Noetix Robotics
尽职调查报告:消费级人形机器人先行者
Noetix 是中国高速迭代的人形机器人先锋,靠消费级低价和政策顺风切入市场,但单位经济还没跑通,拥挤赛道里的泡沫风险很高。
封面要素
公司概况
Noetix Robotics(松延动力)是一家总部位于北京的人形与仿生机器人公司,2023 年 9 月由 Jiang Zheyuan(CTO)和 Zhang Shipu(CEO)创立。公司开发消费级双足人形机器人(Bumi,$1,370;N2;E1)和仿生头部平台(Hobbs 系列,32 个面部表情自由度)。在 CATL 关联的 CD Capital 和北京机器人产业基金支持下,Noetix 在 2026 年 2 月以央视春晚人形仿生机器人独家合作伙伴身份进入大众视野,并已完成 9 轮融资,累计超过 RMB 1.5 billion。公司面向教育、家庭陪伴和商业服务市场,产能约 1,000 台 / 月。
- 成立时间
- 2023-09-01
- 创始人
- Jiang Zheyuan, Zhang Shipu
- 创立地点
- Beijing, China
- 总部
- Beijing, China
- 产品
- 多产品人形机器人组合覆盖消费级双足机器人(Bumi,$1,370)、敏捷研究平台(N2,$5,500)、高阶商用人形机器人(E1,$5,570-10,000)和仿生头部交互系统(Hobbs 系列)。所有产品都兼容 ROS2,并开放编程接口。
- 客户
- 教育机构、消费家庭(STEM / 陪伴)、商业场所和研究实验室
- 商业模式
- 以分层价位($1,370-$10,000)销售人形机器人硬件,并可能通过内容生态和维护获得经常性收入;早期商业牵引来自电商(JD.com)和直销。
- 阶段
- Series B
- 融资情况
- Series B 轮(约 RMB 1 billion / ~$145M)于 2026 年 3 月完成,由 CD Capital(CATL)领投。9 轮累计融资 >RMB 1.5B。
执行摘要
主要优势
- 已商业交付的全双足消费级人形机器人里,Bumi 价格最低($1,370)
- CCTV 春晚独家合作带来 10 亿+ 观众级品牌曝光
- 北京机器人产业基金和十五五机器人任务提供强政策顺风
- 双足机器人 + 32 DOF 仿生头双线打法独特,竞品尚未复刻
- 执行速度突出:2.5 年完成 9 轮融资,43 天造出首台机器人
主要风险
- 市场泡沫风险:140+ 公司在收入未起量前拿到独角兽估值,2027-2028 年或进入整合
- 收入和单位经济完全未披露;公开收入数据近乎为零,却估值 $1.39B
- AI 自主能力缺口限制消费机器人在预设动作之外的日常实用性
- Unitree R1 引发激烈价格战,12-18 个月内侵蚀成本领先护城河
- 技术愿景和 IP 高度依赖创始人 Jiang Zheyuan,关键人风险集中
未决问题
- 绝对收入、利润率和单位经济仍未公开
- 实际终端销售率与预订 / 产量之间的差距不明
- 完整董事会构成和治理结构未披露
- 消费者留存和日活使用数据缺失
- 专利组合防御力尚无独立评估
目录
01公司概览
1.1 身份、创立与公司结构
Noetix Robotics(官方注册名:松延动力(北京)科技集团股份有限公司,Songyan Power [Beijing] Technology Group Co., Ltd.)是一家总部位于北京的人工智能公司,聚焦人形机器人本体和仿人仿生面部的研发、制造与销售。公司由 Jiang Zheyuan 和 Zhang Shipu 于 2023 年 9 月创立;Jiang Zheyuan 任董事长兼 CTO,Zhang Shipu 任联合创始人兼 CEO。Jiang Zheyuan 从清华大学博士项目退学创业,使他成为中国人形机器人赛道最年轻的创始人之一。核心技术团队来自清华大学、中国科学院、浙江大学和香港大学。2026 年初,公司改制为股份制集团,为规模化前的治理补强。Jin Zhiyong 任运营总监。截至 2026 年,公司在北京、常州和东莞等地拥有 100 多名员工。不到三年从零扩到百人以上,反映出当前政策环境下中国深科技创业公司的速度和强度。[CO001, CO002, CO003, CO004, CO005, CO006]
| 人物 | 职务 | 背景 | 创始人-市场匹配 | 关键人物依赖 |
|---|---|---|---|---|
| Jiang Zheyuan (姜哲源) | 创始人、董事长兼 CTO | 清华大学博士辍学;中国人形机器人领域最年轻创始人 | 顶尖高校出身,机器人 / AI 技术积累深 | 高——技术愿景与 IP 归属 |
| Zhang Shipu (张世璞) | 联合创始人兼 CEO | 投资与国际教育背景 | 商业化与融资能力强,能连接资本市场 | 高——融资与战略 |
| Jin Zhiyong (晋志勇) | 运营负责人 | 运营管理 | 硬件制造的运营爬坡 | 中——运营执行 |
核心团队包括清华、中科院、浙江大学和 HKU 校友。完整高管团队和董事会构成未公开披露。
[CO001, CO002, CO003, CO004, CO005]1.2 融资历史与投资人格局
Noetix Robotics 自 2023 年 9 月成立以来已完成 9 轮融资;对一家硬件机器人创业公司来说,这个节奏异常快。最近一轮是 2026 年 3 月完成的近 RMB 1 billion(约 USD 140–145 million)Series B,由 CATL 旗下产业投资平台 Chen Dao Capital(CD Capital)领投。联合投资方包括 Guoke Investment、Jingguosheng Fund 和 Jiuhe Venture Capital。Series B 之前,公司在 2025 年末从 Fangguang Capital 和 CICC Capital 获得近 RMB 500 million 的 Pre-B 和 Pre-B+ 轮融资。由 Shoucheng Holdings 管理的北京机器人产业发展投资基金曾两次投资(2024 年 3 月、2025 年 9 月),构成带有国资关联的背书,也释放出强政策契合信号。Series B 后,媒体报道公司的投后估值接近 RMB 10 billion。公司累计融资超过 RMB 1.5 billion。北京机器人基金、Guoke、Jingguosheng 等国资关联资本,与 CICC、Fangguang、Jiuhe 等民营成长资本共同构成混合投资人基础,降低了单一投资人依赖。[CO008, CO009, CO010, CO011, CO012, CO013]
| 利益相关方 | 角色 | 轮次 / 日期 | 意义 | 尽调问题 |
|---|---|---|---|---|
| Chen Dao Capital(CD Capital / CATL 产业资本) | Series B 轮领投方 | Series B 轮,March 2026 | 全球最大 EV 电池制造商背书,释放供应链协同信号 | 确认是否有战略条款,而非纯财务投资 |
| Beijing Robot Industry Fund(Shoucheng Holdings 管理) | 战略投资人 | 2024 + Sep 2025 | 国资关联基金(RMB 10B 目标规模)释放政策同频和优先准入信号 | 确认治理权利与监督义务 |
| CICC Capital | Pre-B+ 轮领投方 | Pre-B+ 轮,Nov 2025 | 中国头部投行背书,验证财务可信度 | 评估后续跟投承诺 |
| Fangguang Capital | Pre-B 轮领投方 | Pre-B 轮,Oct 2025 | 押注机器人方向的成长期 VC | 核实基金周期与退出时间线 |
| Guoke Investment | Series B 轮共同投资人 | Series B 轮,March 2026 | 国资背景资本进一步强化政策同频 | 厘清共同投资条款 |
| Jingguosheng Fund | Series B 轮共同投资人 | Series B 轮,March 2026 | 市 / 省级基金参与 | 确认承诺出资规模 |
| Jiuhe Venture Capital | Series B 轮共同投资人 | Series B 轮,March 2026 | 早期 VC 在成长期继续持仓 | 评估历史入场价格 |
成立以来共九轮融资;表内仅列公开公告中具名的投资人。更早期种子 / 天使投资人未公开披露。
[CO008, CO009, CO010, CO011, CO012, CO013]Noetix Robotics 的身份、产品、客户、资本和依赖关系如何连接。
[CO001, CO006, CO019, CO032]1.3 产品组合与关键里程碑
Noetix 采取双轨产品策略:一端是双足人形机器人,另一端是仿生人形系统。旗舰消费产品 Bumi 是一台 94 cm、12 kg 的双足人形机器人,定价 RMB 9,998(约 USD 1,370),面向教育、家庭陪伴和 STEM 编程。Bumi 拥有 21 个自由度,兼容 ROS2,搭载 Rockchip 处理器,并开放编程接口。N2「Athlete」型号(118 cm、30 kg、18 DOF)面向敏捷运动研究,价格约 USD 5,500。E1 型号(136 cm、21+ DOF)服务高阶研究和商业导览,价格 USD 5,570–10,000。Hobbs 系列覆盖仿生头部平台,具备 32 个主动面部自由度,用于更有情绪共鸣的交互。关键里程碑包括:2023 年 9 月公司成立;43 天造出第一台机器人;N2 在 2025 北京人形机器人半程马拉松中获得亚军;Bumi 于 2025 年 10 月发布,并在 JD.com 获得 500 多个预订单;2026 年 2 月 16 日以人形仿生机器人独家合作伙伴身份登上央视春晚;2026 年 3 月完成 Series B。公司拥有 30 多项核心专利,截至 2025 年 7 月单月交付超过 100 台机器人。N2 还成为全球首款运行 Huawei OpenHarmony 操作系统的消费级人形机器人,目标是接入智能家居生态。[CO016, CO017, CO018, CO019, CO020, CO021]
| 日期 | 事件 | 类型 | 金额 / 状态 | 参与方 | 影响 |
|---|---|---|---|---|---|
| 2023-09 | 公司成立 | 成立 | Jiang Zheyuan, Zhang Shipu | 公司在北京成立;首台机器人 43 天完成 | |
| 2024-03 | Beijing Robot Fund 首次投资 | 融资 | 未披露 | Beijing Robot Industry Fund | 早期即获得国资关联背书 |
| 2025-04 | N2 获北京人形机器人半程马拉松亚军 | 产品 | 21km,3h37m | Noetix N2 | 验证敏捷运动能力 |
| 2025-05 | Hobbs 3 发布及生态战略大会 | 产品 | Noetix Robotics | 推出 32 DOF 仿生头部平台 | |
| 2025-07 | 单月交付超过 100 台机器人 | 规模 | >100 台 | Noetix Robotics | 量产能力得到验证 |
| 2025-08 | 世界人形机器人运动会夺冠 | 产品 | 金牌与银牌 | N2、E1 机器人 | 国际赛事获奖提升可信度 |
| 2025-09 | Beijing Robot Fund 第二次投资(HKEX 文件) | 融资 | 未披露 | Shoucheng Holdings | 国资继续跟投;HKEX 文件披露 |
| 2025-10 | Bumi 消费机器人以 RMB 9,998 发布 | 产品 | RMB 9,998 价格点 | Noetix Robotics | 万元以下价格打开大众消费市场 |
| 2025-10 | Pre-B 轮交割 | 融资 | ~RMB 300M | Fangguang Capital | 成长期资金支持规模化 |
| 2025-11 | Pre-B+ 轮交割 | 融资 | ~RMB 200M | CICC Capital | Pre-B 系列合计约 RMB 500M |
| 2025-12 | Hobbs Mini 1 仿生头部平台取得突破 | 产品 | Noetix Robotics | 面向行业的标准化仿生头部模块 | |
| 2026-02-16 | CCTV 春晚表演 | 合作 | 独家合作伙伴 | CCTV、Noetix Robotics | 借数十亿观众获得大众品牌曝光 |
| 2026-03 | Series B 轮以约 RMB 1B 交割 | 融资 | ~RMB 1B (~USD 140-145M) | CD Capital (CATL)、Guoke、Jingguosheng、Jiuhe 参投 | 估值接近独角兽;进入 CATL 生态 |
时间线来自 Noetix 官方网站、HKEX 文件和媒体报道。更早期种子 / 天使轮未公开标注日期。
[CO008, CO016, CO019, CO020, CO021, CO022]从创立到 Series B 的关键里程碑,涵盖产品发布、赛事获奖、融资轮次和媒体曝光。
[CO020, CO021, CO022, CO023, CO024, CO033]1.4 封面指标与信息缺口
对一家成立两年的公司而言,Noetix Robotics 的增长叙事很有吸引力,但几项关键指标仍未披露。Series B 后估值据报接近 RMB 10 billion(约 USD 1.39 billion)。累计融资超过 RMB 1.5 billion。员工数已确认超过 100 人。产能正爬坡至每月 1,000 台。但年收入、ARR、毛利率、净收入留存率和精确客户数仍属私有信息。公司的披露画像是私有且未披露;除投资人和媒体报道外,没有公开财务文件。投资人文件确认公司已有收入生成,但没有绝对金额。快速融资与财务不透明之间的落差,是当前增长阶段中国深科技硬件创业公司的典型模式。更重要的是,在几乎没有披露收入的情况下,据报 RMB 10 billion 的估值引出一个问题:市场是否纯粹在按技术潜力和政策顺风定价,而不是按商业基本面定价?多位覆盖该赛道的分析师已提示这种泡沫风险。[CO026, CO027, CO028, CO029, CO030, CO035]
| 指标 | 数值 | 日期 | 置信度 | 缺口 |
|---|---|---|---|---|
| 投后估值 | ~USD 1.39B | 2026-03 | 中 | 精确数字未披露 |
| 融资总额 | >RMB 1.5B (~USD 210M) | 2026-03 | 中 | 根据各轮融资公告汇总 |
| 员工人数 | >100 | 2026-Q1 | 中 | 精确人数未披露 |
| 产能 | ~1,000 units/month(目标) | 2026-H1 | 低 | 爬坡时间线不清楚 |
| 收入 | 未披露 | 低 | 已确认有收入;绝对金额未披露 | |
| 客户数 | 500+ 份预订(仅 Bumi) | 2025-Q4 | 中 | 全型号合计未知 |
| 专利 | >30 项核心专利 | 2026-Q1 | 中 | 具体申请细节未公开 |
| 成立日期 | September 2023 | 2023-09 | 高 |
收入和精确估值仍属私有信息;估算来自融资公告和投资人披露。产能是公司给出的目标,不是已确认的实际吞吐量。
[CO026, CO027, CO028, CO029, CO030]概括 Noetix Robotics 成熟度和牵引力的关键绩效指标。
[CO007, CO035, CO036]1.5 图表要点
02市场分析
2.1 市场边界与定义
人形机器人市场覆盖为人类环境交互而设计的全身双足机器人,不同于工业机械臂、自动驾驶汽车和纯软件 AI。Noetix 具体竞争的是消费与教育人形机器人细分市场:价格低于 $20,000,面向家庭陪伴、STEM 教育、商业导览和轻服务任务。相邻市场包括工业协作机器人(cobots)、四足机器人,以及没有双足运动能力的社交 / 陪伴机器人。现状替代品包括教育平板、智能家居助手(Alexa、Google Home)、玩具机器人(没有高级运动能力)和真人家教 / 看护。市场边界排除工业制造机器人、手术机器人和自动无人机。中国人形机器人市场的特点是国家深度参与——十五五规划(2026-2030)明确把机器人和具身智能列为战略性国家技术,一方面通过政府采购要求拉动需求,另一方面通过补贴研发和制造推动供给。纳入的支出包括价格从 $1,000 到 $100,000、具备双足运动能力的机器人硬件,以及配套软件平台、维护合同和内容生态。排除的支出包括纯软件 AI 助手、非双足工业机器人和医疗机器人系统。[CM001, CM002, CM003, CM004]
| 细分市场 | 覆盖支出 | 排除支出 | 买方 / 付款方 | 对 Noetix 的重要性 |
|---|---|---|---|---|
| 消费级人形机器人 | 家庭陪伴、教育、娱乐机器人,价格低于 $20K | 工业机械臂、手术机器人、自动驾驶车辆 | 家庭、家长 | 核心(Bumi) |
| 教育 / 研究人形机器人 | 高校实验室、STEM 项目、机器人竞赛平台 | 纯软件仿真、非双足平台 | 机构、科研经费 | 核心(N2、E1) |
| 商业服务人形机器人 | 接待、导览、品牌活动、礼宾 | 后台物流、制造 | 企业、营销预算 | 次要(E1、Hobbs) |
| 仿生交互平台 | 用于 HRI 研究和治疗的情感表达头部 | 全身工业机器人、远程医疗 | 医疗健康、娱乐 | 新兴(Hobbs 系列) |
市场细分按最终用途和价格带定义。Noetix 的多产品策略覆盖消费到商业服务,但不包含重工业。
[CM001, CM002, CM003]2.2 TAM/SAM/SOM 与多视角测算
多家可信分析机构给出了人形机器人机会的边界。RBC Capital Markets 预计,到 2050 年,全球人形机器人总可寻址市场将达到 $9 trillion,覆盖制造、物流、医疗、家庭和国防应用。近端来看,Morgan Stanley 预计中国人形机器人市场 2026 年约为 $2 billion,出货约 50,000 台;因商业部署得到验证,这一估计较此前上调 79%。到 2030 年,仅中国市场预计就将达到 $15 billion,年出货 446,000 台,隐含 65%+ CAGR。Noetix 主攻的消费 / 教育人形机器人可服务可寻址市场(SAM),是中国人形机器人总市场的一个子集,估计占 2026 年总市场的 15-20%,约 $300-400 million;随着成本曲线跌破 $10,000 消费门槛,到 2030 年有望增长到 $2-3 billion。Noetix 的可获得服务市场(SOM)受当前约 1,000 台 / 月(12,000 台 / 年)产能约束,按 $1,400-$10,000 的平均售价计算,近期最大年收入能力约为 $17-120 million。[CM005, CM006, CM007, CM008, CM009, CM010]
| 发布方 | 年份 | 地域 | 规模 | CAGR | 方法 | 置信度 | 局限 |
|---|---|---|---|---|---|---|---|
| RBC Capital Markets | 2026 | 全球 | 2050 年达 $9T | ~25% 隐含 | 自上而下劳动力替代 | 中 | 周期很长;不确定性高 |
| Morgan Stanley | 2026 | 中国 | $2B (2026) | 65%+ (2026-2030) | 自下而上出货跟踪 | 高 | 仅近端;不含软件 |
| Morgan Stanley | 2026 | 中国 | $15B (2030) | 65%+ (2026-2030) | 产能 + 部署要求 | 中 | 假设政策延续 |
| ChinaBizInsider 估算 | 2026 | 中国消费 / 教育子集 | $300-400M (2026) | ~70% | 中国人形机器人 TAM 总量的 15-20% | 低 | 作者估算;无直接来源 |
| Noetix SOM 估算 | 2026 | Noetix 可触达市场 | $17-120M 收入产能 | N/A | 1,000 units/month × ASP 区间 | 低 | 基于产能;需求未经验证 |
不同时间跨度和地域范围下,TAM 估算差异极大。Morgan Stanley 的中国近端数据最有行动价值。SOM 受产能约束。
[CM005, CM006, CM007, CM008, CM009]面向 Noetix 的人形机器人总可用市场(TAM)/ 可服务市场(SAM)/ 可获取市场(SOM)分层。
SAM 和 SOM 是作者估算,来自 Morgan Stanley 对中国总市场的测算。长期 TAM 具备推测性。
[CM005, CM009, CM010]2026 年中国人形机器人市场规模的低 / 基准 / 高估算。
区间来自 Morgan Stanley(基准)和 RBC(高位)预测。低位估算来自更保守的升级前预测。
[CM006, CM007, CM008]2.3 买方、用户与付款方分层
对 Noetix 的产品组合而言,人形机器人买方可以分成四类。第一类是教育机构(学校、STEM 项目、机器人竞赛),购买 N2 和 E1 作为教学平台;预算归学校管理层或教育技术采购部门,课程现代化要求触发采用。第二类是消费家庭,购买 Bumi 作为陪伴和编程教育设备;父母付款,儿童是主要用户,触发因素是 STEM 教育需求以及低于 $1,500 价位带来的新奇价值。第三类是商业企业(酒店、商场、活动场馆),购买 E1 和 Hobbs 型号用于接待、导览和品牌激活;营销或设施管理预算出资。第四类是研究机构和大学,采购 N2 和 E1 作为开发平台;资金来自研究经费和院系设备预算。消费产品购买流程通过电商(JD.com)完成,摩擦较低;机构销售则需要直销触达、演示和 1-3 个月采购周期。[CM011, CM012, CM013, CM014, CM015]
| 细分 | 买方 | 用户 | 付款方 | 流程 | 预算负责人 | 采纳触发因素 |
|---|---|---|---|---|---|---|
| 消费 / 家庭 | 家长 | 儿童、家庭 | 家庭收入 | 电商购买(JD.com) | 家长可支配预算 | STEM 教育需求、新奇感、价格突破 |
| 教育 / STEM | 学校、高校 | 学生、研究人员 | 机构预算 | 采购周期(1-3 个月) | EdTech 或院系预算 | 课程现代化、竞赛备战 |
| 商业服务 | 酒店、商场、活动 | 访客、宾客 | 企业资本开支 | 直销、演示、试点 | 市场 / 设施 | 品牌差异化、降低人力成本 |
| 研究 / 开发 | 研究实验室 | 研究人员、工程师 | 科研经费 | 经费资助采购 | PI 或院系负责人 | 平台能力、开放 API 访问 |
消费者(冲动购买 / 低摩擦)和机构(多月采购)的买方旅程差异很大。Noetix 目前在消费和教育场景最强。
[CM011, CM012, CM013, CM014]Noetix 目标细分市场中的买方、用户、付款方关系。
数值代表每个细分市场—因素组合的相对可达性评分(0-1),根据采购复杂度估算。
[CM011, CM013, CM015]Noetix 消费级人形机器人的购买和部署步骤。
漏斗数值为数量级估算。认知基于 CCTV 收视人群;购买量由预订数据外推估算。
[CM015, CM018]2.4 增长驱动与采用约束
增长由五股力量推动。第一,十五五规划下的政府政策要求省级部署目标,并资助商业试点。第二,成本大幅下降——Noetix 的 Bumi 定价 $1,370,相比上一代人形机器人降价 90%+,跨过消费可负担门槛。第三,央视春晚时刻制造了大众文化认知,也让人形机器人在中国家庭里变得正常化。第四,OpenHarmony 生态集成把人形机器人定位为 IoT 节点,而不是孤立设备。第五,中国更广泛的制造供应链支持快速扩产。约束包括:AI 软件缺口限制真实世界自主性;首批国家标准在 2026 年 3 月推出后,安全和监管仍不确定;家用联网机器人存在网络安全顾虑;过度投资带来泡沫风险(140+ 家公司,潜在产能过剩);以及消费者从新奇兴趣到日常效用之间仍有采用缺口。关键约束是自主性缺口——当前机器人擅长预设动作,但还无法泛化到非结构化家庭环境。[CM016, CM017, CM018, CM019, CM020, CM021]
| 驱动 / 约束 | 方向 | 时间 | 影响 | 尽调问题 |
|---|---|---|---|---|
| 十五五规划机器人要求 | 驱动 | 2026-2030 | 政府采购托住确定性需求底部 | 核实省级部署目标细节 |
| 低于 $1,500 的消费级价格点 | 驱动 | 2025 年至今 | 跨过消费者可负担门槛,打开大众市场 | 跟踪实际售出 vs 预订 |
| CCTV 春晚品牌认知 | 驱动 | 2026-Q1 事件 | 让中国家庭把人形机器人视为常态 | 衡量春晚后搜索量和订单转化 |
| AI 自主性差距 | 约束 | 2026-2028 | 机器人受限于预编程任务,日常效用降低 | 评估自主能力研发路线图 |
| 安全 / 监管不确定性 | 约束 | 2026 年起持续 | 2026 年 3 月出台首个国家标准;要求仍在演进 | 跟踪标准定稿和合规成本 |
| 市场泡沫风险 | 约束 | 2026-2028 | 竞争者超过 140 家;若需求跟不上产能,可能产能过剩 | 跟踪行业整合和倒闭率 |
| 网络安全顾虑 | 约束 | 2026 年起持续 | 联网家庭机器人抬高隐私和安全风险 | 审查数据治理和安全架构 |
驱动因素见效快且有政策背书;约束偏中期,可能在 2028 年前触发整合。
[CM016, CM017, CM018, CM019, CM020, CM021]2.5 图表要点
03竞争对手
3.1 竞争格局与关键玩家
2026 年,中国人形机器人市场已有 140 多家活跃公司,但在消费和教育细分市场,Noetix 的主要竞争对手集中在四家:Unitree Robotics、Galbot(Beijing Galbot)、MagicLab 和 UBTech。Unitree 是最直接的竞争对手,其 R1 型号直接瞄准与 Noetix Bumi 相同的低于 $20,000 消费级市场。Unitree 依托四足机器人业务(Spot 对标产品)积累更久,并在 2026 年央视春晚上展示了令人印象深刻的武术编排。Galbot 聚焦交互助手能力和包括物体操作在内的实用服务演示。MagicLab 以 MagicBot 系列的表现性手势和杂技级精度突出。UBTech 是同业中唯一公开上市公司(HKEX: 9880),具备 Noetix 缺少的财务透明度。国际上,竞争集合包括 Figure AI(估值 $39B)、Agility Robotics(聚焦物流)、Tesla Optimus(商业化前)和 Boston Dynamics(Atlas / Spot),但这些公司主要瞄准工业场景,而非消费用例。2025-2026 年,中国市场出货了全球 90% 的人形机器人,因此国内竞争比国际竞争更紧迫。快速资本流入进一步推高竞争强度——仅 2026 年上半年,15 家中国具身 AI 独角兽估值就超过 $1.4 billion。每个竞争者都有不同技术路线:Unitree 强在工业能力,Noetix 强在交互和可负担性,MagicLab 强在编舞式表达,Galbot 强在实用操作。竞争动态仍在快速演变,新进入者几乎每月出现。[CP001, CP002, CP003, CP004, CP005, CP006]
| 公司 | 总部 | 成立时间 | 估值 / 阶段 | 主要产品 | 主要细分市场 | 差异点 |
|---|---|---|---|---|---|---|
| Noetix Robotics | 北京 | 2023 | ~$1.39B(Series B 轮) | Bumi、N2、E1、Hobbs | 消费 / 教育 | 最低价格 + 仿生头部 |
| Unitree Robotics | 杭州 | 2016 | >$1B(Series C+ 轮) | R1、G1、H2、Go2 | 消费 / 工业 | 工业级 + 价格战 |
| Galbot | 北京 | 2023 | ~$1.5-2B | Galbot 系列 | 家庭 / 服务 | 物体操作灵巧度 |
| MagicLab | 深圳 | 2023 | ~$1-2B | MagicBot Gen1、Z1 | 娱乐 / 活动 | 编舞精度 |
| UBTech(HKEX:9880) | 深圳 | 2012 | 上市公司(市值约 $3B) | Walker X、Alpha | 企业 / 教育 | 唯一上市;成熟度最高 |
| Figure AI(美国) | 圣何塞 | 2022 | ~$39B | Figure 02 | 工业 / 物流 | 估值最高;尚未产生收入 |
估值来自媒体报道估算,未必反映当前市场状况。按 Series B 轮报道,Noetix 估值接近 RMB 10B。
[CP001, CP002, CP003, CP004, CP005, CP006]3.2 功能与能力对比
Noetix 在三条轴线上差异化:价格、产品线宽度和仿生情绪交互。定价 $1,370 的 Bumi 是目前商业可得的最便宜全双足人形机器人,比 Unitree R1 低约 30-40%。但 Unitree R1 载荷能力更强,工业级耐久性更好。Noetix 的双轨策略(双足移动 + 32 DOF 面部表情仿生头部)很独特——没有竞争对手同时提供两者。Galbot 在操作灵巧度上领先(实时物体折叠演示),MagicLab 在编舞精度上领先。UBTech 拥有最成熟的企业部署记录。所有竞争对手都共享 OpenHarmony 或 ROS2 兼容性,因此软件生态并不是差异点。Noetix 的关键差异点包括:消费级低于 $1,500 定价的先发者;央视独家合作伙伴身份(未与竞争者共享);同业没有的仿生头部平台;以及最快的研发迭代周期(43 天造出第一台机器人,2.5 年完成 9 轮融资)。相对同业的弱点包括:员工数小于 Unitree 和 UBTech;不像 UBTech 那样公开披露财务;经营记录更短;单型号产量更低。中国如今出货全球 90% 的人形机器人,把整场价格战和创新竞赛集中在国内市场,竞争强度因此进一步升高。[CP009, CP010, CP011, CP012, CP013, CP014]
| 能力 | Noetix | Unitree | Galbot | MagicLab | UBTech |
|---|---|---|---|---|---|
| 低于 $1,500 的消费级机器人 | 是(Bumi) | R1(约 $2,000) | 否 | 否 | 否 |
| 双足行走 | 是(全系型号) | 是 | 是 | 是 | 是 |
| 仿生面部表情(32 DOF) | 是(Hobbs) | 否 | 否 | 否 | 否 |
| 物体操作 | 基础 | 基础 | 高阶 | 基础 | 高阶 |
| 特技 / 体操 | 后空翻(N2) | 武术 | 高精度旋转 | 特技动作 | 有限 |
| OpenHarmony / ROS2 支持 | 是 | 是 | 是 | 部分支持 | 是 |
| 量产(>100 台 / 月) | 是 | 是 | Unknown | Unknown | 是 |
| 公开财务披露 | 否 | 否 | 否 | 否 | 是(HKEX) |
特性评估基于公开规格和媒体演示;能力声称未经独立验证。
[CP009, CP010, CP011, CP012]| 公司 | 入门型号 | 价格(USD) | 目标用户 | 关键规格 |
|---|---|---|---|---|
| Noetix | Bumi | $1,370 | 消费者 / STEM | 94cm、12kg、21 DOF |
| Noetix | N2 | $5,500 | 科研 / 活动 | 118cm、30kg、18 自由度、3.5m/s |
| Unitree | R1 | ~$2,000 | 消费者 | 估算值;规格与 Bumi 相近 |
| Unitree | G1 | ~$16,000 | 企业 / 科研 | 全尺寸,工业级 |
| UBTech | Alpha 系列 | $5,000+ | 教育 | 成熟教育平台 |
| Figure AI | Figure 02 | 尚未定价 | 工业 | 全尺寸,替代劳动力 |
Unitree R1 定价来自行业来源估算。Figure AI 尚未公布商业定价。
[CP009, CP013, CP014]按价格可及性和能力广度定位人形机器人公司。
X 轴:价格可及性(1=最可及)。Y 轴:能力成熟度(1=最成熟)。位置根据产品规格和市场定位估算。
[CP009, CP015]3.3 护城河耐久度与竞争风险
Noetix 的竞争护城河立在四根支柱上:(1)靠垂直整合和供应链控制取得成本领先,支撑低于 $1,500 的价位;(2)央视春晚合作带来的品牌认知,提供难以匹敌的消费者知名度;(3)仿生头部平台具备 32 DOF 面部表情能力,是独特技术差异点;(4)北京机器人基金背书带来的政策契合。但四根支柱都存在被侵蚀的风险。Unitree R1 的价格战,以及拥有更大生产规模、可能做到成本平价的竞争者,正在压迫成本领先。品牌护城河有时效性——如果没有持续营销投入,春晚效应会消退。仿生头部平台今天有差异,但底层执行器技术开放,未来可能被复制。政策契合也不是独享——竞争者同样通过各类省级基金获得国资支持。最耐久的护城河元素是双足 + 仿生的一体化产品架构,它同时需要运动能力和面部表情专长,建设周期以年计。最不耐久的是价格,Unitree 和新进入者正在快速逼近。整合风险很高:140+ 家公司参与,预计 2027-2028 年将出现显著行业出清;资金市场收紧前,能否实现可持续单位经济模型决定生存。[CP016, CP017, CP018, CP019, CP020, CP021]
| 护城河要素 | 当前强度 | 侵蚀风险 | 可持续窗口 | 主要威胁 |
|---|---|---|---|---|
| 成本领先(低于 $1,500) | 高 | 高 | 12-18 个月 | Unitree R1 价格战;规模经济 |
| 品牌(央视春晚) | 高 | 中 | 6-12 个月 | 缺乏持续投放,认知度会退潮 |
| 仿生头部技术(32 DOF) | 中 | 低 | 2-3 年 | 复制需要专项技术积累 |
| 政策对齐(北京基金) | 中 | 中 | 规划期(2026-2030) | 多家竞争对手同样享有 |
| 双足 + 仿生一体化架构 | 中 | 低 | 2-4 年 | 需要双重专长;复制难度高 |
可持续性评估来自竞争格局下的主观估算。成本领先最不耐用,价格战已经打起来。
[CP016, CP017, CP018, CP019, CP020]关键人形机器人维度上的相对能力评分。
1-10 分评分来自公开规格和演示估算,未做独立基准测试。
[CP008, CP021, CP030]Noetix Robotics 的关键竞争定位指标。
[CP016, CP018, CP021]3.4 图表要点
04财务
4.1 收入模型、定价与渠道结构
从财务画像看,Noetix 更像一家试图用激进定价压缩采用曲线的年轻硬件公司,而不是靠经常性席位变现的软件公司。公开资料显示,公司有四类可变现产品:大众市场端的 Bumi,票价逐步更高的研究和商用产品 N2、E1,以及作为专业交互模块的 Hobbs 级仿生头部。这个宽度很重要,因为它创造了多条平均售价通道;但它也让收入确认和产品组合分析更难,公开资料没有拆分已签订单、已交付单位或渠道终端售出。最强的可见需求证明仍是 Bumi 在 JD.com 的预订单以及随后与 JD 的战略合作,它们意味着真实的漏斗顶部兴趣。即便如此,直销、合作伙伴渠道和国际扩张都很可能伴随折扣、演示费用和售后服务成本,而这些没有披露。因此,财务上的收入故事是正面的,但仍处早期:Noetix 有变现证据和分销杠杆,却没有足够公开披露来证明收入质量已经成熟。[CI001, CI002, CI003, CI004, CI005, CI006]
| 来源 | 证据 | 主要买方 | 商业状态 | 财务含义 |
|---|---|---|---|---|
| Bumi 硬件销售 | 公开消费级定价和京东预售 | 家庭、学校 | 商业化可见 | 低 ASP 能扩大漏斗,但会挤压利润率 |
| N2 硬件销售 | RBTX 与 Aparobot 上架信息,加上官方产品页 | 研究实验室、教育机构、演示运营方 | 商业化可见 | 更高 ASP 提高单台收入密度 |
| E1 硬件销售 | 官方 E1 页面和第三方价格参考 | 教育、服务、轻商业 | 商业化可见 | 中端产品可能抬高综合 ASP |
| Hobbs 仿生模块 | 公司官网的仿生定位 | 商业服务、陪伴、交互场景 | 商业化可见,但透明度低 | 可能把收入拓展到整机机器人之外 |
收入来源证据基于产品和渠道,因为 Noetix 未按产品系列或地域披露已确认收入。
[CI001, CI002, CI003, CI004, CI005]| 产品 | 公开价位 | 销售渠道 | 变现方式 | 读者提示 |
|---|---|---|---|---|
| Bumi | USD 1,370 / RMB 9,998 | 零售与渠道合作伙伴 | 一次性硬件销售 | 大众市场入口支撑销量叙事 |
| N2 | ~USD 5,500 | 科研、教育、经销商 | 一次性硬件销售 | ASP 更高,但买方池更小 |
| E1 | USD 5,570-10,000 | 机构与商业客户 | 一次性硬件销售 | 区间暗示定制化和套餐差异 |
| Hobbs 产品线 | 未披露 | 项目制或垂直解决方案销售 | 模块 / 专用系统销售 | 未公开 ASP 或附加率披露 |
主力 SKU 定价公开,但折扣、捆绑、维护和软件附加均未披露。
[CI002, CI003, CI004, CI005, CI006, CI007]Noetix 的可见变现栈从消费级硬件,延伸到机构采购和渠道支撑的收入层。
这些数值用于说明变现层级和公开 ASP 锚点,不代表实际收入结构,也不代表各条收入流的审计口径贡献。
[CI006, CI007, CI008, CI011]4.2 单位经济模型、成本驱动与资本强度
公开记录足以估出经济形态,却不足以承销贡献利润率。如果 Noetix 某天用 Bumi 单一产品填满 1,000 台 / 月的产能目标,年收入能力仍只是美元口径的中等八位数;更高端的产品组合可以大幅抬高这一数字。这令人鼓舞,但也暴露出 Noetix 在独角兽估值看起来符合常规收入倍数之前,需要爬到多大的规模。成本结构很可能由执行器、传感器、电池、板载计算和服务义务主导。N2 文档强调快换电池、边缘 AI 硬件、深度感知和动态运动能力;这些都会把 BOM 和现场支持复杂度推得高于简单消费电子。RBTX 的交付周期信号显示,公司更像按订单履约,而不是拥有充足成品库存;这在运营上合理,但可能拉长现金转换周期。由此得到的是典型机器人悖论:更高技术雄心提升定价权,但在规模制造追上来之前,每一次性能升级都会增加资金需求和执行风险。[CI009, CI010, CI011, CI017, CI018, CI019]
| 视角 | Bumi | N2 | E1 | 含义 |
|---|---|---|---|---|
| ASP 可见度 | 高 | 高 | 中 | 有公开定价,但实际 ASP 未知 |
| 可能的 BOM 压力 | 中 | 高 | 高 | 性能抬升时,传感器、算力、执行器和电池会推高成本 |
| 服务负担 | 中 | 高 | 高 | 能力越强,机器人可能越需要现场支持和培训 |
| 利润率判断把握 | 低 | 低 | 低 | 未按 SKU 披露毛利率或保修准备金 |
本表比较单位经济性的形态,不是已披露毛利率。单元格由公开规格和支持义务推断。
[CI020, CI021, CI022, CI023, CI024, CI025]| 项目 | 公开证据 | 状态 | 重要性 |
|---|---|---|---|
| 近期融资 | 2026 年 3 月 Series B 轮接近 RMB 1B | 强 | 缓解短期流动性压力 |
| 累计融资 | >USD 210M,九轮融资 | 强 | 说明投资人多次愿意为扩张买单 |
| 工厂爬坡 | 三城合计目标 1,000 台 / 月 | 要求高 | 意味着资本开支和营运资金需求仍会持续 |
| 现金披露 | 未公开现金、烧钱速度或债务细节 | 可见度弱 | 无法判断真实资金续航 |
相较创业公司同业,资本充足性看起来扎实,但月度烧钱和债务可见度仍然缺失。
[CI013, CI014, CI016, CI018, CI019, CI030]公开证据显示,Noetix 机器人从标价走到真实贡献毛利,中间会吃掉多项成本。
指数值只是方向性示意,依据公开产品复杂度和渠道结构推导,不是公司披露的 COGS 占比。
[CI023, CI024, CI025, CI026]Noetix 落在高投入、现金回报尚不透明的象限,这是快速扩张机器人公司的典型位置。
X 轴代表资本强度,Y 轴代表当前公开现金回报可见度;位置为定性估算。
[CI018, CI019, CI030, CI031]4.3 牵引信号、资金充足度与披露缺口
按创业公司标准,Noetix 资金充足。不到三年 9 轮融资、近 RMB 1 billion 的 Series B,以及政策关联资本反复投资,共同降低了眼前的流动性焦虑。但同一组证据也暴露了公开尽调的边界:没有披露现金余额、烧钱速度、债务时间表,也没有把订单桥接到收入确认的说明。投资人能看到融资成功,却看不到财务质量。最乐观的解释是,战略资本正在为广泛人形机器人采用前的时间敏感型产能圈地提供资金。更怀疑的解释是,民间资本正在掩盖一个负毛利抢份额竞赛,而这个赛道的竞争者还在更快降价和扩供。两种解释可以同时成立。因此,短期运营的资金充足度看起来可以接受,但这只是因为外部投资人持续开票。没有审计财报,甚至没有一致 KPI 披露,公开证据仍让核心承销问题悬而未决。[CI013, CI014, CI015, CI016, CI027, CI028]
| 缺失指标 | 已公开信息 | 数据缺失为何重要 | 最佳尽调路径 |
|---|---|---|---|
| 已确认收入 | 有收入,但未披露 | 无法测算估值倍数或增长质量 | 要求提供按 SKU 和渠道拆分的月度收入桥 |
| 毛利率 / 贡献利润率 | 未公开利润率披露 | 无法验证低价策略在经济性上是否成立 | 要求提供 BOM、保修准备金和服务附加分析 |
| 现金消耗和资金续航 | 未公开现金流量表 | 无法评估下一轮融资时点风险 | 要求提供月度烧钱和现金余额历史 |
| 积压订单转化与流失 | 预订单和渠道协议可见 | 无法区分宣传热度与已付费交付 | 要求提供订单簿账龄及退款 / 取消数据 |
最大的投资判断缺口在运营财务指标,而不是公司融资时间线;后者的公开信息已经足够。
[CI012, CI031, CI032, CI033, CI035, CI036]公开可支撑的财务区间很宽:产能目标和估值已知,但收入兑现仍未知。
收入产能区间是基于公开 SKU 定价和月产 1,000 台目标推算的情景估计,不等于确认收入。
[CI011, CI015, CI034, CI036]4.4 财务结论与尽调含义
财务结论不是 Noetix 缺少商业牵引,而是公开数据还没有证明牵引能否转化为有吸引力的单位经济模型。公司显然拥有产品市场好奇心、融资渠道和足以影响赛道的制造雄心。但投资人若要用基本面支撑 RMB 10 billion 估值,仍需要硬数字;公开证据还停在这些数字之前:收入结构、各产品毛利率、服务附加率、保修准备金、渠道返利、净营运资本和月度烧钱。实践中,合适的尽调姿态是把 Noetix 视作一个重融资的硬件爬坡项目:如果制造和渠道执行守得住,上行空间不对称;如果价格战先于成本曲线改善到来,下行也很大。下一步尽调不应再要一份市场叙事演示材料,而是一份管理层数据包,包含月度出货、实际 ASP、按 SKU 计算的贡献利润率,以及按渠道拆分的订单到现金转换。在此之前,证据天平支持继续关注,而不是给出清洁的承销舒适度。[CI030, CI031, CI032, CI033, CI034, CI035]
05产品与技术
5.1 产品组合与客户工作流定义
Noetix 卖的不是一个只有单一买方故事的人形机器人。公开产品面覆盖面向消费者、适合教育场景的本体机器人,如 Bumi;能力更高的运动平台,如 N2 和 E1;以及仿生交互系统,如 Hobbs 3 和 Hobbs W1。这个组合设计很重要,因为产品是按工作流来定义的:有些单元设计用来移动、表演并进行物理互动;有些则设计用来表达情绪、导览并支撑社交存在感。来自公司、RBTX 和 Aparobot 的公开清单都强化了这一点:Noetix 瞄准研究、教育、展示和服务场景,而不是狭窄工业利基。这个宽度在战略上有用,因为它扩大了客户可以运行的部署实验集合;但它也意味着每个模块可能处在不同成熟度。尽调视角下,正确的问题不是「Noetix 有没有产品?」而是「哪些产品面已经足够标准化,可以支撑可重复部署和支持?」[CE001, CE002, CE003, CE004, CE005, CE006]
| 资产 | 形态 | 核心能力 | 主要用户 | 商业角色 |
|---|---|---|---|---|
| Bumi | 紧凑型人形机器人 | 低价交互与 STEM 场景实用性 | 家庭 / 学校 | 入门走量 SKU |
| N2 | 运动型人形机器人 | 动态运动与开发者实验 | 实验室 / 教育 / 演示 | 能力展示与中端平台 |
| E1 | 通用人形机器人 | 更高规格的通用性能 | 机构 / 商业试点 | 中高端平台 |
| Hobbs 3 | 仿生头部 / 人形模块 | 高表现力类人交互 | 服务 / 陪伴 / 体验设计 | 差异化仿生产品 |
| Hobbs W1 | 仿生导览平台 | 导览与互动在场感 | 公共场所 / 接待 | 垂直场景延展 |
矩阵采用 Noetix 及经销商页面的公开定位表述,而非内部产品需求文档。
[CE001, CE002, CE003, CE004, CE005, CE006]| 用例工作流 | 产品锚点 | 运营方 / 买方 | 关键支撑功能 | 成熟度读数 |
|---|---|---|---|---|
| STEM 课堂与实验室实验 | Bumi / N2 | 学校、大学 | 低门槛定价加 API | 商业化可见 |
| 研究与机器人开发 | N2 | 实验室、开发者 | ROS / Linux / Python 支持 | 商业化可见 |
| 服务导览与接待 | E1 / W1 | 公共场所、商业试点 | 交互加具身在场感 | 试点 / 早期部署 |
| 陪伴与情感交互 | Hobbs 系列 | 家庭、疗愈场景、体验设计师 | 重表达能力的仿生头部 | 差异化明确,但规模不透明 |
用例覆盖并不完整,因为 Noetix 没有按工作流发布完整客户部署清单。
[CE006, CE021, CE033]产品体验从产品发现延伸到集成、部署和售后支持。
运行流程基于公开产品、文档和支持界面推断,并非公司发布的客户旅程图。
[CE014, CE017, CE018, CE021]5.2 架构、软件栈与开发者赋能
Noetix 最有意思的技术信号,是硬件定义与明确生态赋能的组合。N2 文档展示了一台具备实质板载栈的机器人:电池、传感器、边缘计算,以及通过 Python、C++、ROS 和 Linux 暴露的运动控制接口。ChinaBizInsider 还补充了一个操作系统说法:N2 使用 OpenHarmony,这让 Noetix 有可能进入更广的中国设备和开发者生态。重要的是,公司没有停在产品手册。它发布了官方开源页面、开发工具界面、产品 SDK 界面和文档中心。这些页面不能证明外部采用已经很广,但显示管理层理解机器人平台的采用瓶颈:如果机器人要超越舞台演示,客户、实验室和集成商需要 API、SDK、文档和支持材料。因此,架构故事强于许多早期机器人创业公司,尽管公司披露的仍更多是意图,而不是已量化的平台使用。[CE007, CE008, CE009, CE010, CE011, CE012]
| 层级 | 公开证据 | 具名组件 | 含义 |
|---|---|---|---|
| 操作系统 | OpenHarmony 报道与 Noetix 新闻稿 | OpenHarmony / Huawei 生态 | 支撑更广泛的设备互操作叙事 |
| 边缘计算 | 经销商与评测描述 | Jetson 级 AI 计算 | 支撑机载感知与控制 |
| 运动控制 | 官方运动定位与多次空翻演示 | 强化学习运动、执行器 | 区分动态运动能力 |
| 开发者层 | 开源、SDK、ROS / Linux API | SDK、Python、C++、ROS | 降低集成商摩擦 |
| 支持物料 | 文档中心与服务页面 | 文档、售后、隐私界面 | 显示演示之外的平台化投入 |
架构层级由公开产品和生态材料归纳而来,不是内部工程图。
[CE007, CE009, CE010, CE012, CE013, CE015]Noetix 的产品栈把硬件、中间件、开发者工具和支持界面连在一起,而不是一具孤立的机器人本体。
图示强调公开可见的架构层,不覆盖所有内部子系统或协议。
[CE010, CE012, CE013, CE015]Noetix 的产品价值依托外部计算、OS、文档和渠道支持层。
依赖项为便于阅读做了分组;详细供应商级映射未公开。
[CE019, CE022, CE032]5.3 信任、质量、安全与合规姿态
信任和质量披露落后于产品叙事。公开材料显示,公司已有隐私、服务支持和售后界面;这比纯实验型创业公司更好,但公司没有发布现场可靠性数据、安全认证状态或量化事故历史。这一点很关键,因为人形机器人是网络物理系统:故障模式不只包括软件 bug 或传感器定价错误,也包括跌倒事件、电池问题、隐私投诉或不安全集成。外部资料也强调,这已经是董事会层面的议题。步行机器人安全标准越来越聚焦部署环境和特定应用认证,而 Gartner、Lockton 和 Autonomy Global 都强调,AI 设备会扩大身份、隐私和责任敞口。Noetix 已有信任界面的雏形,但家庭、学校或公共服务机器人面临的尽调门槛远高于简单 Web 产品。控制存在证据;成熟控制表现还没有证据。[CE019, CE020, CE023, CE024, CE025, CE026]
| 维度 | 已公开内容 | 缺失内容 | 尽调含义 |
|---|---|---|---|
| 安全框架 | 外部标准评论已存在 | Noetix 认证状态未公开 | 需要按部署场景验证安全性 |
| 网络安全 | 通用 AI / 机器人风险指引充足 | 未披露产品级控制措施 | 未证实前,先按网络物理风险处理 |
| 隐私 | 官网已有隐私界面 | 未按工作流披露数据治理细节 | 家庭和教育用例需要更深入审查 |
| 可靠性 | 规格和演示已公开 | 没有 MTBF、正常运行时间或事故指标 | 仅凭公开数据无法给成熟度打分 |
这是披露快照,并不等于断言 Noetix 完全没有这些控制措施。
[CE024, CE025, CE026, CE027, CE028, CE029]当前公开记录中,能力广度领先于信任披露。
评分是基于公开证据密度推导的 0-1 定性成熟度估算,不是内部评分卡。
[CE024, CE029, CE031, CE036]5.4 成熟度地图与技术结论
整体产品技术结论是:能力定义和生态意图偏正面,运营成熟度仍需谨慎。Noetix 的公开证据足以证明它有真实 SKU,而不是占位符;外部清单也足以说明至少部分产品在封闭实验室之外分销;开发者信号界面也足以显示公司认真尝试搭平台,而不是只做一台孤立机器。剩余尽调缺口,正是把吸睛机器人和可投资产品系统区分开的东西:可重复可靠性、可认证部署,以及可衡量的第三方使用。凭借融资和渠道雄心,公司未来很可能跨过这道缺口,但公开记录尚未证明。现阶段,Noetix 应被理解为在产品表达上领先、在生态信号上领先,但相对规模化部署买方最终会要求的信任和质量指标,披露仍落后。[CE031, CE032, CE033, CE034, CE035, CE036]
| 阶段标记 | 证据 | 产品重点 | 读数 |
|---|---|---|---|
| 2025 发布潮 | 官方产品线与早期发布可见度 | Bumi / Hobbs / N2 家族 | 产品组合定义成型 |
| 2026 CES 曝光 | FutureTekNow 报道 | N2 | 从概念转向量产级品牌表达 |
| 2026 OpenHarmony 重点 | ChinaBizInsider 与文档生态 | N2 软件层 | 平台集成上升为战略重点 |
| 2026 开源 / SDK 浮出 | 官方开发者页面 | 开发者生态 | 平台化投入可见 |
| 需要的下一里程碑 | 尚无公开可靠性指标 | 全部 SKU | 需要现场性能与认证深度证明 |
时间线来自公开发布和生态信号,而非内部路线图材料。
[CE012, CE013, CE016, CE020, CE031, CE036]06客户
6.1 客户分层与采用证明质量
以当前阶段的创业公司标准,Noetix 的客户证据异常偏渠道。公司显然瞄准家庭、教育买方、机器人实验室、公共服务场所和分销伙伴,但最强公开证明并不来自一长串具名终端客户,而是来自 JD.com 预订单、JD 战略合作,以及 RBTX 等分销商清单。这些证据仍有价值,因为它证明有人愿意营销、上架,并且至少在一个案例中为产品预订单位。它也显示 Noetix 已经把产品形态匹配到不同客户原型:Bumi 用于低摩擦入门,N2 面向研究和演示买方,仿生系统用于导览和交互场景。公开证据尚未证明的是深度账户渗透或已披露合同金额。因此,客户故事是真实的,但仍早到需要投资人区分可见需求信号与成熟客户基础披露。[CU001, CU002, CU003, CU004, CU005, CU010]
| 客群 | 典型买方 | 主要产品 | 采购动作 | 当前证明强度 |
|---|---|---|---|---|
| 家庭 / 家长 | 个人消费者 | Bumi | 零售 / 预订 | 最强的可见消费者证明 |
| 学校 / 大学 | 教育机构 | Bumi / N2 | 采购或伙伴销售 | 定位和上架信息提供中等强度证据 |
| 研究实验室 / 开发者 | 技术团队 | N2 | 经销商 / 规格驱动销售 | RBTX 提供中等强度证据 |
| 商业场所 | 零售、导览、活动 | E1 / W1 / Hobbs | 试点或伙伴主导销售 | 概念上可见,具名证明薄 |
| 国际渠道伙伴 | 经销商 / 集成商 | N2 及更广产品线 | 伙伴市场 / 代理 | RBTX 与生态页面提供中等强度证据 |
客群地图来自公开渠道和产品证据,因为 Noetix 没有按垂直行业或地区发布客户普查。
[CU001, CU002, CU003, CU010, CU011, CU012]| 具名实体 | 角色 | 证明类型 | 证明了什么 | 未能证明什么 |
|---|---|---|---|---|
| JD.com | 零售与战略渠道伙伴 | 公开合作 + 预订证据 | 真实零售需求与渠道投入 | 收入占比、复购或流失 |
| JoyInside(JD 平台) | 设备生态 / 分发平台 | 官方媒体披露 | 平台集成与品类可见度 | 按品牌统计的终端销量 |
| RBTX | 国际经销商 / 市场伙伴 | 伙伴与产品页面 | 海外技术买家触达 | 大规模签约机队需求 |
| JD MALL / 618 促销 | 线下 / 促销零售触点 | People's Daily 品类报道 | 更宽的商业化路径 | 持久客户留存 |
具名证明并不完整,因为公开记录强调渠道和平台,而不是已披露的企业客户名单。
[CU004, CU006, CU008, CU010, CU016, CU028]最强证据把具名实体和可观察客户动作结合起来,但企业客户持续性证据仍然稀疏。
矩阵值代表 0-1 的证据密度,而不是商业规模。
[CU004, CU008, CU023, CU029]6.2 采用轨迹与客户旅程
公开资料呈现的采用曲线,从产品可见度开始,并加速进入渠道化。Bumi 在 JD 上 500 多个预订单,是最清晰的硬消费者数据点。随后,Noetix 通过更广的 JD.com 关系深化叙事;这段关系如今覆盖产品设计、用户运营和渠道扩张,而 People's Daily 则把 Noetix 放进 JD 的 JoyInside 设备生态和更广的人形机器人零售推进之中。RBTX 增加了第二层证据,显示 Noetix 不只面向中国消费者,也可以被包装销售给海外技术买方。合在一起,这些信号意味着一条客户旅程:从认知和好奇,到上架、预订、集成、部署和售后支持。仍不透明的是,这条漏斗有多频繁地转化为持久收入或复购。公开记录支持真实的漏斗顶部动能,但转化质量和分群耐久度仍需要直接向管理层尽调。[CU005, CU006, CU007, CU008, CU009, CU010]
| 日期 | 公开信号 | 客户含义 | 证据质量 |
|---|---|---|---|
| 2025-10 | JD.com 上 500+ 个 Bumi 预订单 | 首个实打实的消费者需求数据点 | 直接公开证明 |
| 2026-Q1 | 融资后强调家庭和教育用例 | 重点客群变得明确 | 公司 / 媒体证明 |
| 2026-Q2 | 签署 JD 战略合作 | 渠道关系深化,不止于上架 | 伙伴证明 |
| 2026-Q2 | People's Daily 披露 JoyInside 集成 | 平台分发拓宽 | 官方第三方证明 |
| 2026-Q2 | RBTX 伙伴与产品页面上线 | 国际技术买家触达可见 | 经销商证明 |
轨迹跟踪外部可见客户信号,而不是内部出货或留存指标。
[CU005, CU006, CU007, CU008, CU010, CU015]可见客户旅程从认知和预订开始,经合作伙伴支持的部署,走到支持环节。
旅程根据公开零售、合作伙伴和支持界面推断,不来自管理层漏斗看板。
[CU006, CU013, CU014, CU018]公开证据在漏斗顶部最强,在重复使用和扩张上最弱。
数值是定性证据强度指数,不是披露的转化率。
[CU015, CU020, CU021, CU027]6.3 留存、重复使用与集中度风险
公开披露在这里变薄。本章审阅的资料中,没有来源提供客户数量、活跃装机量、流失率、复购率、续约率,或具名账户内的扩张率。公司确实有维护、售后、法律和隐私界面,这表明管理层预期客户关系会持续,而不是一次性发货。但支持界面只能证明系统存在,不能证明系统运行良好。集中度也是同理。JD.com 在公开证据中占比很高,明显构成客户或渠道集中度风险;但公开材料没有量化按渠道或伙伴拆分的收入占比。这让本章核心出现一种张力:Noetix 因为有可见需求和支持脚手架,看起来强于纯原型故事;但它的留存和集中度指标仍属私有,因此又不如规模化硬件业务成熟。[CU017, CU018, CU019, CU023, CU024, CU029]
| 指标视角 | 公开状态 | 正向信号 | 缺口 |
|---|---|---|---|
| 复购 | 未披露 | 支持基础设施已存在 | 没有复购队列 |
| 续约 / 流失 | 未披露 | 持续渠道关系可见 | 没有合同或流失数据 |
| 满意度 | 未披露 | 售后与维护条款已存在 | 没有 NPS / CSAT / 投诉 KPI |
| 装机基础健康度 | 未披露 | 支持入口显示售后互动 | 没有活跃机队使用指标 |
本快照刻意区分两件事:有支持系统,不等于客户健康指标强。
[CU017, CU018, CU019, CU030, CU033]| 风险 | 当前判断 | 重要性 | 尽调索取项 |
|---|---|---|---|
| JD.com 集中度 | 集中度高且公开可见 | 具名证明集中在一个主导渠道 | 要求按渠道提供收入占比 |
| 企业客户证明缺口 | 重大 | 难以区分试点和规模化部署 | 要求提供前 20 大账户及合同状态 |
| 国际执行风险 | 中等 | 扩张由伙伴主导,而非公司直接披露 | 要求提供伙伴管线和服务覆盖 |
| 价格战观望 | 中等 | 客户可能等待更便宜或更好的机器人 | 要求按发布队列提供转化数据 |
尽管具体收入集中度未公开,集中风险的方向已经清楚。
[CU022, CU023, CU024, CU029, CU032, CU034]公开证据更能证明支持基础设施存在,而不是证明留存已被量化。
柱状值是 10 分制证据可得性评分,不是实际客户留存率。
[CU018, CU019, CU030, CU033]6.4 客户结论与尽调含义
因此,客户结论是建设性但不完整。公开证据足以说明 Noetix 已经走出纯假设阶段:消费者预订了 Bumi,JD.com 承诺更深的渠道关系,第三方市场也在为技术买方定位 N2。这个组合很重要,因为它同时显示低摩擦零售和高摩擦伙伴主导渠道的兴趣。与此同时,几乎所有会让故事变得耐久的机构指标——活跃客户、复购、分群留存、账户集中度、退款、支持负载,以及具名账户内扩张——仍未披露。下一步尽调不应问客户是否存在,而应要求一份分群数据包,按渠道对订单、已交付单位、取消、支持工单和重复活动进行对账。在这些信息可得之前,Noetix 应被视为一家拥有可信采用动能、但耐久性未知数重大的公司。缺失的分群数据层,是把客户基础称为耐久之前的主要门槛项。[CU023, CU025, CU026, CU027, CU028, CU034]
07风险
7.1 监管与法律风险栈
人形机器人面临的法律和监管图景,正从概念走向运营。外部标准评论如今把步行人形机器人视作一个特殊类别,具有跌倒、电池和特定应用部署风险;法律顾问也越来越把联网机器人定义为网络物理系统,其设计、软件更新、集成和现场行为都会影响责任。对 Noetix 来说,这很重要,因为公司明确瞄准家庭、学校、面向公众的服务场景,以及潜在企业空间;在这些环境里,隐私、物理安全和客户损害可能汇合。公司公开页面显示,Noetix 已开始建设隐私、维护和售后条款等治理界面。这比什么都没有好,但不同于发布认证状态、风险评估或事件响应指标。因此,当前风险姿态只有在投资人假设公司及其渠道伙伴内部仍有大量控制工作要做时,才算可管理。[CR001, CR002, CR003, CR004, CR005, CR006]
| 风险 | 可能性 | 影响 | 当前证据 | 缓释路径 |
|---|---|---|---|---|
| 应用层安全不合规 | 中 | 高 | 人形机器人标准正围绕部署场景收紧 | 获取认证文件和部署风险评估 |
| 隐私与同意机制失效 | 中 | 高 | 家庭和教育场景放大数据处理暴露 | 审查数据流、同意 UX 与留存政策 |
| 网络物理责任划分 | 中 | 高 | 法律评论把机器人事故看作设计、软件、运营共同参与的事件 | 厘清公司与伙伴之间的合同责任划分 |
| 跨境合规差异 | 中 | 中 | 海外扩张增加司法辖区复杂性 | 梳理目标国家安全、隐私和保修规则 |
| 叙事 / 披露错配 | 中 | 中 | 对外传播可能跑在经审计的运营证明前面 | 让对外主张对齐成文指标和控制措施 |
登记表优先列出未来 12-24 个月最可能改变投资判断或部署速度的法律监管风险。
[CR001, CR003, CR004, CR005, CR026, CR027]Noetix 最高严重度风险来自安全、执行和商业脆弱性的叠加,而不只是技术新颖性。
评分是基于公开证据密度和行业条件的 0-1 定性估算,不是内部风险评分。
[CR008, CR009, CR011, CR024, CR025]7.2 运营、产品质量与技术依赖风险
Noetix 正试图同时做到几件难事:硬件快速规模化、公众可见度和年轻品类里的技术雄心。即使还没有出现任何单一缺陷,这也会抬高运营风险。制造目标、多城市布局、电池、传感器、边缘计算和带 API 的产品栈,都比简单消费设备制造出更多故障点。交付周期信号和公开支持页面显示,生产、履约和售后能力必须同时跟上。除此之外,公司似乎依赖 OpenHarmony 等外部软件和生态层;它们可以加速采用,也会引入上游兼容性和安全风险。实践中,核心技术风险问题不是机器人能不能移动、能不能在舞台上吸睛——它们显然可以——而是当部署和用户多样性扩大时,质量体系、事件监测和服务物流是否成熟到足以顶住。公开证据还不能自信回答这个问题。[CR008, CR009, CR010, CR011, CR012, CR013]
| 风险 | 发生可能性 | 影响 | 已观察信号 | 缓解路径 |
|---|---|---|---|---|
| 爬坡期制造质量漂移 | 中 | 高 | 1,000 台目标和多城布局 | 审核 QA 体系和工厂流程控制 |
| 电池 / 传感器 / 计算故障 | 中 | 高 | N2 硬件栈复杂,涵盖移动供电和感知 | 要求提供现场故障历史和供应商验证 |
| 网络安全和机器身份缺口 | 中 | 高 | AI 智能体和联网设备攻击面在扩大 | 要求提供安全架构和事件处置预案 |
| 交付和售后瓶颈 | 中 | 中 | 交付周期和支持义务可见,但未量化 | 测试服务人员配置、备件流程和工单 SLA |
| 监控盲区 | 中 | 中 | 未公开遥测或事件 KPI | 要求提供机队可观测性和升级处理看板 |
运营风险聚焦最可能叠加放大安全、成本和声誉影响的故障模式。
[CR013, CR014, CR015, CR017, CR018, CR028]| 依赖项 | 风险 | 重要性 | 缓解路径 |
|---|---|---|---|
| OpenHarmony 生态 | 上游兼容性或安全问题 | OS 依赖会传导到产品可靠性和更新 | 审查抽象层和备选方案 |
| 渠道合作伙伴和代理商 | 执行不一致 | 合作伙伴质量影响客户体验和转化 | 审核伙伴赋能和合同标准 |
| 零售 / 平台集中 | 商业议价失衡 | 少数可见平台主导公开证据 | 分散渠道组合,并在内部披露集中度 |
| 共享供应商生态 | 冲击传导 | 供应网络密集,短缺或缺陷会快速扩散 | 梳理双来源覆盖和关键部件库存 |
逐个供应商的映射并未公开,因此依赖项按平台层和合作伙伴层分组。
[CR012, CR017, CR019, CR020]少数上游失灵就可能沿安全、融资和声誉结果级联放大。
图示展示行业和公司证据所暗示的因果逻辑,不是量化贝叶斯模型。
[CR027, CR028, CR034, CR035, CR037]Noetix 的风险姿态取决于上游软件、硬件、渠道和支持执行。
依赖分组保持高层级,因为供应商级合同和冗余安排未公开。
[CR014, CR017, CR019, CR020, CR029, CR039]7.3 人员、伙伴与融资风险
即便势头很强,Noetix 仍像一家组织深度可能落后于外部预期的公司。创始人集中度高,招聘活跃可见,公司又在同时扩产品线、渠道和地理范围。这会在管理带宽、支持质量和流程纪律上制造执行风险。伙伴可以吸收一部分负荷,但也会增加依赖:公开商业化记录高度依赖外部平台和分销商,而不是大量已披露的直接企业账户。财务上,硬件制造、库存、支持和伙伴赋能都会消耗现金,业务仍暴露在资本强度之下。公司已融资到足以降低近期生存焦虑,但没有发布烧钱、债务或杠杆数据,无法排除融资风险。这正是那种一旦价格战或质量失误出现,增长故事会迅速变成再融资故事的设置。[CR019, CR020, CR021, CR022, CR023, CR024]
| 风险 | 发生可能性 | 影响 | 证据 | 缓解路径 |
|---|---|---|---|---|
| 创始人集中度 | 中 | 高 | 技术领导权似乎集中在创始人身上 | 评估梯队厚度和授权归属 |
| 招聘执行压力 | 高 | 中 | 多个招聘入口显示公司需要主动扩张团队 | 审查招聘漏斗、留存和入职流程 |
| 员工增长中的流程漂移 | 中 | 高 | 100+ 名员工和多地规模可能跑在体系前面 | 检查组织设计和 QA 治理 |
| 叙事跑在证据前面 | 中 | 中 | 传播节奏强于已披露指标 | 用可追溯的客户和质量数据校准市场推进主张 |
人员风险是执行乘数:硬件公司扩张时,它们会放大或压低每一类风险。
[CR021, CR022, CR023, CR030]| 领域 | 当前缓解信号 | 缺失证据 | 放弃触发条件 |
|---|---|---|---|
| 安全治理 | 公司显示出标准意识,也公开了支持入口 | 认证材料包和部署审计 | 发生严重现场安全事件,且没有可信整改 |
| 渠道经济性 | 合作伙伴和代理网络可见 | 渠道毛利和转化数据 | 大型合作伙伴表现不佳且没有备用渠道 |
| 质量体系 | 支持和维护入口存在 | 现场故障和遥测指标 | 支持负担上升且没有遏制证据 |
| 融资韧性 | 近期大额融资 | 烧钱速度、债务和现金跑道可见性 | 产品质量稳定前就需要新资金 |
| 管理层厚度 | 活跃招聘可见 | 梯队厚度和负责人分工图 | 无法把产品、运营和支持领导职责下放 |
放弃标准以事件组合呈现,因为早期机器人市场里,单个数据点噪音太大。
[CR024, CR025, CR036, CR037, CR040]7.4 风险结论与否决标准
最终风险结论不是 Noetix 独特危险,而是公司正处在四件难事的交叉点:新监管、硬科技执行、渠道依赖和赛道级商业不稳定。单独看,任何一个风险都可能扛过去;叠加之后,它们要求的尽调深度远高于产品页面或融资头条能提供的内容。优先级最高的工作流是安全治理、现场质量监测、伙伴经济模型和管理层深度。如果在更广的人形机器人市场仍在降价时,出现重大安全事件、支持崩溃或渠道转化不达预期,融资缓冲可能比公开叙事暗示的缩得更快。因此,投资人应把控制,而不是演示,视作真正的门槛。令人鼓舞的信号是,Noetix 已经展示出一些治理脚手架和伙伴意图。缺失的信号,是这些脚手架能在规模压力下良好运转的证明。缺失的证据不是叙事信心,而是规模下可衡量的控制表现。[CR036, CR037, CR038, CR039, CR040]
08估值
8.1 投资论点与反论点
Noetix 故事里有吸引力的部分很容易看到。公司拥有真实产品面、可见早期需求、有意义的战略和政策契合投资人,并且暴露在中国扩张最快的工业叙事之一上。它不是纯概念。公开资料显示,Bumi 有消费者兴趣,JD.com 和 RBTX 带来分销牵引,运营叙事围绕可负担性、生态集成和量产展开。反论点同样清楚:这一切不会自动说明当前私有估值便宜。公开证据仍缺少已确认收入、毛利率、流失率和资本效率指标。市场产出正在爆发、价格压缩已经可见时,缺少这些指标,比多一条产品视频更重要。公司可以具有战略重要性,但在没有更深尽调前,仍可能难以按当前价格买入。可见牵引与不可见经济性之间的不对称,正是这里需要估值纪律的原因。[CV001, CV002, CV003, CV004, CV005, CV006]
| 字段 | 评估 | 原因 |
|---|---|---|
| 建议 | 继续研究 | 叙事强度真实存在,但基本面不透明 |
| 信心 | 中 | 关键运营指标仍未披露 |
| 风险评级 | 高 | 硬件爬坡、价格战和披露缺口叠加 |
| 估值立场 | 偏高 | 当前价格已经把执行成功计入了 |
| 最佳下一步 | 在 NDA 下索取 KPI 包 | 需要直接证明利润率和需求质量 |
摘要只反映公开证据,也刻意不假装能精确推断缺失的私有指标。
[CV026, CV027, CV028, CV037]| 视角 | 投资逻辑 | 反向逻辑 |
|---|---|---|
| 市场 | 中国人形机器人市场快速放大 | 高速增长也会推高产能过剩和竞争风险 |
| 产品 | Noetix 有真实 SKU 和生态信号 | 可靠性和留存指标仍缺失 |
| 客户 | JD.com 和经销商证明市场确实有需求兴趣 | 具名企业客户能否持续仍未证明 |
| 资本 | 战略和政策投资者降低近期融资风险 | 战略资本不等于单位经济模型透明 |
| 估值 | 如果执行足够强,估值可能变得合理 | 以当前披露基本面看已经偏贵 |
投资逻辑和反向逻辑刻意成对呈现,让每个正面信号都接受最贴近反面论点的压力测试。
[CV004, CV006, CV012, CV014, CV023, CV028]推荐逻辑从市场吸引力开始,一层层推到估值纪律。
金字塔各层是逻辑依赖,不是加权评分输出。
[CV004, CV012, CV014, CV026]8.2 情景框架与可比估值逻辑
合理的估值视角必须按情景思考,而不是押一个英雄式单一结果。牛市情景假设 Noetix 利用资金和渠道支持填满产能、证明产品质量,并保持足够定价权,在下一次融资事件前展示利润率路径。基准情景假设采用稳步但不均衡,战略价值有意义,公司继续依赖民间资本,同时运营指标仍部分不透明。熊市情景假设价格压缩、支持负担或质量失误在复购需求被证明前就到来。可比公司分析有用但不完美。Unitree 相关,因为它结合了出货规模和可见的定价进攻。公开市场或准公开同业可以帮助界定成熟度可能是什么样子,但整个人形机器人品类仍受披露不一致和叙事驱动估值基准拖累。因此,可比公司能帮助讨论,却不能解决讨论。换句话说,情景比点估计更重要,因为已披露证据基础仍然偏斜。[CV007, CV008, CV009, CV015, CV016, CV017]
| 情景 | 核心假设 | 估值含义 | 概率信号 |
|---|---|---|---|
| 乐观 | 产能吃满,利润率路径浮现,合作伙伴渠道转化 | 当前估值介于合理和有吸引力之间 | 需要 KPI 证明,且没有重大事故 |
| 基准 | 采用率上升,但透明度仍有限 | 当前估值介于偏高和合理之间 | 最符合当前公开证据 |
| 悲观 | 价格战、支持负担或质量问题拖慢变现 | 当前估值显得昂贵 | 如果下一轮融资早于 KPI 证明出现,概率更高 |
情景框架为定性判断,因为公开来源没有提供管理层指引或经审计财务模型。
[CV007, CV008, CV009, CV029, CV030]| 可比公司 | 重要性 | 公开信号 | 对 Noetix 的启示 |
|---|---|---|---|
| Unitree | 规模、价格进攻性和逐步浮现的公开市场可见度 | TrendForce 引述 60% 合并毛利率和 IPO 可见度 | 显示规模化人形机器人经济性可以很强,但也加剧竞争 |
| UBTech | 公开市场透明度标杆 | 中国人形机器人赛道至少有一个公开参照点 | 凸显 Noetix 目前直接披露很少 |
| Figure AI | 叙事丰富的全球可比公司 | 公开收入细节有限,但估值很高 | 显示板块情绪会跑在基本面前面 |
| Noetix | 平价消费级 + 平台化切入 | Series B 后私有市场估值 USD 1.39B | 需要比当前公开资料更强的 KPI 证明 |
可比集合用于示意而非穷尽,因为多数赛道同业仍是私营公司,披露也不均衡。
[CV018, CV019, CV031, CV032, CV033, CV034]Noetix 估值对收入兑现和毛利证据高度敏感,不只是被出货新闻牵动。
敏感性区间是作者根据公开产能和定价证据估算,不是管理层指引。
[CV016, CV017, CV018, CV028]Noetix 若能证明毛利守得住、需求能持续,回报会有吸引力;否则可能很差。
区间是根据公开叙事和执行风险搭出的情景启发,不是 DCF 或市场导出的估值标记。
[CV007, CV008, CV009, CV030, CV040]8.3 建议、信心与否决触发器
章节层面最站得住脚的建议是继续研究(research-more),不是因为 Noetix 看起来弱,而是因为估值已经计入了相当多成功。战略和政策资本降低了近期融资恐惧,平台故事也在单一硬件 SKU 之外增加了一些选择权。即便如此,信心最好也只保持在中等,因为公开数据集在收入质量和单位经济模型上仍太薄。如果渠道证明转化为耐久出货,利润率改善,公司仍可能长进这个估值。若赛道价格战加速,或支持和质量经济性差于预期,公司也可能令人失望。正确姿态是让信念保持有条件。这意味着要明确否决触发器:重大安全或支持失败,曝光转化为付费需求较弱,或有证据显示下一轮融资必须在运营 KPI 稳定前完成。[CV021, CV022, CV023, CV024, CV025, CV026]
| 触发条件 | 重要性 | 直接含义 |
|---|---|---|
| 扩张期出现安全或支持故障 | 打破合作伙伴和消费者信任 | 控制措施被证明前,暂缓形成投资信心 |
| 从声量到付费需求的转化弱 | 暴露叙事大于实质 | 下调估值判断 |
| 利润率证明前就需要下一轮融资 | 显示资本效率偏弱 | 假设本轮没有买到足够执行时间 |
| 渠道高度集中或合作伙伴严重不达预期 | 压缩市场进入选择 | 降低对增长叙事的信心 |
放弃触发条件被设计成实际监控测试,而不是抽象风险标签。
[CV030, CV035, CV036, CV040]最重要的可投资性 KPI,仍是公开资料没有直接披露的那些。
[CV001, CV002, CV013, CV037]8.4 最终尽调要求与估值结论
最终估值判断:Noetix 值得密切跟踪,但现阶段还不能按表面估值轻松承销。公开证据足以支撑关注——公司有真实产品,渠道动能也在跑,投资人质量强;但公开证据不足以支撑放松,因为最关键的价值驱动仍未验证:实际 ASP、毛利率、复购需求、退款行为、支持成本和伙伴经济性。它们不是次要细节,而是战略叙事通向真实私募回报的桥。下一轮尽调的最低资料包,应包括按 SKU 和渠道拆分的月度出货、实际毛利、取消订单、服务负担,以及按情景拆解的现金规划。若这些数字过硬,当前估值最终可能显得合理,而不是偏贵;若数字疲弱,估值已经留不出多少犯错空间。在私募标记可能跑得比审计证明更快的品类里,这种纪律尤其重要。[CV026, CV027, CV028, CV037, CV038, CV039]
免责声明
本报告仅基于截至 2026 年 7 月的公开信息,不构成投资建议。收入、盈利能力和详细财务指标尚未披露,也无法独立核实。中文来源可能包含在翻译中流失的细微含义。
证据索引
| 编号 | 陈述 | 可信度 | 来源 |
|---|---|---|---|
| CO001 | Noetix Robotics was founded in September 2023 by Jiang Zheyuan (Chairman/CTO) and Zhang Shipu (CEO). | 高 | SO009, SO011, SO016 |
| CO002 | Jiang Zheyuan left a PhD program at Tsinghua University to found the company, making him one of the youngest founders in China's humanoid robotics sector. | 中 | SO009, SO011 |
| CO003 | Zhang Shipu serves as Co-founder and CEO with an investment and international education background. | 中 | SO009, SO010 |
| CO004 | Jin Zhiyong serves as Director of Operations at Noetix Robotics. | 中 | SO010 |
| CO005 | The core technical team includes alumni from Tsinghua University, Chinese Academy of Sciences, Zhejiang University, and the University of Hong Kong. | 中 | SO016 |
| CO006 | The company's official registered name is 松延动力(北京)科技集团股份有限公司 (Songyan Power [Beijing] Technology Group Co., Ltd.). | 高 | SO009, SO017 |
| CO007 | Noetix Robotics employs over 100 people as of early 2026. | 中 | SO011, SO009 |
| CO008 | Noetix Robotics completed nine funding rounds between September 2023 and March 2026. | 高 | SO001, SO003, SO017 |
| CO009 | The Series B round of nearly RMB 1 billion (~USD 140-145M) closed in March 2026, led by Chen Dao Capital (CD Capital), an industrial investment platform under CATL. | 高 | SO001, SO002, SO003, SO017 |
| CO010 | Series B co-investors included Guoke Investment, Jingguosheng Fund, and Jiuhe Venture Capital. | 高 | SO001, SO002, SO017 |
| CO011 | The Pre-B round of approximately RMB 300 million closed in October 2025 led by Fangguang Capital. | 中 | SO016 |
| CO012 | The Pre-B+ round of approximately RMB 200 million closed in November 2025 led by CICC Capital. | 中 | SO016 |
| CO013 | The Beijing Robot Industry Development Investment Fund invested in Noetix Robotics in March 2024 and again in September 2025. | 高 | SO017, SO019 |
| CO014 | The Beijing Robot Fund is managed by a wholly-owned subsidiary of Shoucheng Holdings and has a target size of RMB 10 billion. | 高 | SO017, SO019 |
| CO015 | Total lifetime capital raised by Noetix Robotics exceeds RMB 1.5 billion (approximately USD 210 million). | 中 | SO001, SO016 |
| CO016 | The Bumi consumer humanoid robot is priced at RMB 9,998 (approximately USD 1,370), measures 94 cm tall, weighs 12 kg, and has 21 degrees of freedom. | 中 | SO012, SO013, SO014 |
| CO017 | The N2 Athlete humanoid robot measures 118 cm, weighs 30 kg, has 18 DOF, peak joint torque of 150 Nm, top speed 3.5 m/s, and costs approximately USD 5,500. | 中 | SO011, SO012 |
| CO018 | The E1 humanoid robot measures 136 cm, has 21+ DOF, walking speed of 1.2 m/s, and costs USD 5,570–10,000. | 中 | SO012 |
| CO019 | Noetix operates manufacturing facilities in Beijing, Changzhou, and Dongguan with a target capacity of approximately 1,000 units per month. | 中 | SO011, SO023 |
| CO020 | The company's first robot was built in just 43 days after founding. | 中 | SO016, SO001 |
| CO021 | In July 2025, Noetix Robotics delivered over 100 robots in a single month, validating mass production capability. | 中 | SO016 |
| CO022 | The N2 robot finished runner-up in the 2025 Beijing Humanoid Robot Half Marathon, completing 21 km in 3 hours 37 minutes. | 中 | SO016, SO011 |
| CO023 | Bumi surpassed 500 pre-orders on JD.com within days of its October 2025 launch. | 中 | SO024 |
| CO024 | On February 16, 2026, Noetix Robotics appeared at the CCTV Spring Festival Gala as Exclusive Partner for Humanoid Bionic Robots. | 高 | SO006, SO007, SO016 |
| CO025 | Four Noetix robots (Bumi, N2, E1, and a bionic unit) performed in the skit Grandma's Favorite at the Gala, performing side flips and backflips on a 12 m2 stage. | 中 | SO006, SO007, SO008 |
| CO026 | Noetix Robotics post-money valuation is reported near RMB 10 billion (approximately USD 1.39 billion) following the March 2026 Series B. | 中 | SO001, SO003, SO005 |
| CO027 | Noetix Robotics holds over 30 core patents as of early 2026. | 中 | SO016, SO001 |
| CO028 | Noetix's production capacity target is approximately 1,000 units per month across three manufacturing facilities. | 中 | SO011, SO023 |
| CO029 | Noetix Robotics has confirmed revenue generation but has not disclosed absolute revenue figures. | 中 | SO001, SO017 |
| CO030 | The company restructured into a joint-stock group in early 2026 to strengthen governance and prepare for large-scale growth. | 中 | SO009 |
| CO031 | The Hobbs series bionic head platform features 32 degrees of active freedom for micro-expression simulation and synchronized lip movements. | 中 | SO016 |
| CO032 | Noetix pursues a dual-track strategy combining bipedal humanoid robots for mobility tasks and bionic humanoids for emotionally resonant interaction. | 中 | SO001, SO016, SO021 |
| CO033 | The N2 robot uses Huawei's OpenHarmony OS as the world's first consumer-grade humanoid on this platform. | 中 | SO021 |
| CO034 | By mid-2026, Noetix plans to deliver its first batch of 1,000 overseas units targeting North America, Europe, and the Middle East. | 低 | SO011, SO001 |
| CO035 | Noetix Robotics is among 15+ Chinese embodied AI unicorns that surpassed USD 1.4 billion valuation in H1 2026. | 中 | SO022 |
| CO036 | China's 15th Five-Year Plan (2026-2030) elevates robotics and embodied intelligence to strategic national priorities. | 高 | SO018, SO019 |
| CM001 | The humanoid robotics market encompasses full-body bipedal robots for human-environment interaction, distinct from industrial arms and autonomous vehicles. | 高 | SM003, SM001 |
| CM002 | Noetix competes in the consumer and education humanoid segment with robots priced under $20,000 for home, STEM, and commercial service use. | 中 | SM014, SM015 |
| CM003 | Status-quo substitutes for consumer humanoid robots include educational tablets, smart home assistants, toy robots, and human tutors. | 中 | SM011 |
| CM004 | China's 15th Five-Year Plan (2026-2030) explicitly prioritizes robotics and embodied intelligence as strategic national technologies. | 高 | SM004, SM009 |
| CM005 | RBC Capital Markets projects a $9 trillion total addressable market for humanoid robotics globally by 2050. | 中 | SM003 |
| CM006 | Morgan Stanley projects China's humanoid robot market at approximately $2 billion in 2026 with approximately 50,000 units shipped. | 高 | SM001, SM002 |
| CM007 | Morgan Stanley's 2026 forecast represents a 79% upgrade from prior estimates driven by verified commercial deployments. | 高 | SM001, SM002 |
| CM008 | China's humanoid robot market is projected to reach $15 billion by 2030 with 446,000 annual unit shipments at 65%+ CAGR. | 中 | SM001, SM002 |
| CM009 | The consumer/education humanoid subsegment represents approximately 15-20% of total China humanoid market, or $300-400 million in 2026. | 低 | SM001, SM014 |
| CM010 | Noetix's maximum near-term revenue capacity is $17-120 million annually based on 1,000 units/month at ASPs of $1,400-$10,000. | 低 | SM014, SM015 |
| CM011 | Education institutions purchase N2 and E1 models as STEM teaching platforms with adoption triggered by curriculum modernization. | 中 | SM007, SM014 |
| CM012 | Consumer households purchase Bumi as a companionship and programming education device with parents as payers and children as users. | 中 | SM016, SM015 |
| CM013 | Commercial enterprises purchase E1 and Hobbs models for reception, guiding, and brand activation from marketing budgets. | 低 | SM014 |
| CM014 | The buying process for consumer products runs through e-commerce (JD.com) with low friction versus institutional multi-month procurement cycles. | 中 | SM016 |
| CM015 | Bumi surpassed 500 pre-orders on JD.com within days of launch, indicating strong early consumer demand signal. | 中 | SM016 |
| CM016 | The 15th Five-Year Plan requires provincial deployment targets for humanoid robots and funds commercial pilots. | 高 | SM004, SM009 |
| CM017 | Noetix's Bumi at $1,370 represents a 90%+ price reduction versus prior-generation humanoids, crossing the consumer affordability threshold. | 中 | SM015, SM014 |
| CM018 | The CCTV Spring Festival Gala appearance created mass cultural awareness of humanoid robots for Chinese households. | 中 | SM005, SM010 |
| CM019 | OpenHarmony OS integration positions humanoid robots as IoT ecosystem nodes rather than standalone devices. | 中 | SM007 |
| CM020 | Current humanoid robots excel at pre-programmed routines but cannot generalize to unstructured home environments. | 高 | SM011, SM010 |
| CM021 | Over 140 companies compete in China's humanoid robotics space, raising overcapacity concerns. | 中 | SM005, SM012 |
| CM022 | Chinese government has issued warnings about market bubbles in the humanoid robotics sector. | 中 | SM005, SM013 |
| CM023 | China introduced its first national safety standards for humanoid robots in March 2026. | 中 | SM008 |
| CM024 | China shipped more humanoid robots than the US in 2025-2026, establishing manufacturing and deployment leadership. | 高 | SM006, SM010 |
| CM025 | Fifteen Chinese embodied AI startups surpassed $1.4 billion valuation in H1 2026, indicating intense capital concentration. | 中 | SM012 |
| CM026 | AI² Robotics and X Square Robot each reached $2.9 billion valuations in 2026, setting the upper bound for Chinese humanoid peers. | 中 | SM017 |
| CM027 | Connected humanoid robots in homes raise cybersecurity and data privacy risks that are currently unregulated. | 中 | SM010, SM011 |
| CM028 | The humanoid robot price war between Unitree R1 and Noetix Bumi is driving entry-level prices below $2,000. | 中 | SM015 |
| CM029 | Long-term global humanoid robotics TAM estimates range from $1 trillion to $9 trillion by 2035-2050 depending on assumptions. | 中 | SM003, SM001 |
| CM030 | Many humanoid robot orders in China still come from state-owned enterprises and research labs, not true mass-market consumers. | 中 | SM005, SM011 |
| CM031 | China's humanoid robotics industry is projected to grow at over 65% CAGR between 2026 and 2030. | 中 | SM001, SM002 |
| CM032 | The Beijing Economic-Technological Development Area provides incentives for robotics commercialization and manufacturing. | 中 | SM009 |
| CM033 | Switching costs from educational tablets to humanoid robot platforms include teacher training, curriculum redesign, and hardware investment. | 低 | SM011 |
| CM034 | The bubble risk is amplified by the gap between production capacity scaling faster than verified end-user demand. | 中 | SM005, SM013 |
| CM035 | Morgan Stanley's forecast upgrade was driven by large-scale verified deployments in logistics and manufacturing sectors. | 中 | SM001, SM002 |
| CP001 | China's consumer humanoid robotics market features over 140 active companies as of 2026. | 中 | SP011, SP012 |
| CP002 | Unitree Robotics is the most direct competitor with its R1 model targeting the same sub-$20,000 consumer segment as Bumi. | 中 | SP002, SP001 |
| CP003 | Galbot focuses on interactive assistant capabilities and real-time object manipulation demonstrations. | 中 | SP001, SP006 |
| CP004 | MagicLab differentiates through expressive gestures and acrobatic precision with its MagicBot series. | 中 | SP001, SP006 |
| CP005 | UBTech is the only publicly listed humanoid robot company (HKEX: 9880) among Chinese peers. | 高 | SP007, SP008 |
| CP006 | Figure AI is valued at approximately $39 billion, making it the highest-valued humanoid robot company globally. | 中 | SP008, SP009 |
| CP007 | China shipped 90% of all humanoid robots globally in 2025-2026, making domestic competition more immediate than international. | 中 | SP007, SP017 |
| CP008 | AI2 Robotics and X Square Robot each reached $2.9 billion valuations in 2026 as the highest-valued Chinese peers. | 中 | SP009, SP008 |
| CP009 | Noetix's Bumi at $1,370 is the cheapest full-bipedal humanoid robot commercially available, undercutting Unitree R1 by 30-40%. | 中 | SP002, SP013 |
| CP010 | Noetix's dual-track strategy combining bipedal mobility plus bionic heads is unique—no competitor offers both capabilities. | 中 | SP013, SP014 |
| CP011 | The Hobbs series 32 DOF bionic facial expression is a capability no competitor currently offers. | 中 | SP015, SP014 |
| CP012 | Galbot leads in manipulation dexterity with real-time object folding demos; Unitree leads in industrial-grade durability. | 中 | SP001, SP002 |
| CP013 | The Unitree R1 at approximately $2,000 is the closest price competitor to Noetix Bumi, creating a consumer-segment price war. | 中 | SP002 |
| CP014 | UBTech Alpha series targets education market at $5,000+ with an established distribution network. | 中 | SP007, SP012 |
| CP015 | All major competitors share OpenHarmony or ROS2 compatibility, making software ecosystem a non-differentiator. | 中 | SP005, SP020 |
| CP016 | Noetix's cost leadership moat is built on vertical integration and supply chain control that enables sub-$1,500 pricing. | 中 | SP013, SP014 |
| CP017 | The CCTV Gala brand moat is time-limited and will fade without sustained marketing investment. | 中 | SP019, SP006 |
| CP018 | The bionic head platform requires specialized expertise in both actuators and AI expression that takes years to build. | 中 | SP015, SP021 |
| CP019 | Unitree's R1 price war is actively eroding Noetix cost leadership advantage in the consumer segment. | 中 | SP002 |
| CP020 | Policy alignment through Beijing Robot Fund is shared with multiple competitors who also receive state backing. | 中 | SP016, SP025 |
| CP021 | Noetix built its first robot in 43 days and completed 9 funding rounds in 2.5 years, demonstrating exceptional iteration speed. | 中 | SP015, SP014 |
| CP022 | Significant industry shakeout is expected by 2027-2028 with 140+ companies competing for limited verified demand. | 中 | SP011, SP010, SP012 |
| CP023 | Tesla Optimus remains pre-commercial and targets industrial labor substitution rather than consumer/education. | 中 | SP017, SP007 |
| CP024 | Robot makers paid millions of yuan for stage time at the CCTV Spring Festival Gala, creating a marketing arms race. | 中 | SP019 |
| CP025 | 15 Chinese embodied AI startups surpassed $1.4 billion valuation in H1 2026, indicating extreme capital competition. | 中 | SP008 |
| CP026 | Noetix's Exclusive Partner status at the CCTV Gala was not shared with Unitree, Galbot, or MagicLab. | 中 | SP006, SP015 |
| CP027 | Barriers to entry include actuator technology development, safety certification, manufacturing scale, and capital requirements. | 中 | SP020, SP012 |
| CP028 | China's first national humanoid robot safety standards introduced in March 2026 create compliance barriers for smaller entrants. | 中 | SP020 |
| CP029 | Customer switching costs between competing humanoid platforms are low due to standardized ROS2/OpenHarmony interfaces. | 低 | SP005, SP020 |
| CP030 | Noetix is the only company to combine CCTV exclusive partnership, sub-$1,500 pricing, and bionic head technology in one portfolio. | 中 | SP002, SP015, SP006 |
| CP031 | Chinese humanoid companies are positioned for global expansion against US, Japanese, Korean, and European competitors. | 中 | SP017, SP007 |
| CP032 | Mass production capability (>100 units/month) has been achieved by Noetix, Unitree, and UBTech but not confirmed for Galbot or MagicLab. | 中 | SP015, SP007 |
| CP033 | UBTech education market distribution network and curriculum partnerships represent a competitive advantage Noetix has not matched. | 低 | SP007, SP012 |
| CP034 | The humanoid robotics bubble risk is amplified by pre-revenue valuations exceeding $1B for multiple Chinese startups. | 中 | SP010, SP011 |
| CP035 | International companies like Figure AI command much higher valuations ($39B) despite lower shipment numbers than Chinese peers. | 中 | SP008, SP009 |
| CI001 | Noetix monetizes through direct hardware sales of Bumi, N2, E1, and Hobbs-class bionic products rather than through a disclosed recurring software subscription. | 中 | SI008, SI009 |
| CI002 | Bumi is positioned as a low-price consumer and education entry product priced around USD 1,370 or RMB 9,998. | 中 | SI021, SI022 |
| CI003 | N2 is positioned as a higher-ticket research and education humanoid priced around USD 5,500. | 中 | SI017, SI018, SI019 |
| CI004 | E1 is positioned as a mid-to-high-end general humanoid with public pricing references between roughly USD 5,570 and USD 10,000. | 中 | SI020, SI021 |
| CI005 | Hobbs-class bionic heads broaden monetization beyond full-body robots into emotion and interaction modules that can be sold into service and companionship use cases. | 中 | SI008, SI009 |
| CI006 | Noetix's direct-sales motion spans consumer retail, education procurement, and partner-led commercial deployments rather than a single enterprise sales channel. | 中 | SI009, SI012, SI023 |
| CI007 | JD.com is a distribution amplifier for Noetix because the partnership covers product design, user operations, and channel expansion across domestic and international storefronts. | 中 | SI012, SI013 |
| CI008 | Bumi recorded more than 500 JD.com pre-orders ahead of release, providing proof that low-ticket hardware can attract consumer demand before large enterprise adoption is visible. | 中 | SI011 |
| CI009 | If Noetix sold only Bumi at its announced consumer price, a 1,000-unit monthly run rate would imply roughly USD 16.4 million of annualized hardware revenue capacity before services. | 低 | SI021, SI022, SI008 |
| CI010 | If Noetix filled a 1,000-unit monthly run rate entirely with N2 units, annualized hardware revenue capacity would be roughly USD 66 million before accessories or service revenue. | 低 | SI017, SI018, SI008 |
| CI011 | A blended mix that includes Bumi, N2, and E1 implies a realistic 2026 annual revenue capacity band closer to USD 25 million to USD 70 million than to unicorn-scale software revenue. | 低 | SI017, SI020, SI021, SI008 |
| CI012 | Noetix has confirmed commercialization and revenue generation, but public sources do not disclose recognized revenue, backlog conversion, or gross margin. | 中 | SI006, SI007, SI008 |
| CI013 | Noetix closed a Series B round of nearly RMB 1 billion in March 2026 led by CATL-linked CD Capital. | 高 | SI005, SI006, SI007 |
| CI014 | Public reporting indicates Noetix has completed nine financing rounds since founding and has raised more than RMB 1.5 billion in aggregate. | 中 | SI005, SI006, SI007 |
| CI015 | The post-Series-B valuation is reported near RMB 10 billion, implying investors are underwriting scale before audited revenue is visible. | 中 | SI005, SI006 |
| CI016 | The Beijing Robot Fund's follow-on investment indicates outside capital has been used not only for growth but specifically for continued R&D intensity. | 中 | SI007 |
| CI017 | Noetix says it delivered more than 100 robots in a single month in July 2025, showing some throughput but not yet proving stable mass-production economics. | 中 | SI008 |
| CI018 | Noetix targets approximately 1,000 units of monthly capacity across Beijing, Changzhou, and Dongguan, which points to a capital-intensive manufacturing ramp. | 中 | SI008, SI010 |
| CI019 | The company is building a flexible production and supply-chain system that integrates ERP and MES workflows, suggesting cash is being deployed into factory operating infrastructure rather than pure research alone. | 中 | SI005 |
| CI020 | Noetix's low consumer price point is itself a gross-margin risk because it pushes the company into the same price-compression zone that competitors like Unitree are attacking. | 中 | SI002, SI004 |
| CI021 | TrendForce reports Unitree's 2025 prospectus showed humanoid revenue exceeding quadruped revenue and combined gross margin around 60 percent, demonstrating that scaled humanoid vendors can achieve attractive hardware margins. | 中 | SI001 |
| CI022 | The existence of a profitable peer benchmark does not imply Noetix enjoys similar economics because Noetix has not disclosed bill-of-materials cost, warranty reserves, or service burden. | 中 | SI001, SI008 |
| CI023 | RBTX lists N2 with a usual delivery time of six to eight weeks, implying a make-to-order or constrained-availability model rather than fully stocked consumer inventory. | 中 | SI017 |
| CI024 | N2 relies on quick-swap batteries, edge AI compute, and multi-sensor perception, all of which increase component cost versus toy or smart-speaker substitutes. | 中 | SI017, SI018 |
| CI025 | Noetix's service promise and after-sales terms imply future field-support, spare-parts, and repair liabilities that public sources do not quantify. | 中 | SI024, SI025 |
| CI026 | JD.com and channel partners can improve sales efficiency, but they can also absorb economics through discounts, promotion subsidies, and customer-support obligations. | 中 | SI012, SI013, SI023 |
| CI027 | China's humanoid output is forecast to grow 94 percent in 2026, so component shortages can coexist with industry overcapacity if end demand fails to absorb the extra supply. | 高 | SI001, SI016 |
| CI028 | Chinese vendors already control roughly 80 to 90 percent of global humanoid shipments, which supports export ambition but also means China-centric pricing will set the margin envelope for all local players. | 中 | SI003, SI004 |
| CI029 | The humanoid market remains concentrated in a handful of leaders, with Unitree and AgiBot projected to account for nearly 80 percent of shipments, leaving smaller vendors exposed to pricing pressure. | 高 | SI001, SI016 |
| CI030 | Noetix's headline funding reduces immediate survival risk, but absent disclosed revenue it remains dependent on external financing to bridge commercialization and scale-up. | 中 | SI005, SI006, SI008 |
| CI031 | Public sources do not disclose cash on hand, burn rate, debt, or project-finance obligations, so external observers cannot calculate runway with confidence. | 中 | SI005, SI006, SI007 |
| CI032 | Bumi demand proof and JD channel support make the top of the funnel visible, but no public source discloses conversion from pre-orders into delivered and paid units. | 中 | SI011, SI012 |
| CI033 | Noetix's financial reporting quality is still private-startup grade because there is no audited income statement, cash-flow statement, or disclosed segment mix in public evidence. | 中 | SI005, SI006, SI007 |
| CI034 | The strongest public support for the current valuation comes from industrial investors and policy-aligned funds rather than from transparent revenue multiples. | 中 | SI006, SI007, SI015 |
| CI035 | The main underwriting blocker is not whether robots can be sold at all, but whether Noetix can scale profitable fulfillment before the price war compresses contribution margin. | 中 | SI002, SI011, SI018 |
| CI036 | Until management discloses gross margin, warranty cost, and channel rebates, the fair diligence stance is to treat Noetix as a well-funded but financially opaque hardware ramp. | 中 | SI024, SI025, SI005 |
| CE001 | Noetix markets a multi-SKU product stack spanning Bumi, N2, E1, Hobbs 3, and Hobbs W1 rather than a single flagship robot. | 高 | SE001, SE003, SE004, SE005, SE014 |
| CE002 | The N2 Athlete page positions the robot around lightweight athletic performance and repeated dynamic movements. | 中 | SE001, SE002 |
| CE003 | The E1 page positions the robot as a higher-spec but cost-conscious general humanoid. | 中 | SE003 |
| CE004 | Hobbs 3 is described as a bionic humanoid platform with 32 active degrees of freedom concentrated in the head and expression system. | 中 | SE004 |
| CE005 | Hobbs W1 is positioned around guidance and interaction use cases rather than athletic locomotion. | 中 | SE005 |
| CE006 | Noetix's product architecture is intentionally split between locomotion-first humanoids and emotion-first bionic systems. | 中 | SE004, SE005, SE014 |
| CE007 | RBTX lists the N2 at 118 cm, 30 kg, and 18 degrees of freedom. | 中 | SE012, SE013 |
| CE008 | RBTX says N2 uses a 48V quick-swap lithium battery with up to two hours of runtime. | 中 | SE012 |
| CE009 | RBTX and Aparobot both describe N2 as using edge AI compute plus 3D or depth-vision sensors for dynamic interaction. | 中 | SE012, SE013 |
| CE010 | RBTX says N2 supports Python, C++, ROS, and Linux APIs, indicating developer-facing integration rather than a sealed toy workflow. | 中 | SE012 |
| CE011 | Aparobot describes N2 as using NVIDIA Jetson edge compute for real-time AI processing. | 中 | SE013 |
| CE012 | ChinaBizInsider reports that N2 uses OpenHarmony, positioning it inside Huawei's device ecosystem rather than as a fully isolated stack. | 中 | SE011 |
| CE013 | Noetix maintains an official open-source page, a development-tools page, and a product-SDK page, which together show deliberate ecosystem enablement. | 中 | SE006, SE007, SE008 |
| CE014 | The open-source pages imply Noetix wants third parties to build on its robots instead of consuming them only as fixed-function demos. | 中 | SE006, SE007 |
| CE015 | OpenHarmony's documentation center and Huawei's developer overview show that the underlying OS has a formal technical-docs layer and active ecosystem governance. | 高 | SE009, SE010, SE023 |
| CE016 | Huawei Central reported an OpenHarmony 7.0 beta launch before HDC 2026, indicating the OS layer is still advancing rapidly. | 中 | SE024, SE009 |
| CE017 | Noetix's ecological cooperation page explicitly solicits global agents and partners, showing that channel and integration support are part of the product strategy. | 中 | SE025 |
| CE018 | The document center and service-support structure indicate Noetix has begun formalizing product documentation and downstream support assets. | 中 | SE015, SE014 |
| CE019 | FutureTekNow says Noetix is building a flexible production and supply-chain system integrating ERP and MES, linking productization directly to manufacturing software maturity. | 中 | SE018 |
| CE020 | FutureTekNow also says Noetix showcased a production-grade N2 at CES 2026, which is stronger product-maturity evidence than a concept render but weaker than disclosed field reliability metrics. | 中 | SE018 |
| CE021 | RBTX positions N2 for research, education, and development use cases, while its partner page extends the narrative toward manufacturing, logistics, healthcare, and customer service. | 中 | SE012, SE017 |
| CE022 | The E1 page emphasizes domestically sourced core components, suggesting cost control and supply localization are product design goals. | 中 | SE003 |
| CE023 | Noetix claims more than 30 core patents, so part of the product moat rests on internal technical IP rather than only channel speed. | 中 | SE014, SE018 |
| CE024 | The public product stack demonstrates impressive locomotion and expression features, but it does not publish MTBF, failure rates, or installed-base uptime. | 中 | SE001, SE004, SE015 |
| CE025 | There's A Robot For That argues that 2025/26 humanoid safety certification increasingly applies to the deployment and application, not only the hardware shell. | 中 | SE020 |
| CE026 | Autonomy Global says connected robots turn safety, cybersecurity, and contractual allocation of responsibility into a single cyber-physical governance problem. | 中 | SE021 |
| CE027 | Gartner says agentic AI and machine identities are expanding attack surfaces, which matters for robots that blend autonomy, networking, and multimodal interfaces. | 中 | SE019 |
| CE028 | Lockton identifies privacy, consent, ransomware, and AI-related disclosure risk as emerging technology exposures, which are relevant if Noetix moves deeper into home and education settings. | 中 | SE022 |
| CE029 | Noetix's public site includes privacy-policy, service-support, and after-sales surfaces, indicating basic governance scaffolding even if control maturity is not fully disclosed. | 中 | SE015, SE025 |
| CE030 | The N2 control model is better described as a controlled platform with APIs than as fully open robotics infrastructure. | 中 | SE012, SE006 |
| CE031 | Noetix's strongest product maturity evidence is in demos, specs, and distribution listings rather than in published production SLA or safety-certification metrics. | 中 | SE012, SE016, SE018 |
| CE032 | Critical technical dependencies include OpenHarmony, Jetson-class edge compute, reinforcement-learning motion control, sensors, and after-sales integration partners. | 中 | SE010, SE013, SE018, SE025 |
| CE033 | The dual-track bipedal-plus-bionic architecture differentiates Noetix from single-lane humanoid peers that only optimize locomotion or only optimize social presence. | 中 | SE004, SE005, SE014 |
| CE034 | Open-source and SDK signaling improve developer credibility, but the public materials do not yet quantify external developer adoption or installed integrations. | 中 | SE006, SE008, SE023 |
| CE035 | The product story is therefore technically credible but still pre-disclosure on field reliability, certification depth, and third-party ecosystem usage. | 低 | SE020, SE021, SE023 |
| CE036 | Noetix appears furthest along in customer-visible hardware definition and ecosystem signaling, while trust and compliance disclosure lag the maturity of the demos. | 中 | SE019, SE020, SE022 |
| CU001 | Noetix serves at least four visible customer segments: households, schools and labs, commercial venues, and partner-led international buyers. | 中 | SU014, SU015, SU017, SU018 |
| CU002 | Bumi is aimed at consumer and education buyers rather than heavy-industry purchasers. | 中 | SU022, SU023 |
| CU003 | N2 is positioned toward research, education, and interactive demonstration use cases. | 中 | SU014, SU013 |
| CU004 | JD.com is the clearest named commercial partner in Noetix's customer evidence set. | 高 | SU002, SU003 |
| CU005 | Bumi accumulated more than 500 pre-orders on JD.com before release, showing verified early retail demand. | 中 | SU001 |
| CU006 | The JD.com partnership spans product design, user operations, and channel expansion rather than a one-off marketing cameo. | 中 | SU002 |
| CU007 | Gasgoo says the JD.com partnership targets more than one million users across the Noetix lineup over three years. | 中 | SU002 |
| CU008 | People's Daily says JD's JoyInside platform already includes Noetix among integrated robot brands. | 高 | SU002, SU003 |
| CU009 | People's Daily also says JD Retail wants robot brands on the platform to achieve more than RMB 10 billion of cumulative 2026 sales, implying a large retail ambition for the category. | 中 | SU003 |
| CU010 | RBTX lists Noetix as a partner and markets N2 through its component marketplace, which is evidence of international channel validation. | 中 | SU013, SU014 |
| CU011 | RBTX positions Noetix for manufacturing, logistics, healthcare, and customer-service scenarios, broadening the buyer map beyond hobby users. | 中 | SU013 |
| CU012 | FutureTekNow says the new funding is aimed at household and educational scenarios, confirming those segments are a current commercialization priority. | 中 | SU015 |
| CU013 | The official ecological cooperation page shows Noetix recruiting agents and partners, which implies indirect channel expansion rather than direct-sales-only growth. | 中 | SU018 |
| CU014 | The company website and media pages show persistent outward-facing market communication, which supports awareness-building but does not prove deep deployment penetration. | 中 | SU008, SU017, SU025 |
| CU015 | Noetix's public adoption trajectory moves from product launch and brand exposure in 2025 to retail, partner, and platform expansion in 2026. | 中 | SU001, SU002, SU003, SU025 |
| CU016 | The strongest named-customer-style proof is channel and platform evidence, not a disclosed enterprise roster with contract values. | 中 | SU002, SU003, SU013 |
| CU017 | Public sources do not disclose active customer count, repeat-purchase rate, churn, or renewal rate. | 中 | SU015, SU017 |
| CU018 | Noetix publishes after-sales, maintenance, and legal pages, indicating the company expects a post-sale support relationship with users. | 中 | SU010, SU011, SU012, SU024 |
| CU019 | The existence of support surfaces is not the same as evidence of customer satisfaction because no public NPS, CSAT, or complaint-resolution KPI is disclosed. | 中 | SU010, SU012, SU024 |
| CU020 | Pre-order evidence proves top-of-funnel curiosity, but it does not disclose how many reservations converted into delivered and paid robots. | 中 | SU001, SU021 |
| CU021 | China's humanoid market is scaling fast, so distribution partnerships can accelerate adoption if Noetix has enough supply and support capacity to keep up. | 中 | SU019, SU020 |
| CU022 | Rapid price compression across Chinese humanoids can also increase customer hesitation if buyers expect better robots to become cheaper soon. | 中 | SU020, SU021 |
| CU023 | The customer concentration risk is currently highest around JD.com because it is the most visible named retail and channel relationship in the public record. | 中 | SU002, SU003 |
| CU024 | International expansion risk is tied to partner execution because public overseas proof is channel-led rather than direct end-customer disclosure. | 中 | SU013, SU015, SU018 |
| CU025 | Schools, labs, and community or education environments appear easier early buyers than long-cycle enterprise automation accounts. | 中 | SU014, SU015, SU022 |
| CU026 | Noetix's product mix lets it pursue both low-friction consumer/education sales and higher-friction commercial deployments at the same time. | 中 | SU017, SU018, SU022 |
| CU027 | The company has enough visible demand proof to show adoption is real, but not enough to prove durable retention or recurring fleet expansion. | 中 | SU001, SU002, SU017 |
| CU028 | Media, supplier, and event pages imply a broadening market ecosystem around Noetix, but they still sit one layer away from direct customer reference calls. | 中 | SU008, SU009, SU025 |
| CU029 | No public source names top customers by revenue share, so concentration cannot be quantified even though JD.com clearly dominates the evidence set. | 中 | SU002, SU003 |
| CU030 | Repair and after-sales terms suggest Noetix expects ongoing service interaction, which can support retention but also exposes it to support-quality failure. | 中 | SU012, SU024 |
| CU031 | The retail pathway is more visible than the enterprise pathway in Noetix's public customer evidence. | 中 | SU001, SU003, SU022 |
| CU032 | Because customer proof is concentrated in retail and partners, public evidence for land-and-expand inside named accounts remains weak. | 中 | SU002, SU013, SU021 |
| CU033 | The best public customer-health interpretation is early adoption momentum with thin retention disclosure. | 中 | SU001, SU018, SU024 |
| CU034 | The hardest customer diligence ask is a cohort table linking orders, deliveries, returns, support tickets, and repeat purchases by channel. | 中 | SU001, SU002, SU024 |
| CU035 | Noetix should therefore be treated as having credible adoption proof, incomplete named-customer disclosure, and material concentration and durability unknowns. | 中 | SU003, SU017, SU021 |
| CR001 | Humanoid-robot safety in 2026 is increasingly governed by application-level standards rather than by hardware-only certification. | 中 | SR005 |
| CR002 | Walking robots create distinct fall-zone and battery-management risks that older collaborative-arm frameworks did not fully contemplate. | 中 | SR005 |
| CR003 | Legal commentary from M.L. Taikins and Autonomy Global treats connected robots as combined safety, cyber, and contractual-risk systems. | 中 | SR001, SR002 |
| CR004 | Gartner says agentic AI and machine identities are creating new cybersecurity attack surfaces in 2026. | 中 | SR004 |
| CR005 | Lockton identifies privacy, consent, ransomware, and AI-disclosure risk as material technology exposures. | 中 | SR003 |
| CR006 | No public Noetix source discloses a completed third-party safety certification package for its humanoid products. | 中 | SR015, SR016, SR031 |
| CR007 | No public source discloses a Noetix product recall, but the absence of disclosed incidents is not proof that operational incidents cannot occur. | 中 | SR016, SR024 |
| CR008 | China's policy environment is supportive of humanoid robotics, which lowers market-access risk but can also accelerate competitive crowding. | 高 | SR006, SR007 |
| CR009 | TrendForce projects 94 percent output growth in China humanoids in 2026, creating real overcapacity risk if adoption lags supply. | 中 | SR008 |
| CR010 | TrendForce also expects Unitree and AgiBot to control nearly 80 percent of shipments, which raises concentration risk for smaller vendors. | 中 | SR008 |
| CR011 | Humanoid APAC and AI Business both point to price compression in 2026, making margin erosion a major model risk for Noetix. | 中 | SR010, SR011 |
| CR012 | Robotics Center describes China's supply-chain density as a strategic advantage, but dense dependence on shared suppliers can also spread shocks rapidly across vendors. | 中 | SR009 |
| CR013 | Noetix's push toward 1,000 units per month and multi-city manufacturing raises quality-control and execution risk during ramp. | 中 | SR012, SR031 |
| CR014 | RBTX and Aparobot show N2 depends on batteries, sensors, edge compute, and API layers, any of which can create failure or support bottlenecks. | 中 | SR013, SR014 |
| CR015 | The N2 lead-time signal of six to eight weeks indicates potential fulfillment risk if demand spikes faster than production or service capacity. | 中 | SR013 |
| CR016 | Noetix's privacy, legal, maintenance, and after-sales pages show governance scaffolding, but they do not quantify control effectiveness. | 中 | SR015, SR016, SR017, SR018 |
| CR017 | OpenHarmony improves ecosystem reach but creates platform-dependence risk if security, compatibility, or governance issues emerge upstream. | 中 | SR026, SR027 |
| CR018 | Noetix's official open-source posture can improve adoption, but it also enlarges surface area for versioning, security, and integration support risk. | 中 | SR028, SR026 |
| CR019 | The ecological cooperation model reduces direct-distribution burden but increases partner-management and quality-consistency risk. | 中 | SR029, SR030 |
| CR020 | JD-style channel dependence creates partner concentration risk because a few external platforms dominate public commercialization evidence. | 中 | SR029, SR030 |
| CR021 | The most visible people risk is founder concentration around Jiang Zheyuan's technical leadership in a rapidly scaling organization. | 中 | SR031, SR020 |
| CR022 | The existence of HR, social recruitment, and campus recruitment pages implies talent acquisition is an active execution priority and therefore a live scaling risk. | 中 | SR020, SR021, SR022 |
| CR023 | A fast-growing robotics company with 100-plus staff faces organizational and quality drift risk if training, process, and support scale more slowly than headcount. | 中 | SR031, SR021 |
| CR024 | Capital intensity is a financial risk because robotics factories, support networks, and inventory require ongoing cash even after a large fundraising round. | 中 | SR012, SR025 |
| CR025 | No public evidence discloses cash burn, debt, or runway, so financing risk cannot be bounded precisely. | 中 | SR012, SR025 |
| CR026 | Privacy risk matters especially if Noetix expands deeper into homes, education, and public-facing deployments that collect audio and visual data. | 中 | SR003, SR015 |
| CR027 | Cybersecurity incidents could become disclosure and reputational events as connected robots move into enterprise networks and consumer spaces. | 中 | SR002, SR004, SR003 |
| CR028 | Warranty and maintenance obligations can become a margin and reputation risk if product quality fails at scale. | 中 | SR017, SR018 |
| CR029 | Support accessibility is partially mitigated by the public contact surface, but response quality and spare-parts logistics are undisclosed. | 中 | SR019, SR018 |
| CR030 | The company's heavy use of brand-news pages raises communication risk because market narrative may outrun audited operating proof. | 中 | SR024, SR012 |
| CR031 | Supportive policy can reverse into compliance risk if national or sector-specific standards become stricter faster than product updates. | 中 | SR005, SR007 |
| CR032 | International expansion increases jurisdictional compliance complexity around product safety, privacy, and after-sales obligations. | 中 | SR001, SR003, SR012 |
| CR033 | No public evidence shows a mature installed-base telemetry or incident-reporting program, so early warning capability is uncertain. | 中 | SR015, SR018, SR019 |
| CR034 | The most realistic operational failure chain is battery or sensor instability leading to safety events, support burden, and reputational damage. | 中 | SR002, SR005, SR014 |
| CR035 | The most realistic commercial failure chain is price compression leading to margin stress, more fundraising need, and weaker channel leverage. | 中 | SR010, SR011, SR012 |
| CR036 | The strongest mitigation signals are support surfaces, partner outreach, and developer tooling rather than disclosed hard-control metrics. | 中 | SR018, SR019, SR028, SR029 |
| CR037 | The thesis-break threshold is not a single lawsuit but a combination of safety incident, channel underperformance, and inability to finance the ramp economically. | 中 | SR001, SR011, SR012 |
| CR038 | A public adverse-stance source is already signaling that the sector's price war can undermine weaker players before scale economics mature. | 中 | SR011 |
| CR039 | Noetix is therefore exposed to layered legal, operational, dependency, and execution risks that interact rather than sit independently. | 低 | SR001, SR002, SR009 |
| CR040 | The practical diligence priority is to test safety governance, quality systems, partner economics, and hiring depth before treating current momentum as durable. | 中 | SR005, SR018, SR021, SR029 |
| CV001 | Noetix was valued near RMB 10 billion or roughly USD 1.39 billion after the March 2026 Series B. | 中 | SV005, SV006 |
| CV002 | Total capital raised exceeds RMB 1.5 billion across nine rounds, which meaningfully reduces immediate survival risk. | 中 | SV005, SV006, SV007 |
| CV003 | The company is well funded for a private robotics startup, but public evidence still does not disclose recognized revenue or gross margin. | 中 | SV005, SV006, SV015 |
| CV004 | TrendForce projects 94 percent output growth in China humanoids in 2026, which supports a large addressable market but also raises overcapacity risk. | 高 | SV001, SV009 |
| CV005 | Robotics Center describes China as the global center of humanoid manufacturing and projects aggressive deployment scale, supporting the market-growth side of the valuation case. | 中 | SV002 |
| CV006 | Humanoid APAC and AI Business both show a 2026 price-compression backdrop, limiting how much valuation investors should pay for unproven margin profiles. | 中 | SV003, SV004 |
| CV007 | The bull case depends on Noetix converting product and channel momentum into high-volume shipments before the price war overwhelms contribution margin. | 中 | SV001, SV012, SV013 |
| CV008 | The bear case is that strategic capital financed a fast hardware ramp before product quality and gross margin became durable. | 中 | SV004, SV011, SV015 |
| CV009 | The base case is that Noetix becomes a real but still opaque Chinese robotics platform whose valuation remains sensitive to the next financing round. | 中 | SV005, SV006, SV011 |
| CV010 | Bumi at roughly USD 1,370 creates a low-friction acquisition story that can support user growth faster than premium-only humanoid strategies. | 中 | SV013, SV015 |
| CV011 | N2 and E1 provide higher-ticket lanes that can lift blended ASP if Noetix can scale institutional or distributor demand. | 中 | SV019, SV020, SV015 |
| CV012 | JD.com is the clearest channel-quality signal because it combines preorder proof, platform inclusion, and a multi-year user-growth target. | 高 | SV012, SV013, SV014 |
| CV013 | RBTX distribution support indicates Noetix is not limited to direct domestic retail, which slightly improves the exit and market-expansion narrative. | 中 | SV018, SV019 |
| CV014 | The strongest anti-thesis is not lack of demand interest but lack of transparent proof on margin, retention, and unit economics. | 中 | SV003, SV004, SV015 |
| CV015 | Noetix's 1,000-units-per-month target implies meaningful revenue capacity, but capacity-based revenue is not the same thing as realized demand. | 中 | SV005, SV015, SV019 |
| CV016 | Using public price points and capacity signals, a rough 2026 annualized hardware revenue-capacity range of roughly USD 25 million to USD 70 million is supportable. | 低 | SV005, SV019, SV020 |
| CV017 | At the current private valuation, that capacity range implies a valuation multiple that is still rich for a hardware company without disclosed revenue quality. | 中 | SV005, SV016 |
| CV018 | TrendForce's discussion of Unitree's 60 percent combined gross margin shows that scaled Chinese robotics vendors can be profitable, but it does not prove Noetix is already there. | 中 | SV001 |
| CV019 | The presence of a profitable peer benchmark makes Noetix's valuation less absurd than a pure science project, yet still highly dependent on future execution. | 中 | SV001, SV004 |
| CV020 | ABIT and AI Business both reinforce bubble risk in Chinese humanoids, which argues for valuation discipline rather than momentum chasing. | 中 | SV004, SV011 |
| CV021 | FutureTekNow and CnTechPost show that strategic capital from CATL-linked CD Capital is a core part of the valuation support story. | 中 | SV005, SV006 |
| CV022 | The HKEX filing on repeated Beijing Robot Fund support suggests policy-aligned capital also underpins the company's financing durability. | 高 | SV006, SV007 |
| CV023 | A valuation supported by strategic and policy capital can still be economically rational, but it is less directly anchored to transparent public operating metrics. | 中 | SV006, SV007 |
| CV024 | OpenHarmony and developer-ecosystem surfaces slightly strengthen the strategic-value argument because they make Noetix look more like a platform than a single device. | 高 | SV021, SV022, SV029, SV030 |
| CV025 | The existence of customer portals, support forms, and policy surfaces suggests some product-system maturity, but not enough to erase financial opacity. | 中 | SV023, SV024, SV025, SV026 |
| CV026 | The most defensible recommendation from public evidence is research-more rather than an immediate buy or avoid call. | 中 | SV003, SV005, SV011 |
| CV027 | Confidence in any positive recommendation should remain medium-to-low because core revenue-quality metrics are undisclosed. | 中 | SV005, SV015 |
| CV028 | The valuation stance looks stretched on disclosed fundamentals, even if not obviously impossible under a successful 2027 scale-up. | 中 | SV005, SV016, SV019 |
| CV029 | The best upside scenario requires volume, gross-margin proof, partner execution, and no major safety or quality interruption during ramp. | 中 | SV001, SV012, SV018 |
| CV030 | The most plausible downside scenario is a mix of price compression, support burden, and another funding round at weaker terms. | 中 | SV003, SV004, SV011 |
| CV031 | Comparable valuation work is limited because most Chinese humanoid peers are private and public disclosures are uneven across the sector. | 中 | SV004, SV011 |
| CV032 | Unitree is a relevant comp because it combines shipment scale, pricing aggression, and emerging public-market visibility. | 中 | SV001, SV004 |
| CV033 | UBTech is a relevant comp because it is public and therefore shows what transparency could eventually look like for Noetix, even if the product mix differs. | 中 | SV011 |
| CV034 | Figure AI is a useful sentiment comp because it shows how far capital can outrun disclosed revenue in the global humanoid narrative. | 中 | SV004 |
| CV035 | The biggest thesis-break condition is not a low-growth quarter but evidence that Noetix cannot convert public excitement into economically sound repeat demand. | 中 | SV013, SV014, SV015 |
| CV036 | A second kill trigger would be a material safety, support, or regulatory failure that undermines consumer and partner trust during expansion. | 中 | SV023, SV024, SV028 |
| CV037 | The highest-priority diligence asks are monthly shipments, realized ASP, gross margin by SKU, refunds, partner economics, and support burden. | 中 | SV005, SV012, SV019 |
| CV038 | A credible positive revision to valuation would require verified customer cohorts and channel economics, not just another strategic funding headline. | 中 | SV012, SV013, SV014 |
| CV039 | If Noetix can prove repeat demand, margin progression, and partner-led global distribution, the current valuation could look fair in hindsight. | 中 | SV012, SV018, SV021 |
| CV040 | If those proof points do not materialize, the current valuation should be treated as expensive relative to disclosed fundamentals. | 中 | SV004, SV011, SV015 |