前 | 中 | 重大 | 测试企业赔偿、下架流程和重复侵权者处理机制 |
| 隐私 / 数据使用 | 非企业数据可能按计划类型用于模型训练 | Meshy FAQ 披露训练使用取决于计划类型,企业版除外 | 中 | 中 | 当前 | 中 | 中等 | 验证管理员控制和企业数据使用豁免 |
| 数据驻留 | AWS 美国存储可能给美国以外的受监管客户带来问题 | Meshy FAQ 称使用 AWS 美国存储,但产品面向全球 | 中 | 中 | 当前 | 中 | 中等 | 梳理区域数据驻留、分处理方和 DPA 选项 |
| 安全认证 | 安全水位取决于其宣称的 SOC 2 和 ISO 27001 控制是否落地 | Meshy FAQ 声称已取得 ISO/IEC 27001:2022 和 SOC 2 认证 | 低 | 高 | 当前 | 中 | 中等 | 审阅报告、范围、例外和桥接函 |
| CFIUS / 治理 | 与中国有关联的投资方若拥有敏感信息权或控制权,可能触发审查 | 融资来源点名中国关联投资方,美国财政部说明了 CFIUS 范围 | 中 | 高 | 0-36 个月 | Unknown | 重大 | 审阅股权结构表、附函、董事 / 观察员权利和信息权 |
| 中美 AI 张力 | AI 模型 IP 盗用警告抬高声誉和客户尽调负担 | CNBC 报道了美国国务院警告和内部人风险 | 中 | 中 | 当前 | 低 | 重大 | 留存内部人威胁控制和访问日志证据 |
| 监管监测 | 美国外资管制可能要求缓解措施,或拖慢交易 | 美国财政部、GAO 和 A&O Shearman 描述了缓解措施格局 | 低 | 中 | 0-36 个月 | Unknown | 中等 | 获取律师备忘录,判断当前和未来交易是否会被审查 |
| 客户合同风险 | 企业客户可能要求更宽的 IP、隐私和安全承诺 | 条款和帮助中心区分付费 / 免费以及企业保护 | 中 | 中 | 当前 | 中 | 中等 | 抽样审阅 MSA / DPA,并对标赔偿条款 |
| 诉讼外溢 | AI 版权和解可能抬高文本数据集之外的原告预期 | Bartz / Anthropic 来源给出 $1.5B 和解基准 | 中 | 高 | 0-36 个月 | 低 | 重大 | 跟踪 3D / 图像案件,以及准备金 / 保险安排 |
本风险登记表基于公开法律、监管、政策和负面报道;严重性由分析师依据引用来源和缺失的私人尽调判断。
[CR013, CR014, CR015, CR016, CR029, CR032]剩余风险集中在 IP 来源、估值、竞争和地缘政治审查。
序数评分是基于公开证据的分析师判断;私有尽调可能改变单元格评分。
[CR007, CR032, CR034, CR039, CR043, CR045]按公开证据风险登记表中的类别,统计高或重大剩余暴露数量。
计数来自本章风险登记表,统计重大 / 高剩余暴露,不是事件发生频率。
[CR007, CR013, CR029, CR036, CR043, CR045]7.2 竞争与生存风险
竞争威胁异常严峻,因为 Meshy 同时被三股力量挤压。Hunyuan3D、TRELLIS、TRELLIS.2 和 TripoSR 等开源发布降低了开发者复现部分生成技术栈、或与公开替代方案做基准测试的成本。Tripo 和 Hyper3D Rodin 等直接产品竞争者营销类似的文本或图像生成 3D 体验;Adobe 和 NVIDIA 等既有厂商则掌握根深蒂固的专业工作流、分发和渲染基础设施。Google 和 OpenAI 还带来平台风险:即便其当前公开页面比 3D 资产生成更宽,它们的模型和智能体平台也能吸收相邻创作工作流。投资含义是,Meshy 的护城河必须来自工作流深度、速度、数据权利姿态、社区和企业信任,而不只是拿到一个能用的 3D 基础模型。[CR017, CR018, CR019, CR020, CR021, CR022]
| 威胁类别 | 代表性来源 | 机制 | 证据信号 | 对 Meshy 的风险 | 待验证缓解措施 |
|---|---|---|---|---|---|
| 开源 3D 模型 | Hunyuan3D, TRELLIS, TRELLIS.2, TripoSR | 公开代码、论文和模型产物让基础生成商品化 | 多个开源项目瞄准带纹理或重建的 3D 资产 | 仅靠模型的护城河被侵蚀 | 工作流锁定、专有数据权和质量基准 |
| 直接 AI 3D 对手 | Tripo 和 Hyper3D Rodin | 类似的文本 / 图像转 3D 用户承诺和企业控制 | 竞品产品页主打快速 AI 3D 生成 | 定价和功能竞争升温 | 赢单 / 输单数据和差异化输出质量 |
| 创意软件既有厂商 | Adobe Substance 3D 和 Adobe-NVIDIA Firefly 合作 | AI 功能可嵌入现有设计套件 | Adobe 掌握专业创意工作流 | 分发劣势 | 插件、导出和创作者社区深度 |
| 基础设施既有厂商 | NVIDIA Omniverse 和物理 AI 栈 | 算力、仿真和数字孪生生态集成 | GTC 2026 推广 Omniverse DSX 和物理 AI | 垂直工作流被捕获 | 合作伙伴关系和非 NVIDIA 可移植性 |
| 前沿平台 | Google 和 OpenAI | 通用多模态智能体可吸收 3D 创作工作流 | 两者都有庞大的 AI 产品发布引擎 | 平台替代 | API、速度和领域专用 UX |
| 人类艺术家 / 工具 | 传统 DCC 工作流 | 最大控制力和专业质量仍有价值 | 评测对比 Meshy 速度和传统精度 | 核心资产仍需手工处理 | 已上线生产资产证明 |
竞争行是代表性而非穷尽;所纳入来源覆盖开源、直接对手和既有厂商威胁。
[CR017, CR019, CR021, CR022, CR024, CR025]技术、法律和市场风险会传导到收入韧性、利润率、融资和估值。
连线表示尽调中要验证的因果风险假设,不是已观察到的失败。
[CR017, CR023, CR026, CR029, CR032, CR043]7.3 技术、质量与生产就绪风险
质量风险不是 Meshy 没有有用输出;更强的反向读法是,用于构思和背景资产的有用性,未必能转化为 AAA、主角资产或对变形敏感的生产用途。Meshy 自己的产品页强调速度和易用,2026 年产品报道突出 Smart Topology 和 3D 打印改进,但独立工作流评审仍指出,某些角色生成案例需要手工调整。这个缺口重要,因为专业买方为可靠性、可预测拓扑、IP 安全复用和低清理时间付费。即便模型能生成有吸引力的第一版草稿,如果企业采用需要支持、重新生成或人工清理,毛利率仍可能承压。尽调应要求检查前后对比网格、失败提示词、生成重试率,以及能证明资产已出现在正式产品中的客户证据,而不是演示。[CR001, CR002, CR003, CR043, CR044, CR045]
| 失效模式 | 发生概率 | 严重性 | 证据 | 缓解成熟度 | 剩余敞口 | 未解决缺口 |
|---|---|---|---|---|---|---|
| 核心资产拓扑或绑定缺陷 | 中 | 高 | 独立工作流评测提到,部分角色案例需要手工调整 | 中 | 重大 | 需要失败输出分布和客户 QA 数据 |
| 高频生成带来的 GPU 成本压力 | 中 | 高 | 产品承诺快速生成 3D,融资支持研发和全球扩张 | Unknown | 重大 | 需要单次生成成本、重试率和毛利率 |
| 企业安全证据范围错配 | 低 | 高 | FAQ 声称具备 SOC 2 和 ISO 27001,但未附公开范围 | 中 | 中等 | 需要审计报告和范围边界 |
| 数据泄露或内部人访问 | 中 | 高 | CNBC 报道 AI 初创公司面临网络攻击和内部人定向攻击 | 中 | 重大 | 需要访问控制证据和事件历史 |
| 质量支持负担 | 中 | 中 | 生产评测把速度和最高级别手工控制区分开 | 中 | 中等 | 需要每个付费客户的支持工单数和退款率 |
运营发生概率和影响来自公开产品主张、第三方评测和网络安全报道推断;公司未披露私有运营指标。
[CR012, CR038, CR043, CR044, CR045]7.4 知识产权、版权与监管风险
法律暴露是本章的标准反向来源领域。Meshy 的政策描述了付费用户输出所有权、企业训练数据保护、美国境内 AWS 存储和安全认证,改善了客户叙事。不过,2026 年 AI 版权法仍不稳定:公开追踪器列出活跃训练数据诉讼,Anthropic/Bartz 和解创造了一个大型金额基准,同时保留了合法训练使用与涉嫌盗版获取之间的重要区分。Meshy 的 3D 领域又增加了额外不确定性,因为输入和输出可能包含受版权保护的角色、游戏资产、工业设计、扫描件和用户提供参考图。最重要的缓释不是泛泛的条款语言,而是可审计的训练数据来源、企业赔偿保障范围、下架处理,以及客户或上传资产不会在合同许可之外被使用的证明。[CR013, CR014, CR015, CR016, CR029, CR030]
| 风险 | 可监测触发项 | 阈值或事件 | 行动含义 |
|---|---|---|---|
| 估值偏高 | ARR 质量和净留存 | ARR 显著低于 $30M,或扩张留存偏弱 | 不按 50x ARR 写投资假设;重新定价或放弃 |
| 版权来源 | 数据集审计和赔偿范围 | 无可审计来源,或赔偿范围窄 / 未投保 | 阻断以企业客户为主的投资逻辑 |
| CFIUS / 地缘政治 | 投资方权利和数据访问 | 中国关联权利包含敏感技术信息访问 | 投资前要求律师意见和缓解措施 |
| 生产质量 | 客户已上线资产证明 | 演示占主导,生产客户需要大量手工清理 | 将 Meshy 视为准专业工具,而非企业平台 |
| GPU 经济性 | 毛利率和重试率 | 付费转化后重试 / 支持负载高,或毛利偏弱 | 下调估值倍数,并要求用量控制 |
| 关键人执行 | 领导梯队和运营指标 | Series B 轮后仍没有规模化 GTM / 合规负责人 | 将高级招聘和董事会汇报作为投资条件 |
否决标准把风险登记表转成尽调测试;这些阈值是投资政策触发项,不是公司披露的指引。
[CR007, CR013, CR032, CR039, CR043, CR045]7.5 地缘政治、CFIUS、网络安全与数据驻留风险
公开材料称 Meshy 通过 Meshy LLC 在美国运营,但融资报道列名了数个与中国相关或源自中国的投资品牌,包括 IDG Capital、Matrix Partners China、HongShan、BAI Capital 和 Source Code Capital。这并不证明不当行为、控制权或必然触发审查,但在中美 AI 环境下构成尽调提示;政府和媒体来源正在明确警告 AI IP 盗窃、蒸馏、网络攻击和内部人威胁。美国财政部的 CFIUS 框架和 GAO 关于缓释的工作显示,外资风险可能取决于对敏感技术信息、治理权、数据和控制权的访问。投资 Meshy 时,应完成干净的股权结构权利审查、数据访问地图、内部威胁控制,并准备面向客户的答案,说明生成资产和上传内容驻留在哪里。[CR008, CR010, CR011, CR012, CR036, CR037]
| 依赖 | 交易对手 | 角色 | 集中度信号 | 失效场景 | 严重性 | 缓解措施 | 剩余敞口 |
|---|---|---|---|---|---|---|---|
| 云存储 | AWS 美国 | 资产存储与处理地点 | FAQ 点名美国境内 AWS | 区域买家需要本地数据驻留或分处理方 | 中 | 企业 DPA 和数据驻留选项 | 中等 |
| GPU / 推理供给 | NVIDIA 生态和 GPU 供应商 | 训练和推理产能 | 3D 生成吃算力,NVIDIA 掌握关键基础设施 | GPU 紧缺会压缩毛利,或限制用户使用 | 高 | 产能规划和多云采购 | 重大 |
| 专业工作流集成 | Adobe、Autodesk、Blender、游戏引擎 | 下游资产清理和采用 | 既有厂商掌握工具链和文件格式工作流 | 既有厂商把竞品 AI 打包进现有席位 | 高 | 导出质量、插件和合作伙伴关系 | 重大 |
| 外国投资方 | IDG、Matrix China、HongShan、BAI 与 Source Code | 资本和潜在治理权 | 融资报道点名多家中国关联投资方 | 信息权或控制权触发审查 | 高 | 股权结构表审阅和律师备忘录 | 重大 |
| 用户生成输入 | 创作者和企业客户 | 训练参考、提示词和上传内容 | 条款要求用户持有必要权利 | 客户输入引发侵权或隐私索赔 | 中 | 提示词政策、下架机制和赔偿上限 | 中等 |
依赖表结合已观察到的交易对手和分析师推断的失效场景;控制权和供应商合同属于私有信息。
[CR008, CR011, CR024, CR025, CR037, CR039]Meshy 依赖云 / GPU 供给、干净的投资人与治理结构、专业工具链和用户可控输入。
该图是基于公开证据的依赖模型;合同条款和基础设施供应商仍属非公开信息。
[CR011, CR024, CR025, CR037, CR039, CR041]7.6 财务模型、关键人物与执行风险
财务和执行风险在于,Meshy 的公开叙事已经跑在变现质量的公开证明前面。融资报道指向全球扩张和研发,但没有披露毛利率、烧钱、净留存、客户集中度、付费转化或每个成功资产的 GPU 成本。拥有数百万用户的免费增值产品仍可能经济上脆弱:免费使用可能吃掉推理容量,付费用户可能在新鲜感之后流失,企业合同也可能要求沉重赔偿保障和支持。创始人 / CEO Ethan Hu 是公司故事的核心;当 Meshy 从产品驱动采用走向企业销售、合规和支持时,这会带来关键人物和组织扩张问题。打破投资命题的测试很具体:留存低于基准、来源问题未解、安全控制失败,或下轮融资承压,都会改变投资姿态。[CR004, CR005, CR006, CR007, CR009, CR045]
| 角色 / 职能 | 依赖或缺口 | 发生概率 | 严重性 | 缓解措施 | 尽调路径 |
|---|---|---|---|---|---|
| 创始人 / CEO | Ethan Hu 仍是融资、产品愿景和招聘叙事的核心 | 中 | 高 | 搭建高管梯队和继任计划 | 访谈 CEO、COO / CFO 同等角色和董事会,确认经营节奏 |
| 企业销售 | 免费增值用户能否转成付费企业工作流,公开证据还没有证明这条链路跑通 | 中 | 高 | 分层 GTM、客户成功和定价纪律 | 审阅签约额、销售管线、NRR、分群留存和头部客户敞口 |
| 法务 / 合规 | 企业采用扩大后,训练数据、IP、隐私和 CFIUS 法律支持也要跟上 | 中 | 高 | 专职法务 / 合规负责人 | 审阅律师备忘录、保险、DPA 和客户赔偿例外 |
| 研究团队 | 开源和大厂模型推进,挤压差异化速度 | 高 | 中 | 留住研究人员,强化工作流 / 数据护城河 | 审阅路线图、模型基准和招聘管线 |
人员风险来自公开创始人可见度、融资规模以及缺失的私有组织架构证据推断。
[CR009, CR046, CR049, CR050]7.7 展项
08估值
8.1 估值快照与建议
Meshy 的估值案例在头部价格上异常清晰,在投资细节上异常不完整。公司宣布以 $1.5 billion 估值完成近 $400 million Series B,并另行在 GDC 2026 披露 $30 million ARR 里程碑。这两个事实意味着约 50x ARR;只有持续的品类领导、快速企业转化和仍然友好的 AI 融资市场才能解释这一水平。因此,投资姿态应是跟踪 / 继续研究,而不是不计价格买入。Meshy 有真实证明点:庞大用户基础、超过 1 亿个已生成模型、明确 ARR 披露,以及愿意大规模押注该品类的投资者。但公开文件尚未展示净收入留存、毛利率、付费转化、优先权条款或队列质量。这使公司值得监控,但投资定价对价格高度敏感。[CV001, CV002, CV004, CV006, CV007, CV008]
| 指标 | 证据 | 投资假设解读 | 主张引用 |
|---|---|---|---|
| 最新轮次 | 近 $400M Series B 轮 | 资金充足,但不能证明单位经济性 | CV001 |
| 投后估值 | $1.5B | 私募市场溢价定价 | CV002 |
| ARR 锚点 | GDC 2026 披露 $30M | 基准年化收入口径 | CV004 |
| 隐含 ARR 倍数 | ~50x | 相比大多数 SaaS 偏高,即便按 AI 标准也高 | CV006 |
| 建议 | 跟踪 / 继续研究 | 等留存、毛利和转化证据 | CV040 |
| 估值立场 | 偏高 | 增长或许值得跟踪,但不足以支持无差别买入 | CV039 |
ARR 倍数只是投后估值 / 已披露 ARR 的简单计算;建议对价格敏感。
[CV001, CV002, CV004, CV006, CV039, CV040]| 立场 | 论点 | 改变观点的证据 | 主张引用 |
|---|---|---|---|
| 正向 | 已披露最大 AI-3D 融资和 12M+ 用户规模 | 持续企业 ARR 和付费转化证明 | CV003; CV007 |
| 正向 | 100M+ 生成模型显示工作流需求强劲 | 分群付费使用和留存证据 | CV008 |
| 正向 | 2026 年 AI 融资市场仍高度接纳 | AI 应用退出市场深度证据 | CV017; CV018; CV019 |
| 反论点 | 约 50x ARR 定价把未来数年执行都计入价格 | ARR 达到 $100M+,且倍数不崩 | CV006; CV039 |
| 反论点 | 免费增值规模不一定等于收入质量 | 付费转化、NRR 与毛利率数据 | CV041 |
| 反论点 | 炒作周期与资本效率警示可能压缩倍数 | 已验证的盈利路径与持久护城河 | CV033; CV035 |
论点按投委会框架归组,不按完整风险清单逐项计数。
[CV003, CV007, CV008, CV017, CV018, CV019]Meshy 在规模和市场动能上得分最高,在估值支撑和私有证据完整性上最低。
评分是作者根据引用证据和尽调缺口映射出的判断。
[CV017, CV018, CV032, CV033, CV040, CV043]8.2 可比公司与倍数基准
倍数基准是核心反向张力。Meshy 隐含的 50x ARR 位于多套 2026 年 AI 估值框架上沿, 但远高于传统 SaaS 参照,不能因为 AI 市场火热就把它常态化。最合适的可比组合必然混杂: Baseten 说明投资人愿意为高增长 AI 基础设施买单,Genspark 说明智能体生产力可以拿到溢价, Tripo 和 Luma 界定产品邻近性,Adobe 与 NVIDIA 的文件则给成熟创意软件和 AI 基础设施替代方案定锚。 上述组合只是样本,不是干净的同业组,因为许多私有 AI 融资不披露 ARR、清算优先权和收入质量。 因此,可比公司读数支持 Meshy 以溢价融资的能力,但不能证明溢价天然合理。合适判断是: 估值偏撑;如果 ARR 快速追上,有机会走向合理。[CV011, CV012, CV015, CV016, CV021, CV024]
| 可比对象 | 指标 / 估值 | 参照意义 | 局限 | 论据引用 |
|---|---|---|---|---|
| Meshy | 估值 $1.5B;约 $30M ARR;约 50x ARR | 标的公司,AI-3D 头部公司 | 私募条款与利润率未披露 | CV002; CV004; CV006 |
| AI 初创公司区间 | 广义 AI 区间约 10x-50x;少数前沿案例更高 | 勾勒 AI 溢价定价区间 | 不同数据集的方法口径不一 | CV011; CV012; CV013 |
| 传统 SaaS | 约 3x-7x,或明显低于 AI 区间 | 下行情景倍数锚 | 未按 AI 原生或超高增长调整 | CV015; CV016 |
| Baseten | 2026 年 6 月融资估值 $1.5B;AI 推理可比公司 | 显示市场愿意为高速增长 AI 基础设施出价 | 基础设施模式不同 | CV021; CV022; CV023 |
| Genspark | 2026 年轮次报道估值 $2.6B | 智能体式生产力 / 创意邻近可比公司 | ARR 与条款披露不完整 | CV024; CV025; CV026 |
| Tripo | AI-3D 产品与定价;估值未披露 | 最接近的产品对标 | 未找到公开估值倍数 | CV027 |
| Luma | 创意 AI 与 API 平台 | 创意基础设施邻近可比公司 | 不是以网格为核心的估值可比对象 | CV028 |
| Adobe / NVIDIA 申报文件 | 公开创意软件与 AI 基础设施参照 | 战略买家与公开市场语境 | 成熟上市公司,不是私营 AI-3D 初创公司 | CV029; CV030 |
可比对象混合了私营 AI 融资、市场倍数区间、直接产品竞品和公开申报参照;没有任何一行应被当成完全同业。
[CV002, CV004, CV006, CV011, CV012, CV013]Meshy 隐含 50x ARR 接近 2026 年宽口径 AI 倍数区间顶部,远高于传统 SaaS。
区间综合第三方市场框架;Meshy 倍数为投后估值除以已披露 ARR。
[CV011, CV012, CV014, CV015, CV016, CV039]8.3 情景与敏感性分析
情景视角能看清最新价格埋进了多少执行假设。熊案假设 ARR 翻倍至约 $60 million, 但转化或利润率担忧浮出水面后,退出倍数压缩到 12x;结果价值约 $0.7 billion, 会让 Series B 价格显得昂贵。基准案假设 ARR 达到约 $120 million,并以 25x 退出, 产生约 $3.0 billion 价值,摊薄前具备合理的风险投资回报。牛案要求 Meshy 成为持久的 AI-3D 工作流层,ARR 达到约 $250 million,并守住 35x 的溢价倍数,带来约 $8.8 billion 价值。上述情景刻意让倍数随收入规模扩大而压缩,因为今天的 50x 依赖异常增长。因此, 实际承销问题不是 Meshy 好不好,而是多大的价格纪律能补偿仍未公开的经营经济性。[CV036, CV037, CV038, CV039, CV045, CV046]
| 情景 | ARR 假设 | 退出倍数 | 隐含估值 | 概率信号 | 关键风险 |
|---|---|---|---|---|---|
| 悲观 | $60M ARR | 12x ARR | $0.7B | 增长放缓或转化走弱 | 下轮降价风险与稀释 |
| 基准 | $120M ARR | 25x ARR | $3.0B | 12x 增长放缓但仍强劲 | 执行与企业客户留存 |
| 乐观 | $250M ARR | 35x ARR | $8.8B | 品类领导力与 AI 倍数韧性 | 护城河与基础设施成本 |
情景估值是示意性的承销敏感性,不是预测;数值四舍五入到十亿美元一位小数。
[CV036, CV037, CV038, CV045, CV046]示例估值结果从低于最近价格的熊市情景,到需要 AI 溢价倍数持续存在的牛市情景。
区间围绕情景表数值套用敏感性带,并已取整。
[CV037, CV038, CV043]估值逻辑从已披露 ARR 出发,先看增长、品类稀缺性和倍数耐久性,再扣除不利风险。
瀑布图为方向性桥接,混合 ARR 和价值单位;合计值仅展示估值逻辑。
[CV036, CV039, CV041, CV042, CV045]8.4 估值驱动因素与下行风险
正向估值驱动来自品类稀缺、产品宽度、使用规模,以及 2026 年市场对 AI 增长的奖励。 AI 应用正进入 3D 工作流,Meshy 也吃到了这条叙事红利。反向驱动同样关键:免费增值漏斗可以做出亮眼的 注册用户和模型生成指标,却不一定带来同比例的付费留存;在位厂商和大型 AI 平台可能削弱差异化; 更广泛的 AI 市场仍容易遭遇炒作周期重置。持怀疑态度的估值资料强调资本效率和盈利路径, 上述字段正是 Meshy 尚未公开披露的内容。核心尽调问题,是从免费用户到账户付费再到留存 ARR 的收入质量桥接, 并配上工作流级毛利率。没有这条收入桥接,投资人承销的可能只是用户规模观感,而不是经济耐久性。[CV032, CV033, CV034, CV035, CV041, CV042]
| 触发项 | 阈值 / 事件 | 传导机制 | 行动含义 |
|---|---|---|---|
| ARR 增长放缓 | 下次承销检查时 ARR 低于 $60M | 50x 入场倍数无法安全消化 | 转向回避 |
| 付费转化弱 | 用户增长上升但付费账户停滞 | 免费增值规模失去估值意义 | 要求价格重置 |
| 利润率拖累 | 推理或训练成本撑不起软件型利润率 | 退出倍数滑向较低 SaaS / 基础设施区间 | 按悲观情景承销 |
| 护城河侵蚀 | 现有创意或 AI 平台追平工作流质量 | 竞争压缩估值倍数 | 延后入场或压低入场价 |
| 优先权压力 | Series B 条款显著损害普通股或新资金上行空间 | 即便公司增长,回报结构仍恶化 | 要求拿到资本结构条款 |
这些触发项是从估值传导风险推导出的尽调监控点,并非已知已经触发的事项。
[CV041, CV042, CV044, CV045, CV046]8.5 退出路径、流动性与最终尽调问题
从公开证据看,Meshy 的退出路径存在,但不近在眼前。2026 年融资与退出市场新独角兽多、 IPO/M&A 活动更强,给可选性提供支撑;创意软件或 AI 基础设施领域的战略买家也有合理理由关注 3D 资产生成。不过,IPO 准备度需要经审计的收入质量、治理、留存、利润率,以及可预测的企业客户扩张; 并购准备度需要证明 Meshy 拥有工作流层,而不是一个可替换的模型功能。最终尽调清单应聚焦六项: 收入质量、毛利率、股权结构条款、企业客户牵引、护城河耐久性和退出准备度。如果六项均得到验证, ARR 扩大后,当前估值可以从偏撑走向合理。如果结果令人失望,同一价格会变得昂贵, 因为倍数压缩会直接传导为下行风险。[CV017, CV018, CV029, CV030, CV040, CV043]
| 主题 | 缺失证据 | 重要性 | 尽调路径 |
|---|---|---|---|
| 收入质量 | 付费客户数、NRR、GRR 与 ARR 桥 | 验证 12M 用户能否变现 | 索取分群 ARR 瀑布 |
| 毛利率 | 按工作流拆分的推理、训练和云成本 | 决定可持续软件倍数 | 审阅云账单与利润率桥 |
| 股权结构 | 清算优先权、期权池与按比例认购权 | 决定入场价与回报瀑布 | 审阅融资文件 |
| 企业客户进展 | 具名企业客户、ACV 与续约行为 | 检验通往 $100M+ ARR 的路径 | 客户访谈与合同样本 |
| 护城河韧性 | 对标竞品的质量、延迟与工作流锁定 | 决定退出倍数韧性 | 做盲测资产质量基准测试 |
| 退出准备度 | 审计准备度、收入确认与治理 | 决定 IPO 或并购可选性 | CFO 尽调与买家格局复盘 |
这些问题定义了从跟踪上调到买入前必须确认的事项。
[CV040, CV041, CV042, CV043, CV044, CV045]8.6 附录
免责声明
本报告仅供参考,不构成投资建议。
证据索引
| 编号 | 陈述 | 可信度 | 来源 |
|---|---|---|---|
| CO001 | Meshy is an AI-powered 3D content creation platform for turning text, images, and creative concepts into exportable 3D assets. | 高 | SO001, SO002 |
| CO002 | Meshy positions its product around text-to-3D, image-to-3D, AI texturing, animation, and API workflows. | 高 | SO002, SO006 |
| CO003 | Meshy is rooted in Silicon Valley with a global team according to its about page. | 中 | SO002 |
| CO004 | A third-party profile states that Meshy AI was founded in 2021 by Ethan Hu in San Jose, California. | 中 | SO017 |
| CO005 | Meshy describes Ethan Hu as the founder and CEO of Meshy. | 中 | SO014, SO012 |
| CO006 | Ethan Hu is described by Meshy as an MIT-trained Ph.D. known for creating the Taichi GPU programming language. | 中 | SO014, SO021, SO022 |
| CO007 | Public sources reviewed for this chapter did not identify a second named Meshy founder or a named CFO, COO, or CTO. | 中 | SO002, SO012, SO014, SO015 |
| CO008 | The concentration of public leadership references around Ethan Hu creates a key-person diligence dependency. | 中 | SO012, SO014, SO015 |
| CO009 | Meshy raised nearly $400 million in a Series B round announced in July 2026. | 高 | SO012, SO013 |
| CO010 | Meshy disclosed a $1.5 billion valuation in the July 2026 Series B announcement. | 高 | SO012, SO013 |
| CO011 | The Series B announcement called the financing the largest funding round to date for a company built specifically for AI 3D. | 中 | SO012, SO013 |
| CO012 | The Series B was backed by IDG Capital, Matrix Partners China, and Monolith Management as lead or named major investors. | 中 | SO012, SO016 |
| CO013 | Existing or participating investors named in public coverage include Granite Asia, Sequoia China or HongShan, BAI Capital, and Source Code Capital. | 中 | SO012, SO016 |
| CO014 | Meshy said the Series B proceeds would be used primarily for research and development and global market expansion. | 中 | SO012 |
| CO015 | The SaaS News reported Meshy at over $1.38 billion valuation, which is lower than the company-distributed $1.5 billion figure. | 中 | SO016 |
| CO016 | The canonical report-wide valuation used in this chapter is $1.5 billion because the company-distributed July 2026 release and Yahoo reprint give that value. | 中 | SO012, SO013, SO016 |
| CO017 | Meshy reported more than 12 million registered users as of July 2026. | 中 | SO012, SO013 |
| CO018 | Meshy reported more than 100 million models created by July 2026. | 中 | SO012, SO013, SO014 |
| CO019 | Meshy announced at GDC 2026 that annual recurring revenue had doubled to $30 million in three months. | 中 | SO014 |
| CO020 | 36Kr reported that Meshy ARR exceeded $40 million in April 2026, creating a higher private-market data point than the $30 million GDC milestone. | 中 | SO015 |
| CO021 | This chapter treats approximately $30 million ARR as the canonical public milestone and flags later ARR values as unverified private-company reporting. | 中 | SO014, SO015 |
| CO022 | No reliable public source reviewed disclosed Meshy headcount as an exact current employee count. | 中 | SO002, SO012, SO015, SO018, SO019, SO020 |
| CO023 | 36Kr quoted an internal-letter concept of a 150-person company delivering the output of a 1,500-person team, but that is not a verified headcount disclosure. | 中 | SO015 |
| CO024 | Meshy runs a freemium subscription and credit model with Free, Pro, Premium, Studio, Ultra, and Enterprise tiers. | 中 | SO003, SO029 |
| CO025 | Meshy lists Pro at $20, Premium at $40, Studio at $60, and Ultra at $100 in its pricing surface reviewed for this run. | 中 | SO003 |
| CO026 | Meshy provides a REST API for programmatic 3D generation workflows. | 中 | SO004, SO008, SO009 |
| CO027 | Meshy documentation describes Text to 3D and Image to 3D API endpoints as separate generation paths. | 中 | SO008, SO009 |
| CO028 | Meshy supports export or workflow formats including FBX, OBJ, GLB, USDZ, STL, 3MF, and BLEND across official and help documentation. | 中 | SO001, SO007 |
| CO029 | Meshy integrates or advertises workflows with Blender, Unity, Unreal Engine, Maya, Godot, and multiple 3D-printing slicers. | 中 | SO001, SO035 |
| CO030 | Meshy 3D Agent turns conversations, text, photos, or sketches into print-ready 3D models according to the July 2026 announcement. | 中 | SO012, SO032 |
| CO031 | Auto Split is a one-click part-splitting feature for 3D printing that Meshy says is available to registered users. | 中 | SO012, SO033 |
| CO032 | Meshy launched Meshy Labs and Black Box: Infinite Arsenal at GDC 2026 as an experimental AI-native gameplay initiative. | 中 | SO014 |
| CO033 | Meshy 6 was released shortly before the GDC 2026 Meshy Labs announcement. | 中 | SO014, SO034 |
| CO034 | Meshy presents enterprise controls including SOC 2 Type II, ISO 27001, GDPR, SSO, separate enterprise data storage, and dedicated account support. | 中 | SO001 |
| CO035 | Meshy reported that five of the world's ten largest technology companies have teams building with Meshy. | 中 | SO012 |
| CO036 | Meshy named Nexon, NetEase Games, 37 Interactive Entertainment, Bambu Lab, Creality, Elegoo, FlashForge, xTool, Hugo Boss, and Sweden's museum of art and design as customers or partners. | 中 | SO012 |
| CO037 | The public evidence reviewed does not disclose Meshy's exact cap table, ownership percentages, or prior-round terms. | 中 | SO012, SO016, SO018, SO019, SO020 |
| CO038 | Restricted investor-database sources were discovered for Crunchbase, Tracxn, and PitchBook, but they did not provide readable corroboration through the approved fetch path. | 中 | SO018, SO019, SO020 |
| CO039 | The company's publicly named investor base includes several China-linked investors despite Meshy being presented as a Silicon Valley company. | 中 | SO002, SO012, SO016 |
| CO040 | No public legal, sanctions, or regulatory adverse event for Meshy was found in the reviewed chapter sources. | 中 | SO012, SO015, SO018, SO019, SO020 |
| CO041 | A prior milestone record indicates Meshy publicly launched Meshy-1 on October 19, 2023. | 中 | SO017 |
| CO042 | Meshy's about page reports 10 million-plus users while the July 2026 funding release reports 12 million-plus registered users, so this chapter uses the fresher July 2026 scale figure. | 中 | SO002, SO012 |
| CO043 | The Meshy x Formlabs tutorial is product-workflow evidence rather than a contractual partnership disclosure. | 中 | SO036 |
| CO044 | Meshy's growth narrative depends on company-reported private SaaS metrics rather than audited financial statements. | 中 | SO012, SO014, SO015 |
| CM001 | Meshy positions itself as an AI 3D modeling platform that transforms text descriptions, 2D images, and conversational prompts into production-ready 3D assets. | 高 | SM015, SM020 |
| CM002 | Meshy's official documentation lists Image, 3D Model, 3D Printing, Animate, Scene, and Video modules in its web workspace. | 高 | SM020, SM015 |
| CM003 | Meshy says every generation capability is available through REST API endpoints for developers building at scale. | 高 | SM019, SM020 |
| CM004 | Meshy sells packaged access through a public pricing page, making subscriptions and credits part of the observable payer path. | 高 | SM018, SM015 |
| CM005 | Meshy's text-to-3D feature is presented as a core workflow for generating 3D models from prompts. | 高 | SM016, SM020 |
| CM006 | Meshy's image-to-3D feature is presented as a core workflow for generating 3D models from images. | 高 | SM017, SM020 |
| CM007 | Research and Markets frames generative AI for 3D assets as a distinct market with component, deployment, asset-type, and end-user segmentation. | 中 | SM001 |
| CM008 | 3D AI Studio reports that the AI 3D market reached about $3.23 billion in 2026 and is projected near $9.4 billion by 2030. | 高 | SM002, SM001 |
| CM009 | GII reports that AI for 3D asset generation and texturing is projected to reach USD 12.84 billion by 2036 at a 20.8% CAGR from 2026 to 2036. | 中 | SM003 |
| CM010 | The Business Research Company reports the generative AI in gaming market will grow from $2.21 billion in 2026 to $5.09 billion in 2030. | 中 | SM004 |
| CM011 | The Business Research Company estimates a 23.1% CAGR from 2025 to 2026 for generative AI in gaming and a 23.2% CAGR to 2030. | 中 | SM004 |
| CM012 | GII identifies games, metaverse, and VFX as end-user categories for AI 3D asset generation and texturing. | 中 | SM003 |
| CM013 | GII says environment and props are estimated to hold the largest AI 3D asset-generation share in 2026. | 中 | SM003 |
| CM014 | GII says text-to-3D diffusion models are estimated to dominate the AI-model segment in 2026. | 中 | SM003 |
| CM015 | GII says plugin and API integration is expected to account for the largest integration share in 2026. | 中 | SM003, SM019 |
| CM016 | MarketsandMarkets estimates the metaverse market at $83.9 billion in 2023 and $1,303.4 billion by 2030. | 中 | SM005 |
| CM017 | MarketsandMarkets estimates the digital twin market will grow from $21.14 billion in 2025 to $149.81 billion in 2030. | 中 | SM006 |
| CM018 | Grand View Research estimates the digital twin market at $49.5 billion in 2026 and $328.5 billion by 2033. | 中 | SM009 |
| CM019 | Mordor Intelligence estimates the 3D rendering market will grow from $5.23 billion in 2026 to $13.92 billion by 2031. | 中 | SM007 |
| CM020 | Mordor Intelligence identifies gaming as the strongest 3D rendering end-use momentum segment at a 23.95% CAGR. | 中 | SM007 |
| CM021 | Mordor Intelligence reports that AR/VR and metaverse rendering workflows are the fastest-growing 3D rendering applications at 28.10% CAGR. | 中 | SM007 |
| CM022 | Autodesk Maya remains an incumbent professional alternative for 3D animation, modeling, simulation, and rendering workflows. | 中 | SM012 |
| CM023 | Blender is an open-source 3D creation suite and a no-license-fee substitute for parts of Meshy's workflow. | 中 | SM013 |
| CM024 | Adobe Substance 3D apps represent incumbent texturing and materials workflows adjacent to AI generated assets. | 中 | SM014 |
| CM025 | Sketchfab and TurboSquid represent asset-store substitutes where buyers can buy or download prebuilt 3D models instead of generating new ones. | 中 | SM027, SM028 |
| CM026 | NVIDIA Omniverse and Unreal Engine evidence that 3D content workflows extend into simulation, industrial digital twins, and real-time engines. | 中 | SM029, SM030 |
| CM027 | Meshy's July 2026 release says work that once took specialized skills, expensive software, and weeks can take about a minute and a dollar. | 中 | SM021 |
| CM028 | 3D AI Studio describes the traditional model-to-export pipeline as three to five days versus minutes with AI-assisted generation. | 中 | SM002 |
| CM029 | 3D AI Studio says open-source 3D models reached genuine production quality in 2025-2026, creating cost and commoditization pressure. | 中 | SM002 |
| CM030 | Gartner says generative AI entered the Trough of Disillusionment as organizations learned its potential and limits. | 中 | SM032 |
| CM031 | Gartner says organizations face governance challenges including hallucinations, bias, fairness, and regulation that can impede generative AI productivity applications. | 中 | SM032 |
| CM032 | Mordor Intelligence notes professional 3D software cost, piracy, talent scarcity, and workflow-skill gaps as constraints on the 3D rendering market. | 中 | SM007 |
| CM033 | PR Newswire reported in July 2026 that Meshy raised nearly $400 million in Series B financing at a $1.5 billion valuation. | 高 | SM021, SM023 |
| CM034 | PR Newswire reported in July 2026 that Meshy had more than 12 million registered users and over 100 million models created. | 高 | SM021, SM023 |
| CM035 | PR Newswire reported that Meshy's annual recurring revenue was growing about 12x year over year as of July 2026. | 高 | SM021, SM023 |
| CM036 | PR Newswire's GDC 2026 release ties Meshy Labs to AI-native gameplay and a $30 million ARR milestone. | 高 | SM022, SM021 |
| CM037 | Meshy's disclosed customer and partner examples span game companies, 3D-printing brands, and global consumer brands. | 中 | SM021 |
| CM038 | Indie game developers are likely user-buyer-payers when one creator can generate draft assets using public web pricing and a self-serve workflow. | 中 | SM016, SM018, SM020 |
| CM039 | AAA and mid-market studios are likely team or department buyers because workflows require pipeline integration, asset governance, and engine compatibility. | 中 | SM003, SM019, SM021, SM030 |
| CM040 | Product-design, manufacturing, and digital-twin buyers are adjacent SAM rather than core TAM because their budgets often attach to simulation and visualization suites. | 中 | SM006, SM009, SM029 |
| CM041 | Public evidence does not isolate Meshy's paid conversion rate, paying customer count by segment, or exact revenue mix across games, printing, design, and ecommerce. | 低 | |
| CM042 | Public market reports use different boundaries, so Meshy's serviceable market should be presented as a range rather than a single TAM figure. | 中 | SM001, SM002, SM003, SM004, SM007 |
| CM043 | The most conservative direct lens in this chapter is the $2.21 billion 2026 generative-AI-in-gaming estimate, not the broader metaverse or digital-twin totals. | 中 | SM004, SM005, SM006, SM009 |
| CP001 | Meshy positions itself as a free AI 3D model generator that converts text and images into 3D models in seconds. | 高 | SP001, SP003 |
| CP002 | Meshy disclosed a nearly $400 million Series B at a $1.5 billion valuation with more than 12 million registered users and over 100 million models created. | 高 | SP009, SP010 |
| CP003 | Meshy said at GDC 2026 that ARR doubled to $30 million in three months and that the platform passed 10 million global users. | 中 | SP010 |
| CP004 | Meshy pricing lists Free, Pro at $20 per month, Studio at $60 per month, and higher creator/team tiers on its current pricing page. | 中 | SP002 |
| CP005 | Meshy text-to-3D lets users select model type, pose, and generation count and typically produces a preview in about one minute. | 中 | SP003 |
| CP006 | Meshy image-to-3D supports Meshy 6, high-fidelity output near 600,000 faces, printability checks, and exports including FBX, OBJ, GLB, USDZ, STL, BLEND, and 3MF. | 中 | SP004 |
| CP007 | Meshy animation advertises auto-rigging in under 30 seconds and a library of more than 600 preset motion clips. | 中 | SP005 |
| CP008 | Meshy API access requires Pro tier or above and uses credit costs, rate limits, and queued-task limits by plan. | 中 | SP006 |
| CP009 | Meshy 3D Agent accepts text, photos, sketches, or rough ideas and returns downloadable 3D models in formats such as FBX, OBJ, GLB, USDZ, STL, BLEND, 3MF, and DXF. | 中 | SP007 |
| CP010 | Meshy describes a Formlabs Form Now workflow in which creators can print and ship Meshy-generated models from the workspace for US creators. | 中 | SP008 |
| CP011 | Tripo offers a text/image-to-3D product surface with public model examples oriented toward stylized game and asset generation. | 中 | SP011 |
| CP012 | Tripo pricing includes a free plan with 200 monthly credits and a Pro plan advertised around $19.90 monthly before annual discounts. | 中 | SP012 |
| CP013 | Tripo paid plans advertise Smart Mesh, ultra-high mesh quality, multi-view to 3D, batch generation, and bulk export. | 中 | SP012 |
| CP014 | Luma positions itself around creative agents, physical-world intelligence, image generation, and video generation rather than a pure text-to-3D asset workflow. | 中 | SP013, SP014 |
| CP015 | Luma APIs emphasize image and video generation pipelines with Ray and Uni models instead of Meshy-style downloadable mesh-first workflows. | 中 | SP014 |
| CP016 | Hyper3D describes Rodin as a high-quality controllable AI 3D model generator producing meshes, UVs, textures, and editable 3D models. | 中 | SP015 |
| CP017 | Hyper3D pricing includes a free generate-before-confirmation workflow and a Creator plan advertised at $30 monthly or $24 monthly on annual billing. | 中 | SP016 |
| CP018 | Rodin paid packaging advertises Smart Low-Poly, HD Texture, Custom Texture, baked normals, and more polycount options. | 中 | SP016 |
| CP019 | Kaedim positions itself as a production platform that turns sketches, reference packs, photos, briefs, and art direction into 3D assets teams can inspect and revise. | 中 | SP017 |
| CP020 | Kaedim emphasizes professional review loops, unlimited revisions, customer IP ownership, and no training on customer IP. | 中 | SP017 |
| CP021 | Spline is a web-based collaborative platform for production-ready interactive 2D and 3D experiences. | 中 | SP018 |
| CP022 | Spline AI generates 3D objects from text prompts and images inside the dashboard or editor. | 中 | SP019 |
| CP023 | Spline pricing shows paid seats at $12 and $20 per month billed annually, with AI credits included in higher plans. | 中 | SP020 |
| CP024 | 3D AI Studio says Meshy alternatives include aggregators, premium quality specialists such as Rodin, game-optimized tools such as Tripo, and template-based tools. | 中 | SP021 |
| CP025 | 3D AI Studio criticizes Meshy for locking users into a single AI model and argues multi-model platforms offer more flexibility. | 中 | SP021 |
| CP026 | 3D AI Studio identifies Rodin as the stronger choice for geometry quality and Tripo as the stronger choice for raw speed. | 中 | SP022 |
| CP027 | The Indie Hackers benchmark ranked Hyper3D Rodin first and Meshy second, with Meshy strongest on iteration speed and plugins. | 中 | SP023 |
| CP028 | The Indie Hackers benchmark described Rodin as stronger for production-ready geometry, clean quad topology, UVs, and PBR textures. | 中 | SP023 |
| CP029 | Blender remains a free and open-source full 3D creation suite with modeling, sculpting, UV, rendering, and ecosystem advantages. | 中 | SP024 |
| CP030 | Autodesk Maya remains a professional 3D modeling and animation tool with APIs, scripting, procedural Bifrost workflows, and AI-controlled motion tools. | 中 | SP025 |
| CP031 | ZBrush remains a specialized digital sculpting, modeling, and painting tool with more than 200 proprietary brushes. | 中 | SP026 |
| CP032 | Adobe Substance 3D focuses on professional-quality 3D materials and texturing rather than full text-to-3D generation. | 中 | SP027 |
| CP033 | NVIDIA Omniverse targets simulation-ready assets, OpenUSD workflows, synthetic data, neural reconstruction, and scene optimization for physical AI pipelines. | 中 | SP028 |
| CP034 | Hunyuan3D publicly releases open-source 3D generation models, Blender add-ons, texture modules, and training-code updates. | 中 | SP029 |
| CP035 | TRELLIS is an open large 3D asset generation model that outputs radiance fields, 3D Gaussians, and meshes from text or image prompts. | 中 | SP030 |
| CP036 | Stability AI says Stable Fast 3D can transform a single image into a 3D asset in 0.5 seconds with mesh, materials, albedo, and optional remeshing. | 中 | SP031 |
| CP037 | Google DeepMind Genie 3 generates interactive environments from text prompts at 24 frames per second, illustrating big-tech movement toward world models. | 中 | SP032 |
| CP038 | Meshy differentiation is strongest when speed, scale, API access, animation, plugins, and 3D-printing workflow breadth matter more than maximal production topology. | 中 | SP001, SP003, SP005, SP006, SP008, SP010 |
| CP039 | Meshy is most exposed when buyers prioritize photorealistic production geometry, clean quad topology, or multi-model fallback instead of fastest iteration. | 中 | SP021, SP022, SP023 |
| CP040 | Price compression risk is credible because Tripo, Spline, open-source models, and Stability-style fast reconstruction all offer low-cost or free entry points. | 中 | SP012, SP020, SP029, SP030, SP031 |
| CP041 | Switching costs in AI 3D are moderate because teams can multi-home across generators but may retain Meshy for API integrations, asset history, and engine workflows. | 中 | SP006, SP021, SP022 |
| CP042 | Manual tools defend through precision, artist control, plugins, and established pipelines even when AI generators accelerate first drafts. | 中 | SP024, SP025, SP026, SP027 |
| CP043 | CSM AI was not scored in the feature matrix because current public product and pricing evidence was not reliably retrievable in this run. | 低 | |
| CI001 | Meshy announced a nearly $400 million Series B financing in July 2026. | 高 | SI001, SI005 |
| CI002 | Meshy announced a $1.5 billion post-money valuation for the July 2026 Series B. | 高 | SI001, SI005 |
| CI003 | IDG Capital, Matrix Partners China and Monolith Management led Meshy's Series B financing. | 高 | SI001, SI004 |
| CI004 | Granite Asia, HongShan, BAI Capital and Source Code Capital participated in the Series B syndicate. | 高 | SI001, SI006 |
| CI005 | Meshy described the Series B as oversubscribed. | 中 | SI001 |
| CI006 | Meshy said the Series B proceeds will fund multimodal 3D foundation-model R&D, infrastructure and global enterprise expansion. | 高 | SI001, SI006 |
| CI007 | Meshy announced a $30 million ARR milestone at GDC 2026. | 高 | SI002, SI005 |
| CI008 | 36Kr Europe headlined Meshy as having over $300 million ARR. | 低 | SI003 |
| CI009 | The $300 million ARR headline is a low-confidence outlier because it conflicts with Meshy's official $30 million ARR milestone. | 中 | SI002, SI003 |
| CI010 | Meshy said ARR grew approximately 12x year over year by GDC 2026. | 中 | SI002 |
| CI011 | Meshy reported more than 12 million users in the Series B announcement. | 高 | SI001, SI005 |
| CI012 | Meshy reported more than 100 million generated 3D models in the Series B announcement. | 高 | SI001, SI005 |
| CI013 | Meshy stated that half of the world's top ten technology companies by market capitalization are customers. | 中 | SI001 |
| CI014 | Meshy offers a Free plan with 100 credits per month. | 高 | SI013, SI014 |
| CI015 | Meshy lists a Pro plan at $20 per month. | 高 | SI013, SI014 |
| CI016 | Meshy lists a Studio plan at $60 per month. | 高 | SI013, SI014 |
| CI017 | Meshy lists an Ultra plan at $100 per month. | 高 | SI013, SI014 |
| CI018 | Meshy lists Enterprise pricing as custom rather than publicly posted. | 中 | SI013 |
| CI019 | Meshy's docs state that plan credits can be spent across generation workflows. | 中 | SI014 |
| CI020 | Meshy requires Pro tier or above to use its public API. | 中 | SI015 |
| CI021 | Meshy documents Text to 3D API access as an integrable product capability. | 中 | SI018 |
| CI022 | Meshy documents Image to 3D API access as an integrable product capability. | 中 | SI019 |
| CI023 | Official pricing pages disclose list prices but not realized net revenue, discounts or enterprise contract values. | 中 | SI013, SI014 |
| CI024 | Meshy is a private company and did not disclose gross margin in the reviewed public sources. | 中 | SI001, SI002, SI030 |
| CI025 | Meshy did not disclose CAC, payback period or sales-efficiency metrics in the reviewed public sources. | 中 | SI001, SI013, SI030 |
| CI026 | Meshy did not disclose net revenue retention or gross revenue retention in the reviewed public sources. | 中 | SI001, SI002, SI030 |
| CI027 | Meshy did not disclose monthly burn or runway in the reviewed public sources. | 中 | SI001, SI006, SI030 |
| CI028 | The nearly $400 million Series B materially improves Meshy's capital adequacy absent contrary burn evidence. | 中 | SI001, SI006 |
| CI029 | Using the official $30 million ARR milestone and $1.5 billion valuation implies an approximate 50x ARR valuation multiple. | 中 | SI001, SI002, SI022 |
| CI030 | AI-native software can command premium revenue multiples when growth and defensibility are exceptional. | 中 | SI022, SI023, SI028 |
| CI031 | Several 2026 SaaS benchmark sources report ordinary SaaS multiples far below Meshy's implied 50x ARR multiple. | 中 | SI024, SI025, SI026, SI027 |
| CI032 | The valuation case depends on whether Meshy's ARR is durable, retained and expandable rather than merely top-of-funnel freemium conversion. | 中 | SI022, SI029, SI013 |
| CI033 | Meshy's freemium packaging creates a conversion-risk diligence issue because free users and generated models are not equivalent to paying accounts. | 中 | SI013, SI014, SI029 |
| CI034 | The official public materials do not disclose paid customer count separately from registered users. | 中 | SI001, SI013 |
| CI035 | The official public materials do not disclose revenue mix among subscriptions, enterprise contracts, API usage and credit consumption. | 中 | SI001, SI013, SI015 |
| CI036 | Meshy's API monetization adds usage-based revenue potential beyond seat subscriptions. | 中 | SI015, SI018, SI019 |
| CI037 | Enterprise custom pricing may support higher contract values but cannot be underwritten from public list prices alone. | 中 | SI013, SI015 |
| CI038 | PitchBook and Seedtable identify public funding-profile records for Meshy, but some details remain behind restricted or summary surfaces. | 中 | SI008, SI010 |
| CI039 | Tracxn provided an independent company-profile cross-check for Meshy before the July 2026 Series B. | 中 | SI009 |
| CI040 | No public SEC filing result for Meshy financial statements was obtained through the fetched SEC endpoint. | 中 | SI030 |
| CI041 | Crunchbase access was rate-limited during this run. | 低 | SI011 |
| CI042 | CB Insights returned a page-not-found response for the retained Meshy company-profile URL during this run. | 低 | SI012 |
| CI043 | G2 pricing access required JavaScript during this run, limiting independent pricing corroboration from that review surface. | 低 | SI020 |
| CI044 | MeshyReview independently summarized Meshy pricing tiers shortly before the run date. | 低 | SI021 |
| CI045 | The most important financial diligence blocker is not top-line traction but undisclosed margins, burn, retention, conversion and enterprise net revenue. | 中 | SI001, SI013, SI022, SI030 |
| CI046 | The funding announcement positions Meshy as having grown from early research into a unicorn within roughly five years. | 中 | SI001, SI006 |
| CI047 | A $30 million ARR base with 12x year-over-year growth implies the prior-year ARR base was much smaller than the current run-rate. | 中 | SI002 |
| CI048 | Absent disclosed revenue recognition policy, credits and API usage should be treated as monetization mechanics rather than audited revenue. | 中 | SI014, SI015, SI030 |
| CE001 | Meshy presents itself as an AI 3D platform that generates 3D assets from text and images. | 高 | SE001, SE002 |
| CE002 | The public product catalog includes Text-to-3D, Image-to-3D, AI texturing, animation, remesh, Meshy Agent, Auto Split, and API surfaces. | 高 | SE001, SE027 |
| CE003 | Meshy’s Text-to-3D feature claims users can generate fully textured 3D models from a text prompt in under one minute. | 中 | SE003 |
| CE004 | Meshy’s Image-to-3D feature claims it can infer a 3D structure from a single image in less than one minute. | 中 | SE004 |
| CE005 | Meshy’s texture feature supports texture generation from text or image prompts and higher-detail outputs. | 高 | SE005, SE020 |
| CE006 | The Text-to-3D API exposes PBR maps including metallic, roughness, and normal maps when PBR is enabled. | 高 | SE017, SE020 |
| CE007 | Meshy documents an animation surface for rigging models and applying animation-library motions. | 高 | SE006, SE023 |
| CE008 | Meshy’s public materials describe an animation library with 500-plus game-ready motions. | 中 | SE006 |
| CE009 | Meshy supports export workflows that include GLB, FBX, OBJ, STL, BLEND, and USDZ in public help or review sources. | 高 | SE029, SE040 |
| CE010 | Meshy’s animation API returns animation outputs in GLB and FBX and also includes USDZ-related processed URLs. | 中 | SE023 |
| CE011 | Meshy positions use cases across game development, 3D printing, ecommerce, education, film production, and XR. | 中 | SE008 |
| CE012 | The official Unity plugin page frames Meshy as a way to push generated 3D models directly into Unity workflows. | 中 | SE015 |
| CE013 | Meshy’s help center states the product has plugins for external workflows. | 中 | SE028 |
| CE014 | Meshy describes a ComfyUI partner node that brings generated, textured, and rigged assets into node-based workflows. | 中 | SE014 |
| CE015 | Meshy’s API documentation covers reference and guide material for programmatic 3D generation. | 中 | SE016 |
| CE016 | Meshy provides a Text-to-3D API endpoint for programmatic text-driven model generation. | 中 | SE017 |
| CE017 | Meshy provides Image-to-3D and Multi-Image-to-3D API endpoints for image-based model generation. | 高 | SE018, SE019 |
| CE018 | Meshy provides Retexture, Remesh, Rigging, and Animation API endpoints for post-generation asset operations. | 高 | SE020, SE021, SE022, SE023 |
| CE019 | Meshy’s webhook documentation allows task status updates to be sent automatically to configured HTTPS payload URLs. | 中 | SE024 |
| CE020 | Meshy’s Balance API retrieves current credit balance for accounts using Meshy services. | 中 | SE025 |
| CE021 | Meshy API pricing documentation ties API usage to credit consumption. | 中 | SE026 |
| CE022 | The Text-to-3D API documentation shows a preview and refine style task flow for generated assets. | 中 | SE017 |
| CE023 | Meshy’s remesh documentation includes smart-topology options and target face-count controls. | 中 | SE021 |
| CE024 | Meshy’s Auto Split feature is intended to segment 3D models for 3D printing and add watertight caps. | 高 | SE010, SE031 |
| CE025 | There’s An AI For That states most generated 3D models are not print-ready as generated because they may exceed build plates, need color separation, or contain open mesh surfaces. | 中 | SE039 |
| CE026 | Costbench reports user-friction concerns that retries can reproduce the same errors and consume credits inefficiently. | 中 | SE036 |
| CE027 | Independent directory pages classify Meshy as an AI 3D tool rather than a full traditional DCC replacement. | 中 | SE037, SE043, SE044 |
| CE028 | Public reviewed sources do not provide a rigorous third-party Meshy-versus-Rodin benchmark for geometry quality. | 低 | |
| CE029 | Meshy claims enterprise-grade security and certifications including SOC2 Type II, ISO 27001, and GDPR on its public surface. | 高 | SE001, SE002 |
| CE030 | Meshy says payment details are processed by third-party payment gateways and are not stored directly by Meshy. | 中 | SE041 |
| CE031 | Meshy says uploaded Image-to-3D concept art is not used to train its models without consent. | 中 | SE042 |
| CE032 | Meshy 3D Agent is described as a beta conversational workflow for ideation, concept generation, and model output. | 高 | SE009, SE030 |
| CE033 | Meshy’s public GitHub organization includes MCP and 3D-agent repositories related to the Meshy generation platform. | 中 | SE032, SE033 |
| CE034 | PR Newswire reported that Meshy unveiled Meshy Labs at GDC 2026. | 中 | SE034 |
| CE035 | The same PR Newswire release reported a $30M ARR milestone in connection with the Meshy Labs announcement. | 中 | SE034 |
| CE036 | Meshy 5 was announced with PBR texture improvements and reliability improvements. | 中 | SE012 |
| CE037 | Meshy 6 was announced as improving geometry and faster workflows. | 中 | SE011 |
| CE038 | Meshy’s product architecture remains proprietary beyond public API parameters, product docs, and marketing descriptions. | 中 | SE016, SE017, SE021 |
| CE039 | Product Hunt and other directory sources provide community-facing discovery evidence but not production-quality validation. | 中 | SE038, SE037 |
| CE040 | The API surface is broad enough to support asset generation, asset post-processing, event callbacks, and credit monitoring. | 高 | SE017, SE018, SE020, SE021, SE022, SE023, SE024, SE025 |
| CE041 | Meshy’s public product is strongest for rapid ideation and prototyping workflows where speed matters more than guaranteed clean topology. | 中 | SE003, SE004, SE023, SE025 |
| CE042 | The lack of public architecture papers or third-party benchmarks leaves model internals, training data composition, and repeatability unverified. | 低 | |
| CU001 | Meshy publicly targets game and media teams, 3D printing users, XR and education users, and browser-based creators. | 高 | SU002, SU012 |
| CU002 | Meshy’s customer page presents Stratton Studios, Thorns Tavern, and Jupiter as named customer references. | 中 | SU001 |
| CU003 | Stratton Studios said Meshy changed work that took weeks of modeling into hours of exploration. | 中 | SU001 |
| CU004 | Jupiter said a workflow that took one week now takes two hours using Meshy. | 中 | SU001, SU020 |
| CU005 | Jupiter’s one-week-to-two-hour case-study outcome is equivalent to a 98% production-time reduction. | 中 | SU001, SU020 |
| CU006 | HackerNoon disclosed that Jupiter, 37 Interactive Entertainment, and Thorns Tavern are Meshy customers in the discussed production cases. | 中 | SU020 |
| CU007 | 37 Interactive Entertainment reportedly reduced high-poly sculpting workload by 30% to 40% using a part-based Meshy workflow. | 中 | SU020 |
| CU008 | Thorns Tavern embedded the Meshy API into a consumer custom-miniature workflow. | 中 | SU020 |
| CU009 | Thorns Tavern reported modeling time dropping from one to two weeks to a few minutes after using Meshy. | 中 | SU020 |
| CU010 | Thorns Tavern reported an 80% per-model cost reduction in the HackerNoon production case. | 中 | SU020 |
| CU011 | Meshy’s July 2026 financing announcement stated that the company had more than 12 million registered users. | 高 | SU032, SU035 |
| CU012 | Meshy’s July 2026 financing announcement stated that users had created more than 100 million models. | 高 | SU032, SU035 |
| CU013 | Meshy’s March 2026 GDC announcement stated that it served more than 10 million users at individual and enterprise scale. | 中 | SU034 |
| CU014 | Analytics Insight reported Meshy had 10 million users and more than 100 million generated 3D models in 2026. | 中 | SU023 |
| CU015 | AIxploria described Meshy as having more than 10 million creators and more than 100 million generated models. | 中 | SU030 |
| CU016 | Meshy’s public traction metrics are registered-user and generated-model figures, not active-user or paying-user figures. | 中 | SU032, SU034, SU035 |
| CU017 | Meshy pricing lists a Free plan, Pro at $20 per month, Studio at $60 per month, and custom Enterprise pricing. | 中 | SU003 |
| CU018 | Meshy says free users receive 100 credits each month. | 中 | SU003 |
| CU019 | Meshy says Pro provides 1,000 credits per month and API access. | 高 | SU003, SU004 |
| CU020 | Meshy says non-Enterprise API-generated models are retained for a maximum of three days. | 高 | SU004, SU009 |
| CU021 | Meshy says Enterprise API users have a 100 RPS rate limit and a customizable queued-task allowance that defaults to 50. | 高 | SU004, SU008 |
| CU022 | Meshy says API access requires Pro tier or above. | 高 | SU004, SU006 |
| CU023 | Meshy says premium-plan users own assets they create, while free-plan assets are licensed under CC BY 4.0. | 高 | SU003, SU013, SU014 |
| CU024 | The CC BY 4.0 free-tier condition means commercial free-tier users must credit Meshy when using generated assets. | 高 | SU003, SU013 |
| CU025 | 3D Printing Industry reported that Meshy connected model creation directly to Formlabs’ Form Now print-on-demand service. | 中 | SU021 |
| CU026 | 3D Printing Industry reported that Form Now users can receive a manufactured part in as little as two days. | 中 | SU021 |
| CU027 | 3D Printing Industry reported the prompt-to-confirmed-order sequence takes under five minutes in the Form Now integration. | 中 | SU021 |
| CU028 | 3D Printing Industry reported xTool and Snapmaker as companies building custom creative tools on Meshy’s API. | 中 | SU021 |
| CU029 | 3D Printing Industry reported Flashforge was working with Meshy on full-color model compatibility for a Q2 2026 printer launch. | 中 | SU021 |
| CU030 | 3D Printing Industry reported Meshy partnered with MakerWorld to embed AI model generation into Bambu Lab’s ecosystem. | 中 | SU021 |
| CU031 | Meshy documentation presents plugins or workflows for Blender, Unity, Unreal, Roblox, and Godot. | 中 | SU010, SU011, SU016, SU017 |
| CU032 | Meshy’s official help center includes a dedicated education-plan article for students and educators. | 中 | SU012 |
| CU033 | Meshy’s Help Center says it can support game or project workflows. | 中 | SU015 |
| CU034 | SaaSHub summarized public opinion as positive on accessibility but critical of Meshy’s depth for advanced users. | 中 | SU025 |
| CU035 | SaaSHub reported critiques around performance and scalability in demanding design environments. | 中 | SU025 |
| CU036 | Meshy’s help center includes guidance for fixing hollow Meshy models for 3D printing. | 中 | SU018 |
| CU037 | Meshy’s help center includes guidance for checking and fixing model printability. | 中 | SU019 |
| CU038 | TopAI.tools reports 13 reviews for Meshy AI and says 92.3% of users recommend it. | 中 | SU028 |
| CU039 | Toolify’s Meshy profile reports 38.8K users for the tool page. | 中 | SU029 |
| CU040 | SaaSworthy lists Meshy with freemium, limited-feature free access and paid credit packages. | 中 | SU031 |
| CU041 | Slashdot presents Meshy as a 3D generative AI production suite and lists alternative products. | 中 | SU024 |
| CU042 | Future Tools describes Meshy as a freemium platform for 3D content, texturing, and modeling. | 中 | SU027 |
| CU043 | There’s An AI For That describes Meshy as serving game developers, designers, makers, and indie developers. | 中 | SU026 |
| CU044 | Meshy does not publicly disclose NRR, GRR, churn, renewal rate, or contract length in the sources reviewed for this chapter. | 低 | |
| CU045 | Meshy does not publicly disclose paid-subscriber count, free-to-paid conversion, active-user cohorts, or enterprise account count in the sources reviewed for this chapter. | 低 | |
| CU046 | Meshy does not publicly disclose top-customer concentration or channel revenue mix in the sources reviewed for this chapter. | 低 | |
| CU047 | The public customer evidence is strongest for production workflow outcomes and weakest for retention durability. | 中 | SU001, SU020, SU021, SU025 |
| CU048 | The named-customer record is a sample of public references rather than an exhaustive customer list. | 中 | SU001, SU020, SU022 |
| CR001 | Meshy presents itself as a text-and-image-to-3D generator that creates editable 3D models in seconds. | 高 | SR001, SR002 |
| CR002 | Meshy states that its platform supports text-to-3D, image-to-3D, AI texturing, animation, and API workflows. | 中 | SR002 |
| CR003 | Meshy reports more than 100 million generated assets on its about page. | 中 | SR002 |
| CR004 | 3Dnatives reported that Meshy had more than 12 million registered users and over 100 million models created as of July 2026. | 中 | SR005 |
| CR005 | 36Kr and 3Dnatives reported that Meshy raised nearly $400 million in Series B financing in July 2026. | 中 | SR004, SR005 |
| CR006 | 3Dnatives and Ohsem reported Meshy's first publicly disclosed valuation as $1.5 billion. | 中 | SR005, SR006 |
| CR007 | At an assumed $30 million ARR base, a $1.5 billion valuation would equal roughly 50 times ARR. | 中 | SR005, SR006 |
| CR008 | 36Kr reported Series B participation from IDG Capital, Matrix Partners China, Monolith Management, Granite Asia, HongShan, BAI Capital, and Source Code Capital. | 中 | SR004 |
| CR009 | Meshy describes Ethan Hu as founder and chief executive in funding coverage. | 中 | SR005, SR006 |
| CR010 | Meshy's privacy policy identifies Meshy LLC as the operator of the service. | 中 | SR033 |
| CR011 | Meshy's Help Center states that assets and user files are stored with Amazon Web Services in the United States. | 中 | SR035 |
| CR012 | Meshy's Help Center states that Meshy maintains ISO/IEC 27001:2022 and SOC 2 certifications. | 中 | SR035 |
| CR013 | Meshy's data FAQ says non-enterprise user data may be used for future AI model training depending on plan. | 高 | SR035, SR033 |
| CR014 | Meshy's data FAQ says enterprise customer data is not used for model training. | 中 | SR035 |
| CR015 | Meshy's commercial-use help article says paid plan users own generated assets outright, subject to rights in their inputs. | 高 | SR036, SR034 |
| CR016 | Meshy's commercial-use help article says free-plan outputs are licensed under CC BY 4.0 attribution terms. | 高 | SR036, SR034 |
| CR017 | The Hunyuan3D 2.0 paper describes high-resolution textured 3D asset generation. | 中 | SR007 |
| CR018 | Tencent-Hunyuan publishes Hunyuan3D-2 code on GitHub. | 中 | SR008 |
| CR019 | Microsoft's TRELLIS project page describes structured 3D latents for scalable and versatile 3D generation. | 中 | SR009 |
| CR020 | Microsoft's TRELLIS.2 project page describes native and compact structured latents for 3D generation. | 中 | SR010 |
| CR021 | The TripoSR repository describes fast 3D object reconstruction from a single image. | 中 | SR011 |
| CR022 | Tripo markets a text-and-image AI 3D model generator. | 中 | SR012 |
| CR023 | Hyper3D Rodin markets an AI 3D model generator with text or image input and enterprise controls. | 中 | SR014 |
| CR024 | Adobe Substance 3D remains an incumbent professional 3D design software suite. | 中 | SR037 |
| CR025 | NVIDIA's GTC 2026 materials highlighted Omniverse digital-twin blueprints and physical-AI infrastructure. | 中 | SR015 |
| CR026 | Adobe and NVIDIA announced a 2026 strategic partnership for next-generation Firefly models and creative workflows. | 中 | SR016 |
| CR027 | Google's I/O 2026 announcement list emphasized AI agents and platform-wide AI features. | 中 | SR017 |
| CR028 | OpenAI maintains an active company-announcements page for frontier AI releases and partnerships. | 中 | SR018 |
| CR029 | Presenc AI tracks AI training-data lawsuits as a 2026 legal-risk category. | 中 | SR019 |
| CR030 | Axis Intelligence describes an AI copyright lawsuits tracker covering live case status. | 中 | SR020 |
| CR031 | is4.ai frames AI copyright lawsuits as a complete 2026 legal guide for the sector. | 中 | SR021 |
| CR032 | The Authors Guild describes the Anthropic settlement as a $1.5 billion copyright settlement. | 高 | SR022, SR024 |
| CR033 | TechCrunch reported that the $1.5 billion Anthropic copyright settlement was approved in July 2026. | 高 | SR023, SR022 |
| CR034 | TechCrunch reported that the Anthropic ruling distinguished AI training fair use from the acquisition of pirated books. | 高 | SR023, SR024 |
| CR035 | Kluwer Copyright Blog described the Bartz settlement class as covering reproduction-right owners of books in pirated datasets. | 中 | SR024 |
| CR036 | JPMorgan Chase Center for Geopolitics frames U.S.-China AI competition as a systemic strategic contest. | 中 | SR025 |
| CR037 | CNBC reported that the U.S. State Department ordered a global warning about alleged China AI thefts by DeepSeek and others. | 中 | SR026 |
| CR038 | CNBC reported that China-linked actors target AI companies through cyberattacks and insider-risk vectors. | 中 | SR027 |
| CR039 | The Treasury CFIUS page states that CFIUS reviews certain foreign investment transactions for national-security implications. | 高 | SR028, SR029 |
| CR040 | GAO reported that U.S. foreign-investment mitigation efforts address national-security risks. | 高 | SR029, SR028 |
| CR041 | A&O Shearman reported that CFIUS launched a known-investor pilot program while maintaining scrutiny of foreign-adversary transactions. | 高 | SR030, SR028 |
| CR042 | A&O Shearman reported that CFIUS required mitigation for about 9% of notices filed in 2024. | 中 | SR030 |
| CR043 | Medium's Meshy workflow review reported that some character-generation use cases still need manual adjustments. | 中 | SR032 |
| CR044 | Medium's Meshy review contrasted traditional 3D modeling's maximum control with Meshy's speed and beginner-friendly workflow. | 中 | SR032 |
| CR045 | Meshy's official product claims imply inference-dependent operations that can pressure GPU capacity as user volume grows. | 中 | SR001, SR015 |
| CR046 | Meshy's financing proceeds are reported to target research and global market expansion. | 中 | SR005, SR004 |
| CR047 | Meshy's global user base and U.S. AWS storage create a data-residency diligence item for non-U.S. customers. | 中 | SR002, SR035, SR033 |
| CR048 | Meshy's China-linked investor base creates a plausible CFIUS and U.S.-China scrutiny risk for future governance or control rights. | 中 | SR004, SR028, SR030, SR026 |
| CR049 | Open-source 3D models reduce defensibility if Meshy's differentiation depends mainly on model availability rather than workflow, data rights, or distribution. | 中 | SR007, SR008, SR009, SR011 |
| CR050 | The absence of public gross margin, burn, retention, and customer-concentration disclosures prevents a complete residual-risk score for financial execution. | 低 | |
| CV001 | Meshy announced a nearly $400 million Series B financing in July 2026. | 高 | SV001, SV003 |
| CV002 | Meshy announced a $1.5 billion valuation for the July 2026 Series B. | 高 | SV001, SV003 |
| CV003 | Meshy described the round as the largest funding round to date for an AI-3D company. | 高 | SV001, SV003 |
| CV004 | Meshy disclosed a $30 million ARR milestone at GDC 2026. | 高 | SV002, SV001 |
| CV005 | Meshy said ARR had grown about 12x year over year. | 中 | SV002, SV003 |
| CV006 | A $1.5 billion valuation divided by $30 million ARR implies about 50x ARR. | 高 | SV001, SV002 |
| CV007 | Meshy reported more than 12 million users in the Series B announcement. | 高 | SV001, SV003 |
| CV008 | Meshy reported more than 100 million generated 3D models in the Series B announcement. | 高 | SV001, SV003 |
| CV009 | IDG Capital, Matrix Partners China and Monolith Management were named among the Series B backers. | 中 | SV001 |
| CV010 | Granite Asia, HongShan, BAI Capital and Source Code Capital were named among existing investor participants. | 中 | SV001 |
| CV011 | ValueAdd VC frames 2026 AI startup multiples around 10x to 50x revenue versus 3x to 7x for SaaS. | 中 | SV004 |
| CV012 | TLDL reports that foundation-model startups can command roughly 20x to 50x ARR in 2026. | 中 | SV010 |
| CV013 | Finro’s Q1 2026 AI multiples research is a market-data reference for private AI multiples. | 中 | SV011 |
| CV014 | SaaSRise identifies AI software valuation multiples as materially above classic SaaS levels in 2026. | 中 | SV012 |
| CV015 | Acquiry’s 2026 SaaS multiple work supports a classic SaaS benchmark far below Meshy’s implied 50x ARR. | 中 | SV015 |
| CV016 | ScaleXP’s SaaS ARR multiple discussion reinforces that normal recurring-software multiples sit well below frontier AI marks. | 中 | SV016 |
| CV017 | TechCrunch reported that almost 40 new unicorns had been minted so far in 2026. | 中 | SV006 |
| CV018 | Crunchbase News reported a record H1 2026 global startup investment environment with stronger exit activity. | 中 | SV007 |
| CV019 | Crunchbase News reported that AI helped push Q1 2026 venture funding to record levels. | 中 | SV008 |
| CV020 | Agent Market Cap’s 2026 landscape describes investor willingness to fund high-growth AI companies at premium valuations. | 中 | SV009 |
| CV021 | Baseten announced a $1.5 billion financing to power AI inference infrastructure in June 2026. | 高 | SV021, SV022 |
| CV022 | TechCrunch reported Baseten was raising $1.5 billion months after its last mega-round. | 中 | SV022 |
| CV023 | Sacra provides a revenue and valuation profile for Baseten that is useful for AI infrastructure benchmarking. | 中 | SV023 |
| CV024 | Economic Times reported Genspark was valued at $2.6 billion in a 2026 funding round. | 中 | SV024 |
| CV025 | Axios Pro reported Genspark reached a $2.6 billion valuation with a $100 million extension. | 中 | SV025 |
| CV026 | SaaSRise reported Genspark’s Series B extension at a $2.6 billion valuation. | 中 | SV026 |
| CV027 | Tripo operates an AI 3D product and publishes pricing, making it a direct product comp but not a disclosed valuation comp. | 中 | SV027, SV028 |
| CV028 | Luma positions around creative AI and APIs, making it an adjacent creative-infrastructure comp rather than a direct mesh-first comp. | 中 | SV029, SV030 |
| CV029 | Adobe’s SEC submissions identify it as a public creative-software filing comp for mature software valuation context. | 高 | SV017, SV019 |
| CV030 | NVIDIA’s SEC submissions identify it as a public AI-infrastructure filing comp for strategic-buyer context. | 高 | SV018, SV020 |
| CV031 | CB Insights’ AI 100 provides a private AI company benchmark universe but does not provide a full AI-3D comp set. | 中 | SV031 |
| CV032 | Andreessen Horowitz argues AI applications are being built in 3D, supporting a large creative-workflow thesis. | 中 | SV032 |
| CV033 | Gartner’s AI hype-cycle framing is an adverse warning that adoption timing and inflated expectations can diverge. | 中 | SV033 |
| CV034 | TLDL’s 2026 valuation discussion warns that AI multiples remain far above SaaS norms despite moderation. | 中 | SV010 |
| CV035 | Flippa’s 2026 AI valuation analysis emphasizes capital efficiency and profitability path as checks on high private-market prices. | 中 | SV013 |
| CV036 | The base case assumes Meshy can grow ARR from about $30 million to about $120 million while the exit multiple compresses to 25x. | 中 | SV001, SV002, SV004 |
| CV037 | The bear case assumes ARR reaches about $60 million and the multiple compresses to 12x, implying a value below the latest post-money. | 中 | SV002, SV013, SV033 |
| CV038 | The bull case assumes ARR reaches about $250 million and a 35x premium multiple persists, implying meaningful upside. | 中 | SV002, SV004, SV010 |
| CV039 | Meshy’s current valuation stance is stretched because the 50x ARR multiple prices in several years of fast execution. | 中 | SV001, SV002, SV004, SV010 |
| CV040 | The recommendation is track rather than buy because public evidence supports momentum but not retention, margin, preference stack or paid conversion. | 中 | SV001, SV002, SV013, SV033 |
| CV041 | Freemium conversion uncertainty should reduce willingness to underwrite the headline user count as revenue quality. | 中 | SV001, SV002, SV013 |
| CV042 | Big-tech and incumbent creative-tool competition can compress Meshy’s exit multiple if workflow control shifts to incumbents. | 中 | SV019, SV020, SV033 |
| CV043 | A plausible exit path is a strategic acquisition or IPO only after enterprise ARR scale, retention and margin evidence become public or diligence-proven. | 中 | SV007, SV017, SV018 |
| CV044 | Private cap-table terms, liquidation preferences and dilution from the large Series B remain undisclosed in reviewed public sources. | 低 | |
| CV045 | A move to at least $100 million ARR with durable enterprise retention would make the current price easier to treat as fair. | 中 | SV002, SV004, SV010 |
| CV046 | Failure to prove paid conversion or a material growth slowdown would move the recommendation toward avoid. | 中 | SV013, SV033 |
| CV047 | A scenario range, comps bar chart and KPI scorecard are the clearest visual summaries for IC review. | 中 | SV004, SV010, SV013 |
| CV048 | The comparable-company table is a sample because many private AI rounds do not disclose ARR, valuation or preference terms. | 中 | SV023, SV025, SV031 |
| CV049 | AI investors in 2026 are emphasizing market capture and growth durability over current profitability for the strongest companies. | 中 | SV005, SV007, SV008, SV009 |
| CV050 | Meshy leads the narrow AI-3D private-round sample by disclosed round size and valuation among reviewed AI-3D specialists. | 中 | SV001, SV027, SV028, SV031 |
| 编号 | 出版方 | 标题 | 引文 |
|---|---|---|---|
| SO001 | Meshy | Meshy homepage | Meet the world's most popular and intuitive free AI 3D model generator. |
| SO002 | Meshy | About Us | Meshy is an AI-powered 3D content creation platform that supports Text to 3D, Image to 3D, AI Texturing, Animation, and API workflows. |
| SO003 | Meshy | Pricing | Free, Pro, Premium, Studio, Ultra, and Enterprise plans are presented on the pricing page. |
| SO004 | Meshy | API platform | Tap directly into Meshy’s generation power via a robust, well-documented REST API. |
| SO005 | Meshy Help Center | What is Meshy? | Meshy is a tool to generate 3D models from text prompts and images. |
| SO006 | Meshy Help Center | What features does Meshy have? | Meshy supports text to 3D, image to 3D, AI texturing, animation, and API workflows. |
| SO007 | Meshy Help Center | What 3D file formats do you support? | The help page lists supported 3D file formats. |
| SO008 | Meshy Docs | Text to 3D API | The Text to 3D API documentation describes task creation and model generation. |
| SO009 | Meshy Docs | Image to 3D API | The Image to 3D API documentation describes image-based model generation. |
| SO010 | Meshy Docs | Animation API | The Animation API documentation covers animation task endpoints. |
| SO011 | Meshy Docs | Analyze Printability API | The printability endpoint analyzes whether a generated model is printable. |
| SO012 | PR Newswire / Meshy | Meshy raises nearly $400 million at a $1.5 billion valuation | Meshy announced it has raised nearly $400 million in a Series B round at a $1.5 billion valuation. |
| SO013 | Yahoo Finance | Meshy raises nearly $400 million at a $1.5 billion valuation | As of July 2026, the company's annual recurring revenue is growing about 12x year over year, with more than 12 million registered users and over 100 million models created. |
| SO014 | PR Newswire / Meshy | Meshy unveils Meshy Labs at GDC 2026 | The platform doubled its annual recurring revenue to $30 million in just three months. |
| SO015 | 36Kr Europe | Exclusive: Meshy AI is redefining 3D generation | 36Kr reported April 2026 ARR exceeded $40 million and described a 150-person AI-native organization. |
| SO016 | The SaaS News | Meshy Raises $400M Series B at $1.38B Valuation | The SaaS News reported a nearly $400 million Series B and a valuation over $1.38 billion. |
| SO017 | Grokipedia | Meshy AI | Meshy AI was founded in 2021 by Ethan Hu in San Jose, California. |
| SO018 | Crunchbase | Meshy AI organization profile | Crunchbase profile was not readable through the fetch workflow. |
| SO019 | Tracxn | Meshy.ai company profile | Tracxn returned a blocked or not-found profile in the fetch workflow. |
| SO020 | PitchBook | Meshy company profile | PitchBook profile fetch exposed only a tracker page, not profile content. |
| SO021 | GitHub | Taichi programming language repository | Taichi is an open-source programming language for high-performance computer graphics. |
| SO022 | Yuanming Hu | Yuanming Hu personal website | Yuanming Hu's website describes his research in computer graphics and physical simulation. |
| SO023 | Taichi Docs | Taichi documentation | Taichi documentation describes the programming language and its GPU-oriented workflow. |
| SO024 | The Decoder | Meshy raises $400 million for AI-powered 3D model generation | The Decoder article was retained as independent funding coverage. |
| SO025 | TechFundingNews | From MIT research to $1.5B unicorn | TechFundingNews was discovered but blocked during fetch. |
| SO026 | VoxelMatters | Meshy raises $400 million Series B | VoxelMatters was discovered but blocked during fetch. |
| SO027 | BAI Capital | BAI Capital website notice | BAI Capital page fetch returned investment-risk notices and broker-dealer disclaimers. |
| SO028 | HongShan | HongShan website | HongShan website lists offices and investment platform context. |
| SO029 | Meshy Help Center | How do credits work? | Meshy Help explains credit consumption mechanics. |
| SO030 | Meshy | AI texture generator feature | The feature page describes AI texture generation for 3D models. |
| SO031 | Meshy | AI animation generator feature | The feature page describes AI animation generation for 3D models. |
| SO032 | Meshy Blog | Meshy 3D Agent | Meshy describes Meshy 3D Agent as an AI agent for 3D creation. |
| SO033 | Meshy Blog | Auto Split 3D printing | Auto Split divides 3D models into printable parts. |
| SO034 | Meshy Blog | Meshy 6 launch | Meshy 6 launch coverage describes a newer model generation release. |
| SO035 | Meshy Blog | AI 3D models for Unity official Meshy plugin 2026 | Meshy describes an official Unity workflow for generated 3D assets. |
| SO036 | Meshy Tutorials | Meshy x Formlabs workflow | The tutorial presents an industrial-grade 3D printing workflow with Meshy and Formlabs. |
| SM001 | Research and Markets | Generative Artificial Intelligence (AI) for Three-Dimensional (3D) Assets Global Market Report | |
| SM002 | 3D AI Studio | The State of AI 3D Generation in 2026 | |
| SM003 | Global Information, Inc. | AI for 3D Asset Generation & Texturing Market Forecasts (2026-2036) | The AI for 3D asset generation and texturing market is projected to reach USD 12.84 billion by 2036. |
| SM004 | The Business Research Company | Generative AI In Gaming Global Market Report 2026 | |
| SM005 | MarketsandMarkets | Metaverse Market Size & Share, Global Forecast | |
| SM006 | MarketsandMarkets | Digital Twin Market Size, Share & Trends - Global Forecast to 2030 | |
| SM007 | Mordor Intelligence | 3D Rendering Market Size and Share Analysis | |
| SM008 | Statista | AR & VR - Worldwide | |
| SM009 | Grand View Research | Digital Twin Market Size, Share & Trends Analysis Report | |
| SM010 | Newzoo | Global Games Market Report 2025 | |
| SM011 | Unity | Gaming Report | |
| SM012 | Autodesk | Maya: 3D Computer Animation, Modeling, Simulation, and Rendering Software | |
| SM013 | Blender Foundation | About Blender | |
| SM014 | Adobe | Substance 3D Apps | |
| SM015 | Meshy | Meshy homepage | |
| SM016 | Meshy | Text to 3D | |
| SM017 | Meshy | Image to 3D | |
| SM018 | Meshy | Pricing | |
| SM019 | Meshy | Meshy API | |
| SM020 | Meshy Docs | Meshy Documentation | |
| SM021 | PR Newswire | Meshy Raises Nearly $400 Million at a $1.5 Billion Valuation | Meshy has raised nearly $400 million in a Series B round at a $1.5 billion valuation. |
| SM022 | PR Newswire | Meshy Unveils Meshy Labs at GDC 2026 | |
| SM023 | Yahoo Finance | Meshy raises nearly $400 million | |
| SM024 | 36Kr Europe | Meshy raises nearly USD 400 million Series B | |
| SM025 | The Decoder | Meshy raises $400 million for AI-powered 3D model generation | |
| SM026 | The SaaS News | Meshy Raises $400M Series B | |
| SM027 | Sketchfab | 3D Models | |
| SM028 | TurboSquid | 3D Models for Professionals | |
| SM029 | NVIDIA | NVIDIA Omniverse | |
| SM030 | Unreal Engine | Unreal Engine | |
| SM031 | Andreessen Horowitz | AI Apps Are Being Built in 3D | |
| SM032 | Gartner | Hype Cycle for Artificial Intelligence | Gen AI enters the Trough of Disillusionment as organizations gain understanding of its potential and limits. |
| SP001 | Meshy | Meshy homepage | Meet the world's most popular and intuitive free AI 3D model generator. |
| SP002 | Meshy | Meshy pricing | |
| SP003 | Meshy | Text to 3D feature page | |
| SP004 | Meshy | Image to 3D feature page | |
| SP005 | Meshy | AI animation generator feature page | |
| SP006 | Meshy | Meshy API overview | |
| SP007 | Meshy | Meshy 3D Agent blog | |
| SP008 | Meshy | Meshy x Formlabs 3D printing tutorial | |
| SP009 | PR Newswire | Meshy raises nearly $400 million at a $1.5 billion valuation | more than 12 million registered users and over 100 million models created |
| SP010 | PR Newswire | Meshy unveils Meshy Labs at GDC 2026 and $30M ARR milestone | |
| SP011 | Tripo AI | Tripo AI homepage | |
| SP012 | Tripo AI | Tripo Studio pricing | |
| SP013 | Luma AI | Luma homepage | |
| SP014 | Luma AI | Build with Luma APIs | |
| SP015 | Hyper3D | Hyper3D homepage | |
| SP016 | Hyper3D | Rodin pricing | |
| SP017 | Kaedim | Kaedim homepage | |
| SP018 | Spline | Spline homepage | |
| SP019 | Spline | Spline AI | |
| SP020 | Spline | Spline pricing | |
| SP021 | 3D AI Studio | Comprehensive guide to Meshy.ai alternatives | If you need maximum quality and have the budget, go with Rodin AI ($99/mo). For game development, Tripo AI ($24/mo) is solid. |
| SP022 | 3D AI Studio | Best Meshy alternatives for AI-powered 3D modeling | Tripo is the best pick for raw speed, Rodin for geometry quality, and Hunyuan3D for image-to-3D detail. |
| SP023 | Indie Hackers | Best AI 3D model generator in 2026: tested nine tools | Hyper3D.ai (Rodin) kept ending up at the top. |
| SP024 | Blender Foundation | About Blender | |
| SP025 | Autodesk | Maya overview | |
| SP026 | Maxon | ZBrush overview | |
| SP027 | Adobe | Substance 3D | |
| SP028 | NVIDIA | Omniverse developer page | |
| SP029 | Tencent Hunyuan | Hunyuan3D-2 GitHub repository | |
| SP030 | Microsoft | TRELLIS GitHub repository | |
| SP031 | Stability AI | Introducing Stable Fast 3D | |
| SP032 | Google DeepMind | Genie 3 frontier world models | |
| SI001 | PR Newswire / Meshy | Meshy Raises Nearly $400 Million at a $1.5 Billion Valuation, the Largest Round to Date in AI 3D | Meshy announced it raised nearly $400 million at a $1.5 billion valuation and described the round as the largest to date in AI 3D. |
| SI002 | PR Newswire / Meshy | Meshy Unveils Meshy Labs at GDC 2026 -- Breakthrough AI-Native Gameplay and $30M ARR Milestone | Meshy said at GDC 2026 that it had reached a $30M ARR milestone. |
| SI003 | 36Kr Europe | Silicon Valley Unicorn with Over $300M ARR Becomes Popular in Global 3D Generative AI Field | 36Kr Europe headlined Meshy as having over $300M ARR, conflicting with Meshy's own $30M ARR milestone. |
| SI004 | Value Add Pulse | Meshy Raises Nearly $400M for AI 3D Generation | Value Add summarized the nearly $400M Series B and $1.5B valuation as an AI 3D funding record. |
| SI005 | Yahoo Finance | Meshy Raises Nearly $400 Million at a $1.5 Billion Valuation, the Largest Round to Date in AI 3D | Yahoo Finance republished the funding announcement and labeled it a paid press release. |
| SI006 | TechFundingNews | From MIT research to $1.5B unicorn: Ethan Hu’s Meshy raises $400M for AI-powered 3D creation | TechFundingNews framed the round as a move from MIT research to a $1.5B unicorn. |
| SI007 | BigGo Finance | Meshy Closes Nearly $400 Million Series B, Shattering AI 3D Generation Funding Record at Over $1.5 Billion Valuation | BigGo Finance described the Series B as nearly $400M and at over $1.5B valuation. |
| SI008 | Seedtable | Meshy — Funding, Investors & Team | Seedtable listed Meshy funding, investors and team information. |
| SI009 | Tracxn | Meshy - 2026 Company Profile, Team & Competitors | Tracxn profile content was used as an independent funding-profile cross-check. |
| SI010 | PitchBook | Meshy 2026 Company Profile: Valuation, Funding & Investors | PitchBook profile is a restricted-access funding and valuation profile for Meshy. |
| SI011 | Crunchbase | Meshy AI organization profile | Crunchbase blocked automated access, so it is retained only as a restricted diligence target. |
| SI012 | CB Insights | Meshy AI company profile | CB Insights returned a page-not-found response for the Meshy profile URL during this run. |
| SI013 | Meshy | Meshy Official Pricing: Free, Pro, Studio & Enterprise Plans | Meshy lists Free, Pro, Studio, Ultra and Enterprise packaging on its official pricing page. |
| SI014 | Meshy Docs | Meshy Pricing & Credits: Plans and Usage | Meshy Docs explains that plans include credits spent across generations and related features. |
| SI015 | Meshy | 3D Model Generation API — Text & Image to 3D | Meshy says users need a Meshy account and Pro tier or above to use the API. |
| SI016 | Meshy | Privacy Policy - Meshy | Meshy privacy policy states a January 26, 2026 revision date. |
| SI017 | Meshy | Blog - Meshy | Meshy official blog shows recent product updates and release cadence. |
| SI018 | Meshy Docs | Text to 3D API | Meshy Docs | The Text to 3D API documentation describes integration of Meshy Text to 3D capabilities. |
| SI019 | Meshy Docs | Image to 3D API | Meshy Docs | The Image to 3D API documentation describes integration of Meshy Image to 3D capabilities. |
| SI020 | G2 | Meshy Pricing 2026 | G2 pricing content required JavaScript during this run. |
| SI021 | MeshyReview | Meshy Pricing 2026: Which Plan Should You Choose? | MeshyReview summarized Meshy official pricing and credits shortly before the run date. |
| SI022 | ValueAddVC | AI Company Valuation Multiples Framework 2026: How Investors Price Pre-Revenue AI | ValueAddVC describes high AI valuation multiples but emphasizes diligence on revenue quality and defensibility. |
| SI023 | SaaSRise | The AI Software Valuation Report 2026 | SaaSRise compares venture and M&A revenue multiples across AI software categories. |
| SI024 | Acquiry | SaaS Valuation Multiples in 2026: What the Data Actually Shows | Acquiry notes the SaaS valuation correction and lower median public-market revenue multiples. |
| SI025 | ScaleXP | SaaS ARR & Revenue Valuation Multiples 2026 | ScaleXP summarizes 2026 SaaS valuation takeaways for finance teams. |
| SI026 | Windsor Drake | 2026 SaaS Valuation Multiples by ARR Band | Windsor Drake discusses private lower-middle-market SaaS multiples by ARR band. |
| SI027 | KCENav | SaaS Valuation Multiples 2026: Median 4.5x ARR, Top Quartile 8.1x+ | KCENav reports a private mid-market SaaS median of about 4.5x ARR and top quartile above 8.1x. |
| SI028 | SaaS Valuation Multiple | AI SaaS Valuation Multiples 2026 | The AI SaaS multiples article describes premium valuation bands for AI-native software. |
| SI029 | Recurly | The 2026 State of Subscriptions report | Recurly frames subscription performance around subscriber behavior and recurring billing trends. |
| SI030 | U.S. Securities and Exchange Commission | EDGAR full-text search endpoint for Meshy | The SEC endpoint returned no usable Meshy filing result in the fetched response, supporting a public-filing gap rather than a filed financial history. |
| SE001 | Meshy | AI 3D Model Generator: Create 3D from Text & Images | Meshy presents an AI 3D generator for creating 3D from text and images. |
| SE002 | Meshy | 3D Model Generation API — Text & Image to 3D | Meshy markets an API for 3D model generation from text and images. |
| SE003 | Meshy | Free Text to 3D AI Generator 2026: Prompts to Models | Generate fully textured 3D models from a simple text prompt in under 1 minute. |
| SE004 | Meshy | Free Image to 3D Model 2026 — Photo to 3D in a Minute | Transform a single image into a three-dimensional model in less than 1 minute. |
| SE005 | Meshy | AI Texture Generator: Create texture from image and text easily | Meshy describes AI texturing from text or image prompts and HD output. |
| SE006 | Meshy | AI 3D Animation Generator Online | Meshy describes rigging generated models and using an animation library. |
| SE007 | Meshy | Meshy Official Pricing: Free, Pro, Studio & Enterprise Plans | Meshy lists Free, Pro, Max, Max Unlimited, Studio, and Enterprise plan surfaces. |
| SE008 | Meshy | Meshy AI Use Cases | Professional AI 3D Modeling Solutions | Meshy positions its workflow for games, printing, commerce, education, film, and XR. |
| SE009 | Meshy | Meshy 3D Agent: The World’s First AI Agent for 3D Creation (Beta) | Meshy describes a beta chat-based 3D creation agent. |
| SE010 | Meshy | Auto Split Is Here: Make 3D Models Printable in Seconds | Auto Split returns a split result in approximately 40 seconds and adds watertight caps. |
| SE011 | Meshy | Meshy-6: Smarter Geometry, Faster Workflows, Limitless 3D Creativity | Meshy-6 is described as improving geometry and workflow speed. |
| SE012 | Meshy | Introducing Meshy 5: New PBR Textures, Higher Reliability, and Smarter AI Tools | Meshy 5 introduced PBR texture improvements, reliability, and smarter AI tools. |
| SE013 | Meshy | How to Choose GLB vs FBX vs OBJ in 2026 | Meshy explains GLB, FBX, OBJ, and STL export choices for 2026 workflows. |
| SE014 | Meshy | Meshy in ComfyUI: Official Partner Node for AI 3D Generation | Meshy describes an official partner node that brings generation into ComfyUI. |
| SE015 | Meshy | AI 3D Models for Unity — Official Meshy Plugin 2026 | Meshy describes an official Unity plugin for generating Unity-ready 3D models. |
| SE016 | Meshy Docs | Meshy API Docs: Reference & Guides | The Meshy API documentation is a reference and guide surface. |
| SE017 | Meshy Docs | Text to 3D API | Text-to-3D API docs describe preview and refine task objects and PBR outputs. |
| SE018 | Meshy Docs | Image to 3D API | Image-to-3D API docs describe creating 3D tasks from images. |
| SE019 | Meshy Docs | Multi-Image to 3D API | Multi-image API docs describe generating a model from multiple images. |
| SE020 | Meshy Docs | Retexture API | Retexture API docs describe applying textures to 3D assets. |
| SE021 | Meshy Docs | Remesh API | Remesh API docs describe topology, target polycount, and smart-topology options. |
| SE022 | Meshy Docs | Rigging API | Rigging API docs describe creating rigging tasks for 3D characters. |
| SE023 | Meshy Docs | Animation API | Animation API docs include GLB, FBX, USDZ, armature, and animation result URLs. |
| SE024 | Meshy Docs | Webhooks | Meshy allows at most five active webhooks per account and requires HTTPS payload URLs. |
| SE025 | Meshy Docs | Balance API | The Balance API retrieves current credit balance for Meshy services. |
| SE026 | Meshy Docs | Pricing | Meshy API pricing docs describe credit consumption for API operations. |
| SE027 | Meshy Help Center | What Features does Meshy have? | The help center enumerates Text to 3D, Image to 3D, Text to Texture, and Animation features. |
| SE028 | Meshy Help Center | Does Meshy have plugins? | Meshy help describes plugins for external creation workflows. |
| SE029 | Meshy Help Center | Choosing the Right 3D File Format to Download Your Meshy Models | The help center explains file-format choices for Meshy downloads. |
| SE030 | Meshy Help Center | Getting Started with Meshy Agent (Beta) | The help center describes Meshy Agent as a beta conversational workflow. |
| SE031 | Meshy Help Center | How Does Auto Split Work in Meshy? | The help center describes Auto Split for preparing models for printing. |
| SE032 | GitHub | meshy-dev/meshy-mcp-server | The repository is an MCP server for the Meshy AI 3D generation platform. |
| SE033 | GitHub | meshy-dev/meshy-3d-agent | The repository contains AI agent skills for the Meshy AI 3D generation platform. |
| SE034 | PR Newswire | Meshy Unveils Meshy Labs at GDC 2026 -- Breakthrough AI-Native Gameplay and $30M ARR Milestone | The release says Meshy unveiled Meshy Labs at GDC 2026 and cited a $30M ARR milestone. |
| SE035 | G2 | Meshy Pricing | The G2 page was bot-blocked in this run, so it is retained only as restricted-access review provenance. |
| SE036 | Costbench | Meshy Pricing 2026: 6 Plans from Free-$100/month | Costbench notes retries can reproduce the same errors and that extra credits may be expensive relative to plan credits. |
| SE037 | FutureTools | Future Tools - Meshy AI | FutureTools classifies Meshy as an AI tool for 3D generation. |
| SE038 | Product Hunt | Meshy: Create stunning 3D models with AI | Product Hunt archives a Meshy product page with community-facing product positioning. |
| SE039 | There’s An AI For That | Meshy v6 - AI Tool For 2D to 3D image conversion | The page states most generated models are not print-ready because they can exceed build plates, need color separation, or contain open mesh surfaces. |
| SE040 | Toolify | Meshy: 3D AI platform for generating 3D models from text or images | Toolify describes Meshy as generating 3D models from text or images and exporting FBX, OBJ, STL, BLEND, and USDZ. |
| SE041 | Meshy Help Center | How does Meshy ensure the security of payment information? | Meshy says payment details are processed by third-party payment gateways and not stored directly by Meshy. |
| SE042 | Meshy Help Center | Are concept art images uploaded in Image to 3D used to train your model? | Meshy says uploaded image-to-3D concept art is not used for training without consent. |
| SE043 | AIxploria | Meshy AI: Reviews, Price, Info & 60 Alternatives AI Tools | 2026 | AIxploria lists Meshy AI among 2026 AI tools with reviews, pricing, and alternatives. |
| SE044 | TopAI.tools | Meshy AI - AI 3D Tool | TopAI.tools describes Meshy AI as an AI 3D tool. |
| SU001 | Meshy | Customer Stories | How Leading Teams Scale 3D Content Creation with Meshy. |
| SU002 | Meshy | Use Cases | Game engines, slicers, motion pipelines, AR viewers. Meshy fits in. |
| SU003 | Meshy | Pricing | Free: $0; Pro: $20/mo; Studio: $60/mo; and Enterprise: custom pricing. |
| SU004 | Meshy | API | You will need to first create a Meshy account and be on the pro tier or above to use the API. |
| SU005 | Meshy | About Meshy | 100M+ |
| SU006 | Meshy Docs | Meshy API Documentation | |
| SU007 | Meshy Docs | API Pricing | |
| SU008 | Meshy Docs | API Rate Limits | |
| SU009 | Meshy Docs | API Asset Retention | |
| SU010 | Meshy Docs | Blender Plugin Introduction | |
| SU011 | Meshy Help Center | Does Meshy have plugins? | |
| SU012 | Meshy Help Center | Education Plan for Students and Educators | |
| SU013 | Meshy Help Center | Can I sell the models on other platforms? | |
| SU014 | Meshy Help Center | If I cancel my subscription will models revert to CC BY 4.0? | |
| SU015 | Meshy Help Center | Can Meshy support my game or project? | |
| SU016 | Meshy Help Center | Meshy to Blender: A Complete Workflow | |
| SU017 | Meshy Help Center | Integrating Meshy Assets into Unity/Unreal Engine | |
| SU018 | Meshy Help Center | How do I fix a hollow Meshy model for 3D printing? | How do I fix a hollow Meshy model for 3D printing? |
| SU019 | Meshy Help Center | How to check and fix your model’s printability | |
| SU020 | HackerNoon | AI 3D Generation in Production: Real Workflow Results from Game, Hardware, and Print Studios | Jupiter integrated Meshy's base mesh generation and cut basic model production time from 7 days to 2 hours. |
| SU021 | 3D Printing Industry | Meshy closes the 3D printing loop with AI-to-physical manufacturing | The Form Now integration extends that pipeline one step further. |
| SU022 | FeaturedCustomers | Meshy Case Studies | |
| SU023 | Analytics Insight | Meshy AI in 2026: The Platform Redefining 3D Design and Digital Content Production | Meshy has garnered 10 million users and powered more than 100 million 3D models. |
| SU024 | Slashdot | Meshy Reviews and Product Profile | |
| SU025 | SaaSHub | Meshy AI reviews. Is Meshy AI good? | Some technical reviews suggest that while Meshy AI excels in user-friendliness, it may not offer the depth of features demanded by advanced users. |
| SU026 | There’s An AI For That | Meshy v6 - AI Tool For 2D to 3D image conversion | Recognized in the 2026 G2 Best Software Awards for Best Design Software and Highest Customer Satisfaction. |
| SU027 | Future Tools | Meshy AI | |
| SU028 | TopAI.tools | Meshy AI - AI 3D Tool | Based on 13 reviews, 92.3% of users recommend Meshy AI. |
| SU029 | Toolify | Meshy Product Profile | 38.8K users |
| SU030 | AIxploria | Meshy AI: Reviews, Price, Info & Alternatives | Meshy has attracted more than 10 million creators, who have collectively generated over 100 million models. |
| SU031 | SaaSworthy | Meshy Product Overview | |
| SU032 | PR Newswire | Meshy Raises Nearly $400 Million at a $1.5 Billion Valuation | more than 12 million registered users and over 100 million models created |
| SU033 | 36Kr Europe | Meshy raises nearly $400 million at a $1.5 billion valuation | |
| SU034 | PR Newswire | Meshy Unveils Meshy Labs at GDC 2026 | serves more than 10 million users at the individual and enterprise scale |
| SU035 | Yahoo Finance | Meshy Raises Nearly $400 Million at a $1.5 Billion Valuation | more than 12 million registered users and over 100 million models created |
| SR001 | Meshy | AI 3D Model Generator: Create 3D from Text & Images | |
| SR002 | Meshy | About Us — Meshy AI 3D Generator | |
| SR003 | Meshy | Best AI Tools for 3D Printing in 2026 | |
| SR004 | 36Kr | Meshy completes nearly $400 million Series B financing | |
| SR005 | 3Dnatives | Meshy Raises Nearly $400 Million in Series B, Valued at $1.5 Billion | |
| SR006 | Ohsem | Meshy Raises Nearly $400 Million At A $1.5 Billion Valuation | |
| SR007 | arXiv | Hunyuan3D 2.0: Scaling Diffusion Models for High Resolution Textured 3D Assets | |
| SR008 | Tencent-Hunyuan | Hunyuan3D-2 GitHub repository | |
| SR009 | Microsoft | TRELLIS: Structured 3D Latents for Scalable and Versatile 3D Generation | |
| SR010 | Microsoft | TRELLIS.2: Native and Compact Structured Latents for 3D Generation | |
| SR011 | VAST-AI-Research | TripoSR: Fast 3D Object Reconstruction from a Single Image | |
| SR012 | Tripo AI | AI 3D Model Generator from Text & Images | Tripo 3D | |
| SR013 | Luma AI | Luma | AI Agents for Creative Work | |
| SR014 | Hyper3D | Hyper3D Rodin - Best AI 3D Model Generator | |
| SR015 | NVIDIA Newsroom | GTC 2026 News | |
| SR016 | Adobe Newsroom | Adobe and NVIDIA Announce Strategic Partnership to Deliver the Next Generation of Firefly Models | |
| SR017 | 100 things we announced at I/O 2026 | ||
| SR018 | OpenAI | OpenAI Newsroom | Recent news | |
| SR019 | Presenc AI | AI Training Data Lawsuit Tracker 2026 | |
| SR020 | Axis Intelligence | AI Copyright Lawsuits Tracker 2026 — Every Case, Live Status | |
| SR021 | is4.ai | AI Copyright Lawsuits 2026: Complete Legal Guide | |
| SR022 | Authors Guild | What Authors Need to Know About the $1.5 Billion Anthropic Settlement | |
| SR023 | TechCrunch | Anthropic's landmark $1.5B copyright settlement is approved | |
| SR024 | Kluwer Copyright Blog | The Bartz v. Anthropic Settlement: Understanding America's Largest Copyright Settlement | |
| SR025 | JPMorgan Chase Center for Geopolitics | Beyond the Benchmarks: A Systemic View of U.S.-China AI Competition | |
| SR026 | CNBC / Reuters | U.S. State Department orders global warning about alleged China AI thefts by DeepSeek, others | |
| SR027 | CNBC | China-linked actors target more than technology as AI competition with U.S. intensifies | |
| SR028 | U.S. Department of the Treasury | The Committee on Foreign Investment in the United States (CFIUS) | |
| SR029 | U.S. Government Accountability Office | Foreign Investment in the U.S.: Efforts to Mitigate National Security Risks Can Be Strengthened | |
| SR030 | A&O Shearman | Navigating the evolving U.S. national security investment landscape | |
| SR031 | AI Indigo | Meshy AI Review: Is it the Right Choice for Your 3D Workflow in 2026? | |
| SR032 | Medium | Meshy AI 3D Generator Review 2026: The Complete Production Workflow Tested | |
| SR033 | Meshy | Privacy Policy - Meshy | |
| SR034 | Meshy | Terms of Use - Meshy | |
| SR035 | Meshy Help Center | Is Meshy Safe and Private? Data and Training FAQ | |
| SR036 | Meshy Help Center | Can I use my generated assets for commercial projects? | |
| SR037 | Adobe | 3D design software - Adobe Substance 3D | |
| SV001 | PR Newswire | Meshy Raises Nearly $400 Million at a $1.5 Billion Valuation | raised nearly $400 million in a Series B round at a $1.5 billion valuation |
| SV002 | PR Newswire | Meshy Unveils Meshy Labs at GDC 2026 and $30M ARR Milestone | $30M ARR milestone |
| SV003 | ValueAdd VC | Meshy $400M Series B AI 3D 2026 | |
| SV004 | ValueAdd VC | AI Startup Valuation Multiples 2026 | AI trades at 10-50x vs SaaS at 3-7x |
| SV005 | ValueAdd VC | AI Startup Statistics 2026 | |
| SV006 | TechCrunch | Almost 40 New Unicorns Have Been Minted So Far This Year | |
| SV007 | Crunchbase News | Global Startup Investment Hit Record $510B In H1 2026 | |
| SV008 | Crunchbase News | Q1 2026 Shatters Venture Funding Records As AI Boom Pushes Startup Investment Higher | |
| SV009 | Agent Market Cap | AI Startup Valuation Landscape: Hottest Companies Funding Growth 2026 | |
| SV010 | TLDL | AI Startup Metrics and Valuations 2026 | Multiples have moderated from peak madness but remain far above SaaS norms |
| SV011 | Finro | AI Valuation Multiples Q1 2026 | |
| SV012 | SaaSRise | The AI Software Valuation Report 2026 | |
| SV013 | Flippa | AI Startups Valuation Multiples: Key Considerations for 2026 | |
| SV014 | Qubit Capital | How AI Company Valuations Work: Multiples and Benchmarks | |
| SV015 | Acquiry | SaaS Valuation Multiples 2026 | |
| SV016 | ScaleXP | SaaS ARR Revenue Valuation Multiples | |
| SV017 | U.S. Securities and Exchange Commission | Adobe submissions metadata | |
| SV018 | U.S. Securities and Exchange Commission | NVIDIA submissions metadata | |
| SV019 | Adobe | Adobe Substance 3D product page | |
| SV020 | NVIDIA Developer | NVIDIA Omniverse developer page | |
| SV021 | Business Wire | Baseten Raises $1.5 Billion to Power the Next Era of AI Inference | |
| SV022 | TechCrunch | AI Inference Startup Baseten Reportedly Raising $1.5B | |
| SV023 | Sacra | Baseten revenue, valuation and funding | |
| SV024 | Economic Times Entrepreneur | AI startup Genspark valued at $2.6 billion in latest funding round | |
| SV025 | Axios Pro | Genspark hits $2.6B valuation with $100M extension | |
| SV026 | SaaSRise | Genspark.ai closes $100M Series B extension at $2.6B valuation | |
| SV027 | Tripo | Tripo AI official website | |
| SV028 | Tripo | Tripo AI pricing | |
| SV029 | Luma AI | Luma AI official website | |
| SV030 | Luma AI | Luma AI API | |
| SV031 | CB Insights | AI 100 2026 report | |
| SV032 | Andreessen Horowitz | AI apps are being built in 3D | |
| SV033 | Gartner | Hype Cycle for Artificial Intelligence |