初创公司尽调
尽调报告 Robotics / Embodied AI Series C-equivalent (late-stage private) 2026-06-14

Galbot

Galbot:中国资本最充足的人形机器人创业公司——真实部署,财务不透明

Galbot 是战略可信但财务不透明的中国人形机器人头部公司,已有真实工业部署、世界级 embodied AI 和国资背书;在承诺投资前,应继续深度研究和尽调。

封面要素

最近一轮融资 01
RMB 2.5B (~$350M) [CO019]
估值 03
3000 USD M [CO017]
成立时间 04
2023-05-19 [CO001]
工业订单 05
Thousands of units [CO021]
零售覆盖 06
30+ cities [CO021]

公司概况

Galbot(Beijing Galbot AI Co., Ltd.)是一家北京具身人工智能公司,2023 年 5 月 19 日由 He Wang(一位 Stanford 训练背景的 Peking University AI 教授)和 Zhang Zhizheng 创立。公司打造 G1,这是一款采用轮足混合移动结构的通用人形机器人,同时自研完整技术栈——数据集(100 亿+ 数据点)、具身基础模型(GraspVLA、TrackVLA、GroceryVLA)和机器人硬件。Galbot 已将 G1 部署到工业制造(CATL、Mercedes-Benz、Zeekr)、智慧零售(Galbot Store、30+ 城市)和医疗健康(Xuanwu Hospital、北京药房)。公司累计融资超过 $1.15B,其中包括 2026 年 3 月由国家背景资金参与的人民币 25 亿元融资;截至 2025 年 12 月,Galbot 估值 $3B,巩固了中国估值最高未上市人形机器人公司的位置。

官网
www.galbot.com
成立时间
2023-05-19
创始人
He Wang, Zhang Zhizheng
创立地点
Beijing, China
总部
Beijing, China
产品
Galbot G1 是一款通用人形机器人(轮足混合,标准姿态 173 cm,约 92.5 kg,IP54,48V 30Ah 电池,10 小时运行时间),由端到端具身 AI 平台驱动。G1 服务于工业装配与物流、自主零售门店运营(Galbot Store)和医疗健康辅助(药房发药、医院导诊)。
客户
企业客户集中在工业制造(汽车、电池、电子)、智慧零售运营商和医院 / 医疗健康机构,目前重点放在中国市场部署。
商业模式
硬件整机销售和企业部署合同(工业、医疗健康);通过 Galbot Store 开展自主零售运营(可能采用收入分成或 RaaS);潜在的软件 / AI 模型授权。公司未公开披露收入。
阶段
Series C-equivalent private
融资情况
2026 年 3 月完成人民币 25 亿元($350M)融资(由 National AI Industry Investment Fund 领投);此前各轮合计约 $800M(截至 2025 年 12 月累计,包括 2025 年 6 月 CATL 领投的 $153M 和 2025 年 12 月 $300M+ 融资)。截至 2026 年 3 月,累计资本约 $1.15B+。
[CO001, CO002, CO003, CO005, CO015, CO017, CO018, CO019]

执行摘要

主要优势

  • 全栈自有平台:10B+ 数据点数据集、GraspVLA/TrackVLA/GroceryVLA 模型和 G1 硬件,让 Galbot 区别于纯硬件同业。
  • 国资背景资本底座:National AI Fund、Sinopec、CITIC、Bank of China 和 SAIC 投资,带来政策触达和规模化采购优势。
  • 工业部署已验证:CATL、Mercedes-Benz、Zeekr、Bosch JV 以及数千台订单,证明商业牵引已越过试点阶段。
  • 学术创始人信誉强:He Wang 的 Stanford 博士和 PKU 教授身份,加上 BAAI/PKU 联合实验室,说明研究到产品的管线够深。

主要风险

  • 财务不透明:收入、毛利率和单位经济均未披露;当前 $3B 估值完全压在叙事和投资人预期上。
  • CATL 集中度:最大投资人也是披露口径下最大客户,这种关联方关系带来治理和集中度风险。
  • 关键人依赖:He Wang 同时担任学者和 CEO,带来接班与领导带宽风险。
  • 监管和地缘敞口:中国科技硬件在西方市场面临出口管制和数据本地化风险;新的国家标准也会增加合规成本。
  • 技术商品化:多家中美公司都在收敛到类似 VLA 架构;护城河能维持多久,取决于自有数据和客户锁定。

未决问题

  • 任一期间经审计收入、ARR 和毛利率
  • G1 部署的单机 P&L 和回本周期
  • CATL 合同条款、期限和排他性约定
  • 员工数和 burn rate 轨迹
  • IP 组合范围和 freedom-to-operate 分析
  • 医疗(患者 PII)部署的数据隐私合规状态
  • cap table 结构、清算优先权堆栈和稀释路径

目录

Chapter 01

01公司概览

1.1 身份、创立版图与当前产品位置

Galbot 的公开身份比许多快速崛起的机器人创业公司更清楚。公司把自己定义为一家具身 AI 机器人公司,2023 年 5 月创立于北京;核心叙事不是把人形或移动操作机器人停留在实验室演示,而是让它们进入真实商业环境。官方页面指向北京总部,也显示其研发版图延伸到深圳、苏州和香港;这符合一家试图同时拼前沿模型、硬件集成和商业部署支持的公司画像。旗舰 G1 平台面向工业、零售和医疗健康工作流,开发者门户和应用套件也表明 Galbot 想围绕机器人搭软件和生态层,而不是把它当作一次性硬件展示。结果是,Galbot 更像一套全栈商业化工程,而非单纯研究项目;不过不少产品和财务细节仍来自公司自控渠道。[CO001, CO002, CO003, CO004, CO005, CO006]

快照 KPI 表
指标数值 / 状态日期置信度缺口
成立2023-05-19,北京2023-05-19公开来源确认日期,但本章未纳入公司注册细节。
总部 / 研发北京总部;深圳、苏州和香港设研发2026-06-14各地员工相对分布未披露。
阶段Series B+/C 等效私营独角兽2025-12-01优先股具体条款和董事会权利未公开。
旗舰产品面向工业、零售和医疗工作流的 G1 人形 / 移动操作机器人2026-06-14定价和部署单位经济性未公开。
2025 年末最新估值标记~$3.0B2025-12-01估值由公司相关报道支撑,而非经审计财务披露。
总融资额标记2025 年末约 $800M;另有 2026 年 3 月 RMB2.5B2026-03-02累计完全摊薄股权结构计算未公开。
公开规模标记数千台订单;30 多个城市门店覆盖2026-05-28订单转收入的独立核对不可得。
数据资产声称10B+ 具身 AI 数据点2026-04-09数据集构成和标注方法未披露。
财务披露收入和员工人数未披露2026-06-14已审计收入、利润率或员工人数包未出现在已审阅材料中。

公开指标综合官方页面、媒体报道和公司相关发布;估值、订单和数据规模数字的可验证性弱于身份和产品事实。

[CO001, CO003, CO005, CO011, CO017, CO018]
FO003: 快照 KPI

紧凑的承销信号解释了 Galbot 为什么吸引资本,同时仍需要更深尽调。

[CO011, CO017, CO019, CO024, CO029, CO030]

1.2 创始人、领导层集中度与机构网络

创始人集中度既是 Galbot 最大优势之一,也是最清晰的风险之一。公开材料把公司归于 He Wang 和 Zhang Zhizheng,这两位创始人在官方与媒体报道中仍是主要具名运营者。集中度之所以重要,是因为 Galbot 的产品故事取决于高速整合机器人、具身 AI 数据和垂直部署关系的能力;反过来,外部观察者也难看清管理层冗余或接班安排。公司伙伴页面和第三方报道提供了一些安慰:Galbot 与 Peking University、BAAI、Xuanwu Hospital 等重要研究和临床机构,以及工业和汽车利益方都有联系。即便如此,公开记录对独立董事构成、正式治理和创始人之外的高管厚度仍着墨很少。从尽调角度看,Galbot 机构连接强、技术可信,但仍高度创始人主导,关键人物依赖被放大。[CO002, CO008, CO009, CO010, CO021, CO032]

领导层与创始人表
人物 / 节点角色背景职能覆盖或创始人-市场匹配关键人物依赖
He Wang联合创始人官方和媒体材料公开确认其为创始人。支撑公司身份和面向伙伴的可信度。高依赖,因为创始人之外的公开领导层深度有限。
Zhang Zhizheng联合创始人官方和媒体材料公开确认其为创始人。与 He Wang 一起支撑技术和商业化叙事。高依赖,因为公开继任梯队未披露。
研究与实验室网络外部领导节点伙伴页面和报道提到 PKU、BAAI 以及医院合作者。增加科研可信度和领域验证。依赖仍是间接的,因为伙伴机构不是内部管理层。
临床部署网络外部执行节点Xuanwu Hospital 和药房部署提供医疗运营语境。支持工厂试点之外的高风险真实世界验证。依赖为中等,因为具体内部医疗领导梯队未被点名。
工业赞助方组合战略支持节点CATL、Bosch 和国资背景投资者在融资报道中反复出现。拓宽商业化入口和资本可得性。依赖为中等,因为支持可能绑定具体关系,而非系统化能力。

公开记录清楚列出创始人和外部机构节点,但未披露独立董事会、完整高管梯队或正式治理委员会。

[CO002, CO008, CO009, CO010, CO032, CO035]

1.3 融资历史、估值跃升与利益相关方图谱

Galbot 的融资轨迹是本章最强的外部验证点。公司早期资本路径据称从 2023 年 6 月种子轮开始,随后在当年完成天使轮和天使+轮,2024 年 3 月又完成数亿元人民币融资。到 2025 年中,多家媒体描述了一笔约人民币 11 亿元或约 $151-153M 的融资,并与 CATL 关联资本和 Bosch 相关战略合作挂钩。2025 年底融资节奏再次加快,公司关联和独立报道均称其新增融资超过 $300M,估值约 $3B,累计融资约 $800M。2026 年 3 月,国家背景投资者据称又领投人民币 25 亿元融资,把 Galbot 从纯风险资本背书推入国家战略资本通道。这套资本栈很亮眼,但估值逻辑仍重度依赖预期,因为收入、利润率和员工数仍未向公众详细披露。[CO011, CO012, CO013, CO014, CO015, CO016]

利益相关方或投资者地图
利益相关方角色控制 / 经济重要性公开关联尽调问题
National AI Industry Investment Fund(领投基金)牵头国资背景融资方释放国家战略相关性信号,并可能影响长期规模决策。据报道牵头 2026 年 3 月 RMB2.5B 轮。确认治理权、董事会席位和政策条件。
CATL 相关资本战略产业投资者把 Galbot 连接到电池和制造生态,并带来真实部署路径。与 2025 年约 RMB1.1B / $151M-$153M 融资相关。核实商业承诺与纯财务赞助的区别。
Bosch 投资部门 / JV 渠道战略投资者和商业化伙伴同时提供背书,并可能提供进入工业客户的路径。在 2025 年融资和合资报道中被点名。厘清关系范围、排他性和经济性。
Sinopec Capital / CITIC / Bank of China / SAIC 组合国资和产业跟投方暗示广泛的国内机构支持和潜在企业入口。出现在 2026 年轮次参与方名单中。要求提供分批规模、持股比例以及任何客户采购绑定。
Mercedes-Benz / Toyota / BAIC / Zeekr / SAIC 等车企战略客户或试点组合如果转化为重复项目,这些客户标识会实质性强化承销。公开报道和伙伴材料中提到。区分已签生产项目与试点或概念验证工作。
Xuanwu Hospital 和药房渠道临床和零售部署利益相关方提供最清晰的真实服务环境使用公开证明。医院和机器人药师报道中提到。确认付费部署规模、留存和合规义务。
PKU / BAAI 研究生态科学验证利益相关方提升模型和机器人可信度,但未必直接转化为收入。伙伴材料和公开报道中点名。将研究声望与已签约商业需求分开。

对一家私营机器人创业公司来说,Galbot 的公开利益相关方组合异常强,但经济条款、持股比例和从试点到生产的转化率大多仍未公开。

[CO015, CO016, CO019, CO020, CO021, CO025]

1.4 商业化证据与运营里程碑

Galbot 的公开商业化案例靠部署证据支撑,而不是靠已披露财务指标。2025 年和 2026 年的多方来源把公司与 CATL、Mercedes-Benz、Zeekr、Bosch、Toyota、BAIC、SAIC、Xuanwu Hospital 以及连锁药房场景联系起来,说明其商业足迹横跨工业和服务场景。最清晰的垂直证据来自医疗健康和药房自动化:围绕 G1 机器人的报道把 Galbot 放在北京真实发药或零售辅助环境中。其他来源提到数千台订单、覆盖 30 多个城市、100 亿+ 具身 AI 数据资产;不过这些规模指标更接近公司口径,尚未独立审计。里程碑顺序依然重要:创立、早期融资、2025 年工业背书、2025 年底独角兽估值、2026 年医疗部署,以及国家人形机器人标准同步出现,共同说明 Galbot 处在有利的商业化窗口,而不只是靠演示视频堆出的炒作周期。[CO015, CO017, CO019, CO021, CO022, CO023]

里程碑表
日期事件类型金额 / 估值 / 状态参与方含义
2023-05-19公司在北京成立创立新的具身 AI 机器人创业公司成立He Wang;Zhang Zhizheng确立此后所有融资和部署说法的起点。
2023-06-01种子轮被报道融资种子融资完成创始人和早期支持者显示公司成立后很快启动资本形成。
2023-08-01天使轮被报道融资天使融资完成早期风险投资和战略支持者表明市场很早就快速相信团队。
2023-10-01天使加轮被报道融资追加早期融资完成复投和新进入支持者暗示规模化部署前已有基于里程碑的跟投资金意愿。
2024-03-01数亿元人民币轮次被报道融资中期扩张资本风险投资和产业投资者资助产品化和部署扩张。
2025-06-24CATL 相关融资和 Bosch 相关合作广泛浮现合作报道描述约 RMB1.1B / $151M-$153M 融资CATL 相关资本;Bosch 投资部门标志公司从创业承诺转向产业支持下的商业化。
2025-12-23超过 $300M 的新融资被报道融资约 $3.0B 估值;描述累计融资约 $800MGalbot 和参与投资者在 2026 年政策顺风前确立独角兽以上状态。
2026-03-02国资背景 RMB2.5B 融资宣布融资RMB2.5B 新轮次National AI Industry Investment Fund、Sinopec、CITIC、Bank of China、SAIC 等资本方确认异常强的国资和产业赞助。
2026-03-14G1 机器人药师在北京部署被报道产品真实世界医疗用例公开Galbot;Xuanwu Hospital;药房运营方提供应用商业化最清晰的证明之一。
2026-03-22中国制定人形机器人国家标准监管行业标准化里程碑中国监管机构和行业组织改善 Galbot 所处的政策框架。
2026-04-21全球投资者对中国人形机器人关注升温负面行业热度夹杂地缘政治审视国际投资者和媒体凸显 Galbot 的估值背景不仅暴露于执行,也暴露于情绪和地缘政治。

2023-2024 年早期融资阶段来自后续公司和媒体摘要重建;2025-2026 年里程碑则有更多同期报道直接记录。

[CO001, CO012, CO013, CO014, CO015, CO016]
FO001: 公司里程碑时间线

融资、部署和政策拐点显示,Galbot 不到三年就从成立走到国家资本加持的规模化阶段。

[CO001, CO012, CO014, CO015, CO016, CO017]

1.5 负面背景、监管与承销边界

核心承销问题不是 Galbot 是否有意思,而是证据基础是否成熟到足以支撑其估值和战略叙事。公开报道支持强劲投资者需求、国家背书和可见试点,但公司仍未发布详细收入、毛利率、盈利能力或员工数指标,已审阅材料中也没有经审计财务包。因此,估值主要靠订单、部署、战略股东和对具身 AI 采用的预期来支撑。外部风险也真实存在。中国监管标准可能帮助人形机器人市场规范化,但医疗健康和零售部署会带来责任、安全与合规问题;与此同时,公司以中国为中心的资本和供应链暴露,在与全球知名汽车品牌合作时会产生地缘政治敏感性。投资者可以合理把 Galbot 视作动能上的品类领导者,但在把隐含估值视为充分承销之前,仍需要更深的治理、经济性和法律韧性尽调。[CO027, CO028, CO029, CO030, CO031, CO032]

FO002: 公司快照逻辑

Galbot 的承销逻辑把创始人、数据、战略资本、部署和监管连成一条商业化投资逻辑,失败点也很清晰。

[CO006, CO008, CO016, CO021, CO025, CO027]

1.6 图表

Chapter 02

02市场分析

2.1 市场边界、纳入支出与替代品

Galbot 的相关市场比“所有机器人”窄,也比“人形硬件出货”宽。实际边界是部署到物理工作流中的企业级人形系统,以及让这些系统真正有用所需的具身 AI 软件、训练数据、集成和服务。因此,本章纳入工业制造、仓储物流、零售服务和医疗健康部署,因为这些细分领域反复出现在当前市场报道和 Galbot 自身商业化口径中。除非人形形态是买方用例的核心,否则通用工业自动化、AGV、协作机器人和纯软件 AI 类别不纳入。由此,现状竞争不只是其他人形机器人创业公司,还包括人工、固定自动化、特定工作流机器人和被推迟的自动化支出。这个定义很重要:一个万亿美元级广义 TAM 可以方向上成立,却仍可能夸大 Galbot 近期可服务市场。[CM001, CM002, CM003, CM028, CM029, CM030]

市场定义表
细分 / 类别纳入支出排除支出买方 / 付款方与 Galbot 的相关性
企业人形机器人系统人形硬件加机器人端软件,用于物理工作流非人形机器人和通用企业 AI工厂、物流、零售、医疗运营方核心可服务类别
具身 AI 软件和服务数据采集、训练、集成、机队运营、维护没有机器人部署的纯软件助手自动化、IT 和运营预算让硬件产生生产力所必需
工业制造工作单元辅助、物料搬运、检测、重复性体力任务不需要人形形态的固定自动化工厂运营和自动化团队最强的近期滩头阵地
仓储物流拣选、内部移动、接近分拣的劳动力替代无需人形机器人即可解决任务的 AMR 或输送线物流运营和仓库工程重要,但 ROI 要与许多替代方案比较
零售、服务与医疗面客辅助、接待、货架或地面工作流、养老支持纯自助终端软件或仅远程呈现系统门店运营、服务创新、医院或护理管理者有用的证明点细分,但往往扩张更慢
消费 / 家庭机器人长期家庭辅助用例近期企业生产力工作流消费者基本在 Galbot 近期可服务市场之外

边界以企业人形机器人部署为重心;只有当软件和服务绑定机器人生产力,而非通用 AI 支出时,才纳入其中。

[CM001, CM002, CM003, CM028, CM029, CM030]

2.2 多视角测算与相互矛盾的估计

已发布的市场数字分歧很大,因为它们衡量的东西不同。基于 IDC 的报道描述了 2025 年已实现硬件市场:约 18,000 台、约 $440M 收入,这锚定了商业化仍处早期的现实。MarketsandMarkets 和 SkyQuest 给出更宽的人形机器人品类前瞻收入预测,而 People's Daily 引用的具身 AI 预测范围更宽。Morgan Stanley 和 UBS 把时间线拉到 2050 年,并把机会描述为数万亿美元级,但这些长期 TAM 视角情景假设很重,不能当作近期市场出清需求。中国本土规模也不确定:CCID 相关报道称国内人形市场到 2026 年可能超过人民币 200 亿元,但公开方法论比标题更薄。尽调结论不该是一项估计正确、其他都错误;正确结论是,Galbot 的可触达市场会随着视角从硬件收入、全栈人形系统扩展到更广义具身 AI 栈而急剧放大。[CM004, CM005, CM006, CM017, CM018, CM019]

TAM / SAM / SOM 或规模测算视角表
发布方年份地区数值CAGR方法 / 视角置信度局限
基于 IDC 的报道2025全球~18,000 台;~$440M 硬件收入~508% 同比出货增长已实现年度出货和硬件收入视角仅硬件快照,不是全栈市场价值
MarketsandMarkets2025-2030全球2025 年 $2.92B 至 2030 年 $15.26B39.2%更广义人形机器人市场预测,包含软件 / 服务摘要商业类别定义比仅硬件 IDC 视角更宽
SkyQuest2025-2033全球2025 年 $1.47B 至 2033 年 $35.41B48.9%全球人形机器人预测与其他分析机构相比,时间范围和纳入规则可能不同
人民日报引用的行业报告2025-2030全球具身 AI2025 年 $4.44B 至 2030 年 $23B~39%比人形硬件本身更宽的具身 AI 视角不是纯人形机器人市场指标
CCID 相关报道2026中国>20B 元n/a中国国内近期商业化视角可访问报道中看不到完整公开方法
UBS2035 / 2050全球2035 年 >2M 台人形机器人;2050 年 >300M 台;2050 年 $1.4T-$1.7Tn/a长期保有量与 TAM 情景远期情景分析,不是近期需求预测
Morgan Stanley2050全球$5T TAM;约 930M 台工业 / 商用人形机器人n/a通用人形机器人长期 TAM 情景情景假设占比重,对采用率假设高度敏感

表格有意保留彼此不完全可比的口径,因为它们衡量的边界不同:已实现硬件收入、更宽的人形机器人系统、具身智能、仅限中国的商业化,以及长期 TAM 情景。

[CM004, CM005, CM006, CM017, CM018, CM019]
FM001: 市场规模视角

从长期全球 TAM 情景,一层层下探到 Galbot 当前企业切入点。

长期和中期层使用的发行方边界不同;Galbot 切入点层使用订单数量证据,而不是收入,因为目前没有公开的、经独立验证的分部收入披露。

[CM017, CM020, CM021, CM023, CM032, CM033]
FM002: 市场估算区间

已发布的市场价值视角横跨早期已实现硬件收入、中期商业预测和长期 TAM 情景,并统一为 USD billions。

中期行混合了相邻但并不完全相同的市场定义;长期行把 UBS 区间与 Morgan Stanley 更高情景放在一起,是为了呈现不确定性,而不是暗示精确共识。

[CM005, CM017, CM020, CM021, CM022, CM024]

2.3 买方分层、预算所有者与 Galbot 相关性

最可信的近期买方,是拥有重复性物理工作流、且面临劳动力短缺或强烈自动化试验意愿的企业。工业制造是最清晰的滩头阵地:工作流结构化,安全边界可管理,预算通常在工厂运营、工业工程或自动化团队手里。仓储物流类似,但 ROI 审查更严,因为买方已经会拿人形机器人与输送线、AMR 和其他专用系统比较。零售和服务部署能制造可见证据,但许多仍属于品牌或体验预算,而非硬生产力预算。医疗健康和养老长期拉力很强,因为人口老龄化是结构性需求驱动,但认证、责任和工作流敏感性会让规模化更慢。Galbot 公开的工业制造、零售和医疗健康部署说明,它真正的近期市场是逐个垂直行业推进的企业采用路径,而不是泛泛的消费机器人故事。[CM007, CM008, CM028, CM029, CM030, CM031]

细分市场 / 买方图谱
细分市场买方用户付款方工作流预算归属采用触发点
工业制造OEM 或大型制造商产线工人或工业技师工厂或自动化预算物料搬运、检测、重复性体力支持工厂运营 / 工业工程劳动力短缺,加上可衡量的吞吐量或安全收益
仓储物流3PL 或大型托运方仓库作业员运营或资本开支预算拣选、内部转运、补货仓库运营 / 自动化负责人用工波动,且愿意测试柔性自动化
零售 / 服务零售商、商场运营方、酒店餐饮连锁门店或场地员工运营或创新预算接待、货架、客服、场内协助门店运营 / 创新团队品牌差异化,或在可见岗位替代人力
医疗 / 养老医院、诊所、照护机构护士、护工、支持人员设施运营或照护创新预算转运、监测协助、重复性支持任务照护管理 / 临床运营人口老龄化叠加员工短缺
教育 / 研究大学或实验室研究人员和学生研究预算教育、实验、数据采集首席研究员 / 实验室经理需要开发平台,而不是规模化运营
消费 / 家庭家庭消费者消费者钱包家务协助家庭决策者价格大幅下降,且安全信任显著提升

预算归属字段综合了分析师、部署和公司证据;它们显示 Galbot 当前部署最可能落在哪些位置,而不是声称覆盖了企业采购的全部场景。

[CM028, CM029, CM030, CM031, CM032, CM033]
FM003: 买方准备度热力图

矩阵按劳动力痛点、合规摩擦和 ROI 证明,而不是重复买方—付款方地图,比较各细分场景规模化采用人形机器人的准备度。

序数评分综合了公开部署、分析师和公司证据;目的在于比较细分场景准备度,而不是量化市场份额。

[CM033, CM037, CM043, CM047, CM049, CM050]

2.4 增长驱动因素与为什么中国先重要

最强的增长驱动因素在中国比多数其他市场更早汇合。政策支持很明确,从标准化工作到地方产业基金,再到把具身 AI 视为战略产业。供应链深度同样重要:随着零部件生态、电池、执行器和制造能力提升,单位经济性应当下降,商业化也会更少依赖展示性试点。TrendForce 认为 2026 年下半年是市场重心从基础能力转向用户价值的节点,这与 IDC 式证据一致:出货量已经大到足以形成真实装机基础。需求侧拉力也在上升,来自老龄化、劳动力短缺,以及制造、物流和服务工作流中的企业自动化兴趣。对 Galbot 来说,中国不只是方便的本土市场;它是技术进步、政策支持和买方试验最早同时发生的大型地理市场。[CM009, CM010, CM011, CM037, CM038, CM039]

增长驱动因素与约束表
驱动因素 / 约束方向时间影响尽调问题
人口老龄化与劳动力短缺驱动因素2026-2035支撑照护、物流和服务岗位的自动化需求量化 Galbot 哪些细分市场当前用工痛点最尖锐
政府支持与地方基金驱动因素2025-2026降低中国商业化摩擦核实哪些省级项目直接惠及 Galbot 客户
单机成本下降与规模扩大驱动因素2026-2030改善 ROI,扩大买方池跟踪 BOM、执行器、电池和集成成本曲线
具身智能与 LLM 进步驱动因素2026 年以后扩大可用任务集和部署灵活性区分演示进展与生产稳健性
中国供应链成熟驱动因素当前帮助国内厂商更快放量测试供应链规模是否也提升现场可靠性
单机成本高、资本强度大约束当前让许多部署停留在试点或展示阶段按垂直行业和任务询问回本周期
任务泛化受限约束当前把机器人限制在狭窄、脚本化的工作流里要求提供跨场地、跨任务迁移证据
数据稀缺、真实世界训练有限约束当前放慢具身模型可靠性提升审核数据采集、遥操作和 sim-to-real 闭环
触觉感知与手部灵巧性瓶颈约束当前限制具备经济价值的操作广度检查抓取成功率、恢复能力和易碎物体表现
安全、责任和合规不确定性约束2026 年以后提高买方尽调和部署开销审查认证路径、召回和可追溯流程
需求落后于制造产能约束2026-2027产出扩张可能快于付费买方增长对比订单积压、产能利用率和续约数据

各行混合了结构性增长驱动因素和执行约束;尽调核心不是市场是否存在,而是每个买方细分市场里的约束能多快消退。

[CM009, CM010, CM012, CM013, CM014, CM015]
FM004: 采用漏斗或价值链地图

人形机器人采用会从广泛技术兴趣压缩到小得多的池子:部署必须具备规模、合规和经验证的 ROI。

漏斗宽度是指数化、方向性的,并非来自已发布转化数据集;它概括了证据所显示的流失位置:技术、安全和 ROI 要求越高,部署越容易掉队。

[CM014, CM016, CM040, CM042, CM043, CM044]

2.5 约束、监管与剩余不确定性

市场大到值得重视,但也仍不成熟,标题式 TAM 容易遮住运营风险。高单价和资本强度让许多部署停留在试点阶段。更重要的是,行业仍卡在任务泛化、真实世界数据收集、灵巧操作、安全验证以及接入客户工作流等问题上。这些不是表面问题;它们正是决定需求能否跟上当前制造产能建设速度的因素。监管影响双向。中国 2026 年 3 月标准和 2026 年 5 月数字 ID 制度,可能通过改善可追溯性和召回纪律来降低买方犹豫,但也抬高合规预期。关于 Galbot 自身可获得份额的公开证据仍不完整:公司披露了订单和目标行业,却没有足够独立数据说明各细分的转化、利用率或留存。因此,投资者应把当前市场证据视为足以证明值得关注,但噪音仍太大,无法在没有额外尽调的情况下支撑单一精确的 TAM-to-SOM 桥。[CM012, CM013, CM014, CM015, CM016, CM023]

2.6 图表

Chapter 03

03竞争对手

3.1 竞争格局与真正重要的玩家

评估 Galbot,不能只对标一小撮知名人形机器人品牌。真实格局包括直接的全栈人形机器人同行、车企背景或上市公司入局者、可能抽象掉硬件差异的模型层竞争者、固定自动化和人工零售 / 工厂劳动力等替代品,以及可借汽车或电子生态供应链规模进入的理论新玩家。在直接具身 AI 竞争中,AgiBot 和 Unitree 因当前中国出货证据最重要,Figure 因资本强度和全球叙事能力最重要,Physical Intelligence 则重要在于:如果通用机器人基础模型能跨硬件移植,它可能在模型层压缩产品差异化。相邻压力来自 XPENG、UBTech、Boston Dynamics 和 1X,它们分别进攻工业、商业或家庭机器人不同切片。实际的现状替代品仍是人工加特定任务自动化,这让买方审查聚焦可靠性和劳动力替代,而不是单纯奖励人形新奇感。[CP001, CP002, CP004, CP006, CP009, CP012]

竞争格局
竞争者总部成立时间阶段融资估值产品关键客户 / 垂直行业部署状态
Galbot中国北京2023私营成长期已披露累计 $1.15B+最新披露 $3BG1 具身智能机器人平台工业、零售、医疗在工厂、医院有具名部署,并覆盖 30+ 城市零售网络
AgiBot中国上海2023私营成长期未披露未披露A2、G1、X2、D1 人形和四足产品线工业自动化 OEM 平台引用来源称 2025 年出货 5,100 台,2026 年 3 月生产第 10,000 台机器人
Unitree中国杭州2016私营后期或 IPO 前未披露未披露G1 和 H1 人形机器人,以及四足机器人研究开发者、工业试点声称全球出货至 30+ 个国家,并公开列出 G1 定价
Figure AI美国2022私营后期Series C 承诺资本 >$1B投后 $39BFigure 01、02、03,搭载 Helix 和 BotQ劳动力自动化与未来家庭市场商业与家庭路线图,并在扩建制造能力
Physical Intelligence美国2024私营模型公司未披露未披露π0 和 π0.5 通用机器人模型跨硬件模型层在 8 款机器人上验证软件模型,并发布开源版本
XPENG IRON中国广州上市公司项目上市在位企业母公司资助母公司公开市场估值IRON 人形机器人,加 VLA 2.0 技术栈零售导览、移动出行和汽车邻近机器人目标 2026 年底量产,2027 年 Q1 起用于门店
UBTech Walker S1中国深圳上市公司上市成长期上市公司融资公开市场估值Walker S 系列人形机器人工业和服务机器人计划 2026 年 5,000 台、2027 年 10,000 台

表格合并了公司声明和独立报道中的已披露融资、估值和部署信号;几家竞争者没有发布可比融资或估值数据,因此空白反映的是披露缺口,而不是没有资本。

[CP002, CP004, CP006, CP009, CP010, CP012]
FP001: 竞争对手定位象限

按当前商业证明和与 Galbot 当前投资逻辑的直接重叠度,对主要同行做序数地图。

X 和 Y 位置是基于引用证据综合得出的 1-10 序数判断,而不是已发布市场份额或收入数据。

[CP004, CP009, CP012, CP014, CP021, CP026]

3.2 高信号竞争者画像

AgiBot、Unitree、Figure 和 Physical Intelligence 代表四类不同竞争者原型。AgiBot 是最清晰的中国规模同行:它结合了宽硬件组合、激进量产里程碑和超越单一机器人本体的 OEM 平台叙事。Unitree 是最清晰的公开价格锚点,也是最可见的低端商业化故事,G1 更便宜,国际发货覆盖广。Figure 是资本最充足的风险投资背书标杆,把头部估值与制造和模型栈故事绑定,目标同时覆盖劳动力和家庭用例。Physical Intelligence 与其说是直接硬件卖家,不如说是模型层风险;其通用 π0 系列和开源发布让未来硬件更可互换变得更容易想象。XPENG 和 UBTech 是重要快速跟随者,因为它们带来汽车或公开市场资源;但当前最大的承销压力仍来自中国规模双雄和美国资本 / 模型双雄。[CP002, CP003, CP004, CP005, CP006, CP007]

GTM 与定价对比
竞争者价格区间部署模式目标客户分销验证状态
Galbot企业定价未披露直接部署与集成工厂、零售商、医院国资关联投资者和具名企业客户多场地具名部署,但没有公开 ASP
AgiBot未披露硬件加 Powered by AgiBot OEM 平台工业客户和 OEMCES 首秀与国内规模化推进出货证据强,但经济性未披露
Unitree公开 G1 锚定价 $13,500机器人销售,全球发货研究开发者和轻量商业用途在线品牌触达和 30+ 个国家同业中公开价格透明度最高
Figure AI未披露企业部署与未来家庭推出商业运营方和家庭风险投资网络和战略投资者资本和路线图已验证,价格未公开
Physical Intelligence开源、模型主导软件和模型分发机器人开发者和实验室openpi 仓库和研究社区模型验证强,但直接变现不太可见
XPENG未披露借母公司渠道商业化推出零售和移动出行关联用户汽车品牌和零售渠道路线图公开,但规模化交付仍在未来

定价对比只提供方向感,因为除 Unitree 外,公开标价很少;多数同业更该比较买方合同模式、渠道强度和验证成熟度,而不是逐台标价的苹果对苹果比较。

[CP007, CP010, CP014, CP023, CP024, CP027]
FP003: 融资 / 估值格局

头部资本和估值基准塑造了竞争耐力叙事。

图表混合了估值和资本基准,因为只有部分同行同时披露两者;目的在于展示续航不对称,而不是逐项比较企业价值。

[CP010, CP020, CP030]

3.3 能力、定价、GTM 与信任对比

能力对比的重点不是原始演示有多抓眼球,而是谁能把操作、导航、部署成熟度和企业集成放进同一个包里。Galbot 已发布的技术栈异常垂直:它声称自有数据、模型、硬件,以及一组面向抓取、导航和零售工作流的 VLA。AgiBot 和 Figure 在同时掌握身体与智能层方面最相似,Unitree 在公开入门价上更透明,Physical Intelligence 作为通用软件抽象最强。GTM 上,Galbot 与 CATL、Mercedes-Benz、Zeekr、医院以及多城市零售版图的具名部署,说明它更可能靠标杆客户和政策通道做企业销售,而非走消费或研究导向漏斗。信任姿态同样重要:随着监管和责任框架追上工厂、医院和公共零售空间的人形机器人应用,买方会比头条式基准视频更看重具名部署、安全信号和标准对齐。[CP018, CP019, CP020, CP021, CP023, CP024]

能力对比矩阵
维度GalbotAgiBotUnitreeFigurePhysical Intelligence
技术栈所有权数据、模型、硬件自研本体加平台架构硬件主导,配控制技术栈本体、模型和制造技术栈跨第三方机器人的模型层
公开核心差异化GraspVLA、TrackVLA、GroceryVLA,加 Sim2RealOne Robotic Body Three Intelligences(一体三智架构)低成本公开人形机器人标杆Helix VLA 加 BotQ通用 π0 基础模型
操作成熟度零售和工厂演示较强产品族宽,但公开任务细节参差公开移动能力和基础操作验证较好聚焦劳动力场景的操作叙事取决于所连接的机器人本体
导航或自主性TrackVLA 和多场地运营OEM 平台定位公开运动能力强家庭与劳动力自主路线图强调跨平台泛化
部署成熟度具名工厂、医院和门店出货量声明高出货覆盖广,但企业披露较少叙事势能强,但公开量级细节较少软件成熟,但没有直接部署规模
定价透明度none

矩阵是定性判断,反映所给来源明确支持的内容,不是实验室基准排名;定价透明度低或无,意味着公开定价有限或缺失。

[CP007, CP011, CP018, CP019, CP020, CP024]
FP002: 竞争能力条形图

基于公开证据,从技术栈深度、部署和自主性比较竞争对手能力广度,并做指数化处理。

分数是 0-100 归一化综合分,来自技术栈所有权、部署成熟度和公开产品广度的定性证据。

[CP007, CP011, CP012, CP018, CP020, CP024]

3.4 切换成本、锁定与分销不对称

Galbot 的护城河不是经典软件网络效应,而是一组运营包:数据集深度、现场集成、客户工作流调校和采购通道入口。客户一旦在真实门店、医院或工厂验证部署,这能形成有意义的切换成本,但并不意味着多供应商并行不可能。如果人形本体逐渐收敛到相似 VLA 架构和任务 API,客户可能同时测试不止一家供应商,再按可靠性、支持质量和经济性分配任务。因此,分销变得异常重要。Galbot 的国家背景投资者组合以及与工业冠军的关系,可能改善其进入中国试点项目、建立采购可信度和获得伙伴引荐的能力,这些是小型纯风险资本创业公司难以匹配的。但不对称也有反面:XPENG 或其他大型制造商这类拥有汽车或电子供应链的既有玩家,如果人形品类变得更由规模驱动而非模型驱动,可能在制造杠杆上追平甚至超过 Galbot。[CP021, CP022, CP027, CP033, CP034, CP035]

护城河评估
护城河维度Galbot 评估持久性关键风险
数据集规模10B+ 数据点声明支撑学习闭环优势对手可能很快积累可比具身数据
垂直技术栈所有权自研硬件加 VLA 技术栈,降低依赖Figure 和 AgiBot 也在推进类似全栈打法
具名企业部署工业、零售和医疗场景验证强中-高部署可能仍以试点为主,尚未深度规模化
分销与政策入口国资背景投资者组合可能改善采购入口政策优势可能停留在国内,也可能被大型在位企业追平
定价权公开 ASP 和服务经济性未披露,判断不清晰成本更低或资本更厚的同行可能挤压利润率
架构独特性Sim2Real 加任务专用 VLA 目前看起来有差异化VLA 商品化会很快削弱独特性

耐久性标签是基于截至 2026-06-14 已披露证据给出的承销判断;它们不是数值化的市场份额预测,拿到输赢单和定价数据后需要重新审视。

[CP019, CP021, CP022, CP025, CP030, CP034]

3.5 护城河耐久性与反面案例

反对 Galbot 的理由不是没有竞争者,而是强竞争者太多,并且 VLA 叙事重叠、广泛买方付费意愿尚未充分证明。外部批评仍很尖锐。TechXplore 引用观察者称,许多人形机器人仍然表演性强于功能性,真实用例仍很窄。市场领导权说法也有噪音:AgiBot 在一些 2025 年出货数据集中排名第一,但 Unitree 用自己的出货说法质疑这种领先。最后,中国估值基准仍大幅低于 Figure 等美国同行,这说明即便部署叙事看起来可比,资本市场仍在折价中国人形玩家的耐久性或全球变现能力。最稳妥的判断是,Galbot 在中国特定部署入口和垂直栈所有权上拥有真实近期护城河;但若要长期护城河超过中等耐久度,它还必须证明比那些正冲向同一架构的对手拥有更好经济性、更快学习闭环和更高真实世界利用率。[CP029, CP030, CP031, CP032, CP034, CP036]

3.6 图表

Chapter 04

04财务

4.1 收入模型、定价与确认问题

虽然 Galbot 没有发布完整财务拆分,但其公开商业故事暗示了几条收入流。最可见的是 G1 平台硬件销售,不过这些销售很可能与集成工作、部署配置和支持承诺绑定,模糊了设备收入与实施服务之间的边界。第二条收入流出现在零售和药房运营中,Galbot Store 和药房部署说明其更像运营商或托管服务模式,而不是一次性产品销售。与 CATL、Mercedes-Benz 和 Bosch 相关伙伴的工业项目,意味着更大的企业合同和更长验收周期;医疗健康部署则带来服务质量和可靠性义务,可能让收入确认时点更复杂。同样重要的是缺失项:Galbot 未披露 ASP、合同期限、门店经济性是否基于收入分成,或软件 / 模型授权是否与硬件分开入账。因此,收入模型的形状可读,但真实组合和会计处理仍不透明。[CI008, CI009, CI010, CI011, CI012, CI013]

收入模式和定价
收入流定价单位经济性代理指标规模成熟度置信度
硬件销售企业 ASP 未披露可能是单次部署中最大的名义合同价值真实存在但不透明
工业部署和集成按项目或里程碑计价ACV 更高,但周期更长且有验收风险已通过具名客户得到印证
维护和支持未披露经常性附加收入可稳定客户生命周期价值可能存在,但未单独量化
Galbot Store 运营不清晰:运营方、托管服务,或收入分成劳动力替代和门店吞吐是核心价值代理指标商业上可见,但会计处理不清晰
医疗和药房自动化未披露可靠性和受监管工作流可能支撑更高溢价的服务收入新兴垂直场景
数据集或模型授权公开信息未确认若真实存在,可能改善软件毛利仅属推测

收入流来自部署描述和公开产品界面的推断;Galbot 没有发布正式的收入结构、价格表或收入确认政策。

[CI008, CI009, CI010, CI011, CI012, CI013]
定价变现表
项目公开定价信号备注置信度
G1 机器人价格未披露Galbot 硬件没有公开 ASP
Galbot Store 经济性未披露运营方或托管服务逻辑可见,但未定价
工业部署费用未披露可能按站点和工作流协商项目价格
医疗部署定价未披露可靠性要求意味着服务范围可能更高端
外部价格锚点Unitree G1 价格 $13,500可作为低端背景,但不能直接对比 Galbot

投资人最需要价格可见性,但这里最薄弱;表格把 Galbot 价格披露缺失和外部市场锚点分开。

[CI008, CI009, CI013, CI014]

4.2 GTM 动作与销售效率代理指标

Galbot 的 GTM 看起来由企业销售和标杆客户驱动,而非自助式增长。公开牵引力体现在具名部署、战略投资者关系,以及工厂、医院和零售环境中的类案例叙事,这意味着高接触销售、更长试点和比软件式产品驱动增长更复杂的实施。大工业和医疗健康 logo 能建立信任和可复制性,这是优势;但销售效率只能从代理指标推断。公开证据显示 Galbot 围绕劳动力替代或吞吐经济性优化:据称一台机器人可以运营 50 平方米门店,并在三年内替代三个班次劳动力;医院药房工作流则引用 99.5% 处理成功率。这些数据点支持价值创造,却没有披露 CAC、回本周期、留存或扩张。GTM 含义是,Galbot 可能赢下高 ACV 客户,但投资者仍缺少判断获客是否高效、可重复且资本足够轻的指标,无法据此完全支撑当前融资规模。[CI010, CI015, CI016, CI017, CI018, CI019]

单位经济性代理指标
指标数值 / 估算来源置信度
单台机器人覆盖门店面积每台 50 平方米Galbot JS bundle 和二级评述
劳动力替代代理指标三年覆盖三个班次Galbot JS bundle 和二级评述
年度劳动力价值代理指标按每小时 $15 且满负荷使用,约 $131K / 年基于劳动力替代声明的外部测算
药房处理成功率99.5%ChinaTechNews
工业订单累计数千台China Daily 和公司关联报道
毛利率区间20-40% 合理但未验证仅基于行业背景

这些是代理指标和背景锚点,不是审计后的经济性;劳动力价值一行是明确的外部估算,毛利率区间是品类启发式判断,不是 Galbot 披露。

[CI017, CI018, CI019, CI020, CI023, CI034]

4.3 成本结构、利润率驱动因素与资本强度

Galbot 的成本结构几乎肯定比软件投资者本能想象的更偏硬件。一次人形机器人部署包含物料清单成本、机电组件、电池、传感器、算力、工厂人工、现场安装和持续服务支持。因此,毛利率不仅由定价能力决定,也受良率、利用率、保修负担和服务效率支配。公开来源没有披露 Galbot 毛利率,但更广泛的物理 AI 和机器人背景显示,其利润率区间很可能远低于纯 SaaS,并依赖制造成熟度。营运资本也是潜在拖累:机器人和零部件必须在库存和部署周期中先被融资,现金要到之后才能完全收回,尤其当企业买方谈判里程碑付款时。2026 年 3 月融资很可能正是为这些压力供血——制造扩张、模型开发和现场部署扩张——因此 Galbot 应被当作资本密集型物理 AI 公司承销,而不是轻资产 AI 软件供应商。[CI022, CI023, CI024, CI025, CI026, CI027]

FI002: 财务画像条形图

指数化财务画像显示,Galbot 今天在尽调中哪里最强、哪里最弱。

分数是归一化到 0-100 的定性尽调指数,并不代表经审计财务比率。

[CI013, CI020, CI023, CI026, CI029, CI032]

4.4 公开牵引力真实,但财务披露很薄

Galbot 的公开牵引力在运营层面亮眼,在财务层面偏弱。运营侧,来源指向截至 2025 年底覆盖 30 多个城市的零售存在、截至 2026 年初 100 多个药房或门店部署、累计数千台工业订单,以及药房工作流中高度具体的任务成功率声明。这足以说明 Galbot 不是只待在实验室的创业公司。但若没有订单到交付的转化、合同金额、验收时点或服务绑定细节,上述指标无法干净转化为已确认收入。公司未披露收入、ARR、EBITDA、毛利率、烧钱速度或客户集中度。即便利用率或劳动力替代叙事很强,也仍是代理指标,而非已确认财务结果。因此,正确框架是不对称的:部署证据有意义,但公开材料呈现运营远多于会计,收入质量仍未验证。[CI017, CI019, CI020, CI021, CI032, CI033]

财务缺口台账
指标公开可得缺口描述置信度尽调索取项
收入未披露年度收入或 ARR按垂直行业和季度索取月度确认收入
毛利率未披露按产品或服务线拆分的毛利率索取单位经济性和毛利率桥表
EBITDA 或经营亏损未披露盈利能力或烧钱速度索取管理报表和现金流摘要
客户集中度有具名客户 logo,但没有收入集中度数据索取前 10 大客户收入占比和 backlog
CAC 和回本期企业销售动作可见,但效率指标缺失索取漏斗、销售周期和回本数据
订单转化部分已引用数千台订单,但交付节奏未知对齐订单、验收台数和确认收入
现金余额和 runway公开信息披露了融资金额,但没有当前现金索取银行余额、烧钱速度和前瞻运营计划

这张台账把运营证明和财务证明拆开:许多头部部署声明已经公开,但承销需要的会计指标仍是私有信息。

[CI013, CI016, CI021, CI026, CI027, CI028]
FI003: 财务披露 KPI 快照

紧凑呈现公开披露内容,以及仍需尽调补齐的内容。

[CI004, CI006, CI020, CI032]

4.5 资本充足性与融资依赖

从资本充足性看,Galbot 按私营创业公司标准很强。融资路径从未披露金额的 2023 年种子轮和天使轮,升级到 2025 年 6 月人民币 11 亿元机构轮,再到 2025 年 12 月估值 $3B、金额超过 $300M 的融资,随后是 2026 年 3 月人民币 25 亿元融资,使累计披露资本达到约 $1.15B+。这个资本规模应能给 Galbot 留出有意义空间,用于扩大制造、商业部署和模型训练,短期不至于碰到融资悬崖。两个但书是烧钱和融资依赖。烧钱速度未披露;对一家研发和现场运营都很重的人形机器人公司来说,消耗可能相当大。融资依赖也不仅是数字问题,而是战略问题:Galbot 因国家和产业资本已经入局,看起来比许多创业公司暴露更低;但同一支持也可能把预期锁向国内战略目标。总体看,资本充足性不错,但在现金和烧钱披露前,跑道仍只能估算。[CI001, CI002, CI003, CI004, CI005, CI006]

融资轮次时间线
轮次日期金额领投方累计融资估值关键条款
Seed2023-06-01未披露未披露未披露未披露公开机构轮前的创始融资
Angel2023-08-01未披露未披露未披露未披露早期天使融资,公开信息未披露规模
Angel+2023-10-01未披露未披露未披露未披露2024 年规模化前的过桥融资
机构成长期轮2024-03-01数亿元人民币(估算)未披露公开口径尚未完全对齐未披露首个主要机构轮;公开金额仍为近似数
CATL 领投轮2025-06-25RMB 1.1B (~$153M)CATL Capital / Puquan Capital累计约 $500M独角兽(>$1B)共同投资方包括 China Development Bank、Beijing Robotics Industry Fund、Granite Asia
新一轮融资2025-12-01>$300M来自中国、新加坡和中东的投资方累计融资约 $800M$3B资金用于扩大部署和具身 AI 开发
National AI Fund 轮2026-03-02RMB 2.5B (~$350M)National AI Industry Investment Fund(Phase III,领投基金)累计融资超过 $1.15B未单独披露共同投资方包括 Sinopec、CITIC Investment Holdings、Bank of China、SAIC Financial Holdings

早期轮次公开披露但未披露金额,2024 年 3 月融资仍是媒体估算;后续行使用公司和新闻来源披露的金额及累计融资口径。

[CI001, CI002, CI003, CI004, CI005, CI006]
FI001: 融资瀑布图

从早期融资逐步推导 Galbot 截至 2026 年 3 月后的近似累计资本基础。

早期轮和 2024 年 3 月数字为估算值,因为只有后续轮次公开了规模;图表用于展示数量级,而不是经审计的累计募资。

[CI001, CI002, CI004, CI005, CI006]
FI004: 资本强度 / 现金流地图

融资如何转化为制造、部署,以及仍待回答的现金效率问题。

该流向图抽象呈现融资到运营的桥梁;现金余额、烧钱速度和营运资本周转仍未披露。

[CI007, CI024, CI025, CI026, CI028, CI036]

4.6 财务结论与尽调要求

Galbot 的财务图景是强资产负债表代理指标与弱披露质量的组合。多头案例很直接:公司融到足够重要的钱,在多个垂直领域有可信部署案例,并且看起来是在战略支持而非绝望中为制造和具身 AI 研发供血。空头案例同样清楚:投资者还无法验证收入、毛利率、烧钱、回本、客户集中度,甚至无法看清硬件、服务和运营商收入之间的拆分。这一点很关键,因为该品类已经面临真实买方需求质疑,并且相较 Figure 等美国同行存在严重估值差。财务结论因此是:资本充足性谨慎正面,透明度中性偏负,收入质量未解决。核心尽调要求不是再看一个愿景演示,而是一条带日期的财务桥:从订单到已交付单元再到已确认收入,并按业务线披露利润率和现金消耗。[CI026, CI030, CI031, CI032, CI033, CI034]

4.7 图表

Chapter 05

05产品与技术

5.1 用客户工作流定义产品

Galbot 的产品更应被理解为具身劳动自动化,而不是通用人形机器人展示。放到客户工作流里,G1 替代的是结构化室内环境中的重复性取放搬运、扫描分拣配送和导引服务任务;这些场景里,触达灵活性和连续运行时间比富有表现力的双足行走更重要。在药房,工作流是货架扫描、药品识别、取回和交接;在自主便利零售中,是 50 平方米紧凑门店里的补货、拣单、无人收银支持和夜间运营;在工厂中,是常规搬运、零部件移动,最终走向精密装配。因此,这台机器人位于移动操作臂与人形服务工之间:双臂、躯干升降和大垂直工作空间让它能操作为人类设计的货架和柜台,轮足底盘则优先保证稳定性和续航。这个框架很重要,因为 Galbot 的公开部署强调的是狭窄但经济性可读的工作,而不是无约束家庭自主。[CE001, CE002, CE008, CE026, CE027, CE030]

FE001: 产品模块流

从客户环境进入 Galbot 具身 AI 模块和运营任务的工作流地图。

该流程抽象的是公开产品叙事,而不是厂商发布的系统图;它反映了官方描述和部署报道所暗示的工作流角色。

[CE001, CE009, CE012, CE013]

5.2 硬件平台与运行架构

Galbot 官方材料给出了异常具体的 G1 硬件边界。平台标准姿态高 1730 mm,躯干可升降 650 mm,垂直工作空间覆盖 0–2100 mm,手臂长 710 mm、臂展约 190 cm,可覆盖标准人类腰部上下的货架和柜台。双臂载荷列为 5 kg,足够搬运药盘、零食袋、瓶子和轻型工业零件,而非重型制造负载。电力来自 48V 30Ah 锂电池,宣称最长运行 10 小时,并配套 WiFi、Ethernet、USB 和云集成,用于机队监督。最影响部署经济性的设计选择是轮足移动结构:Galbot 似乎在优化稳定室内导航和更长续航,同时通过躯干升降和长臂几何保留人类空间触达能力。6.25 英寸触摸屏提供本地操作员交互,多模态视觉、触觉和深度感知则支撑硬件之上的具身控制栈。仍有一个规格冲突:官方套件引用约 92.5 kg 机身重量,而二级评测有时引用 85 kg。[CE002, CE003, CE004, CE005, CE006, CE007]

G1 硬件规格
参数数值备注对比背景
高度(标准姿态)1730 mm官方产品 bundle 和产品页都指向约 173 cm 站立高度。人形尺度服务机器人,尺寸适配标准货架和柜台。
躯干升降650 mm大行程躯干升降扩大了高低货架触达,不改变底盘占地。室内零售和药房工作更看重这一点,而不是腿部表现力。
臂长710 mm官方 bundle 已披露。长臂双臂几何弥补了稳定轮式底盘的触达不足。
垂直工作空间0–2100 mm官方 bundle 给出标准范围,并称部分姿态可触达更高位置。覆盖地面料箱到高位零售货架。
双臂负载5 kg负载看起来针对物品搬运,而非重型装配。适合 SKU、托盘、瓶装物和轻型零件。
电池48V 30Ah 锂电池电池规格来自官方;循环寿命和热插拔设计未披露。支撑最长 10 小时的标称运行时间。
运行时长最长 10 小时公司宣称的续航;未披露 duty-cycle 假设。单班次室内运营具备竞争力。
防护等级IP54仅提供基础防尘和防溅水能力。低于更严苛工业或临床清洁制度常要求的认证深度。
传感器视觉、触觉、深度模态已经公开,但传感器供应商和冗余层未披露。足以支撑抓取、跟踪和避障感知操控。
连接WiFi、Ethernet、USB、云集成指向车队管理定位和本地接口选项。降低接入联网零售和企业网络的 rollout 难度。
机身重量官方约 92.5 kg;部分评测约 85 kg建模楼面载荷或运输前,需要先解决不同来源的重量冲突。影响搬运、移动能耗和安全规划。

各行只汇总截至本次报告日期已公开披露的规格。二级来源与官方 bundle 不一致时,先列官方数值,并在备注中披露差异。

[CE003, CE004, CE005, CE006, CE007]
FE002: 技术栈图

分层展示公开 Galbot 技术栈中的传感器、控制、基础模型和开发者 / 机群界面。

Galbot 尚未发布标准技术栈图,因此本图根据官方资料包、开发者门户和相互印证的新闻报道重构架构。

[CE005, CE014, CE015, CE016, CE018]

5.3 AI 栈、数据引擎与开发者界面

Galbot 的软件故事围绕垂直化视觉-语言-动作模型展开,而不是单一通用基础模型。GraspVLA 被定位为核心具身抓取模型,训练基础是数十亿次模拟交互,并由 DexGraspNet 规模抓取数据支撑;商业承诺是对未见过的物体和任务实现零样本处理。TrackVLA 把技术栈延伸到导航和跟随行为:通过视觉跟踪人或物体、接受语音指令,并在目标短暂丢失后重新捕获。GroceryVLA 则把抽象收窄到零售操作,声称可在杂乱门店中处理可变形零食袋、刚性瓶子和易碎罐子,无需逐 SKU 重新编程。跨这些模块,Galbot 描述了一套“大脑-小脑-神经控制”架构,把感知、决策和低延迟反馈控制压缩进比传统机器人栈更端到端的管线。数据护城河主张建立在 100 亿+ 数据点和 Sim2Real 闭环之上;该闭环大量依赖合成数据生成,据称使用 NVIDIA Isaac Sim,随后进行有限真实世界微调。公开开发者平台说明真实集成界面存在,但公开文档仍比全球领先具身 AI 同行更薄。[CE009, CE010, CE011, CE012, CE013, CE014]

AI / 软件栈
组件描述宣称能力差异化证据质量
GraspVLA端到端具身抓取基础模型无需额外训练即可 zero-shot 泛化到新物体和新任务把大规模仿真预训练与自有机器人执行闭环配在一起中高:官方和行业媒体相互印证,但没有公开 benchmark 套件
TrackVLA导航和目标跟踪模型跟随人或物,接受语音指令,遮挡后恢复跟踪在复杂空间里把移动能力和意图跟随接起来中:公司材料有描述,但外部技术细节有限
GroceryVLA零售专用操控模型无需逐件重编程即可处理柔性、刚性和易碎商品针对真实门店库存做垂直专门化,而非泛化机器人 demo中高:官方描述和部署报道支撑
Brain-cerebellum-neural control architecture(脑-小脑-神经控制架构)感知、决策、控制一体化栈把多模态输入转成低延迟动作闭环相比传统模块化机器人栈,声称端到端控制程度更高中:架构停留在营销层描述,没有系统论文
开发者平台公开开发者门户和手册提供集成和二次开发界面表明 Galbot 想做伙伴可扩展性,而不是只卖封闭设备中:门户存在是公开事实,API 深度仍不清晰
Sim2Real 数据引擎合成预训练加少量真实世界微调减少人工重标注,加快迁移到新场景若仿真质量足够高,可能形成数据飞轮优势中高:有独立报道,但未 benchmark

证据质量取决于每项声明有多少一手技术细节支撑、多少来自二级报道。目前没有公开 benchmark 仓库或可复现实验框架评估这些模型。

[CE009, CE010, CE012, CE013, CE014, CE016]

5.4 部署成熟度、可靠性与支持信号

公开部署证据显示 Galbot 已走出实验室演示,但还没有发布能证明工厂级可靠性的机队运营台账。最强量化证据来自医疗健康:北京药房部署的媒体报道提到 10 多个运营点、24/7 运行模式和 99.5% 药品处理成功率;零售报道则称单台 G1 可自主运营 50 平方米门店,铺设目标扩展到 100 多个药房或门店点位。这些是有用的成熟度信号,因为它们意味着机器人反复接入有货架、SKU 和员工工作流的真实环境。工业成熟度更像有潜力,而非已被证明。Bosch 和 UAES 合资公告表明,流程制造和汽车伙伴认真看待 Galbot,但公开文件仍未发布产线级吞吐、MTBF、恢复时间或服务人员配比。竞争含义是,Galbot 今天在半结构化室内运营中看起来可信;但 99.9%+ 工业准确率的路线图可信度,仍取决于围绕正常运行时间、异常处理和部署支持工具的私下尽调。[CE019, CE020, CE025, CE026, CE027, CE028]

部署场景
垂直行业用例部署阶段已验证指标证据质量
药房货架扫描、药品取放、引导配送已在北京 10+ 家药房运营引用药品处理成功率 99.5%;24/7 运行中高:多家独立报道,但没有官方案例仪表盘
自主零售 / Galbot Store在紧凑空间内补货、拣选、运营门店商业化 rollout单台机器人可运营 50 平方米门店;引用 100+ rollout 目标中:公司和二级报道支撑,第三方财务验证有限
即时零售仓连续拣选和库存移动运营中 / 规模化试点融资新闻稿称稳定 24/7 运行超过一年中:官方 PR 声明,缺少站点级利用率数据
汽车 / 复杂装配常规作业和未来装配自动化试点 / 合资扩张没有公开吞吐数字;已披露 Bosch 和 UAES 合作中:伙伴背书提升可信度,但量化披露少
电池制造CATL 关系牵头的常规工厂作业试点到早期商业化组合层面声称拿到数千台工业订单中高:官方融资稿加外部报道
医院服务流程病房支持、药房、导诊系统合作 / 试点点名宣武医院合作,但没有公开 SLA 指标中:官方稿确认范围,但未披露结果

阶段判断反映目前可见的最强公开证据,不代表私人合同状态。除药房成功率和宽泛铺设数量外,量化指标很少,因此证据质量仍低于成熟工业自动化供应商通常会披露的水平。

[CE008, CE019, CE020, CE026, CE027, CE029]
路线图和里程碑
事项状态目标日期证据风险
GraspVLA 发布已交付2025-01-01Robotics & Automation News 报道,并出现在公司技术口径中虽然目标很高,公开基准透明度仍有限
药房覆盖超过 10+ 门店铺设中2025-12-31Aparobot 和后续铺设报道提到 100+ 目标跨站点扩张运营和合规,可能比试点更难
Bosch BOYIN 工业联盟已签约 / 实施阶段2025-06-01融资报道和合作方材料宣布 JV工厂经济性和产线级 KPI 仍未披露
UAES RoboFab 汽车实验室已启动2026-03-01国家背书和融资报道提到该实验室实验室活动还不能证明已实现多站点生产部署
两年内实现工厂车间人形机器人商业化管理层目标2027-07-01TechNode 和 China Daily 引述公司领导层说法激进目标要靠准确率、安全性和服务可靠性追上公司宣称

目标日期来自公开里程碑锚点或管理层表述,不是经审计的交付承诺。风险聚焦已宣布合作或发布,与可重复生产经济性证据之间的落差。

[CE009, CE019, CE020, CE028, CE029, CE030]
FE003: 按场景划分的部署成熟度

基于已披露运营证据,比较 Galbot 公开部署场景的相对成熟度。

分数是分析师基于公开部署证据数量给出的 1–4 成熟度判断:4 = 有指标的重复运营;3 = 持续部署但指标稀疏;2 = 具名试点或 JV;1 = 仅概念。

[CE026, CE027, CE029, CE030, CE038]

5.5 差异化、安全、合规与路线图可信度

Galbot 的主要差异化主张不是单一组件,而是栈级组合:自有数据、具身基础模型、机器人硬件,以及进入零售、医疗健康和制造部署环境的通道。与 Peking University、BAAI、Bosch 和 UAES 的伙伴关系强化了这样一种判断:Galbot 在搭建数据与制造飞轮,而非一次性产品。2025 World Humanoid Robot Games 和药品分拣挑战的第三方验证,增加了一些信号,说明该技术栈能泛化到基准任务。尽管如此,安全与合规只被部分去风险。IP54 有用,但只是有限防尘防水,并不能替代详细医疗、工厂或隐私认证。中国 2026 年 3 月人形机器人标准和 2026 年 5 月机器人数字 ID 制度会抬高基础合规义务;法律评论也指出责任、自主性和数据处理仍有未解决问题。最尖锐的信任缺口是隐私:一份负面报告明确指出,Galbot 没有解释药房场景中的患者个人可识别健康信息如何被保护。这个遗漏不否定产品,但意味着更广泛医疗健康渗透的路线图可信度,仍取决于尚未公开的治理和安全披露。[CE021, CE022, CE023, CE024, CE031, CE032]

信任 / 质量 / 合规表
控制 / 认证 / 质量指标状态范围缺口
IP54 防护等级已公开披露机器人外壳耐轻度粉尘和泼溅不能替代详细医疗认证或严苛工厂认证
药品处理成功率药房场景引用 99.5%医疗 / 药房拣选流程方法论和样本量未公开披露
中国人形机器人标准适用于 2026 年监管制度国家安全 / 合规基线Galbot 具体符合性文件未公开
机器人数字 ID 注册中国自 2026 年 5 月起要求机队注册和可追溯性运营合规流程未公开描述
隐私和患者数据控制未公开详述医疗部署PHI/PII 安全架构仍是重大尽调缺口

本表区分已披露控制和缺失披露。有意把公开隐私和认证细节缺位视为缺口,而不是推定已经合规。

[CE006, CE026, CE031, CE032, CE034, CE035]
FE004: 产品成熟度 / 能力地图

覆盖 Galbot 主要模块和部署环境的能力成熟度视图。

评级是基于截至运行日公开证据作出的定性判断。强,意味着已有实际部署或详细规格支撑;弱,意味着披露大多停留在营销层面。

[CE013, CE018, CE026, CE030, CE038]
Chapter 06

06客户

6.1 客户群分层与公开足迹

与其用经典 SaaS 客户列表理解 Galbot,不如用买方、用户和付款方角色来拆。工业制造中,买方和付款方是大型企业或战略伙伴,例如 CATL、Bosch 关联实体、BAIC、SAIC、Toyota 相关客户和 UAES;日常用户是工厂操作员、生产团队和自动化工程师。医疗健康中,医院或药房运营商是买方,药剂师和支持人员是用户,直接运营受益者是患者工作流。零售中,Galbot 通过 Galbot Store 和 Galaxy Space Capsule 式自主便利业态,部分扮演自己的标杆客户,在一些点位既是供应商也是运营方。公开证据显示,所有垂直领域都集中在中国:具名部署、国家背书报道、监管语境和城市级铺设引用都以中国为中心。这给了 Galbot 一条连贯的本土市场切入点,但也意味着当前客户基础在地理和政策制度上比垂直列表乍看起来更集中。[CU001, CU002, CU003, CU022, CU024, CU035]

按垂直行业划分的客户分群
垂直行业代表客户部署状态单元量代理指标收入模式
工业制造CATL、BAIC、SAIC、Toyota、Mercedes-Benz、Zeekr、Bosch/UAES 生态试点到早期商业化扩张数千台工业订单声称是主要代理指标机器人销售加部署 / 服务,可能由合作方协助
医疗Xuanwu Hospital、北京药房运营方运营站点加旗舰合作10+ 药房;医院范围公开但未量化部署合同和服务 / 支持收入
零售 / 便利Galbot Store、Galaxy Space Capsule 网络已运营并扩张中2025 年 30+ 城市;2026 年 20+ 城市 100+ 单元公司自营门店和 / 或托管自动化解决方案
仓储 / 物流未具名自主仓储客户已运营,但归属披露稀疏引用超过一年 24/7 运行;未披露地点数量部署加持续运营 / 支持

分群依赖公开部署叙事,而非披露的收入拆分。单元量代理指标采用各垂直行业最强的公开数量,不能解读为按收入加权的份额。

[CU001, CU003, CU015, CU021, CU022]
FU001: 客户部署柱状图

按垂直场景展示公开可识别的部署足迹,每个板块采用目前最强的单位数或账户代理指标。

不同柱子代表不同公开代理指标,并非同一套标准化分母。图表想说明证据密度落在哪里,而不是直接比较各垂直场景收入。

[CU003, CU011, CU013, CU016]

6.2 采用轨迹与部署台账

Galbot 的采用轨迹通过部署数量可见,而不是通过已披露收入或 cohort 指标可见。最重要的商业信号是 2025 年 12 月融资稿称,来自 CATL、Toyota 和 BAIC Group 领衔工业客户的累计订单达到数千台。这个说法足够大,意味着真实管线,而不是少数试点;不过约束性订单、框架协议和分阶段铺设的组合并未公开。零售方面,公司称 Galbot Store 到 2025 年底已扩展至 30 多个城市,之后报道又指出到 2026 年 3 月,20 多个城市已有 100 多台。医疗健康方面,北京至少 10 家药房投入运营,并引用了 99.5% 药品处理成功率和 24/7 运行。仓储又提供了另一个耐久性代理指标:官方融资表述称已稳定全天候运行超过一年。合在一起,这些信号显示公司正从展示性安装走向可重复部署模板;但公开采用情况仍需谨慎解读,因为 Galbot 没有披露利用率、单机器人经常性收入,或从试点转向扩展机队的转化。[CU005, CU011, CU012, CU013, CU015, CU024]

采用指标台账
指标数值 / 估计日期置信度缺口
累计工业订单数千台2025-12-16中高绑定订单和分阶段框架的混合未公开拆分
零售城市覆盖30+ 城市2025-12-16未附带同一日期的单元数量
零售单元覆盖20+ 城市 100+ 单元2026-03-02二手报道;未拆分自有站点和第三方站点
北京运营药房10+2026-03-14中高精确门店名单和重复经济性未披露
药品处理成功率99.5%2026-03-14方法论和样本量未发布
仓储场景连续运行超过一年 24/72025-12-16地点和停机日志未披露
募集资本总额累计 $800M+2026-03-02资本是信心信号,不是直接客户指标
公开留存披露NRR、GRR 或流失率均无2026-06-14重大客户耐久性缺口

台账混合直接采用指标和一条明确未披露记录,因为留存数据缺位会实质影响本章。除非来源本身使用“数千”等近似表达,否则不引入估计。

[CU005, CU011, CU012, CU013, CU015, CU018]
FU002: 采用 / 部署漏斗

从宽泛商业主张一路下探到较小一组公开量化部署证据。

该漏斗衡量公开证据质量,而不是内部销售漏斗转化率。它显示 Galbot 具名关系不少,但有独立可审计商业指标的部署很少。

[CU016, CU023, CU032, CU038]

6.3 具名客户证据与证据质量

当 Galbot 或高可信融资报道明确把客户 logo 与真实工作流相连时,具名客户证据最强。CATL 是最清晰的工业锚点,因为它被描述为领投方和客户,并且关系中包含工厂常规运营和大规模累计订单。Xuanwu Hospital 是最清晰的医疗健康锚点,因为公司公开把病房、药房和医院导诊纳入这项合作。BAIC、SAIC 和 Toyota 是有意义的 logo,但证据权重并不相同;它们主要出现在融资或画像报道中,被列为具名下单或对齐的工业客户,而非深度案例研究。Mercedes-Benz 和 Zeekr 出现在 TechNode 关于本地工厂轮式机器人的报道中,但缺少详细运营说明。Bosch 和 UAES 最适合被视为混合伙伴-客户通道:合资合作验证了市场需求和工厂入口,但公开材料尚未证明这些实体已形成常态化机队采购。零售证据很特殊,因为 Galbot 自有门店既是部署证据,也是部分自运营渠道;这改善了运营反馈闭环,但在集中度分析上弱于独立客户 logo。[CU004, CU006, CU007, CU008, CU009, CU010]

具名客户部署
客户垂直行业部署类型规模结果指标证据质量日期
CATL电池制造面向生产的常规工厂作业战略客户;属于数千台工业订单池的一部分没有公开站点 KPI;最强证明是投资方—客户绑定关系存在的证据强;运营细节证据中等2025-12-16
Xuanwu Hospital医疗覆盖病房、药房、导诊系统的医院合作具名旗舰机构工作流范围已确认;没有发布 SLA 仪表盘具名证明强;量化结果证据中等2025-12-16
北京海淀药房医疗 / 药房运营中的药房机器人北京 10+ 运营站点药品处理成功率 99.5%;24/7 运行中高:独立媒体加多次引用2026-03-14
Bosch / BOYIN alliance(产业联盟)工业制造JV 牵头的工厂自动化扩张平台渠道,而非已确认机队数量没有公开吞吐量 KPI中:合作方信号强,采购细节透明度低2025-07-03
UAES / RoboFab汽车制造具身 AI 制造联合实验室实验室启动 / 扩张载体没有公开机队或生产率 KPI中:项目具体,但运营深度仍早期2026-03-02
BAIC Group汽车制造具名工业订单客户纳入数千台订单叙事未披露站点 KPI中:仅在融资 / 画像报道中反复出现2025-12-16
SAIC Motor汽车制造具名工业订单或对齐客户2026 年报道提及未披露站点 KPI中:仅二手报道2026-03-02
Toyota汽车制造具名工业订单客户纳入官方订单名单未披露站点 KPI中高:官方点名,但无案例研究2025-12-16
Mercedes-Benz汽车制造工厂机器人部署提到本地工厂使用没有公开量化结果低中:单篇二手报道2025-06-25
Zeekr汽车制造工厂机器人部署提到本地工厂使用没有公开量化结果低中:单篇二手报道2025-06-25
Galbot Store / Galaxy Space Capsule(零售空间)零售 / 便利公司自营自主门店网络2025 年 30+ 城市;2026 年 20+ 城市 100+ 单元披露了门店覆盖和铺设数量中:运营证明真实,但部分是自我客户证据2026-03-02

本清单收录截至运行日公开点名的部署,或与合作方相连的运营场景。它并不穷尽全部客户,因为 Galbot 不发布标准客户总账;部分 logo 只出现在融资或画像报道里,没有独立案例研究。

[CU004, CU005, CU006, CU007, CU008, CU009]
具名客户证明表
客户细分部署 / 用例生产 vs 试点结果局限
CATL工业制造常规工厂作业和战略订单项目面向生产的早期商业化数千台订单声称的一部分没有公开工厂级 KPI 或续约数据
Xuanwu Hospital医疗病房、药房、导诊合作试点到早期生产具名旗舰医疗部署没有公开 SLA 或扩张细节
北京药房运营方医疗药品取送和药房自动化运营中的生产站点引用 99.5% 处理成功率;24/7 运行方法论未披露
Galbot Store / Galaxy Space Capsule(零售空间)零售自主便利零售网络生产铺设先到 30+ 城市,后到 20+ 城市 100+ 单元部分自营,独立性较弱
Bosch / UAES 渠道工业制造JV 牵头的工厂扩张和汽车实验室试点 / 渠道建设验证了需求和进入工厂车间的通道尚未证明标准化机队采购

这张面向验证器的枚举表,把最强具名客户证明整理成标准形态。它补充上方更宽的用户要求具名部署表,而非替代。

[CU004, CU010, CU011, CU013, CU016, CU033]
FU003: 客户证据矩阵

矩阵展示 Galbot 在当前垂直场景中哪里有具名客户、规模信号和量化结果。

评级是定性判断,只反映截至运行日可获得的公开证据。「强」意味着公开记录中直接具名且量化;「中等」意味着有一定支撑,但还不足以完成商业验证。

[CU023, CU032, CU037]
FU004: 客户旅程图

从战略关系到现场部署,再到潜在队列扩张的高层路径。

这是基于公开部署叙事重构的概念性旅程图;并非供应商发布的销售流程图。

[CU022, CU028, CU029]

6.4 留存代理指标、扩张动作与渠道动态

Galbot 不披露 NRR、GRR、logo 留存、流失率或合同期限数据,客户黏性只能从结构性信号推断。最正面的代理指标是运营连续性:官方材料提到仓库 24/7 使用超过一年,药房部署也在继续,说明至少部分客户没有在试点后撤掉机器人,而是把它们留在工作流里。第二个代理指标是战略绑定。CATL 既是投资方又是客户,关系横跨资本、背书和工厂使用场景,可能加深锁定;但这种亲近也带来关联方风险。扩张动作看起来走三条渠道:直接向标杆工业和医疗客户做企业销售;借 Bosch 和 UAES 进入制造业;自营零售业态,先验证自身经济性、积累数据,再把模板向外销售。药房从 10 多个点位推向 100 多个点位的目标,也暗示了“先落地、再复制”的打法,只要每个已验证工作流能复制到更多地点。仍然缺失的是商业质量披露:公开信息没有说明续约时点、机队增购率、软件附加销售,或已安装基数中有多少是付费生产、有多少是补贴性的战略铺设。[CU018, CU019, CU026, CU027, CU028, CU029]

客户集中度和风险
因素评估证据
CATL 关联方集中度头部投资方与头部工业客户的关系,在官方和独立报道中反复出现
地域集中度截至运行日,所有主要公开部署和具名 logo 都以中国为中心
垂直行业集中度中高虽然有医疗和零售多元化,工业制造似乎仍是最大的订单池
留存披露风险没有公开 NRR、GRR、流失率或续约数据
证据质量风险多个 logo 只出现在融资报道中,而非详细案例研究
需求成熟度风险中高独立负面报道称,全行业真实买方需求和用例仍有限

评估是基于本章最强公开证据作出的分析师判断。风险等级只给方向;一旦有集中度和续约记录的数据室指标,应以那些指标替换。

[CU020, CU024, CU030, CU031, CU038]

6.5 集中度风险与负面信号

客户故事仍有吸引力,但风险还没有完全出清。CATL 同时是 Galbot 最具战略意义的投资关系,也是最清晰的工业客户锚点,这带来集中度和治理问题;对一家私营公司来说,公开记录还回答不了。公开部署也高度偏向中国,Galbot 因此暴露在单一监管环境、单一人才生态和单一人形机器人早期采用市场里。独立负面报道让这种谨慎更尖锐:TechXplore 认为,可用生产场景仍然有限,人形机器人供给可能跑在真实买方需求前面;CNBC 也把这个行业描述为投资热、商业化尚不成熟。这些批评与 Galbot 的证据形态相吻合:公开客户标识和铺设数量真实存在,但详细机队经济性、续约、经第三方验证的生产率结果仍然稀疏。多垂直场景和伙伴渠道确实降低了一部分单一市场风险;但相对于 Galbot 融资轮次和制造叙事隐含的规模野心,数千台订单说法仍显早期。简言之,采用动能真实存在,但集中度和商业化质量还没有达到后期工业平台投资者想看到的证明水平。[CU017, CU020, CU021, CU025, CU030, CU031]

Chapter 07

07风险

7.1 风险概览与排序

Galbot 的风险画像并非由某一个生死缺陷主导。公司融资可信、部署可见,又站在国家市场顺风上;但这些正面因素也会遮住一个事实:人形机器人经济性要真正站稳,许多依赖项必须同时跑通。最严重的风险落在政策、商业化和集中度的交叉处。中国的标准体系和数字 ID 制度长期可能帮助可信供应商,但短期会制造具体的合规关口和召回义务。与此同时,公开报道仍显示,赢下买方比造出机器人更难,因此即使技术演示好看,规模假设也可能落空。再叠加 CATL 集中度、硬件资本强度高、收入指标不披露,下行情形就是累积式的,而不是孤立风险。[CR001, CR002, CR014, CR020, CR026, CR027]

缓释措施与否决标准表
风险可监测指标触发项 / 阈值重要性行动含义
监管合规拖累数字 ID 注册和标准认证的公开证据商业部署在 2026-2027 年扩张时仍没有清晰合规证明将意味着政策风险在限制规模化,而不是推动规模化视为重大尽调阻碍
CATL 集中度可见部署或收入中与 CATL 绑定的占比CATL 实质性减少订单、试点范围或投资人支持会同时打击收入证明和融资信号大幅降低估值容忍度
买方需求疲弱旗舰参考之外具名的复购客户机器人仍停留在展示型部署,缺少广泛续约或扩张将表明需求没有复利增长按试点较重的硬件公司承销,而非可扩展平台
安全 / 网络事件重大现场故障、召回或安全漏洞新数字 ID 制度下出现一起严重事件可能同时触发责任、召回成本和需求犹豫暂停投资,直到根因和响应清楚
单位经济性不及预期机器人无法可靠替代目标劳动班次的证据支持成本或运行时间短板侵蚀客户 ROI将削弱需求和利润率假设转入熊市情景框架
政治 / 出口冲击关键算力或出口渠道受限新出口管制摩擦影响性能路线图可能拖慢模型迭代和国际化可选项降低对长期倍数扩张的信心

这些触发项都写成可观察指标,好让本章直接进入投资、等待或放弃决策,而不是停留在泛泛提醒。

[CR008, CR020, CR031, CR032, CR038, CR039]
FR001: 风险热力图

序数矩阵按发生可能性、影响、缓释成熟度和剩余暴露,对 Galbot 的主要风险桶排序。

评分是截至 2026-06-14 基于引用证据综合出的承保序数判断,而不是预测概率。

[CR001, CR003, CR013, CR014, CR016, CR020]
FR004: 风险类别柱状图

集中度和监管与尚未证实的规模经济性叠加时,严重程度最高。

柱值是委员会式的 1-5 序数严重程度评分,不是概率。

[CR010, CR020, CR026, CR029, CR040]

7.2 监管、法律与安全风险

监管图景对 Galbot 异常重要,因为中国政府不只是旁观人形机器人,而是在围绕它们建立可执行框架。2026 年 3 月国家标准体系和 2026 年 5 月数字 ID 制度,让人形机器人更接近一个受治理的工业产品类别;这有利于长期市场发展,但会让准备不足的供应商付出高成本。法律风险比合规清单更宽。Hill Dickinson 的分析有说服力,因为它把责任、隐私和问责视为未解问题,即使在完全自主到来之前也是如此。如果机器人在药房或医院弄伤员工、误识别人、误处理敏感数据,责任可能横跨制造商、运营方和软件栈。也就是说,一起重大事故可以同时触发商业犹豫、监管审查和直接成本。[CR001, CR002, CR003, CR004, CR005, CR006]

监管 / 法律风险登记表
司法辖区 / 议题当前状态要求 / 暴露Galbot 合规姿态缺口严重性
中国国家人形机器人标准体系2026 年 3 月宣布框架制造商必须对齐安全、伦理、测试和互操作性预期Galbot 受益于国内对齐,但仍要完成落地工作未披露公开认证包或合规路线图
中国机器人数字 ID 制度2026 年 5 月起运行注册成为市场准入和可追溯要求Galbot 在中国运营,最终应登记已部署单元还没有单元级注册或召回流程的公开证明
缺陷召回和转售限制嵌入数字 ID 制度缺陷单元可能需要召回;翻新或转售受约束抬高硬件缺陷和服务失误成本未披露公开缺陷准备金或召回就绪情况
事故后的责任分配法律分析仍未定型造成损害后,制造商、运营方和软件提供方都可能被索赔Galbot 混合 B2B 环境让运营方与制造商责任更难切分未披露公开赔偿或保险框架中高
隐私 / 生物识别处理人形机器人可处理职场和客户数据跨境隐私和生物识别规则尚未统一医疗和零售部署让数据最小化变得重要未披露医疗部署的公开隐私架构中高
地缘政治技术管制芯片与市场准入仍高度受政治影响限制可能拖慢先进算力获取或出口增长Galbot 已谈到供应链多元化,但风险仍在模型训练层面尚无公开的多供应商缓释细节

登记表聚焦政策和法律约束:即使机器人本身看起来技术上可用,这些约束也可能卡住部署。

[CR001, CR002, CR003, CR004, CR005, CR008]
FR002: 风险传导图

主要下行路径从监管收紧和需求集中出发,传导到采用放慢、单位经济性走弱和融资压力。

[CR002, CR013, CR014, CR026, CR030, CR031]

7.3 运营可靠性与产品化风险

运营风险仍是亮眼原型叙事和韧性业务之间最大的桥。公开来源支持一个判断:人形机器人系统正在快速变好;但这些来源也显示,买方要看到可重复节省,前面必须有太多环节不出错。工厂部署需要极高准确率和正常运行时间,而具身 AI 的错误不会只是悄悄拉低仪表盘指标,可能直接变成物理事故。Deloitte 关于物理 AI 的警告在这里比在纯软件里更关键,因为幻觉和感知错误会让执行器在人周围移动。电池完整性、触觉感知和网络加固又增加更多层。TechXplore 和 Associated Press 的报道同样有价值,因为它们把当下商业瓶颈框定为需求与信任,而不是原始生产速度;这正是那种在大量资本花掉之后才会暴露的风险。[CR013, CR014, CR015, CR016, CR017, CR018]

运营 / 质量 / 安全风险登记表
失效模式重要性可能性影响当前缓释措施剩余风险
可靠性低于工厂级门槛准确率或运行时间达不到生产容差,工业 ROI 就会失效试点部署与全栈优化公开运行时间数据仍缺失
买方需求落后于硬件产出机器人造得比卖出或续约更快,规模经济就跑不出来标杆客户与国家背书下的可见度需求形成在大规模层面仍未被证明
具身 AI 幻觉或感知错误真实环境中,软件错误会变成安全事故中高仿真、测试与受限任务设计意外边界场景仍难彻底消除
联网机群遭网络攻陷远程攻陷可能造成数据泄露和物理安全事故企业级控制与托管环境未看到公开的安全保证报告
电池热失控或充电问题人形机器人靠近人群,电池起火和服务风险都会放大电池设计与标准锂电安全实践未披露公开事故或准备金
触觉感知与灵巧度瓶颈利用率靠细腻人类任务拉动,机器人可能仍做不好中高任务专门化与全栈软件迭代通用能力叙事可能跑在现场现实前面

运营条目强调会直接伤害运行时间、安全和真实客户价值的失效模式,而不是泛泛的制造业公司风险。

[CR006, CR007, CR013, CR014, CR015, CR016]

7.4 伙伴依赖与财务模型风险

公司最强的外部证明点,也同时是集中度风险。CATL 给了 Galbot 一个有分量的工业参考客户和融资信号,但也把需求可信度和投资者信心集中到一段关系上。Bosch 相关合作、国资背景资本和 NVIDIA 相关工具,让公司看起来战略连接紧密;但一旦条款变化、政治风向转变或平台访问收紧,每条关系都会压缩自由度。财务上,公司仍难以干净承销,因为公开证据给出了估值和融资数字,却没有可比的收入、烧钱速度或支持成本披露。这种组合会制造一个具体下行情形:如果买方需求比头条部署集合所暗示的更窄,Galbot 仍可能显得很突出,却需要在不确定条款下继续融资。因此,Galbot 的 $3 billion 估值与美国同行 Figure 更高标记之间的差距,应部分理解为风险折价,而不只是可选上行。[CR020, CR021, CR022, CR023, CR024, CR025]

合作伙伴 / 依赖风险登记表
依赖项合作伙伴 / 供应商依赖性质锁定程度可替代性风险等级
锚定客户 + 投资人集中CATL需求证明、资本信号和工业验证部分押在同一交易对手身上低中
制造 / 合资渠道Bosch 相关合资公司与生态伙伴合作伙伴可能影响分销、经济性与路线图对齐中高中高
国家资本与银行国家人工智能基金、中国银行、Sinopec、CITIC、SAIC 相关资本政策准入与融资纵深部分取决于政治对齐低中中高
算力与仿真栈NVIDIA Jetson Thor / Isaac 相关工具与云算力训练和开发流程可能依赖集中度较高的生态中高中高
中国机器人供应链国内执行器、传感器与集成生态规模化取决于持续供给和跨境部件准入
医疗与零售落地伙伴医院、药房和商业场所通用性证明取决于伙伴是否愿意把试点扩到生产部署

依赖图按单一外部参与方或平台对收入可信度和未来融资的直接伤害程度排序。

[CR020, CR021, CR022, CR023, CR024, CR025]
人员 / 执行风险登记表
因素描述严重程度缓释措施剩余风险
创始人集中王鹤兼具创始人、CEO 和资深学者角色技术可信度强,公众形象突出注意力分散可能拖慢运营节奏
全栈跨度模型、硬件、数据和商业化并行推进一体化架构可减少跨供应商摩擦并行下注过多可能削弱执行焦点
商业证明对技术证明参考部署已经存在,但规模化复购更不清晰工业和医疗客户标识增强可信度相比标杆成功,可重复性证明更弱
财务不透明收入、烧钱速度和单位经济性仍未披露大额融资轮争取时间经济性不透明可能削弱下一轮议价能力
政策绑定的增长路径国家支持可能加速国内采用中高国内生态优势真实存在政策依赖可能增加海外或纯商业扩张难度

执行条目拆出来自领导力集中、披露缺口,以及把试点可见度转成可重复规模的难题。

[CR010, CR011, CR026, CR027, CR028, CR033]
FR003: 依赖关系图

Galbot 最强的依赖,落在锚定客户、资本提供方、政策系统和算力平台的交叉点。

[CR020, CR021, CR022, CR023, CR024, CR039]

7.5 缓释因素、监测项与尽调问题

Galbot 并非没有保护。全栈架构可以降低产品最战略部分对外部供应商的依赖;国资背景投资者,加上工业、零售和医疗场景中的参考部署,也比许多更早期人形机器人创业公司打下了更强基础。监管框架也有有利的一体两面:清晰标准和数字身份工作流一旦内化,资本更充足的供应商可能受益于小竞争对手跨不过去的门槛。但这些缓释因素抹不掉核心尽调问题。投资者仍需要证据证明:合规工作流已经可运营;CATL 集中度没有压倒性;正常运行时间和服务成本能支撑替代劳动力叙事;一场严重现场事故不会级联成召回和融资问题。因此,正确姿态是带监测的信念,而不是一概怀疑。[CR033, CR034, CR035, CR036, CR037, CR038]

7.6 图表

Chapter 08

08估值

8.1 融资背景与入场纪律

评估 Galbot 的正确起点是当前融资标记,而不是传统的贴现现金流。公开证据给出一个强头条:2026 年 3 月,以约 $3 billion 估值融资超过 $300 million;此前融资看起来已把累计资本推到 $1.15 billion 以上。这足以把 Galbot 视为资本最充足的中国人形机器人公司之一,但不足以说明当前价格显然公允。公开来源仍未披露经审计收入、毛利率、烧钱速度,或决定头条投后估值能否转化为有吸引力普通股入场价格的详细股权结构。因此,入场纪律比品类热度更重要。新投资者应假设,结构、集中度和商业化时点至少与市场规模叙事同样重要。[CV001, CV002, CV003, CV011, CV015, CV030]

建议摘要表
维度评估决策含义
建议继续研究证据有吸引力,但以当前价格锚点看,信息仍太不透明,难以干净买入。
信心判断方向比精确价值区间更清楚,因为关键私营公司指标仍未披露。
风险评级政策、集中度、商业化和披露风险都仍在场。
估值立场偏高当前估值可用战略价值解释,但还不能靠已披露的运营证明支撑。
目标回报 / 持有4-6 年内需 >3x以 $3B 进入,该回报要求异常强的执行和条款纪律。
最可能的近期路径再做一轮私募融资,或结构化 IPO 前融资IPO 准备度仍需要经审计的经济性和更广泛的商业证明。

建议明确受价格影响,并反映战略前景与当前已披露运营证明之间的差距。

[CV001, CV011, CV029, CV030, CV031, CV035]

8.2 投资论点:为什么 Galbot 仍可能重要

Galbot 的多头逻辑相当强。中国正在成为人形机器人制造和出货规模中心,这会给能整合软件、硬件和客户入口的公司带来自然本土优势。Galbot 自身定位与这个机会一致。公司声称拥有全栈具身 AI 架构、大量自有数据资产和多层模型,而不是只有一台演示机器人。工业、零售和医疗场景中的公开部署证据也说明 Galbot 已越过纯原型阶段。再加上国资背景投资者、高知名度创始人和 CATL 相关工业验证,公司有理由主张,自己正在搭建通向中国企业采用人形机器人的内侧通道。如果这些要素在未来两到三年转化为可衡量的重复收入,当前估值最终可能比今天看起来更站得住。[CV003, CV004, CV005, CV006, CV007, CV008]

正反论点表
支柱牛市论点熊市论点需解决的问题
市场位置中国的出货领先和制造纵深可让 Galbot 比许多海外同行更快复利增长。国内市场拥挤,加上中国资产倍数较低,即使有规模也可能封顶上行空间。需要优先垂直领域中可持续份额的证据。
产品护城河全栈模型、数据和硬件可形成一体化学习循环。VLA 趋同可能比管理层预期更快压缩差异化。需要外部证据证明数据和模型优势能转化为更好的现场结果。
商业证明CATL 和其他部署说明 Galbot 已越过纯原型阶段。少数参考客户标识仍可能掩盖集中度和复购疲弱。需要复购订单和多客户扩张证据。
资本基础国资背景投资人提供韧性和政策准入。未知优先权和集中度可能让后进入投资人保护不足。需要股权结构、条款和治理细节。
团队质量王鹤的技术履历支撑具身 AI 叙事。商业化拓宽后,创始人集中会抬高执行负荷。需要组织纵深和运营节奏证据。
监管环境标准长期可能提高弱竞争者的门槛。合规、召回和隐私义务可能先拖慢价值兑现。需要合规准备度和医疗 / 隐私控制的实操证据。

本表把每个投资论点支柱与反论点并列,列出投资人必须解决的问题,才能激进承销当前价格。

[CV003, CV004, CV006, CV007, CV009, CV014]
FV001: 推荐逻辑

当前估值下,集中度与披露缺口抵消了产品和市场潜力,因此建议仍偏谨慎。

这张流程图把承保链条压缩成最可能推动委员会决策的几个变量。

[CV004, CV006, CV011, CV016, CV018, CV029]
FV004: 投资 KPI

KPI 面板说明 Galbot 为什么具备战略吸引力,也说明按当前估值仍未达到清晰买入门槛。

[CV001, CV002, CV008, CV010, CV011, CV029]

8.3 反论点:为什么当前价格仍可能过高

反论点的重点,不在于人形机器人是否重要,而在于投资者是否被要求过早为一个仍不透明的故事付钱。Galbot 的 $3 billion 估值没有公开收入或单位经济性披露支撑。品类中的需求形成仍是真实风险,TechXplore 的报道尤其强调这一点;而 CATL 周围的伙伴集中意味着,一个被高调庆祝的证明点也是单点脆弱性。合规、责任、网络安全和隐私问题也并不抽象;收紧的标准和数字身份系统最终可能帮助强供应商,但首先会抬高证明准备就绪的成本。竞争强度进一步削弱了清晰溢价论。Unitree 的定价、AgiBot 的存在、XPENG 的机器人野心,以及 Physical Intelligence 等软件优先平台,都说明战略稀缺性可能比私募估值所暗示的更快收窄。按今天的入场点,这足以让建议保持谨慎。[CV010, CV011, CV012, CV015, CV016, CV017]

论点破裂与否决触发项表
项目描述紧急程度未解决是否击穿论点
经审计收入证明提供经审计或董事会级别的收入、毛利率和烧钱披露。立即是,因为没有经济性就无法干净承销价格。
CATL 集中度披露合同条款、期限和依赖组合。立即是,如果一个交易对手实际上锚定收入和融资信心。
实际交付排期按主要部署展示实际与承诺的单位交付时间。是,如果出货与商业化叙事明显脱节。
股权结构表与优先权披露清算顺位、参与权和优先级。是,如果该结构在名义估值下明显压低新资金顺位。
合规状态说明数字 ID、隐私和医疗控制的准备度。是,如果监管可能中断关键部署。

这些触发项围绕会直接改变投资建议的问题,而不是泛泛增加谨慎提示。

[CV010, CV018, CV029, CV030, CV037, CV038]

8.4 可比公司组与情景区间

Galbot 需要一个混合估值框架。一端,Figure AI 官方 $39 billion Series C 显示,全球资本市场仍愿意给感知中的品类领导者分配多大资本。另一端,中国同行和上市公司可比对象显示,定价压力、市场折价和披露差异可以很快压缩这种乐观。因此,情景框架比单点估计更重要。多头情形假设 Galbot 成为清晰的国内工业领导者,拥有真实收入规模和更丰富的战略可选性。基准情形假设公司取得有意义进展,但集中度、披露和政策风险仍带来折价。空头情形假设商业化滑坡,或集中度和监管显著削弱未来融资杠杆。这些区间很宽,但宽度反映现实:今天的公开证据对方向的支持强于对精度的支持。[CV013, CV014, CV025, CV026, CV027, CV028]

牛市 / 基准 / 熊市情景表
情景概率关键假设隐含的 5 年价值驱动因素估值区间
牛市25%中国继续成为人形机器人商业化中心,Galbot 赢得工业领先地位,到 2028 年收入至少达到数亿美元。经营杠杆叠加国内品类领导者的战略溢价。$25B-$35B
基准50%Galbot 在两到三个垂直领域规模化,但披露只逐步改善,市场仍给集中度和政策风险打折。收入可见度更可衡量,治理证明也更好,但仍不完美。$10B-$15B
熊市25%商品化、合规拖累或 CATL 收缩阻止广泛规模化,并迫使融资条款更苛刻。下行保护更多取决于资产和战略可选项,而非爆发式增长。$1B-$1.5B

区间是讨论范围,不是管理层指引;锚点是里程碑推进,而不是单一收入倍数。

[CV012, CV026, CV027, CV028, CV031, CV037]
可比估值表
可比公司类型阶段估值($B)收入模式关键差异点估值倍数语境
Galbot私募轮成长 / IPO 前叙事3人形机器人硬件 + 具身 AI 部署中国全栈工业聚焦,并有国资背景资本头部私募轮估值标记,但没有公开收入披露
Figure AI私募轮C 轮 / 品类领导者39通用人形机器人平台同行中已披露的最大私募估值官方 2025 年 C 轮投后估值
AgiBot私营公司估算后期私募 / 扩张期4中国人形机器人部署国内出货可见度强来自市场报道的中点估算,不是官方价格
Unitree公私混合市场标记商业产品规模机器人硬件销售中国市场公布价格点激进是有用的定价锚点,而非已披露私募估值
XPENG公开可比公司上市电动车 / 机器人可选项汽车销售叠加机器人可选项披露深入,具备公开市场流动性使用基于申报文件的上市公司语境,而非直接迁移估值
Boston Dynamics / Hyundai战略现有玩家企业支持的商业化竞争者工业机器人与战略部署现有制造与商业化纵深战略可比,不直接迁移倍数
Physical Intelligence基础模型可比私营 AI 平台通用机器人模型平台显示价值可能流向软件优先的控制层叙事和融资语境,而非干净倍数

这组可比混合了已披露私募估值锚点、上市公司和战略可比公司,因为 Galbot 运营披露不足,无法使用公式化的单一倍数方法。

[CV001, CV013, CV021, CV022, CV023, CV024]
FV002: 可比公司定位

已披露的私募估值锚点显示,Galbot 与美国头部同行仍有多大差距;也凸显这些估值之下,公开运营披露有多稀少。

柱值为百万美元,排除了没有直接可比私募估值锚点的上市公司可比对象。

[CV001, CV013, CV014, CV033]
FV003: 估值情景区间

只有商业化里程碑和财务披露相对公开基线大幅改善时,Galbot 当前价格才显得有吸引力。

情景区间是委员会式讨论区间,单位为百万美元,基于里程碑假设搭建,而不是单一收入倍数。

[CV026, CV027, CV028, CV029]

8.5 建议、退出逻辑与最终尽调

本章结论是继续研究。这不是否定 Galbot 的战略位置,而是判断当前公开记录仍留下太多未解问题,无法把 $3 billion 标记称为有吸引力。公司有真实优势,包括资本厚度、可信的本土市场位置和可见部署;但缺失项正是入场质量最关键的部分:经审计经济性、股权结构条款、交易对手集中度、出货真实性数据集和详细合规姿态。合理持有期是四到六年,因为商业化成熟度和 IPO 准备大概率滞后于融资故事。建立信念的最佳路径不是更多叙事,而是更窄、更硬的证据。如果经审计财务、重复客户扩张、更干净的集中度调整后单位经济性出现,估值立场可能从偏贵走向公允。在那之前,最终尽调应承担大部分工作。[CV029, CV030, CV031, CV036, CV037, CV038]

最终尽调问题表
主题缺失证据重要性负责人 / 尽调路径
经审计收入与单位 P&L季度收入、毛利率、支持成本和现金消耗估值纪律取决于证明业务本身,而不只是证明品类。与管理层和审计师做财务尽调。
CATL 合同结构期限、定价、排他性和终止权客户集中会同时扭曲上行和下行。审查已签协议的商业与法律条款。
交付真实底表各站点实际出货与已宣布部署的对照投资逻辑需要真实规模,而不只是高曝光试点。运营尽调加客户访谈。
毛利率结构硬件毛利与软件或服务贡献决定规模会提升价值,还是只放大支持负担。财务与产品尽调。
员工数与烧钱速度按职能划分的员工数和月度现金消耗用于判断现金跑道和未来稀释压力。HR 与 CFO 尽调。
IP 与隐私状态专利地图、FTO 意见、医疗隐私控制两者若薄弱,都可能制造不易察觉的下行。法务与安全尽调。

把一个具备战略吸引力的私募轮次转化为可完整承销的投资决策,这些问题是最低证据包。

[CV011, CV015, CV018, CV019, CV030, CV038]

8.6 图表

附录 A: 方法论与来源覆盖

本报告完全基于 2025-06-01 至 2026-06-14 期间审阅的公开来源。主要来源包括 Galbot 官方网站(www.galbot.com 和 developer.galbot.com)、PRNewswire 新闻稿,以及中国 MIIT/SCIO/Xinhua 的监管表述。二级来源包括 TechNode、The Robot Report、CnTechPost、Yicai Global、China Daily、TechXplore、CNBC、TrendForce、Deloitte 和 Hill Dickinson。竞争对手数据来自官方公司页面(Figure.ai、AgiBot.com、Unitree.com、XPENG、Physical Intelligence)。未使用任何非公开文件、数据室或管理层访谈。

关键缺口:Galbot 不发布财务报表、收入、员工数、毛利率或单位经济性。订单被报道称为「数千台」,但没有交付时间表。医疗隐私合规姿态没有公开文件。所有财务估计均为分析师和媒体推断;没有可用审计数据。

免责声明

本报告仅基于公开可得信息,不构成投资建议。Galbot 未验证或背书任何内容。估计和预测反映作者对可得来源的分析;实际结果可能出现重大差异。本报告生成于 2026-06-14,随着新信息出现可能过时。

证据索引

结论
编号陈述可信度来源
CO001 Galbot was founded in Beijing on 2023-05-19. SO002, SO007
CO002 Public materials identify He Wang and Zhang Zhizheng as Galbot's founders. SO002, SO007, SO011
CO003 Galbot publicly places its headquarters in Beijing and lists R&D presence in Shenzhen, Suzhou, and Hong Kong. SO002, SO007
CO004 Galbot presents itself as an embodied-AI robotics company rather than a pure software startup. SO001, SO002
CO005 Galbot's flagship public product is the G1 robot for industrial, retail, and healthcare workflows. SO001, SO003
CO006 Galbot operates a developer portal and exposes technical product signals beyond a marketing-only website. SO005, SO006
CO007 Public product materials tie Galbot's commercial story to retail and healthcare task automation rather than consumer robotics. SO003, SO017, SO018
CO008 Galbot's public network includes research and clinical institutions such as PKU, BAAI, and Xuanwu Hospital. SO004, SO016, SO019
CO009 He Wang and Zhang Zhizheng remain the dominant named public leadership figures across reviewed sources. SO002, SO007, SO011
CO010 Reviewed public materials do not identify an independent board or a broad named executive roster beyond the founders. SO002, SO013
CO011 By late 2025 Galbot was being framed publicly as a unicorn-scale private robotics company at roughly a $3 billion valuation. SO013, SO014, SO020
CO012 Later company and media summaries say Galbot completed a seed round in June 2023. SO011, SO013
CO013 Later company and media summaries say Galbot completed angel and angel-plus rounds in August and October 2023. SO010, SO011, SO013
CO014 Later funding summaries say Galbot closed a several-hundred-million-RMB round in March 2024. SO010, SO011, SO013
CO015 Mid-2025 coverage described a roughly RMB1.1 billion or $151-$153 million Galbot financing associated with CATL-linked capital. SO008, SO009, SO015, SO016
CO016 The same 2025 financing wave also linked Galbot to Bosch's investment arm or a Bosch-related joint-venture cooperation path. SO008, SO015
CO017 Late-2025 reporting described Galbot as securing more than $300 million in new funding at about a $3 billion valuation. SO012, SO013, SO014
CO018 Company-linked reporting said Galbot's cumulative funding reached about $800 million after the late-2025 raise. SO013, SO014, SO020
CO019 On 2026-03-02 multiple sources reported that Galbot raised RMB2.5 billion led by the National AI Industry Investment Fund. SO010, SO011, SO021
CO020 Public coverage of the March 2026 round named Sinopec, CITIC, Bank of China, and SAIC among Galbot's backers. SO010, SO011, SO021
CO021 Public sources link Galbot to CATL, Mercedes-Benz, Zeekr, Bosch, Toyota, BAIC, SAIC, Xuanwu Hospital, and pharmacy-chain contexts. SO004, SO012, SO016, SO019
CO022 Company-linked coverage says Galbot has accumulated several thousand unit orders. SO012, SO013, SO016
CO023 Public coverage says Galbot's retail or pharmacy presence extends across more than 30 cities. SO016, SO019
CO024 Company-linked narratives describe Galbot as having accumulated more than 10 billion embodied-AI data points. SO012, SO013, SO020
CO025 In March 2026 Galbot G1 was publicly presented as an AI-powered robot pharmacist in Beijing. SO017, SO019
CO026 G1 is publicly positioned for retail and pharmacy automation tasks rather than a purely experimental humanoid showcase. SO003, SO017, SO018
CO027 China's March 2026 humanoid-robot standards announcement created a more formal regulatory backdrop for companies like Galbot. SO020, SO021, SO022
CO028 Galbot is benefiting from broader Chinese humanoid-robot policy and capital tailwinds rather than operating in isolation. SO020, SO022, SO023
CO029 Galbot's public valuation story depends more on strategic momentum and expectations than on disclosed financial performance. SO013, SO014, SO023
CO030 Reviewed public materials do not disclose detailed revenue, gross margin, or headcount figures for Galbot. SO001, SO002, SO013
CO031 No audited financial statements or equivalent public reporting package appeared in the reviewed source set. SO001, SO023, SO024
CO032 Key-person risk is material because Galbot's public identity and credibility remain tightly tied to its founders and a small named leadership set. SO002, SO004, SO007
CO033 Geopolitical risk is material because Galbot's capital base, policy support, and deployment ecosystem are heavily China-centric even as its reference customers include global automotive brands. SO020, SO023, SO024
CO034 Healthcare and retail robot deployments expose Galbot to real liability, safety, and compliance complexity. SO017, SO019, SO024
CO035 Galbot's partner and lab network strengthens technical credibility but does not by itself prove recurring commercial economics. SO004, SO016, SO019
CO036 Galbot appears to be building a broader platform stack around the robot rather than relying only on hardware demos. SO001, SO005, SO006
CO037 The March 2026 state-backed round suggests Galbot is being treated as a strategically important domestic robotics platform. SO010, SO020, SO021
CO038 Despite strong fundraising momentum, Galbot's commercial economics still need verification in later diligence chapters. SO001, SO013, SO023
CO039 Bosch- and CATL-linked relationships matter because they combine financing credibility with plausible commercialization channels. SO008, SO015, SO021
CO040 Pharmacy deployments are one of the strongest public proofs that Galbot has moved beyond concept-stage robotics demonstrations. SO017, SO018, SO019
CM001 Galbot's relevant market is enterprise humanoid systems for physical workflows rather than all robotics spending. SM012, SM020, SM025
CM002 The relevant spend boundary includes embodied-AI software, data, integration, and services when they are attached to productive humanoid deployments. SM011, SM012, SM020
CM003 Status-quo substitutes for humanoids include manual labor, fixed automation, cobots, and workflow-specific robots. SM009, SM011, SM012
CM004 IDC-based coverage says global humanoid robot shipments were about 18,000 units in 2025. SM017, SM018, SM019
CM005 IDC-based coverage says global humanoid robot hardware revenue was about $440 million in 2025. SM017, SM018, SM019
CM006 IDC-based coverage says 2025 humanoid robot shipments grew by about 508% year over year. SM018, SM019
CM007 Chinese vendors dominated 2025 humanoid commercialization, with China-based firms leading shipment rankings. SM007, SM018, SM019
CM008 IDC ranked AgiBot first and Unitree second in 2025 humanoid shipments. SM017, SM018, SM019
CM009 TrendForce says the humanoid industry enters a critical phase of commercialization in the second half of 2026. SM001, SM002
CM010 TrendForce says China's humanoid robot output could grow by as much as 94% in 2026. SM001, SM002
CM011 MIIT-linked coverage says more than 140 domestic Chinese humanoid manufacturers released more than 330 models in 2025. SM003, SM020, SM028
CM012 China released a national standard system for humanoid robotics and embodied AI in March 2026. SM003, SM004, SM028
CM013 The March 2026 standard system covers the full industrial chain and robot lifecycle, including safety and ethics. SM003, SM005, SM028
CM014 China launched a national humanoid robot digital ID system in May 2026. SM006
CM015 The May 2026 digital-ID framework assigns each humanoid robot a 29-digit code for lifecycle traceability. SM006
CM016 The new digital-ID regime includes a strict "no code, no market access" rule for robots sold or deployed domestically. SM006
CM017 Long-run humanoid market forecasts diverge widely, from UBS's $1.4-$1.7 trillion by 2050 to Morgan Stanley's $5 trillion by 2050. SM014, SM015, SM016
CM018 UBS's base case expects more than 2 million humanoids by 2035 and more than 300 million by 2050. SM015, SM016
CM019 Morgan Stanley says about 930 million of its 2050 humanoid forecast would be used for industrial and commercial work. SM014
CM020 MarketsandMarkets projects the humanoid robot market will grow from about $2.92 billion in 2025 to about $15.26 billion in 2030. SM012
CM021 SkyQuest projects the global humanoid robot market will reach about $35.4 billion by 2033 at 48.9% CAGR. SM013
CM022 People's Daily cites an industry report valuing embodied AI at about $4.44 billion in 2025 and about $23 billion by 2030. SM020
CM023 CCID-linked reporting says China's humanoid robot market is expected to exceed 20 billion yuan by 2026. SM002, SM029
CM024 Published humanoid sizing lenses are not directly comparable because some measure hardware revenue, some broader humanoid systems, and some embodied AI. SM012, SM013, SM014, SM020
CM025 SkyQuest says North America held the largest humanoid robot market share in 2025. SM013
CM026 IDC-based shipment coverage says Chinese firms dominated global humanoid robot commercialization in 2025. SM007, SM018, SM019
CM027 The apparent conflict between North America revenue-share leadership and China shipment dominance reflects different market definitions rather than a settled consensus. SM013, SM018, SM019
CM028 Industrial manufacturing is one of the clearest near-term buyer segments for humanoid robots. SM001, SM019, SM025
CM029 Warehouse logistics is a cited near-term buyer segment for humanoid robots. SM012, SM019, SM020
CM030 Retail and commercial service are cited near-term buyer segments for humanoid robots. SM012, SM019, SM025
CM031 Healthcare and eldercare are cited near-term buyer segments for humanoid robots. SM012, SM013, SM020
CM032 TechNode says Galbot has deployed products in industrial manufacturing, retail, and healthcare scenarios. SM025, SM026
CM033 TechNode says Galbot has cumulative orders totaling several thousand units from industrial clients including CATL, Bosch, Toyota, BAIC Group, and SAIC Motor. SM025, SM027
CM034 Figure raised over $1 billion at a $39 billion post-money valuation in 2025. SM007, SM022
CM035 CNBC says Galbot's valuation is above $3 billion but still below leading U.S. humanoid startups. SM007, SM025, SM027
CM036 CNBC says Chinese humanoid startups are often valued more like industrial hardware companies than broad AI platforms. SM007
CM037 Aging populations and labor shortages are major demand drivers for humanoid robots in care, retail, and logistics. SM012, SM013
CM038 Government policy support and large local industrial funds are important adoption accelerants in China. SM003, SM020
CM039 Falling unit costs and scaled manufacturing are expected to improve humanoid adoption economics over time. SM001, SM014, SM015
CM040 LLM integration and embodied-AI model progress are major enablers of broader humanoid use cases. SM001, SM011, SM020
CM041 China's supply-chain maturity and component ecosystem support faster humanoid commercialization than many foreign peers. SM001, SM020, SM023
CM042 Public coverage describes 2025 as China's first year of humanoid mass production. SM003, SM007
CM043 High unit costs and capital intensity still limit near-term buyer ROI for humanoid deployments. SM008, SM012, SM025
CM044 Limited task generalization beyond polished demos remains a core humanoid commercialization bottleneck. SM009, SM010
CM045 Data scarcity in real-world robot training remains a major constraint on reliable humanoid deployment. SM009, SM010
CM046 Dexterous hands and tactile sensing remain major bottlenecks for commercially useful humanoid manipulation. SM010, SM011
CM047 Safety, liability, and certification uncertainty still slow enterprise humanoid adoption. SM005, SM006, SM011
CM048 Workflow integration is harder than demoing isolated tasks because recovery, safety validation, and site-specific reliability all have to work in the customer environment. SM009, SM010, SM012
CM049 Buyer demand may lag manufacturing capacity as humanoid output scales faster than proven ROI and procurement readiness. SM007, SM008, SM011
CM050 Galbot's near-term serviceable market is narrower than headline global TAM because current demand is concentrated in enterprise pilots and specific vertical workflows rather than household robots. SM014, SM019, SM025
CP001 The relevant competitive set for Galbot spans direct humanoid peers, auto-backed or public-company entrants, model-layer competitors, and labor or fixed-automation substitutes. SP015, SP016, SP024
CP002 AgiBot was founded in 2023 in Shanghai by former Huawei engineers. SP001, SP002
CP003 AgiBot publicly markets the A2 full-size humanoid, G1 industrial robot, X2 compact humanoid, and D1 quadruped. SP001, SP002
CP004 Omdia-based coverage cited by TrendForce and DirectIndustry ranks AgiBot first globally in 2025 humanoid shipments at roughly 5,100 units and 39% share. SP016, SP017
CP005 AgiBot said it reached its 10,000th robot production milestone in March 2026 after moving from 5,000 to 10,000 units in about three months. SP001, SP017
CP006 Unitree was founded in 2016 in Hangzhou and ships products to more than 30 countries. SP003, SP004
CP007 Unitree's public G1 price point is $13,500 and the robot is described at roughly 35 kilograms, 130 centimeters, 23 degrees of freedom, and a two-hour battery life. SP003, SP005
CP008 Independent market coverage says Unitree claims roughly 5,500 humanoid robots shipped in 2025 while also presenting an IPO-related maturity narrative. SP017, SP018
CP009 Figure positions F.02 for workforce use and F.03 for household use across its Figure 01, 02, and 03 generations. SP006, SP009
CP010 Figure announced more than $1 billion of committed Series C capital at a $39 billion post-money valuation. SP007, SP018
CP011 Figure's Helix stack uses a System 1 and System 2 architecture and sits alongside a BotQ manufacturing narrative. SP006, SP008
CP012 Physical Intelligence describes π0 as a generalist robot foundation model trained on more than 10,000 hours of robot data and controlling eight different robots. SP010, SP011
CP013 Physical Intelligence also points to π0.5 as an update focused on stronger open-world generalization. SP010, SP011
CP014 XPENG says its IRON humanoid sits inside a broader physical-AI stack and targets mass production by the end of 2026, with in-store guide use from Q1 2027. SP012, SP013
CP015 UBTech's Walker humanoid line is associated with plans to ramp to roughly 5,000 units in 2026 and 10,000 in 2027. SP014, SP016
CP016 TrendForce characterizes Boston Dynamics' Atlas as beginning commercial deployment in 2026 with an industrial focus. SP016
CP017 External coverage presents 1X as progressing toward home use while deliberately limiting physical capabilities for safety. SP015
CP018 Galbot's published positioning is full-stack and in-house across dataset, embodied foundation models, and hardware. SP019, SP021
CP019 Galbot claims more than 10 billion embodied data points and a Sim2Real method that pre-trains on synthetic data before fine-tuning on limited real-world data. SP020, SP021
CP020 Galbot publicly markets GraspVLA, TrackVLA, GroceryVLA, and a brain-cerebellum-neural-control architecture. SP020, SP021
CP021 Galbot cites named deployments with CATL factories, Mercedes-Benz, Zeekr, Xuanwu Hospital, and Galbot Store locations across 30 or more cities. SP019, SP021
CP022 Galbot's state-backed investor set likely improves domestic procurement access and policy credibility relative to purely venture-backed peers. SP019, SP020, SP025
CP023 Galbot does not publish a public unit price for G1, leaving enterprise buyers without a transparent ASP benchmark. SP003, SP021
CP024 AgiBot pairs hardware sales with a Powered by AgiBot OEM-platform story rather than only selling finished robots. SP001, SP002
CP025 Figure and Physical Intelligence show that the competitive frontier is shifting toward model and platform depth, not just robot-body engineering. SP007, SP008, SP010, SP024
CP026 The direct humanoid battlefield is bifurcated between Chinese scale players with visible shipment momentum and US peers with much larger valuation support. SP016, SP018, SP024
CP027 Galbot's public customer evidence points primarily to industrial, retail, and healthcare operators rather than hobbyist or research buyers. SP020, SP021
CP028 Unitree is the clearest public low-end price anchor, but its buyer mix and product positioning differ from Galbot's enterprise-grade deployment narrative. SP003, SP021
CP029 AgiBot's shipment-lead story is disputed because Unitree separately claims a roughly 5,500-unit 2025 shipment figure. SP017, SP018
CP030 Chinese humanoid valuations are heavily discounted versus US peers, with Figure at roughly $39 billion versus Galbot around $3 billion. SP007, SP018, SP019
CP031 TechXplore quotes critics arguing that most humanoid robots are still performative rather than functional and that real use cases remain limited. SP015
CP032 Chinese humanoid standards activity and outside legal commentary show that trust and liability questions are becoming formal buying criteria rather than future issues. SP023, SP025
CP033 Trust posture increasingly favors vendors that can show named enterprise deployments plus alignment with emerging safety and standards frameworks. SP021, SP022, SP025
CP034 Switching costs in humanoid deployments are meaningful but not absolute because buyers can multi-home when models, tooling, and task interfaces remain immature. SP015, SP016, SP024
CP035 Distribution power in this market favors companies with automotive, battery, industrial, hospital, or retailer channels rather than standalone robotics labs. SP013, SP019, SP020
CP036 Galbot's moat is strongest in China-specific deployment access and stack integration today, but its long-run durability is only medium if VLA capabilities commoditize and pricing stays opaque. SP015, SP018, SP021, SP024
CI001 Galbot's public funding timeline began with seed, angel, and angel+ rounds in 2023 before scaling into larger institutional rounds from 2024 onward. SI002, SI003
CI002 Galbot raised RMB 1.1 billion in June 2025 in a round led by CATL-linked capital with strategic and state-backed co-investors. SI002, SI005, SI006
CI003 The June 2025 round positioned Galbot as a unicorn valued above $1 billion. SI002, SI005
CI004 Galbot's December 2025 round brought in more than $300 million, took total raised to roughly $800 million, and set a $3 billion valuation. SI001, SI007
CI005 Galbot's March 2026 round added RMB 2.5 billion, led by the National AI Industry Investment Fund with Sinopec, CITIC Investment Holdings, Bank of China, and SAIC Financial Holdings participating. SI003, SI004, SI020
CI006 After the March 2026 financing, Galbot's cumulative disclosed capital was approximately $1.15 billion or more. SI001, SI003, SI004
CI007 Public round descriptions say the new capital is intended for embodied-AI model development, manufacturing scale-up, and commercial expansion. SI001, SI003, SI014
CI008 Public evidence supports a revenue model that includes hardware sales, deployment or integration fees, and recurring service elements. SI010, SI011, SI014
CI009 Galbot Store and pharmacy deployments suggest Galbot may sometimes monetize through managed operations or operator-style economics rather than only one-time robot sales. SI010, SI011, SI013
CI010 Industrial customers such as CATL, Mercedes-Benz, and Bosch-linked partners imply large-account enterprise selling with longer cycles and higher implementation scope. SI006, SI010, SI014
CI011 Healthcare deployments such as Xuanwu Hospital and robot-pharmacy operations introduce service-quality requirements closer to regulated operations than to consumer gadget sales. SI010, SI013
CI012 Public materials do not confirm dataset or model licensing as a separate booked revenue stream for Galbot. SI010, SI011
CI013 Galbot does not disclose a public unit price for G1, leaving ASP and revenue-recognition analysis unresolved. SI010, SI011
CI014 Broader market coverage uses Unitree's $13,500 G1 as a visible low-end humanoid price anchor, but that benchmark is not directly comparable to Galbot's industrial-grade deployments. SI016, SI019
CI015 Galbot's public GTM appears enterprise-led, with traction communicated through named deployment sites and partners rather than broad self-serve acquisition. SI010, SI014
CI016 Public materials do not disclose CAC, payback period, net revenue retention, or other direct sales-efficiency metrics. SI001, SI010, SI014
CI017 Galbot materials say a single robot can operate a 50-square-meter store and replace three labor shifts over a three-year span. SI011, SI012
CI018 At $15 per hour for three eight-hour shifts across 365 days, Galbot's labor-replacement claim implies roughly $131,400 of annual labor value per fully utilized robot. SI011, SI012
CI019 Pharmacy deployment coverage cites a 99.5% medication-handling success rate, indicating high task reliability but not disclosing corresponding revenue. SI013, SI020
CI020 Public traction indicators include 30-plus-city retail presence, 100-plus pharmacy or store deployments, and several thousand cumulative industrial orders. SI001, SI013, SI014
CI021 Orders and deployment counts cannot be translated cleanly into ARR or recognized revenue because delivery schedules, cancellations, and acceptance criteria are not disclosed. SI001, SI010, SI014
CI022 Galbot's likely cost structure includes bill of materials, actuators, batteries, sensors, compute, installation, and field maintenance rather than only software hosting. SI019, SI022, SI023
CI023 Galbot does not disclose gross margin, and industry context suggests a 20-40% hardware robotics band is plausible but unverified for the company. SI017, SI022, SI023
CI024 Working capital is likely meaningful because robots, parts, and deployment services must be financed before cash collection fully catches up. SI022, SI023
CI025 Manufacturing scale-up after the 2026 financing likely increases capex needs if Galbot expands in-house production capacity. SI003, SI022
CI026 With about $1.15B+ of disclosed capital raised, Galbot appears adequately capitalized for near-term scale-up even without public profitability data. SI001, SI003, SI020
CI027 Galbot's burn rate is undisclosed, but a hardware AI company at this stage could plausibly burn $5-20 million per month depending on manufacturing pace and R&D intensity. SI016, SI022, SI023
CI028 That burn proxy would imply more than 24 months of runway after the March 2026 round only if a large share of prior capital remained available and losses do not widen materially. SI003, SI022, SI023
CI029 State-backed financing likely lowers Galbot's refinancing risk relative to purely venture-backed humanoid peers. SI003, SI024, SI025
CI030 Galbot's roughly $3 billion valuation is far below Figure's $39 billion benchmark, highlighting a major Chinese-versus-US humanoid valuation discount. SI001, SI015, SI018
CI031 CNBC and TechXplore both report skepticism that current humanoid deployments have yet proven broad buyer depth or practical use-case breadth. SI015, SI017
CI032 Galbot does not publicly disclose revenue, ARR, EBITDA, cash balance, burn, gross margin, customer concentration, or payback. SI001, SI010, SI014
CI033 Revenue quality is promising but unproven because deployment breadth is visible while monetization mix and recognized revenue remain opaque. SI010, SI013, SI014
CI034 Margin improvement depends on manufacturing yield, service efficiency, and utilization rising faster than price compression in a crowded humanoid market. SI017, SI019, SI022
CI035 The highest-priority diligence asks are a dated revenue bridge, gross margin by line, order-to-delivery conversion, service attach rates, burn, runway, and working-capital terms. SI001, SI010, SI017
CI036 Galbot should be underwritten as a capital-intensive physical-AI company rather than a typical asset-light SaaS business. SI003, SI022, SI023
CE001 Galbot positions G1 as an embodied-intelligence worker for repeated indoor pick, carry, scan, sort, and delivery workflows in retail, pharmacy, warehouse, and factory settings. SE001, SE004
CE002 G1 combines a dual-arm upper body with a wheel-foot mobility structure, indicating a design optimized for stable indoor navigation plus human-space reach rather than pure bipedal locomotion. SE001, SE002
CE003 Official Galbot materials list G1 at 1730 mm height with 650 mm torso lift. SE001, SE002
CE004 Official Galbot materials list 710 mm arm length, a 0–2100 mm vertical workspace, and 5 kg dual-arm payload for G1. SE001, SE002
CE005 Official Galbot materials list a 48V 30Ah lithium battery, up to 10 hours of operating duration, a 6.25 inch touchscreen, and WiFi, Ethernet, USB, and cloud connectivity. SE001, SE002
CE006 Official Galbot materials state IP54 ingress protection and multimodal sensing that includes vision, tactile, and depth inputs. SE001, SE002
CE007 Public specification sheets are not fully harmonized: the official bundle cites approximately 92.5 kg body weight, while secondary reviews have cited 85 kg. SE002, SE006
CE008 Galbot states that one G1 can operate a 50 square meter store footprint, supporting its positioning in compact autonomous retail environments. SE002, SE006
CE009 Galbot launched GraspVLA in January 2025 as an end-to-end embodied AI grasping foundation model. SE009, SE011
CE010 Public materials describe GraspVLA as trained on billions of simulated interactions to improve zero-shot generalization on new objects and tasks. SE009, SE011
CE011 Galbot links its grasping stack to DexGraspNet-scale data, citing 1.3 million grasps across more than 5,000 objects. SE011, SE018
CE012 TrackVLA is described as a navigation and tracking model that can follow people or objects via visual cues, accept voice commands, and resume tracking after temporary visual loss. SE002, SE011
CE013 GroceryVLA is described as a retail-specific manipulation model that can handle deformable snack bags, rigid bottles, and fragile jars in cluttered environments without per-item reprogramming. SE002, SE006
CE014 Galbot describes a brain-cerebellum-neural-control architecture that links multimodal perception to real-time feedback control in an end-to-end embodied stack. SE002, SE011
CE015 Galbot claims to have accumulated more than 10 billion data points and frames that corpus as the largest embodied-intelligence dataset among peers. SE011, SE012
CE016 Galbot's Sim2Real method relies on large-scale synthetic pretraining followed by limited real-world fine-tuning and minimal semantic relabeling. SE008, SE009
CE017 Robotics & Automation News reported that Galbot uses NVIDIA Isaac Sim in its training-simulation pipeline. SE009
CE018 The existence of developer.galbot.com indicates Galbot has at least a public-facing developer and secondary-development surface for integrations. SE003
CE019 Galbot and Bosch launched the BOYIN INNOVATION ALLIANCE joint venture to target industrial embodied-AI applications and high-precision manufacturing scenarios. SE009, SE010
CE020 Galbot and UAES launched the RoboFab initiative to apply embodied AI in automotive manufacturing. SE010, SE024
CE021 Galbot publicly references research collaboration with Peking University and BAAI, signaling outside scientific relationships around embodied AI. SE005, SE011
CE022 Galbot presents itself as a full-stack company spanning data, embodied foundation models, and robotic hardware rather than a hardware-only integrator. SE004, SE011
CE023 Galbot cites third-party competition validation including a gold medal in the 2025 pharmaceutical sorting challenge. SE005, SE011
CE024 Public company materials and coverage also cite a gold medal at the 2025 World Humanoid Robot Games Robot Skills Competition with 336 points, 160 ahead of the runner-up. SE005, SE024
CE025 Public reporting says industrial settings often demand 99.9% to 99.99% accuracy, a higher bar than the 99.5% medication-handling success publicly cited for pharmacy deployments. SE007, SE012
CE026 ChinaTechNews reports that Galbot G1 achieved 99.5% medication-handling success in Beijing pharmacy use. SE007
CE027 Independent coverage says Galbot has more than 10 pharmacy deployments in Beijing and can sustain 24/7 operation in those settings. SE006, SE007
CE028 Galbot leadership has publicly said that broad commercial rollout of humanoid robots in factories is achievable within roughly two years. SE008, SE012
CE029 Retail and pharmacy rollout plans moved from 10-plus operating sites toward a 100-plus-site ambition, indicating management sees repeatability in the current deployment template. SE006, SE024
CE030 Bosch and UAES partnerships show Galbot's industrial roadmap is being pursued through partner-backed factory access rather than purely greenfield direct sales. SE009, SE010, SE024
CE031 China's March 2026 humanoid robot standards create a more formal compliance baseline for companies such as Galbot. SE013, SE014
CE032 China's May 2026 robot digital-ID regime requires companies to register robots using 29-digit codes. SE013, SE016
CE033 Legal analysis of humanoid robots identifies unresolved liability, autonomy, and data-privacy exposure that is relevant to Galbot's deployments even if not specific to Galbot alone. SE015, SE017
CE034 An adverse report states Galbot has not publicly explained how patient personally identifiable health information is secured in pharmacy workflows. SE007, SE017
CE035 IP54 protection is a meaningful basic durability signal but does not by itself amount to detailed medical, cleanroom, or harsh-factory certification. SE001, SE017
CE036 Compared with competitors such as Physical Intelligence and Figure, Galbot currently offers less public developer and research transparency even while claiming a similarly full-stack embodied-AI ambition. SE003, SE019, SE020, SE021
CE037 Galbot's moat appears to rely more on in-house data loops, deployment access, and manufacturing partnerships than on a publicly legible patent or open-research corpus. SE005, SE010, SE011
CE038 The current public record supports meaningful pilot and early commercial traction, but independent fleet reliability, failure, and service-economics data remain too thin to fully validate roadmap credibility. SE022, SE023, SE025
CU001 Galbot's customer model spans three role patterns: enterprises buy and pay for robots, frontline staff use them in workflow, and in some retail formats Galbot itself acts as operator as well as vendor. SU001, SU007, SU008
CU002 All major public Galbot deployments and named customer references are China-centered as of the run date. SU003, SU004, SU005, SU019
CU003 Galbot's public customer base spans at least four verticals: industrial manufacturing, healthcare/pharmacy, retail/convenience, and warehouse/logistics. SU001, SU004, SU005, SU013
CU004 CATL is both a strategic investor and a customer anchor for Galbot's industrial business. SU001, SU002, SU023
CU005 Galbot said in late 2025 that it had cumulative orders for several thousand units from industrial clients led by CATL, Toyota, and BAIC Group. SU001, SU013
CU006 TechNode reported that Galbot robots were operating at local Mercedes-Benz and Zeekr factories. SU002
CU007 2026 coverage linked SAIC Motor and BAIC Group to Galbot's industrial customer or order narrative. SU003, SU004
CU008 Bosch-linked partnerships act as both validation and channel expansion routes into factory automation deployments. SU010, SU011, SU025
CU009 UAES-linked RoboFab activity extends Galbot's reach into automotive manufacturing workflows, even though public fleet counts are not disclosed. SU003, SU025
CU010 Galbot publicly named Xuanwu Hospital as a healthcare collaboration covering patient rooms, pharmacies, and hospital guidance systems. SU001
CU011 Independent coverage says Galbot had 10+ pharmacies operating in Beijing with 99.5% medication-handling success and 24/7 operation. SU005, SU006
CU012 Galbot said its Galbot Store retail footprint had expanded to 30+ cities nationwide by December 2025. SU001
CU013 By March 2026, secondary coverage said Galbot had 100+ retail units across 20+ cities, including Galaxy Space Capsule convenience formats. SU004
CU014 Galaxy Space Capsule-style convenience stores function as a consumer-facing reference deployment for Galbot's humanoid retail model. SU004, SU006
CU015 Galbot claimed stable 24/7 operations for over a year in autonomous warehouse settings. SU001, SU013
CU016 The strongest publicly attributable customer proofs are CATL, Xuanwu Hospital, Beijing pharmacy sites, BAIC and Toyota order mentions, and Galbot's own retail network. SU001, SU004, SU005
CU017 A significant portion of Galbot's public customer story still looks like early commercial rollout or controlled pilot scaling rather than mature fleet saturation. SU014, SU015, SU016
CU018 Galbot does not publicly disclose NRR, GRR, logo churn, or renewal-rate metrics. SU001, SU007, SU022
CU019 Galbot also does not publicly disclose typical contract length or renewal structure for enterprise customers. SU001, SU013
CU020 CATL's dual role as both lead investor and leading customer creates a related-party concentration and governance risk. SU001, SU002, SU015
CU021 Industrial manufacturing appears to be Galbot's largest current commercial opportunity and likely its largest revenue pool, based on the several-thousand-unit order claim and the concentration of named logos there. SU001, SU003, SU017
CU022 Galbot's expansion motion appears to combine direct flagship sales, partner-mediated industrial rollout, and self-operated retail references. SU001, SU004, SU025
CU023 Customer evidence quality is strongest where Galbot or credible press names a specific institution and workflow, and weakest where a logo appears only in generalized profile coverage. SU001, SU002, SU005
CU024 Galbot's current public footprint is geographically concentrated in China even where the customer list spans multiple cities and verticals. SU004, SU005, SU018, SU019
CU025 Vertical diversity in healthcare and retail somewhat offsets concentration risk, but it does not eliminate the company's heavy dependence on industrial accounts for scaled order volume. SU001, SU004, SU005
CU026 Because public retention metrics are absent, durability must be inferred from operational continuity, strategic partnerships, and repeat rollout signals rather than from cohort data. SU013, SU022
CU027 CATL's investor-customer alignment likely increases Galbot's switching costs and lock-in relative to a purely arms-length pilot relationship. SU001, SU002
CU028 Bosch and UAES partnerships provide a plausible land-and-expand channel into larger factory networks if initial validations convert into standardized deployments. SU010, SU011, SU025
CU029 The move from 10+ pharmacies toward a 100+ rollout ambition suggests Galbot is testing a multi-site replication playbook rather than one-off showcase installations. SU004, SU005, SU006
CU030 Even if accurate, the public claim of several thousand industrial orders remains early relative to the scale of market opportunity and manufacturing ambition implied by sector narratives. SU001, SU014, SU017
CU031 Independent adverse coverage in 2026 argues that humanoid-robot demand still lags sector capacity because practical buyer use cases remain limited. SU014, SU015
CU032 Outside the pharmacy success rate and long-run operations claim, Galbot has published very few independently auditable customer outcome metrics. SU005, SU013, SU014
CU033 Galbot's company-operated retail formats provide useful reference-customer evidence, but they are weaker than independent third-party logos for assessing concentration and renewal quality. SU004, SU006, SU008
CU034 Galbot's partner page and JV news flow indicate an ecosystem-assisted GTM motion rather than a pure reseller-led or pure direct-sales model. SU025, SU026
CU035 By the run date, Galbot's named public customers are concentrated in large Chinese industrial accounts and public-service healthcare contexts rather than a broad SMB base. SU001, SU003, SU005
CU036 BAIC, SAIC, and Toyota are important logos, but their public evidence mostly comes from financing and profile coverage rather than detailed deployment case studies. SU001, SU003, SU023
CU037 Mercedes-Benz and Zeekr are useful proof-of-interest logos, but their evidence quality is lower because the public record is limited to secondary profile reporting. SU002
CU038 Public sources do not disclose top-customer revenue share, so concentration risk cannot be quantitatively bounded from the outside. SU001, SU022
CU039 Claims of 24/7 operations for over a year are positive retention proxies but do not substitute for actual renewal, expansion, or contract-quality data. SU013, SU022
CU040 There is no strong public evidence of materially international customer traction or scaled non-China deployments as of June 2026. SU007, SU022, SU026
CR001 China’s March 2026 humanoid robot standard system formalized a national compliance framework spanning safety, ethics, core technologies, and testing, raising the baseline for every domestic manufacturer. SR003, SR004, SR005, SR032
CR002 China’s May 2026 digital-ID regime makes registration a practical market-access requirement for humanoid robots and expands traceability obligations after deployment. SR002, SR004
CR003 The new robot digital-ID rules reportedly require recalls for defective humanoids and prohibit refurbishment or resale of retired units, increasing downside from manufacturing defects. SR002, SR004
CR004 Hill Dickinson argues that humanoid liability remains unsettled because responsibility can shift among the manufacturer, operator, and software provider after an incident. SR001, SR033
CR005 Hill Dickinson also highlights privacy and biometric-data risk because humanoids can process facial, behavioral, and workplace data under uneven cross-border legal regimes. SR001
CR006 Documented robot-safety incidents in automotive and industrial settings show that maintenance or control failures can cause serious human injury even before humanoids become fully autonomous. SR001, SR009
CR007 Hill Dickinson cites a reported AgiBot malfunction that struck a refrigerator and nearly hit an employee, illustrating that near-miss evidence is already surfacing in Chinese humanoid deployments. SR001
CR008 Geopolitical and export-control exposure remains material for Chinese humanoid firms because advanced chips, overseas markets, and perception of strategic technology are politically sensitive. SR028, SR030, SR031
CR009 The embodied-AI stack is converging around VLA-like model approaches, which increases the probability of IP disputes or costly differentiation battles. SR006, SR009
CR010 Galbot’s founder He Wang is both founder-CEO and a Peking University professor, concentrating strategic, technical, and public-facing responsibilities in one key individual. SR010, SR012
CR011 Galbot presents itself as a full-stack embodied-AI company spanning proprietary models, hardware, data, and deployment systems rather than as a single-use robot vendor. SR010, SR022
CR012 Humanoid scale-up requires coordinated sourcing and integration of actuators, sensors, processors, batteries, and end-effectors, making manufacturing complexity a core operational risk. SR006, SR009
CR013 Industrial deployment standards imply that factory humanoids must approach extremely high task accuracy and uptime before replacing multiple human shifts economically. SR007, SR019
CR014 TechXplore’s June 2026 reporting argues that Chinese firms can build humanoids at scale faster than they can persuade buyers to adopt them, making demand formation a first-order risk. SR007
CR015 Associated Press coverage from late 2025 captured continuing skepticism that many humanoid demos remain performative rather than commercially functional. SR026
CR016 Deloitte warns that hallucinations, perception errors, and software faults in physical AI can create real-world safety incidents rather than purely digital mistakes. SR006, SR001
CR017 Connected robot fleets create cybersecurity and unauthorized-access risk because compromise can affect both data security and physical human safety. SR006, SR001
CR018 Galbot G1 deployments place a 48V 30Ah lithium battery pack near users and staff, so battery integrity and thermal management are part of the operational risk stack. SR015, SR021
CR019 Tactile sensing remains a bottleneck for many human-like tasks, which limits how quickly humanoids can move from demos to general-purpose work. SR009, SR019
CR020 Galbot’s CATL relationship concentrates both commercial demand and financing because the battery giant has been described as both a major investor and a prominent deployment reference. SR012, SR022, SR024
CR021 Galbot’s Bosch-related joint-venture and investment links expand manufacturing and distribution options but also introduce partner-governance and term-reset risk. SR016, SR018
CR022 State-backed investors and banks can improve procurement access and resilience, but they also increase political-dependency risk if policy priorities shift. SR011, SR013, SR014
CR023 Automation World reporting shows Galbot integrating NVIDIA Jetson Thor and Isaac-related tooling, which ties some development workflows to U.S.-linked compute ecosystems. SR027, SR028
CR024 Embodied-AI training remains dependent on simulation and cloud-scale compute even for hardware-first companies, leaving Galbot exposed to platform, cost, and availability shocks. SR006, SR027
CR025 TrendForce’s shipment and market-share analysis suggests Chinese supply chains are deep, but concentration within that ecosystem still creates substitution risk if controls tighten. SR020, SR030
CR026 Galbot does not publicly disclose audited revenue or detailed financial statements, limiting confidence in the current valuation and burn profile. SR008, SR010
CR027 Galbot’s March 2026 round was publicly described as more than $300 million at a roughly $3 billion valuation. SR011, SR013, SR022
CR028 Coverage of Galbot’s 2025 financing indicates the company had already raised roughly $800 million before the 2026 round, underscoring the capital intensity of the category. SR012, SR016, SR024
CR029 Galbot’s $3 billion mark still sits far below Figure AI’s disclosed $39 billion post-money valuation, implying either upside optionality or a China-specific risk discount. SR008, SR023
CR030 Humanoid robotics remains capex-heavy because productization requires sustained R&D, hardware iteration, software training, and field support before margins are proven. SR006, SR008, SR007
CR031 The “replace three shifts” industrial value proposition only works if hardware reliability, support costs, and deployment uptime hold under real production conditions. SR019, SR021, SR007
CR032 The digital-ID regime increases recall downside because manufacturing defects can now trigger traceable corrective action and resale restrictions. SR002, SR004
CR033 Galbot’s full-stack architecture reduces dependence on outside vendors for core models and hardware design, partially mitigating platform and supplier risk. SR010, SR022
CR034 Public deployment references span industrial, retail, healthcare, and pharmaceutical settings, which partially reduces single-vertical demand concentration. SR015, SR017, SR022
CR035 State backing can cushion funding volatility and improve market access, but it does not eliminate execution or commercial demand risk. SR011, SR014
CR036 Simulation-led training can reduce the amount of costly real-world data collection required before deployment, though it cannot fully replace field validation. SR006, SR027
CR037 A clearer national standards framework can gradually reduce regulatory ambiguity even while near-term compliance costs rise. SR003, SR005
CR038 If CATL meaningfully reduces orders or investment support, Galbot would likely face simultaneous revenue, signaling, and financing pressure. SR012, SR022, SR024
CR039 If Galbot cannot demonstrate safe digital-ID-compliant field performance, regulatory clearance and commercial expansion could stall at the same time. SR002, SR003, SR006
CR040 The risk profile is cumulative: tighter regulation, unproven demand, and customer concentration can amplify each other instead of remaining isolated issues. SR001, SR007, SR020
CV001 Galbot’s March 2026 financing was publicly described as more than $300 million at an approximately $3 billion valuation. SV006, SV007, SV009
CV002 Coverage of Galbot’s 2025 financing indicates the company had already raised roughly $800 million before the 2026 round, making cumulative capital raised roughly $1.15 billion or more. SV011, SV014, SV016
CV003 Galbot’s investor base includes large state-linked institutions and industrial names, which can improve policy access and domestic procurement credibility. SV006, SV007, SV008
CV004 Galbot presents itself as a full-stack embodied-AI company rather than a pure hardware assembler. SV012, SV013
CV005 Official materials indicate Galbot has amassed more than 10 billion data points and multiple embodied-AI models, supporting the claim of a data and software moat. SV012, SV013
CV006 Publicly cited deployments include CATL factories, healthcare sites, and retail/pharmacy environments, giving Galbot more commercial proof than a lab-only startup. SV009, SV012, SV025
CV007 China’s supply-chain depth and manufacturing base are a structural advantage for domestic humanoid vendors that can iterate hardware more quickly than many foreign rivals. SV004, SV005
CV008 TrendForce reported that China accounted for roughly 90% of global humanoid robot shipments in 2025, reinforcing the importance of domestic scale advantages. SV004, SV005
CV009 Founder-CEO He Wang’s Stanford and Peking University credentials strengthen Galbot’s technical credibility with investors and partners. SV011, SV012
CV010 CATL is both a commercial reference and a concentration risk because one counterparty influences demand signaling and financing confidence at the same time. SV009, SV011, SV016
CV011 Galbot’s $3 billion valuation is not anchored to disclosed revenue, audited margin, or public financial statements. SV001, SV012
CV012 TechXplore’s June 2026 reporting argues that demand still lags manufacturing ambition in humanoids, making revenue-ramp assumptions fragile. SV003
CV013 Figure AI’s official Series C announcement put that U.S. peer at a $39 billion post-money valuation, creating a sharp headline gap versus Galbot’s $3 billion mark. SV002, SV029
CV014 The valuation gap with U.S. peers can reflect not only upside potential but also governance, liquidity, and geopolitical discounts applied to Chinese humanoid names. SV001, SV013
CV015 Public evidence still does not disclose Galbot’s burn rate, gross margin, unit economics, or audited revenue trajectory. SV001, SV012
CV016 The CATL relationship creates related-party style concentration risk because one prominent partner influences both commercial optics and investor narrative. SV009, SV011
CV017 Bosch-linked partnerships can accelerate manufacturing and go-to-market execution, but they also introduce partner-term and strategic-priority risk. SV014, SV015
CV018 China’s standards and digital-ID frameworks tighten the operating environment, which can add compliance cost before commercialization reaches steady scale. SV017, SV023, SV024
CV019 Deloitte’s physical-AI analysis implies that safety, cyber, and perception failures can slow adoption and increase liability for embodied-AI vendors. SV017, SV018
CV020 Convergence toward similar VLA-style and full-stack approaches raises the risk that differentiation narrows faster than current valuations imply. SV017, SV018
CV021 Unitree’s published G1 price point shows that aggressive pricing pressure can emerge quickly in Chinese humanoids even before premium use cases are fully stabilized. SV019, SV005
CV022 AgiBot’s visibility supports the view that Galbot competes in a crowded domestic field rather than owning a uniquely open category. SV020, SV005
CV023 XPENG’s robotics activity widens the comparator set beyond startups and reminds investors that capital can also flow to better-disclosed public competitors. SV021, SV026
CV024 Physical Intelligence represents the competing thesis that generalist robot value may accrue to foundation-model platforms rather than to one hardware integrator. SV022
CV025 Because Galbot lacks disclosed revenue and margin inputs, the comparable set must mix private rounds, public comps, and milestone-based reference points rather than rely on one clean multiple. SV001, SV004, SV026
CV026 A bull case for Galbot assumes Chinese industrial humanoid leadership compounds into at least several hundred million dollars of revenue by 2028 and supports a $25 billion to $35 billion value range. SV004, SV006, SV009
CV027 A base case assumes Galbot wins meaningful scale in two to three verticals, develops revenue visibility by 2027, and supports a $10 billion to $15 billion value range. SV004, SV006, SV009
CV028 A bear case assumes commoditization, regulatory drag, or CATL retrenchment and points to a $1 billion to $1.5 billion downside range. SV001, SV003, SV017
CV029 Given the current disclosure gap and concentration profile, the evidence supports a research-more recommendation rather than a clean buy call at $3 billion. SV001, SV011, SV015
CV030 Unknown preferences, seniority, and dilution overhead matter because the post-money headline does not reveal common-equity entry quality. SV006, SV007
CV031 A four-to-six-year hold period is more realistic than a near-term exit because commercialization maturity still lags the financing narrative. SV001, SV003, SV025
CV032 Boston Dynamics and Hyundai show that well-capitalized incumbents are also commercializing humanoids, reducing any scarcity premium for a private Galbot round. SV027, SV030, SV031, SV032
CV033 Reuters and Figure’s own materials show that category leaders can still attract very large funding rounds at much richer valuations than Galbot commands today. SV028, SV029, SV033
CV034 China market leadership can support scale advantages for Galbot even if overseas investors apply a lower valuation multiple to Chinese robotics firms. SV004, SV005, SV001
CV035 Financial opacity is the single largest reason to treat the current mark as stretched rather than obviously attractive. SV001, SV012
CV036 If audited revenue, unit economics, and customer concentration data validate the current narrative, a stretched valuation could move closer to fair. SV006, SV011, SV012
CV037 If digital-ID compliance or privacy controls fail in healthcare-style deployments, valuation downside would widen quickly because both policy and demand confidence would suffer. SV017, SV023, SV024
CV038 The most important final diligence items are audited revenue, unit P&L, CATL contract terms, actual delivery schedules, burn rate, IP freedom to operate, healthcare privacy posture, and cap-table structure. SV006, SV011, SV012, SV017
CV039 Galbot’s deployment breadth across industrial, retail, and healthcare settings supports the core thesis that the company is beyond the pure prototype phase. SV009, SV012, SV025
CV040 The anti-thesis remains that valuation has outrun public proof on revenue, margins, and concentration-adjusted demand quality. SV001, SV003, SV015
来源
编号出版方标题引文
SO001 Galbot Galbot homepage
SO002 Galbot About Galbot
SO003 Galbot Galbot G1
SO004 Galbot Galbot partners
SO005 Galbot Galbot web application bundle
SO006 Galbot Galbot developer portal
SO007 TechNode Galbot company profile and commercialization report
SO008 Robotics and Automation News Galbot raises $151 million to scale embodied AI humanoid robots and partners with Bosch investment arm
SO009 Yicai Global Chinese Robotics Startup Galbot Bags USD153 Million in Latest Fundraiser
SO010 CnEVPost / CnTechPost Galbot secures major state backing
SO011 TechNode Humanoid robot maker Galbot raises RMB 2.5 billion
SO012 The Robot Report Galbot brings in $300M to scale mobile manipulator deployments
SO013 PR Newswire Galbot secures over $300 million in new funding round
SO014 Rocking Robots Galbot valued at $3 billion as it secures over $300 million in new funding round
SO015 Asia Tech Daily China's Galbot secures $153M and launches humanoid robotics joint venture with Bosch
SO016 China Daily Galbot commercialization and deployment update
SO017 China Tech News China introduces AI-powered robot pharmacist Galbot G1 to Beijing pharmacies
SO018 Aparobot Galbot G1: The humanoid robot revolutionizing retail
SO019 Xinhua News Agency China rolls out digital ID system to regulate booming humanoid robot sector
SO020 TrendForce China humanoid robots market: Unitree, Agibot, Galbot
SO021 State Council Information Office China releases first national standard system for humanoid robotics and embodied AI
SO022 Robotics and Automation News China sets national standards for humanoid robots
SO023 CNBC China humanoid robots draw U.S. investor attention amid geopolitical tension
SO024 Hill Dickinson Humanoid robots and the law: preparing for a new legal frontier
SO025 DirectIndustry e-Magazine China humanoid robots market: Unitree, Agibot, Galbot
SM001 TrendForce China’s Humanoid Robot Output to Surge 94% in 2026; Unitree and AgiBot to Capture Nearly 80% Market Share, Says TrendForce the global industry is set to enter a critical phase of commercialization in the second half of 2026
SM002 DirectIndustry e-Magazine China's Humanoid Robot Market: Unitree, AgiBot, UBTech, Leju, and Galbot
SM003 State Council Information Office / Xinhua China releases national standard system for humanoid robotics and embodied AI over 140 domestic manufacturers releasing more than 330 different models
SM004 Robotics & Automation News China sets national standards for humanoid robots
SM005 Robotics & Automation News Why China's new humanoid robot safety standards matter
SM006 Xinhua China rolls out "digital ID" system to regulate booming humanoid robot sector The new standard enforces a strict "no code, no market access" rule.
SM007 CNBC China ships more humanoid robots than the U.S.
SM008 Tech Xplore China can build humanoids at scale. The hard part is finding enough buyers
SM009 Deloitte Tech Trends 2025: Physical AI
SM010 MDPI Electronics Humanoid Robots: Opportunities, Challenges, and Future Directions
SM011 Hill Dickinson Humanoid robots and the law: preparing for a new legal frontier
SM012 MarketsandMarkets Humanoid Robot Market Size, Share & Trends, 2025 To 2030
SM013 SkyQuest Humanoid Robot Market Size, Share | Forecast Report [2033]
SM014 Morgan Stanley Humanoid Robot Market Expected to Reach $5 Trillion by 2050
SM015 UBS Is the world ready for one billion robots?
SM016 Yicai Global World to Have 300 Million Humanoid Robots by 2050, UBS Report Says
SM017 Jiemian Global IDC sees global humanoid robot shipments at about 18,000 units in 2025; AgiBot leads key applications
SM018 InfotechLead Humanoid Robot Market Surges in 2025 as China Leads Large-Scale Commercial Adoption
SM019 CCTV IDC report: China leads the global humanoid robot rise in 2025 Global shipments of humanoid robots surged to around 18,000 units in 2025, up 508 percent year on year.
SM020 People's Daily Online Humanoid robots highlight China's rise in embodied AI industry
SM021 Figure Figure official website
SM022 Figure Figure Exceeds $1B in Series C Funding at $39B Post-Money Valuation
SM023 Unitree Robotics Unitree Robotics official website
SM024 AGIBOT AgiBot official website
SM025 TechNode Humanoid robot maker Galbot raises RMB 2.5 billion Galbot develops humanoid robots powered by its proprietary embodied AI system and has deployed products across industrial manufacturing, retail, and healthcare scenarios.
SM026 Galbot Galbot official website
SM027 PR Newswire Galbot Secures Over $300 Million in New Funding, Breaking Records with $3 Billion Valuation in China's Humanoid Robot Sector
SM028 Xinhua China's first national standard system for humanoid robotics poised to spur industry development
SM029 EEWorld 2026: The humanoid robot industry is expected to exceed 20 billion yuan
SP001 AgiBot AgiBot official website
SP002 PRNewswire AgiBot Makes Its US Market Debut at CES 2026
SP003 Unitree Robotics Unitree G1 humanoid robot
SP004 Unitree Robotics Unitree Robotics official website
SP005 Unitree Robotics Unitree H1 humanoid robot
SP006 Figure AI Figure company overview
SP007 Figure AI Figure announces Series C
SP008 Figure AI Figure Helix
SP009 Figure AI Figure AI official website
SP010 Physical Intelligence π0: A Vision-Language-Action Flow Model for General Robot Control
SP011 GitHub Physical-Intelligence/openpi repository
SP012 XPENG XPENG at CVPR 2026
SP013 XPENG XPENG AI Day 2025
SP014 UBTech Robotics UBTech Robotics official website
SP015 TechXplore China humanoids scale hard but buyers remain limited
SP016 TrendForce Humanoid robot market outlook 2026
SP017 DirectIndustry China humanoid robots market: Unitree, AgiBot, Galbot
SP018 CNBC Chinese humanoid robots are attracting investors but still trade below US peers
SP019 PRNewswire Galbot secures over $300 million in new funding round
SP020 TechNode Humanoid robot maker Galbot raises RMB 2.5 billion
SP021 Galbot Galbot official website
SP022 Robotics & Automation News Why China's new humanoid robot safety standards matter
SP023 Hill Dickinson Humanoid robots and the law: preparing for a new legal frontier
SP024 Deloitte Tech Trends 2025: Physical AI
SP025 SCIO China issues humanoid robot-related standards update
SI001 PRNewswire Galbot secures over $300 million in new funding round
SI002 TechNode Galbot raises RMB 1.1 billion led by CATL-linked capital
SI003 TechNode Humanoid robot maker Galbot raises RMB 2.5 billion
SI004 CnEVPost Galbot secures major state backing
SI005 Yicai Global Chinese robotics startup Galbot bags USD153 million in latest fundraiser
SI006 Robotics & Automation News Galbot raises $151 million to scale embodied AI humanoid robots and partners with Bosch investment arm
SI007 The Robot Report Galbot brings in $300M to scale mobile manipulator deployments
SI008 Rocking Robots Galbot valued at $3 billion as it secures over $300 million in new funding round
SI009 AsiaTechDaily China's Galbot secures $153M and launches humanoid robotics joint venture with Bosch
SI010 Galbot Galbot official website
SI011 Galbot Galbot website JavaScript bundle
SI012 Aparobot Galbot G1: the humanoid robot revolutionizing retail
SI013 ChinaTechNews China introduces AI-powered robot pharmacist Galbot G1 to Beijing pharmacies
SI014 China Daily Galbot expands embodied AI commercial deployments
SI015 CNBC Chinese humanoid robots still trade below US peers
SI016 TrendForce Humanoid robot market outlook 2026
SI017 TechXplore China humanoids scale hard but buyers remain limited
SI018 Figure AI Figure announces Series C
SI019 DirectIndustry China humanoid robots market: Unitree, AgiBot, Galbot
SI020 Xinhua China robotics update with Galbot deployment context
SI021 Hill Dickinson Humanoid robots and the law: preparing for a new legal frontier
SI022 Deloitte Tech Trends 2025: Physical AI
SI023 MDPI Academic review of humanoid and service-robot economics
SI024 SCIO China robotics and AI industrial policy update
SI025 Robotics & Automation News China sets national standards for humanoid robots
SI026 U.S. Securities and Exchange Commission (NVIDIA) NVIDIA Corporation Form 10-K (FY ended January 25, 2026)
SI027 U.S. Securities and Exchange Commission (AMD) Advanced Micro Devices Form 10-K (FY ended December 27, 2025)
SI028 U.S. Securities and Exchange Commission (MongoDB) MongoDB, Inc. Form 10-K (FY ended January 31, 2026)
SI029 Amazon Web Services Fireworks.ai Case Study
SI030 Sacra Fireworks AI revenue, valuation & funding
SI031 Index Ventures Inference is the New Runtime
SI032 Business Wire Fireworks AI Raises $250M Series C
SI033 Tech Funding News Fireworks AI closes $250M at $4B valuation
SE001 Galbot Galbot G1 product page
SE002 Galbot Galbot official JavaScript bundle with embedded product specifications
SE003 Galbot Galbot developer platform
SE004 Galbot Galbot official homepage
SE005 Galbot Galbot about page
SE006 Aparobot Galbot G1: the humanoid robot revolutionizing retail
SE007 ChinaTechNews China introduces AI-powered robot pharmacist Galbot G1 to Beijing pharmacies The company has not revealed how it secures patient personally identifiable health information.
SE008 China Daily Galbot scales embodied AI through Sim2Real
SE009 Robotics & Automation News Galbot raises $151 million to scale embodied AI humanoid robots and partners with Bosch investment arm
SE010 AsiaTechDaily China's Galbot secures $153M, launches humanoid robotics joint venture with Bosch
SE011 PR Newswire Galbot secures over $300 million in new funding round
SE012 TechNode Galbot profile and product development update
SE013 SCIO China Voices: humanoid robot standards update
SE014 Robotics & Automation News China sets national standards for humanoid robots
SE015 Robotics & Automation News Why China's new humanoid robot safety standards matter
SE016 Xinhua China launches digital ID system for robots
SE017 Hill Dickinson Humanoid robots and the law: preparing for a new legal frontier
SE018 MDPI Electronics Academic review of embodied manipulation datasets and control methods
SE019 Physical Intelligence pi0 foundation model launch post
SE020 GitHub OpenPI repository
SE021 Figure Introducing Helix
SE022 TechXplore China's humanoid makers can scale, but buyers remain scarce Use cases are still so limited that demand is not yet matching capacity ambitions.
SE023 Deloitte Tech Trends 2025: physical AI
SE024 CnTechPost Galbot secures major state backing
SE025 TrendForce Humanoid robot market outlook
SE026 GraspNet DexGraspNet project page
SE027 NVIDIA NVIDIA Isaac Sim
SE028 NVIDIA NVIDIA Omniverse platform
SE029 Peking University Peking University official English site
SE030 BAAI Beijing Academy of Artificial Intelligence official English site
SE031 ISO International Organization for Standardization homepage
SE032 MIIT Ministry of Industry and Information Technology of China
SE033 ROS ROS official homepage
SU001 PR Newswire Galbot secures over $300 million in new funding round Galbot said it had secured cumulative orders for several thousand units from industrial clients led by CATL, Toyota, and BAIC Group.
SU002 TechNode Galbot profile and CATL-linked deployment update
SU003 TechNode Humanoid-robot maker Galbot raises RMB 2.5 billion
SU004 CnTechPost Galbot secures major state backing
SU005 ChinaTechNews China introduces AI-powered robot pharmacist Galbot G1 to Beijing pharmacies
SU006 Aparobot Galbot G1: the humanoid robot revolutionizing retail
SU007 Galbot Galbot official homepage
SU008 Galbot Galbot official JavaScript bundle with deployment copy
SU009 China Daily Galbot embodied-AI deployment update
SU010 Robotics & Automation News Galbot raises $151 million to scale embodied AI humanoid robots, partners with Bosch investment arm
SU011 AsiaTechDaily China's Galbot secures $153M, launches humanoid robotics joint venture with Bosch
SU012 Rocking Robots Galbot valued at $3 billion as it secures over $300 million in new funding round
SU013 The Robot Report Galbot brings in $300M to scale mobile manipulator deployments
SU014 TechXplore China's humanoid makers can scale, but buyers remain scarce Use cases are still so limited that demand is not yet matching the sector's production ambitions.
SU015 CNBC China humanoid robots attract investors, but real customers are still emerging
SU016 DirectIndustry eMag China humanoid robots market overview
SU017 TrendForce Humanoid robot market outlook
SU018 Xinhua China launches digital ID system for robots
SU019 SCIO China Voices: humanoid robot standards update
SU021 Hill Dickinson Humanoid robots and the law: preparing for a new legal frontier
SU022 Deloitte Tech Trends 2025: physical AI
SU023 Yicai Global Chinese robotics startup Galbot bags $153 million in latest fundraiser
SU024 Robotics & Automation News China sets national standards for humanoid robots
SU025 Galbot Galbot partner page
SU026 Galbot Galbot about page
SU027 Mercedes-Benz Mercedes-Benz official homepage
SU028 ZEEKR ZEEKR official homepage
SU029 Toyota Toyota Global official homepage
SU030 BAIC Group BAIC Group official English homepage
SU031 SAIC Motor SAIC Motor official English homepage
SU032 Bosch Bosch global official homepage
SU033 UAES UAES official English homepage
SU034 CATL CATL official English homepage
SR001 Hill Dickinson Humanoid robots and the law: preparing for a new legal frontier
SR002 Xinhua China launches digital identity system for humanoid robots
SR003 State Council Information Office China unveils national standard system for humanoid robots
SR004 Robotics & Automation News Why China’s new humanoid robot safety standards matter
SR005 Robotics & Automation News China sets national standards for humanoid robots
SR006 Deloitte Tech Trends 2025: Physical AI
SR007 TechXplore China can build humanoids at scale, but buyers remain hard to find
SR008 CNBC China humanoid robot startups court investors but still trail U.S. peers
SR009 MDPI Electronics Embodied AI and humanoid robot commercialization constraints
SR010 Galbot Galbot official website
SR011 PR Newswire Galbot secures over $300 million in new funding round
SR012 TechNode Galbot deploys robots with CATL and grows embodied AI footprint
SR013 TechNode Humanoid robot maker Galbot raises RMB 2.5 billion
SR014 CnEVPost Galbot secures major state backing in new financing
SR015 ChinaTechNews China introduces AI-powered robot pharmacist Galbot G1 to Beijing pharmacies
SR016 AsiaTechDaily Galbot secures $153M and launches humanoid robotics joint venture with Bosch
SR017 China Daily Galbot expands embodied AI deployment cases
SR018 Robotics & Automation News Galbot raises $151 million and partners with Bosch investment arm
SR019 DirectIndustry e-Magazine China’s humanoid robot market: Unitree, AgiBot, Galbot
SR020 TrendForce China leads humanoid robot commercialization and shipment scale
SR021 Aparobot Galbot G1: the humanoid robot revolutionizing retail
SR022 The Robot Report Galbot brings in $300M to scale mobile manipulator deployments
SR023 Figure AI Figure Series C
SR024 Yicai Global Chinese robotics startup Galbot bags USD153 million in latest fundraiser
SR025 Rocking Robots Galbot valued at $3 billion as it secures over $300 million in new funding round
SR026 U.S. News / Associated Press Humanoid robots take center stage at Silicon Valley summit, but skepticism remains
SR027 Automation World Galbot’s humanoid robot integrates NVIDIA Jetson Thor with potential for manufacturing use
SR028 CNBC Video Nvidia-powered Galbot hedges against U.S. trade risks with a diversified supply chain
SR029 China Biz Insider Galbot deploys heavy-duty humanoid robot at CATL, signaling industrial AI adoption
SR030 U.S.-China Economic and Security Review Commission Humanoid Robots
SR031 Humanoids Daily The GUARD Act: bipartisan bill seeks to ban Chinese robots, threatening the U.S. research baseline
SR032 Robotics & Automation News Why China’s new humanoid robot standards could change the industry
SR033 Hill Dickinson Humanoid robots and the law - preparing for a new era of risk
SV001 CNBC China humanoid robot startups court investors but still trail U.S. peers
SV002 Figure AI Figure Series C
SV003 TechXplore China can build humanoids at scale, but buyers remain hard to find
SV004 TrendForce China leads humanoid robot commercialization and shipment scale
SV005 DirectIndustry e-Magazine China’s humanoid robot market: Unitree, AgiBot, Galbot
SV006 PR Newswire Galbot secures over $300 million in new funding round
SV007 TechNode Humanoid robot maker Galbot raises RMB 2.5 billion
SV008 CnEVPost Galbot secures major state backing
SV009 The Robot Report Galbot brings in $300M to scale mobile manipulator deployments
SV010 Rocking Robots Galbot valued at $3 billion as it secures over $300 million in new funding round
SV011 TechNode Galbot and CATL deployment profile
SV012 Galbot Galbot official website
SV013 Galbot Galbot website application bundle
SV014 AsiaTechDaily Galbot secures $153M and launches humanoid robotics joint venture with Bosch
SV015 Robotics & Automation News Galbot raises $151 million to scale embodied AI humanoid robots, partners with Bosch investment arm
SV016 Yicai Global Chinese robotics startup Galbot bags USD153 million in latest fundraiser
SV017 Hill Dickinson Humanoid robots and the law: preparing for a new legal frontier
SV018 Deloitte Tech Trends 2025: Physical AI
SV019 Unitree Robotics Unitree G1 humanoid robot
SV020 AgiBot AgiBot official website
SV021 XPENG XPENG robotics and embodied intelligence newsroom update
SV022 Physical Intelligence pi0 foundation model announcement
SV023 State Council Information Office China unveils national standard system for humanoid robots
SV024 Xinhua China launches digital identity system for humanoid robots
SV025 Aparobot Galbot G1: the humanoid robot revolutionizing retail
SV026 XPeng Investor Relations XPENG 2025 Annual Report on Form 20-F
SV027 Boston Dynamics Meet the all-new electric Atlas
SV028 Reuters AI robot maker Figure raises $675 million led by Microsoft and Nvidia
SV029 PR Newswire Figure exceeds $1B in Series C funding at $39B post-money valuation
SV030 Hyundai Motor Group Hyundai Motor Group announces AI robotics strategy at CES 2026
SV031 WBUR / Associated Press Hyundai and Boston Dynamics unveil humanoid robot Atlas at CES
SV032 Hyundai Mobis Hyundai Mobis and Boston Dynamics announce global robotics supply chain collaboration
SV033 Robozaps Figure AI review: robots, Helix AI, and complete guide 2026