Zhijian Power / 至简动力 / Simplexity Robotics
Li Auto 精英履历与罕见融资通道加持,但在据报 >US$1B 估值下,仍缺公开客户与经济性证明
理想汽车核心团队背景和早期强赞助让 Zhijian Power 值得跟踪,但公开证据仍停留在 PoC 和未具名合作,无法支撑据报道 >US$1B 估值。
封面要素
公司概况
Zhijian Power / 至简动力 / Simplexity Robotics 是一家注册在杭州的具身 AI 机器人创业公司,由包括贾鹏、王凯、王佳佳在内的 Li Auto 前智能驾驶负责人于 2025-07-31 创立。公开资料一致将其描述为软硬件全栈项目:把自动驾驶式世界模型和 VLA 思路用于通用机器人本体,初期瞄准工厂、超市和物流场景。公司早期资本吸引力异常强——约六个月内完成 5 轮、合计约 RMB2.0 billion 融资,据报估值超过 US$1.0 billion——但公开记录仍缺少具名付费客户、收入、利润率或股权结构证据。
- 成立时间
- 2025-07-31
- 创始人
- Jia Peng, Wang Kai, Wang Jiajia
- 创立地点
- Hangzhou, Zhejiang, China
- 总部
- Hangzhou, Zhejiang, China
- 产品
- 全栈具身智能机器人项目,核心是世界模型 + VLA 基础栈、端侧学习 / 数据闭环和通用机器人本体;公开技术成果包括 LaST0、ManualVLA 和 TwinRL,但商业化部署证据仍停在小批量 / PoC 阶段。
- 客户
- 受控环境里的企业买家,尤其是工厂车间、超市,以及物流或仓储场景;公开来源尚未点名付费客户。
- 商业模式
- 预期为 B2B 机器人平台模式,组合机器人硬件、具身 AI 软件、部署服务和持续学习 / 迭代,但公开来源未披露定价、合同形式或收入结构。
- 阶段
- Private unicorn; precise round nomenclature not fully disclosed publicly
- 融资情况
- 多个公开来源称,公司约六个月内完成 5 轮累计约 RMB2.0 billion 融资,投资方包括 HongShan/Sequoia China、Legend Capital、CAS Star、Gaorong、Tencent 和 Alibaba,据报投后估值超过 US$1.0 billion。具体轮次条款、优先权、稀释和老股转让比例仍未披露。
执行摘要
主要优势
- 创始团队集合了 Li Auto 自动驾驶、系统集成和量产经验。
- 对一家 2025 年成立的公司而言,公开报道的资金获取能力异常强:五轮合计融资约 RMB2.0 billion。
- 官方和技术来源支撑一套连贯的世界模型 / VLA、端侧学习和数据闭环产品逻辑,不只是营销文案。
- 先切工厂、超市和物流场景,比一开始就部署开放环境消费级产品更可信。
- 中国政策与产业顺风支撑具身 AI 和工业机器人实验需求延续。
主要风险
- 没有公开来源点名付费客户、披露客户数量,或证明 PoC 与未具名合作之外已经转化为生产部署。
- 收入、利润率、烧钱速度、现金跑道和轮次条款均未披露,标准经营指标无法承销当前估值。
- 具身 AI 的法律、安全、隐私和产品责任暴露仍然重大,而认证和可靠性指标没有公开证据。
- 出口管制或伙伴集中可能让算力、芯片和生态依赖收紧。
- 行业层面泡沫警示和中国人形 / 具身机器人赛道拥挤,抬高了降估值融资和商品化风险。
未决问题
- 具名付费客户、部署地点、付费与免费试点状态,以及复购或续约证据。
- 收入台账、毛利率路径、物料清单或服务成本结构,以及现金消耗 / 现金跑道细节。
- 股权结构条款、清算优先权、董事会控制权,以及据报道融资后的确切持股比例。
- 承销真实工厂或物流部署所需的安全、在线率、事故和维护指标。
- 员工数、资深制造能力,以及核心创始人之外的独立合规 / 安全负责人。
目录
01公司概览
1.1 身份、法律足迹与叙事定位
本章分析的可投资主体是杭州至简动力科技有限公司,公开也以至简动力、Zhijian Power 和 Simplexity Robotics 出现。官网以「Simplexity Robotics」介绍这家公司,称其成立于 2025 年 7 月下旬,专注用统一模型、数据闭环和机器人硬件打造全场景具身机器人。工商口径把叙事压实为一组法律事实:AiQicha 显示公司成立于 2025-07-31,位于杭州市余杭区,贾鹏任法定代表人,经营范围覆盖 AI 软件、智能机器人研发与销售、工业 / 服务机器人制造,以及数据处理和系统集成等服务。因此,当前更适合按机器人 / 具身 AI 来看。细节很关键:团队可信度来自 Li Auto 的智能驾驶和自动驾驶系统经验,但公开文章描述的是把 VLA、世界模型和自动驾驶能力迁移到机器人,而不是出售智能驾驶平台。[CO001, CO002, CO003, CO004, CO005, CO006]
| 指标 | 数值或状态 | 日期 / 版本 | 置信度 | 缺口或解读 |
|---|---|---|---|---|
| 法律实体 | 杭州至简动力科技有限公司 | 2026-07-03 复核 | 高 | 名称由官网、登记信息和百科页面相互印证。 |
| 英文 / 品牌名称 | Simplexity Robotics / Zhijian Power / 至简动力 | 2026-07-03 复核 | 中 | 各来源别名用法不一;公司官网使用 Simplexity Robotics。 |
| 成立 / 注册 | 2025-07-31 法律注册;公开叙事称 2025 年 7 月下旬成立 | 2025-07-31 | 高 | 精确日期来自 AiQicha;官网称 7 月下旬。 |
| 总部 / 注册地址 | 浙江杭州余杭区 | 2026-07-03 复核 | 高 | 注册办公地址之外的运营地点来自媒体报道。 |
| 主要公开定位 | 使用 VLA / 世界模型能力的具身 AI / 全场景机器人创业公司 | 2026-07-03 复核 | 中 | 尽管团队源自自动驾驶,不应把公司简化为智能驾驶平台。 |
| 阶段 | 未上市独角兽;PoC 和小批量机器人阶段 | 2026-03 报道 | 中 | 商业部署规模仍未披露。 |
| 累计融资 | 据报道五轮累计 RMB 2 billion | 2026-03-09/10 | 中 | 未审阅一手融资备案文件。 |
| 估值 | 据报道 > USD 1 billion | 2026-03 报道 | 中 | 估值来自媒体报道;股权结构表和条款未公开。 |
| 收入 / 客户 / 员工数 | 未公开披露 | 2026-07-03 复核 | 低 | 列为尽调缺口;不要从融资倒推业务牵引力。 |
快照汇总官网、登记和媒体来源;未获支撑的运营指标刻意标为未披露,而不是零。
[CO001, CO002, CO003, CO004, CO008, CO014]公司论点从 Li Auto 自动驾驶人才延伸到具身机器人产品,由资本支撑,但受限于未披露的商业指标。
该流程是尽调逻辑图,不是所有权或数据流图。
[CO003, CO005, CO008, CO009, CO011, CO014]1.2 领导层、创始人-市场匹配与治理信号
领导层叙事异常集中。多篇独立报道都将贾鹏列为 CEO、王凯列为董事长、王佳佳列为 COO,三人均来自 Li Auto 智能驾驶组织。公开报道把贾鹏与 Li Auto 智能驾驶研发、IBM 高性能计算和 Nvidia 自动驾驶架构经历相连;把王凯与 Li Auto CTO 职责和更早的 Visteon 自动驾驶架构经验相连;把王佳佳与智能驾驶量产职责相连。该机器人栈从自动驾驶借用感知、VLA、数据闭环和量产纪律,因此创始人-市场匹配可信。它也带来关键人与治理尽调负担:AiQicha 列出更宽的董事 / 高管名单,Baidu Baike 和第三方工商库披露创始人与机构股东信号,但已审阅来源没有提供签署版股权结构文件、董事会观察员权利、否决权或雇佣锁定条款。后续章节应把 Li Auto 校友论点同时视为优势和依赖项,而不是可持续组织厚度的证明。[CO028, CO029, CO030, CO031, CO032, CO033]
| 人员或角色 | 公开角色 | 据报道背景 | 创始人-市场匹配度 | 依赖项 / 尽调问题 |
|---|---|---|---|---|
| 贾鹏 / Jia Peng | CEO;法定代表人;AiQicha 显示为经理和财务负责人 | 前 Li Auto 智能驾驶研发负责人;IBM HPC;有 Nvidia 自动驾驶架构经历 | 与 VLA、数据闭环和自动驾驶向机器人迁移直接相关 | 核验雇佣条款、股权、发明转让,以及持续技术所有权。 |
| 王凯 / Wang Kai | 董事长 | 据报道曾任 Li Auto CTO;更早为 Visteon 自动驾驶总监 / 首席架构师;Yuanjing 投资合伙人 | 与车规级系统、架构和资本网络相关 | 澄清时间投入、董事会权力,以及投资人 / 创始人冲突管理机制。 |
| 王佳佳 / Wang Jiajia | COO;董事 | 前 Li Auto 智能驾驶量产负责人 | 与机器人产品生产、铺开和运营规模化相关 | 核验运营团队深度和客户交付责任。 |
| 更广泛董事名单 | AiQicha 显示董事包括 Liu Yiran、Cao Wei、Qi Na、Zheng Qingsheng、Wang Lei、Yao Yao | 仅来自登记信息的治理信号 | 暗示融资后投资人 / 董事会扩容 | 获取董事会同意文件、保留事项和观察员名单。 |
| 创始人 / 机构股东组合 | Baidu Baike 和登记资料显示存在创始人、合伙企业 / 机构股东 | 第三方数据显示创始人持股和多个投资载体 | 与大额多轮融资叙事一致 | 索取签署版股权结构表、期权池、SAFEs / 可转债,以及清算优先权。 |
列举不完整;公开来源识别了核心高管和登记高管,但未提供签署版治理文件。
[CO005, CO028, CO029, CO030, CO031, CO032]1.3 融资、阶段、地点与披露规模
融资是公开资料中最一致的规模信号。Sina、Tencent News、36Kr、Yicai Global、Gasgoo、Pandaily、Taibo、Pedaily 等媒体称,至简动力在不到约六个月内宣布 5 轮融资,合计 RMB 2 billion。投资方名单在多个来源中反复出现:Yuanjing/Vision Capital、Lanchi/BlueRun、Sequoia China/HongShan、Legend Capital、CAS Star、Gaorong、Tencent 和 Alibaba。Yicai 将融资额折算为 USD 289.3 million;36Kr 和 Baidu Baike 称估值跨过 USD 1 billion,使公司成为非上市独角兽。产品阶段事实不够成熟,但方向一致:报道提到 45 days 内做出第一代硬件、两代机器人本体、小批量线下生产,以及面向工厂、超市、物流等 B 端封闭场景的 PoC 验证。除杭州外,北京、上海和苏州也被列为战略地点,其中苏州被描述为全球创新中心。已审阅公开来源没有披露收入、客户数量或员工人数;这些是明确的封面指标缺口,而不是零值。[CO014, CO015, CO016, CO017, CO018, CO019]
| 利益相关方 | 角色 | 控制或经济重要性 | 证据状态 | 尽调问题 |
|---|---|---|---|---|
| Jia Peng | 创始人 / CEO / 法定代表人 | 运营和技术关键人 | 由登记信息和媒体印证 | 确认股权、IP 转让、竞业限制状态和继任计划。 |
| Wang Kai | 董事长 / 联合创始人 | 系统架构可信度和资本网络信号 | 由媒体和登记角色印证 | 确认董事长权限、投票权和投资人关联边界。 |
| Wang Jiajia | COO / 联合创始人 | 量产和部署执行负责人 | 由媒体和登记角色印证 | 确认汇报线、KPI 归属和交付组织。 |
| Yuanjing / Vision Capital 与 Lanchi / BlueRun | 财务投资人 | 早期资本和多次投资叙事 | 多家媒体报道 | 索取逐轮认购文件和按比例跟投权。 |
| Sequoia China / HongShan 与 Legend Capital | 财务投资人 | 品牌背书和后续融资信号 | 多家媒体报道 | 确认主体名称、进入轮次日期,以及董事 / 观察员权利。 |
| CAS Star 与 Gaorong | 财务投资人 | 硬科技和成长资本信号 | 多家媒体报道 | 确认投资规模和战略支持承诺。 |
| Tencent | 战略投资人 | 可能关联云、AI、分发或生态 | Sina、Tencent News、Yicai、36Kr 报道 | 澄清商业协议、排他性、数据和合作义务。 |
| Alibaba Group | 战略投资人 | 可能关联云、商业、物流和企业生态 | 与 Tencent 一并被报道 | 澄清投资是否包含商业试点或优惠条款。 |
| Lighthouse / Light Source Capital | 财务顾问 | 最新融资流程信号 | Tencent News 和 36Kr 报道 | 确认委托范围,以及未来融资是否仍在推进。 |
投资人图谱来自公开来源列举;未获得签署版股权结构表、控制权、债务或二级交易条款。
[CO017, CO018, CO019, CO020, CO028, CO029]| 日期 | 事件 | 类型 | 金额 / 状态 | 参与方 | 含义 |
|---|---|---|---|---|---|
| 2025-07-31 | AiQicha 记录法律实体设立 | 成立 | 运营公司成立 | 杭州至简动力科技有限公司;Jia Peng | 公司年龄和尽调时间线的精确锚点。 |
| 2025 年 7 月下旬 | 官网称 Simplexity Robotics 成立于 2025 年 7 月下旬 | 成立 | 品牌 / 公司启动叙事 | Simplexity Robotics | 确认官方具身机器人定位。 |
| 2025 Q3 | 据报道,首名员工到岗后不到 45 天,公司推出第一台自研机器人本体 | 产品 | 原型里程碑(报道) | 核心团队 | 显示执行速度,但需要演示 / 客户验证。 |
| 2025-2026 | 据报道已完成两代 B 端和 C 端机器人本体 | 产品 | 两代本体开发 | 公司团队 | 支撑产品阶段叙事;规格未公开。 |
| 2026-03-09 | 公司公开宣布在不到约六个月内完成五轮融资 | 融资 | 据报道 RMB 2 billion | Yuanjing、Lanchi、Sequoia / HongShan、Legend、CAS Star、Gaorong、Tencent、Alibaba 等投资方 | 将当前阶段定义为重金支持的未上市机器人公司。 |
| 2026-03-09/10 | 媒体报道估值跨过 USD 1 billion | 融资 | 据报道 > USD 1 billion | 多家媒体 | 独角兽标签来自媒体报道;条款仍为私下信息。 |
| 2026-03 | 据报道已进行小批量机器人本体生产和 PoC 验证 | 规模 | PoC / 小批量 | 工厂、超市、物流被列为目标场景 | 运营牵引力仍需客户证据。 |
| 2026-03 | 据报道已布局北京、上海、苏州,并设立苏州全球创新中心 | 规模 | 多地布局 | 公司和地方生态 | 运营足迹超出杭州注册地址。 |
| 2026-04-24 | Baidu Baike 报道其在万物生长大会获浙江 / 杭州独角兽认定 | 治理 | 被认定为独角兽生态参与者 | 杭州活动生态 | 公共身份信号,不是审计后的估值。 |
| 2026-03-22 | NBD 报道具身智能融资热潮和估值担忧 | 负面 | 全行业质疑 | 产业投资人和媒体 | 公司估值应拿商业化证据做压力测试。 |
| 2026-07-03 | 未能从全国门户完整抓取官方负面记录检索结果 | 监管 | 开放尽调项 | GSXT / CreditChina 门户 | 投资交割前,需要实时查询中国登记 / 法律数据库。 |
里程碑按公开来源能提供的最具体日期标注;若公开来源缺少精确日期,近似行已明确说明。
[CO002, CO003, CO014, CO016, CO022, CO023]第一年节奏被压得很紧:法人设立、机器人本体开发、五轮融资、PoC 主张、多地布局搭建,以及赛道层面的估值怀疑。
只有月份 / 季度精度的日期为近似值,因为源文章未给出确切里程碑日期。
[CO002, CO003, CO014, CO016, CO022, CO024]融资和团队信号强,但商业化和治理指标大多仍未公开。
披露融资额和估值来自媒体报道,并非经审计融资条款。
[CO014, CO015, CO016, CO021, CO036, CO037]1.4 里程碑、负面背景与待尽调事项
公开时间线被压得很紧,也带有较强宣传属性:2025 年 7 月完成法律设立;数周内据报推进原型 / 产品;2026 年 3 月宣布融资;随后英文报道将其包装为高速成长的具身 AI 独角兽;2026 年 4 月在浙江 / 杭州生态中获得认可。已审阅页面里最重要的负面证据不是公司特定制裁或诉讼,而是行业层面对具身智能估值、商业化和真实需求风险的怀疑。NBD 称该行业仍处早期,缺少类似 ChatGPT 的已验证范式,并暴露在 FOMO 驱动的估值之下;NetEase 也指出更广泛人形 / 具身机器人热潮中的技术瓶颈、伪需求和盈利担忧。监管尽调在公开检索中也不完整:已审阅国家级和 CreditChina 门户,但静态抓取输出未捕捉到直接面向该公司的官方结果。因此,本章可以核验身份、团队、融资和据报产品阶段里程碑,同时保留经审计财务、签约客户、员工基础、股权结构和官方负面记录检索等缺口。[CO039, CO040, CO041, CO042, CO043, CO044]
| 指标或控制问题 | 公开证据状态 | 重要性 | 下一步尽调 |
|---|---|---|---|
| 收入 / ARR | 已审阅来源均未披露收入或 ARR | 判断估值由业务牵引力支撑,还是由期权价值支撑 | 索取按产品 / 客户拆分的月度收入,以及银行 / 税务佐证。 |
| 客户数 / 已签试点 | 来源提到 PoC 场景,但未点名客户 | 区分实验室演示和企业需求 | 索取已签 LOI、付费 PoC、部署日志、续约条款和客户证明。 |
| 员工数 | 官网列出招聘岗位,但未披露员工数 | 约束烧钱速度、执行能力和团队集中度风险 | 索取薪酬名册、组织架构图、招聘计划和成立以来的离职数据。 |
| 股权结构表 / 治理权利 | 投资人名称公开;权利和持股未公开 | 约束退出经济性和投资人同意风险 | 索取签署版股权结构表、章程、附函和董事会会议记录。 |
| 债务 / 老股转让 | 已审阅来源均未披露债务、授信额度或老股出售 | 影响资金跑道和真实可用的一手资本 | 索取全部融资文件和资金到账 / 用途表。 |
| 官方负面记录 | 静态抓取未捕获公司特定 GSXT / CreditChina 结果 | 监管、诉讼和信用状态是交割条件 | 在 GSXT、CreditChina、法院和税务数据库中实时检索公司名称和统一社会信用代码。 |
本表刻意保留空值和开放项,不从融资新闻倒推运营指标。
[CO036, CO037, CO038, CO039, CO046]1.5 图表
02市场分析
2.1 市场边界:面向受控作业的具身智能机器人
Simplexity Robotics 不应按泛 AI 公司、EV 自动驾驶供应商或广义消费级人形机器人来估算。证据指向一家使用自动驾驶技术栈方法的具身智能机器人公司,初始部署围绕封闭或半封闭的工厂、超市和物流工作流。因此,纳入的支出应覆盖机器人本体、操作或移动模块、车队编排、机器人技能 / 数据迭代、部署集成、维护,以及面向受控工位和场内物流路径的结果挂钩服务模式。排除的支出包括消费聊天机器人 AI、通用云端模型训练、广义 EV 软件、独立 WMS 许可证,以及不具备移动或操作自主性的商品化自动化。现状替代也很宽:人工、输送线、AGV、固定自动化、叉车、外包 3PL 流程和内部自动化团队已经解决了部分工作。这个边界解释了用户背景为何提到智能驾驶:公开报道把创始团队与 Li Auto 自动驾驶和 VLA 工作相连,但可投资市场是实体运营中的机器人部署,不是汽车。[CM001, CM002, CM003, CM004, CM005, CM006]
| 细分 / 类别 | 纳入支出 | 排除支出 | 买方 / 付款方 | 与 Simplexity 的相关性 |
|---|---|---|---|---|
| 面向受控设施的具身智能机器人 | 机器人本体、感知 / VLA 技术栈、操作或移动、技能 / 数据迭代、机队软件、服务 | 与机器人工作无关的通用 AI 软件或云模型支出 | 运营、自动化、工厂、物流负责人 | 核心初始市场边界 |
| 工业内物流与工厂车间 | 物料搬运、装载、巡检支持、线边任务、工厂集成 | 未销售自主层时的传统固定自动化 | 工厂经理 / 工业工程 / 资本开支委员会 | 环境受控,因此是强首个切入点 |
| 仓储与履约机器人 | AMR、AGV、提升机器人、货到人、拣选支持、编排 | 纯 WMS、人工 3PL 劳动、仅输送线升级 | 仓储运营 / 3PL / 电商物流 | 最接近且数据丰富的 SAM 代理 |
| 零售和超市服务运营 | 货架、后仓、补货、库存、客户协助试点 | 消费级家庭机器人和通用零售软件 | 零售运营 / 门店自动化 | Simplexity 点名的初始场景,但公开量化较少 |
| 广义人形和服务机器人 | 仅纳入经济流程清晰的工业 / 服务场景 | 医疗陪伴、教育玩具、消费级人形机器人、演示 | 场景负责人随行业变化 | 是相关政策背景,但作为 TAM 过宽 |
| 自动驾驶技术迁移 | 可复用的感知、VLA、规划、算力、安全和量产方法 | 乘用车软件收入或 robotaxi 车队 | 机器人供应商内部 R&D | 解释团队优势,不是独立买方预算 |
边界把可变现的具身机器人部署支出,与广义 AI、EV 自动驾驶、消费机器人和通用仓储软件区分开。
[CM001, CM002, CM004, CM005, CM006, CM007]同一套感知-规划-量产工具箱能帮助机器人,但操作任务和非结构化作业带来新的证明负担。
该流程概括公开报道和市场逻辑;不主张技术可完整迁移。
[CM004, CM005, CM036, CM037, CM038]2.2 TAM/SAM/SOM 视角,以及为什么数字不能压成一个 TAM
公开数据支持分层测算。全球物流机器人口径是最宽的天花板,Global Market Insights 估算该市场 2026 年为 USD 20.7 billion,2035 年达到 USD 91.4 billion。更窄的仓储机器人口径规模更小,但更接近初始买家:Fortune Business Insights 估算全球仓储机器人 2026 年为 USD 7.35 billion,中国仓储机器人 2026 年为 USD 2.94 billion。MarkNtel 的中国仓储自动化估计相邻但并不完全相同,不过其 2025 年 USD 3.02 billion 基数和 2032 年 USD 9.17 billion 预测,可作为中国自动化 SAM 代理。Simplexity 的公开 SOM 无法计算,因为没有披露付费部署、定价、客户数量、收入或毛利率。正确的尽调姿态,是把中国工业物流机器人视为具身智能里的第一根楔子,再要求管理层证明 PoC 能转化为可重复的付费部署。[CM014, CM015, CM016, CM017, CM018, CM019]
| 发布方 | 年份 | 地域 / 范围 | 数值 | CAGR | 方法 / 用途 | 置信度 | 局限 |
|---|---|---|---|---|---|---|---|
| Global Market Insights 研究机构 | 2026 | 全球物流机器人 | USD 20.7B | 17.9% 至 2035 年 | 物流机器人广义 TAM 上限 | 中 | 包含最后一公里以及中国工业物流之外的全球用例 |
| Global Market Insights 研究机构 | 2035 | 全球物流机器人预测 | USD 91.4B | 17.9% | 长期上限 | 中 | 不是 Simplexity 近期可服务市场 |
| Fortune Business Insights | 2026 | 全球仓储机器人 | USD 7.35B | 16.8% 至 2034 年 | 更贴近仓储机器人 TAM 的口径 | 中 | 全球口径,包含既有 AMR/AGV 品类 |
| Fortune Business Insights | 2026 | 中国仓储机器人 | USD 2.94B | 未单独拆分 | 中国仓储机器人 SAM 代理口径 | 中 | 未覆盖全部具身工厂或超市工作流 |
| MarkNtel Advisors | 2025 | 中国仓储自动化 | USD 3.02B | 17.2% 至 2032 年 | 中国仓储自动化 SAM 替代口径 | 中 | 包含输送机、AS/RS、WMS、服务以及机器人之外的硬件 |
| MarkNtel Advisors | 2032 | 中国仓储自动化预测 | USD 9.17B | 17.2% | 中国自动化长期扩张情景 | 中 | 预测取决于自动化组合和供应商方法 |
| Simplexity 公开资料 | 2026 | 公司特定 SOM | null | null | 未披露 | 低 | 未见公开付费部署、定价、客户或收入 |
数值项以十亿美元计;null 表示公开证据无法针对 Simplexity 单独拆出的规模路径。
[CM014, CM015, CM016, CM017, CM019, CM043]宽口径物流机器人 TAM 先收窄到中国仓储自动化 / 机器人 SAM 代理指标,再落到尚未披露的 Simplexity SOM。
各层数值不可相加;它们是边界和年份不同的嵌套 / 相邻视角。
[CM015, CM016, CM017, CM019, CM043, CM044]可取得的估算把中国相关仓储 / 机器人机会框在区间内:从当下低个位数十亿美元,到本十年后期更大的自动化支出。
所有点位均使用十亿美元,但市场边界混杂;该区间展示估算分歧,不是单一统计模型预测区间。
[CM015, CM016, CM017, CM019, CM045]2.3 买家、用户、付款方与采用路径
买方地图更偏运营,而不是纯技术。在电商履约和 3PL 中,经济买家是物流或仓储运营 VP,日常用户是拣选员、主管和自动化工程师,付款方是物流业务单元或资本开支委员会。在制造业场内物流中,工厂运营和工业工程团队推动采购,IT 与自动化团队则把机器人接入 MES、WMS、安全和设施控制。零售与超市场景增加门店运营和损耗防控干系人;服务场景增加设施或网点运营。JD 的 Zhilang 证据说明买家为什么在意:公开主张包括拣选效率提升超过 3x、存储密度提升 2.5x、回本周期缩短,以及北京一个园区内近 100 台机器人。Geekplus 证据给出在位者标杆:大型 AMR 供应商卖给电商、3PL、服装、医疗、商超、汽车、冷链等高吞吐环境。Simplexity 要么在柔性操作和泛化上胜出,要么找到在位者能力不足的工作流。[CM020, CM022, CM023, CM024, CM025, CM026]
| 细分市场 | 经济买方 | 日常用户 | 付费方 / 预算负责人 | 工作流 | 采用触发因素 |
|---|---|---|---|---|---|
| 电商履约 | 物流 / 履约运营副总裁 | 拣货员、主管、自动化工程师 | 物流资本开支或履约损益 | 货到人、拣选、分拣、存储 | SKU 增长、用工压力、交付 SLA |
| 3PL 与合同物流 | 运营负责人 / 方案设计 | 仓库员工、WMS/RMS 操作员 | 场站资本开支,加上客户合同经济性 | 多客户履约与仓储 | 赢得合同或降低人工成本 |
| 制造业厂内物流 | 工厂经理 / 工业工程 | 产线操作员、物料搬运员、安全团队 | 工厂资本开支 / 精益制造预算 | 线边搬运、配套备料、检测、转运 | 吞吐瓶颈或质量 / 安全需求 |
| 汽车与电子工厂 | 自动化负责人 / 生产工程 | 技术员、工艺工程师、维修人员 | 制造工程预算 | 零部件搬运、精密操作、测试支持 | 柔性自动化与固定工装取舍 |
| 超市与零售运营 | 零售运营 / 门店自动化 | 门店员工、补货团队、库存团队 | 门店运营或连锁级转型预算 | 后仓、货架、库存、服务任务 | 用工短缺、降低缺货、提升门店效率 |
| 冷链与医疗物流 | 冷链运营 / 医药供应链负责人 | 仓库员工、质量 / 合规团队 | 受监管物流或质量预算 | 温控仓储、可追溯搬运 | 合规、准确性和用工约束 |
买方图谱来自物流机器人市场定义、JD 部署证据以及 Geekplus 披露的客户细分。
[CM020, CM022, CM023, CM024, CM025, CM026]采购要让运营发起人、日常用户、集成负责人和财务付款方围绕 ROI 达成一致。
角色映射由买方证据和市场分段定义综合得出;Simplexity 的确切买方头衔未公开。
[CM022, CM023, CM025, CM026, CM046]中国政策有助于催生场景,但买方在规模化部署前仍需要 KPI 验证和经济性证明。
该漏斗是政策和采购流程图,不是转化率估算。
[CM008, CM009, CM040, CM041, CM047]2.4 增长驱动、中国政府支持与自动驾驶技术栈迁移
增长由政策、劳动力、电商、物流复杂度,以及把 AI 转化为实体生产力的诉求共同拉动。中国 2026 年 MIIT/SASAC 行动尤其明确:推动真实场景训练、创新联合体、技能包、验证、常态化部署、融资支持,甚至租赁或 robot-as-a-service 模式。更广的 Robot+ 政策和新的人形 / 具身 AI 标准体系增加合法性,也帮助打开国企和地方场景。市场数据同样支持需求:物流机器人驱动因素包括电商增长、劳动力短缺、AI 导航进步、RaaS、效率和韧性。自动驾驶迁移重要,是因为相同的 VLA / 世界模型、感知、导航、计算、电力和量产纪律可以缩短机器人迭代周期。可是,相比道路,机器人还需要在更不标准化的环境里处理操作、平衡、人机交互和异常场景鲁棒性;汽车技术栈可信度是起跑优势,不是采用证明。[CM008, CM009, CM010, CM011, CM012, CM020]
| 驱动因素 / 约束 | 方向 | 时间 | 含义 | 尽调追问 |
|---|---|---|---|---|
| MIIT 真实场景训练与部署行动 | 驱动 | 2026 | 打开代表性场景和政府关联买方 | 列出 Simplexity 已提交或获接纳的场景 |
| Robot+ 与人形机器人政策支持 | 驱动 | 当前至 2020 年代末 | 为制造和物流场景采用机器人背书 | 梳理地方补贴、试点和采购路径 |
| 电商与全渠道履约 | 驱动 | 当前 | 推高对快速、准确、柔性仓储自动化的需求 | 按细分市场验证客户待交付需求 |
| 用工短缺与工资压力 | 驱动 | 当前 | 改善自动化 ROI,提升 RaaS 吸引力 | 按任务和城市量化回本周期 |
| AI、计算机视觉和自主导航进步 | 驱动 | 当前 | 扩大机器人在低结构化工作流中的任务边界 | 对标既有厂商的任务成功率 |
| 自动驾驶技术栈迁移 | 驱动 | 近期 | 可能缩短感知、规划、安全和生产学习曲线 | 区分可迁移模块和机器人特有缺口 |
| 既有厂商已部署 AMR 车队 | 约束 | 即期 | 抬高 Simplexity 的客户验证门槛 | 对 Geekplus/Hai/Hikrobot/Quicktron 做输赢复盘 |
| 高资本开支和 ROI 不确定性 | 约束 | 即期 | 需要租赁、RaaS 或结果付费 | 要求已签约付费试点经济性 |
| 存量系统集成复杂度 | 约束 | 即期 | 拖慢既有设施采用 | 审核 WMS/MES/安全集成工作量 |
| 算力、数据与出口管制限制 | 约束 | 2026 年起 | 可能推高训练成本、拖慢模型迭代 | 核查算力供应链和数据权利 |
这张表把市场力量接到采用节奏和尽调动作上,而不是把增长视为自动发生。
[CM008, CM009, CM010, CM020, CM021, CM028]| 政策 / 支持层 | 发布方 | 机制 | 市场影响 | 内含约束 |
|---|---|---|---|---|
| 2026 年真实场景训练行动 | MIIT / SASAC | 代表性场景、创新联合体、技能包、部署验证 | 打开工业、服务和特殊场景入口 | 要求安全性、经济可行性和可量化验证 |
| Robot+ 应用计划 | State Council / MIIT 及其他 16 个部门 | 加深机器人在制造、物流、医疗、农业等场景的应用 | 扩大公私部门预算正当性 | 落地取决于地方场景质量 |
| 人形机器人与具身 AI 标准体系 | MIIT 委员会 / Xinhua 报道 | 围绕大脑、计算、肢体、本体、应用、安全、伦理制定标准 | 提升买方信任和互操作性 | 抬高合规预期 |
| 生成式 AI 暂行办法 | CAC | 规范生成式 AI 服务提供、安全和数据治理 | 塑造具身 AI 系统的模型 / 数据合规框架 | 数据来源和隐私可能拖慢部署 |
| World AI Conference 生态 | WAIC | 为 AI 公司提供行业聚集和政策 / 产业平台 | 支撑叙事和合作伙伴密度 | 参会露出不是客户验证 |
| 金融支持与 RaaS 鼓励 | MIIT / SASAC | 股权、债务、保险、租赁、按次付费、机器人即服务 | 可降低买方前期门槛 | 需要风险共担和服务可靠性 |
政府支持力度强,但每个政策杠杆都内嵌验证、安全、合规或商业化要求。
[CM008, CM009, CM010, CM011, CM012, CM013]2.5 约束、矛盾与尽调缺口
采用约束很实质。公开市场估算方向积极,但边界互不兼容;分析师来源混用了仓储自动化、仓储机器人、物流机器人,有时还混入更广义的服务机器人。在位者已经有部署车队、客户证明和专用工作流,因此 Simplexity 不能假设具身 AI 叙事会自动转化为 SOM。技术风险仍在:公开报道强调训练难度高、算力消耗大、数据需求强,以及机器人作业相较驾驶面对更难的随机性。监管和地缘政治约束也重要,因为具身 AI 模型触及数据治理和先进计算供应。最关键的是,已审阅公开来源没有披露 Simplexity 的付费客户数量、价格、回本期、管线转化、在线率、毛利率或支持成本。市场有吸引力,但投资问题在于:Simplexity 能否在在位者吸收柔性自动化用例之前,把政策推动的试点转成可重复付费部署。[CM021, CM030, CM031, CM032, CM033, CM034]
| 缺口 | 为什么重要 | 当前证据 | 解决负责人 | 下一步尽调 |
|---|---|---|---|---|
| Simplexity 付费客户数量 | 决定 SOM 和销售速度 | 未发现公开数量 | 管理层 / 客户 | 按细分市场和地域索取付费部署清单 |
| 单价和毛利率 | 把 TAM 转成收入和估值支撑 | 未发现价格或毛利率 | 管理层财务 | 收集价目表、BOM、支持成本、折扣 |
| 试点到量产转化 | 区分演示和可重复采用 | 仅有公开 PoC / 小批量证据 | 销售 / 客户背调 | 审阅试点管线、转化率、流失率、正常运行时间 |
| 相对既有厂商的任务级 ROI | 买方会把回本周期同 AMR、AGV、固定自动化和人工比较 | 已有 JD 和 Geekplus 基准;Simplexity 基准缺失 | 产品 / 客户工程 | 在工厂和仓库跑并行任务经济性对比 |
| 算力和数据权利 | 影响模型迭代成本和合规 | 已识别 BIS 和 CAC 约束 | 法务 / 基础设施 | 审核算力供应商、训练数据权利、隐私控制 |
这些缺口解释了为什么市场方向可以有吸引力,但公司特定份额获取仍未解决。
[CM021, CM032, CM038, CM039, CM043, CM045]2.6 图表
03竞争格局
3.1 格局:直接具身机器人同业与 AV 技术栈邻近竞争者
最清晰的竞争框架,是把直接具身机器人同业与自动驾驶技术栈在位者和邻近公司分开。直接同业包括 AgiBot、Unitree、UBTECH、Galbot,以及中国人形和具身 AI 创业公司的长尾,因为它们销售或正在商业化实体机器人、机器人本体、机器人开发者平台,或面向工厂、零售、物流、教育和服务场景的任务机器人。Simplexity 处在同一类广义待完成任务中,因为官网将其描述为围绕一个基础模型、数据闭环效率和可靠机器人硬件打造的全场景具身机器人公司。Momenta、DeepRoute.ai、Horizon Robotics、Baidu Apollo、Pony.ai、WeRide、Mobileye 和 AutoX 等邻近竞争者不同:它们主要是自动驾驶软件、芯片、robotaxi 或 ADAS 技术栈。它们不会一对一替代一款人形机器人 SKU,但会争夺人才、资本、训练数据、OEM 信任、安全案例和世界模型可信度。这个区分很重要:Simplexity 可能把直接机器人订单输给价格更低或部署更多的人形机器人,也可能在 AV 技术栈玩家把车队规模实体 AI 资产转成机器人平台时失去战略重要性。[CP001, CP002, CP003, CP004, CP005, CP009]
| 竞争者 / 类别 | 类别 | 规模 / 融资信号 | 目标细分市场 | 差异化 | 相对 Simplexity 或买方需求的主要短板 |
|---|---|---|---|---|---|
| Simplexity Robotics / 至简动力 | 目标公司 | 2025 年 7 月下旬成立;Gasgoo 报道其完成 $50 million 天使轮 | 全场景具身机器人;首个买方尚未公开 | 统一世界模型 / VLA 架构、端侧训练语言、模型定义本体论点 | 尚无公开部署、定价、客户或产量证据 |
| AgiBot / Zhiyuan Robotics | 直接具身 AI 机器人同业 | 官方和 PR 来源称到 2026 年有 15,000 台机器人下线 | 工业和真实运营场景;开发者生态 | 多条机器人产品线、量产叙事、Genie / 开发者生态、部署年份定位 | 厂商自述的生产和部署说法需要客户层面佐证 |
| Unitree | 直接机器人硬件同业 | 公开产品目录和商城价格;TrendForce 预计其与 AgiBot 领先 | 开发者、教育、消费、工业巡检、人形机器人实验 | 入门价格低且可见,四足 / 人形硬件组合广 | 低价硬件可能让本体商品化,但企业任务软件深度不够可见 |
| UBTECH Robotics | 上市机器人既有厂商 | 港股上市 / 上市公司披露面,以及人形 / 服务机器人组合 | 教育、服务、工业人形、物流、医疗 | 上市公司身份、Walker 系列和更长机器人行业历史 | 已审阅页面未清楚披露当前人形机器人详细定价和部署经济性 |
| Galbot | 直接具身 AI 初创同业 | 2026 年报道称新一轮融资 >$300M、累计融资 $800M、估值 $3B | 工业、零售、仓储物流、智慧城市服务 | 全栈具身 AI、工厂 / 零售 / 仓库部署说法、战略工业投资者 | 官网自身可读细节很少;PR 说法需要买方尽调 |
| Momenta / DeepRoute.ai / AV AI 初创公司 | 相邻物理 AI 技术栈 | 具备车队运营和模型经验的自动驾驶软件公司 | 汽车 OEM 与辅助 / 自动驾驶项目 | 道路世界感知、规划、数据基础设施、安全工程 | 已审阅证据中不是直接人形机器人产品 |
| Horizon / Baidu Apollo / Mobileye | 既有 AV/ADAS 平台 | 上市公司或大型平台规模、OEM 渠道、芯片 / ADAS / robotaxi 生态 | OEM、智能驾驶车辆、robotaxi / ADAS 部署 | 分发能力、监管姿态、量产供应链、边缘计算可信度 | 主要争夺物理 AI 技术栈控制权,而不是机器人本体销售 |
| Pony.ai / WeRide / AutoX 自动驾驶对照组 | Robotaxi 与自动驾驶运营商 | Pony 与 WeRide 有公开 IR 披露;AutoX 官网抓取返回不可用的 404 页面 | Robotaxi、robotruck、自动出行 / 物流 | 车队运营、牌照、安全案例、地域、客户合同 | 商业模式仍早期;无法从所给 URL 取回 AutoX 当前公开证据 |
竞争者类别有意把直接机器人本体 / 平台同业与 AV 技术栈相邻玩家分开;规模单元格只使用已抓取的公开证据,不纳入缺乏支持的运营指标。
[CP001, CP004, CP005, CP007, CP009, CP012]该顺序图按机器人具身程度和商业化证据,把直接具身机器人同行与相邻 AV 技术栈竞争者分开。
分数是基于已抓取来源证据给出的 1-10 顺序判断;x 代表机器人具身 / 直接程度,y 代表商业化证据。
[CP001, CP007, CP009, CP012, CP013, CP020]3.2 直接同业画像与商业化信号
直接同业在外部证据上明显领先 Simplexity。AgiBot 自有页面和 2026 年公告强调多条机器人产品线、全栈具身智能架构,以及第 15,000 台机器人下线里程碑;这是一家 2025 年 7 月下旬才成立的公司无法公开匹配的规模信号。Unitree 的差异化不在不透明企业口径,而在可见的产品宽度和价格透明度:其官网列出人形机器人和四足机器人,G1 起售价 $13,500、Go2 起售价 $1,600,给买家和开发者形成低价公开锚。UBTECH 提供上市公司披露和人形 / 服务机器人组合,即便 Walker 的详细定价仍不可得。Galbot 是已审阅来源中融资最强的直接创业同业,2026 年报道称其新融资超过 $300 million、估值 $3 billion,已有自主零售部署、工厂用例和来自产业方的战略资本。Simplexity 的单模型和端侧训练表述技术野心很强,但当前竞争缺口是商业化证据,不是愿景。[CP001, CP002, CP006, CP007, CP008, CP009]
| 购买标准 | Simplexity | AgiBot | Unitree | UBTECH | Galbot | AV 技术栈相邻玩家 |
|---|---|---|---|---|---|---|
| 统一具身基础模型 | 宣称整合世界模型和 VLA | 具身智能和基础模型说法;G2/A2/G1 产品线 | 硬件先行;G1 有 AI 化身 / OTA 说法 | 核心机器人技术和人形机器人产品 | 全栈具身 AI 和 GraspVLA 说法 | Pony/Momenta 等有世界模型和虚拟驾驶员技术栈,但用于车辆 |
| 机器人本体 / 硬件组合 | 规划全场景机器人;无公开 SKU | 多条产品线,包括 A2/G1/X2/X1/G2 | 人形和四足机器人组合强且可见 | Walker 以及服务 / 教育 / 物流产品线 | G1 及其他人形 / 零售 / 仓储说法 | 没有人形本体证据;证据指向车辆平台 |
| 生产与部署证据 | 公开产量未知 | 第 15,000 台机器人里程碑与部署年度叙事 | 公开目录与 TrendForce 规模预测 | 上市公司有产品组合;这里未拆出当前出货量 | 新闻稿来源称有数千台订单和 30 城零售 | 自动驾驶公司有公开道路 / 车队运营和 OEM 渠道 |
| 端侧或边缘执行 | 称可在端侧实时推理 / 训练 | 具身系统可部署;边缘训练的确切经济性未知 | 产品页显示嵌入式机器人控制 | 机器人应用和工业产品线;细节未知 | 仓储 / 工厂自主能力说法;训练栈细节有限 | Horizon / Mobileye 在边缘计算和车端推理上资产很强 |
| 开发者生态 / 平台 | 招聘和模型论点可见;未找到公开 SDK | Genie / 开发者生态和 RMB 2B 生态计划 | 商店、文档 / 下载页、OTA 表述 | 教育套件和机器人产品 | 合作 / JV 路径比开放平台更清楚 | Apollo 开放平台和自动驾驶开发者生态最强 |
| 信任、安全和监管姿态 | 尚未公开 | 有工厂部署和合作伙伴大会说法;未审阅独立安全记录 | 消费者 / 开发者可获得性有利于采用,但安全和责任仍需尽调 | 上市公司披露提高了信任面 | 工业投资方和客户说法支撑信任,但主要来自新闻稿 | 自动驾驶公司具备监管许可、披露文件和安全案例基础设施 |
| 定价透明度 | Unknown | 已抓取页面未披露 | G1 起价 $13.5K;Go2 起价 $1.6K;H1 公开信号冲突 / 需联系 | Walker / 人形企业机器人未知 | Unknown | 多为企业 / 车辆项目定价,未知 |
| 分销 / 渠道杠杆 | Unknown | 合作伙伴 / 开发者生态在扩大 | 全球品牌和商店渠道 | 上市公司和企业机器人渠道 | CATL / Bosch / SAIC 相关投资方和 JV 信号 | OEM、公开市场、无人出租车和车队渠道 |
标为「未知」的单元格被刻意保留:抓取来源无法支持定价、部署或商业条款。
[CP002, CP004, CP005, CP007, CP008, CP009]| 供应商 / 类别 | 公开定价信号 | 打包方式 / 合同模式 | 已证实的包含能力 | 未知项 / 注意事项 | 含义 |
|---|---|---|---|---|---|
| Simplexity | Unknown | 未知;若已商业化,可能是早期企业 / 试点销售 | 统一模型、数据闭环、端侧训练、机器人硬件论点 | 没有公开 SKU、标价、试点费用或服务条款 | 任务 ROI 和正常运行时间未跑出来前,无法支撑价格溢价 |
| Unitree G1 | 官方产品页称起价 $13.5K;商店标价 $13,500 | 自助商店,加合作 / 联系通道 | 人形本体、OTA 更新、灵巧运动表述 | 企业支持、自主栈和实际折扣未知 | 给低成本人形硬件划出可见锚点 |
| Unitree Go2 | 官方产品页称起价 $1,600 | 自助商店 / 产品销售 | 四足机器人,带 OTA 和应用生态表述 | 并非所有任务的人形替代品 | 说明机器人硬件价格可以远低于企业试点 |
| Unitree H1 | 官方 H1 页面未给出清晰标价;商店标题写着「联系我们」,相关列表则显示 $90,000 | 联系销售 / 企业式销售 | 全尺寸通用人形定位 | 公开价格信号冲突,需要直接报价核验 | 尽调模型中按未知 / 需报价处理 |
| AgiBot | 已抓取官方 / 新闻稿来源未披露 | 机器人产品线销售,加合作伙伴 / 开发者生态 | G2 工业机器人、A2/G1/X2/X1 产品线、Genie / 开发者生态 | 没有公开 ASP、租赁、RaaS 或维护经济性 | 规模说法重要,但无法推断毛利率 |
| Galbot | 已抓取来源未披露 | 企业 / 工业、零售解决方案、仓储物流、合作伙伴 | G1 零售 / 工厂 / 仓储说法,自主商店概念 | 新闻稿来源提到订单,但没有单台价格或服务费 | 融资规模可能补贴部署,也可能补贴价格战 |
| UBTECH Walker / 服务机器人 | 已审阅页面未披露 | 企业 / 产品解决方案销售和上市公司报告 | 人形服务、工业 Walker 导航、教育 / 服务产品 | Walker 价格、部署 ARR 和单位经济性未披露 | 上市身份有助于尽调,但不能提高定价透明度 |
定价采取刻意保守口径:没有支持的单元格保留为未知,不从融资、估值或相邻产品页推断。
[CP001, CP010, CP011, CP012, CP035, CP041]能力图展示 Simplexity 宣称的架构,相对于同行的规模、硬件和渠道证据。
定性评级使用直接审阅证据;未知表示未抓取到价格、出货量或合同证据。
[CP002, CP005, CP007, CP009, CP010, CP012]3.3 邻近 AV 技术栈压力:数据、安全、OEM 渠道与实体 AI 人才
自动驾驶技术栈公司不应被误标为直接人形机器人竞争者,但战略上仍然重要。Horizon 的公开申报文件和官网显示量产定点、Journey 系列出货、OEM 关系和 2024 年收入;这些都是具身机器人创业公司想借用的资产:安全工程、边缘推理、感知栈和车规级供应链。Pony.ai 和 WeRide 的公开投资者材料围绕 robotaxi、robotruck、许可、多国运营和牌照;这些不是人形机器人部署,但证明了车队运营、监管穿越和物理世界 AI 商业化能力。Baidu Apollo 和 Mobileye 还增加平台和 ADAS 分发深度。Gasgoo 关于汽车高管流向具身智能的报道尤其值得关注,因为它展示了 AV 与机器人之间的人才和资本桥梁,也包括 Simplexity 自身来自 Li Auto 的创始人谱系。按承销视角,AV 在位者是二阶竞争威胁:它们未必先于 Simplexity 做出机器人原型,但可以在年轻机器人创业公司证明锁定之前,塑造标准、吸走人才,并与 OEM 或工厂结盟。[CP013, CP014, CP020, CP021, CP022, CP023]
| 相邻玩家 | 主要栈 | 规模 / 证据信号 | 如何与 Simplexity 竞争 | 替代直接性 | 尽调要求 |
|---|---|---|---|---|---|
| Horizon Robotics | 智能驾驶计算 / 软件 | 官方页面称 Journey 出货 10M+、定点 400+、覆盖 300+ 车型、40+ OEM 品牌;2024 年收入 RMB2.38B | OEM 渠道、边缘 AI、感知、量产工程 | 相邻,不是直接机器人本体 | 询问 Horizon 或 OEM 合作伙伴是否把边缘 AI 延伸到工厂 / 移动机器人 |
| Pony.ai | Virtual Driver、PonyWorld、无人出租车 / 无人卡车 / 授权 | 投资者关系材料和 F-1 描述无人出租车、无人卡车、授权业务及早期商业化风险 | 世界模型、车队运营、安全、监管可信度 | 相邻物理 AI 竞争者 | 验证 Pony 类栈能否供应机器人自主模块 |
| WeRide | WeRide One 自动驾驶平台 | 投资者关系材料称在 12 个国家 40+ 城运营 / 测试,并在 8 个市场持有许可 | 多产品 L2-L4 平台和监管足迹 | 相邻 | 评估乘用车之外的交叉授权或 OEM 合作 |
| Baidu Apollo | 自动驾驶和智能汽车平台 | Apollo 页面称 Baidu 2013 年开始做自动驾驶,2017 年推出开放平台 | 平台生态、地图 / 数据、OEM 和客户信任 | 相邻在位者 | 判断 Apollo 生态是否进入具身 / 物流机器人 |
| Mobileye | ADAS 和消费级自动驾驶栈 | Mobileye 对外销售 ADAS / 自动驾驶技术和 SuperVision 桥接方案,并有投资者关系披露 | 计算机视觉安全栈和全球车企渠道 | 相邻在位者 | 核查机器人 OEM 是否采用 Mobileye 软件 / 硬件 |
| Momenta / DeepRoute.ai 自动驾驶对照组 | 自动驾驶软件 | 官方页面确认物理 AI / 自动驾驶定位,但可读细节有限 | 人才和模型相邻性更明显,机器人产品证据不足 | 相邻,直接证据较弱 | 作为直接竞争者打分前,先用客户 / OEM 公告刷新证据 |
| AutoX | 无人出租车 / 自动驾驶 | 提供的官方 URL 在抓取输出中返回 404 正文 | 已审阅官方 URL 无法提供当前证据 | 核验前不打分 | 引用当前竞争规模前,先使用其他官方备案 / 页面 |
本表把自动驾驶栈公司视为渠道、数据、安全和资本威胁;除非出现公开机器人本体证据,否则不把它们当作直接人形替代品。
[CP013, CP014, CP020, CP021, CP022, CP023]简要说明为什么 AV 技术栈既不是直接人形替代品,却仍然重要。
KPI 数值是来源报道的公开主张,不是标准化市占率。
[CP013, CP020, CP021, CP022, CP024, CP026]3.4 护城河耐久性、价格压力、锁定与商品化风险
最大负面信号是,中国人形机器人市场已经从演示走向生产规模和价格分层。TrendForce 预计中国人形机器人产量将在 2026 年大幅增长,并预测 Unitree 加 AgiBot 接近出货量的 80%;DirectIndustry 则提醒,许多公开机器人视频只是演示而非耐久生产证明,同时仍称 2025 年头部玩家也只有数千台。这个组合对 Simplexity 很危险:高端模型主张可能重要,但硬件价格底线、供应链能力和部署背书,可能在后来者建立买家锁定之前就把早期功能商品化。公开定价也不对称。Unitree 给出明确入门价格,而 Simplexity、AgiBot、Galbot 和 UBTECH 的企业人形机器人成本大多未知或以销售询价为主。未知定价应保持未知,不能推断。Simplexity 的可防御性因此取决于能否证明统一 VLA / 世界模型、端侧训练和硬件架构能带来可衡量的任务学习、在线率和数据闭环优势,并足以抵消更便宜的机器人本体、更大的同业车队或 AV 技术栈分发能力。[CP010, CP011, CP030, CP031, CP032, CP033]
| 护城河主张 | 威胁 | 严重度 | 当前证据 | 缓释措施 / 尽调要求 |
|---|---|---|---|---|
| 统一 VLA / 世界模型加端侧训练,可形成任务学习优势 | AgiBot、Galbot 和自动驾驶栈公司也称拥有具身或世界模型资产 | 高 | Simplexity 声称采用该架构,但 AgiBot / Galbot / Pony 来源也有类似物理 AI 表述,且规模更强 | 要求提供并排任务学习基准和边缘训练日志 |
| 可靠机器人硬件若与模型定义本体配套,可形成差异化 | Unitree 激进定价机器人本体,并发布可购买的硬件 SKU | 高 | G1 起价 $13.5K、Go2 起价 $1.6K,形成可见硬件价格锚点 | 拆分 Simplexity 硬件 BOM、正常运行时间、载荷、灵巧度和服务成本 |
| 部署量上来后,数据闭环效率可以复利 | AgiBot 和 Galbot 已称拥有数千台设备 / 订单,或 15,000 台下线里程碑 | 高 | AgiBot 新闻稿和 TrendForce 显示其规模领先;Galbot 新闻稿称有数千台订单 | 核验 Simplexity 数据获取计划、专有任务语料和客户排他性 |
| 中国具身 AI 需求增长可托起所有供应商 | 产量增长也会让功能商品化,并压出价格竞争 | 中 | TrendForce 预计 2026 年中国产量增长 94%,Unitree / AgiBot 份额接近 80% | 建模下行情景:ASP 更快压缩,追随者劣势扩大 |
| 创始人 / 汽车栈背景提高可信度 | 汽车高管正涌入同一赛道,履历独特性下降 | 中 | Gasgoo 报道多位自动驾驶 / 汽车高管进入具身智能,包括 Zhijian Power | 核查招聘管线、竞业限制,以及获取 Li Auto 级数据 / 资产的能力 |
| 早期不透明保留战略可选性 | 定价和客户证据不透明,相比公开同行会削弱买方信任 | 中 | 多数直接同行缺少公开价格,但 Unitree 及上市公司 / 披露型同行提供更多外部证据 | 要求提供试点合同、LOI、客户访谈、保修条款和售后支持 SLA |
严重度衡量的是 Simplexity 赢得试点、维持价格、建立专有数据锁定的压力,不是整个具身机器人市场是否增长。
[CP002, CP007, CP010, CP011, CP016, CP019]Simplexity 的护城河目前仍以命题为主,必须经受价格、规模、数据和渠道压力验证。
就绪度分数是基于已抓取公开证据有无得出的定性尽调判断。
[CP001, CP002, CP010, CP011, CP030, CP031]3.5 图表
04财务
4.1 收入模式与定价可见度
Simplexity 的公开财务档案支持一个混合型企业机器人收入论点,而不是一个已经成型的收入模式。公司官网说明其是一家新成立的具身智能公司,正在建设模型、数据闭环和机器人硬件能力;Preqin 是保留来源中唯一明确描述收入来自向企业客户直接销售机器人硬件、配套软件模型和自动驾驶系统解决方案的来源。对一家全栈机器人公司来说,这个方向合理,但不等于公司披露的收入结构。已审阅官方页面没有发布 SKU、机器人 ASP、软件许可费、服务费率、试点转付费转化、折扣或合同期限。 财务含义是,所有价格和收入质量分析都必须保持条件式。据报封闭场景目标包括工厂、超市和物流,指向 B2B 部署而不是消费硬件规模;早期 PoC 验证是有用的牵引信号。但如果没有签署合同金额、付费部署、使用率、续约权和实施成本,PoC 仍不能计入经常性收入。因此,本章把硬件、软件和解决方案交付视为有支撑的收入流,但对 ARR、当前收入、已实现定价、毛利率、CAC 和回本期留为空值。[CI001, CI002, CI009, CI010, CI011, CI012]
| 收入流 | 机制 | 计量单位 | 当前公开数值 / 状态 | 质量 | 尽调要求 |
|---|---|---|---|---|---|
| 机器人硬件销售 | 面向企业部署的自研机器人本体 | 机器人 / 车队 / 部署 | Preqin 提到直接硬件销售;公司收入未披露 | 中 | 提供机器人 ASP、出货量、确认收入、COGS、保修和退货政策 |
| 配套软件模型 | 随机器人部署交付的 LaST / ManualVLA 式模型能力 | 授权、订阅或捆绑软件 | 官方技术展示可见;打包方式和定价未知 | 低 | 披露软件是单独定价、按用量计费、捆绑销售,还是服务带动 |
| 自动驾驶系统解决方案 | 依托自主能力专长和机器人系统的企业解决方案 | 项目 / 解决方案合同 | Preqin 提到该收入流;客户合同未公开 | 低 | 提供已签合同、里程碑、收入确认政策和毛利率 |
| PoC / 试点服务 | 工厂、超市、物流等封闭场景验证 | 试点费或免费验证 | 有 PoC 报道,付费状态未知 | 低 | 拆分付费试点和免费试用,并披露转化率 |
| 数据与模型改进闭环 | 采集人 / 机器人数据,用于改进具身模型 | 内部能力,可能成为服务组件 | 资金用途包括数据采集;变现未披露 | 低 | 量化每项任务的数据采集成本,以及客户是否为定制付费 |
| 售后支持与集成 | 部署、维护、安全支持和现场集成 | 服务费 / 保修 / 捆绑支持 | 未披露支持定价 | 低 | 提供服务附加率、现场人工小时、SLA 条款和保修准备金 |
收入来源是公开来源可支持的机制,不是 Simplexity 已披露的收入结构或瀑布。
[CI001, CI002, CI009, CI010, CI011, CI012]| 定价项 | 公开证据 | 标价与实际成交价 | 折扣 / 未知项 | 财务含义 |
|---|---|---|---|---|
| 机器人硬件 ASP | 没有公开 Simplexity 价目表 | Unknown | ASP、批量折扣、保修和安装条款未披露 | 没有管理层订单簿前,不建模硬件收入 |
| 软件 / 模型授权 | 官方技术展示可见;Preqin 称配套软件模型有销售 | Unknown | 独立定价还是捆绑定价未知 | 软件附加可能抬升毛利率,但尚未证实 |
| 解决方案交付 / 自主系统 | Preqin 提到面向企业客户的自动驾驶系统解决方案 | Unknown | 里程碑计费和验收标准未披露 | 项目收入可能不平滑,且服务占比高 |
| PoC 费用 | 36Kr 和 Sina 报道 PoC 验证 | Unknown | 付费 / 免费状态及转化率未披露 | 把 PoC 视为牵引力,而不是收入 |
| 战略投资方渠道 | 据报道 Tencent 和 Alibaba 参投 | 不是定价证据 | 没有公开转售商或渠道经济性 | 战略资本可能帮助分销,但不应计入收入 |
| 可比公开价格锚点 | TrendForce 提到 Unitree 毛利率;披露文件显示机器人业务亏损 | 仅可比 | 不是 Simplexity 价格点 | 只用于界定尽调问题,不用于预测 |
本表刻意避免编造价格;每个未知定价单元格都是必需的数据室请求。
[CI005, CI007, CI009, CI010, CI012, CI024]公开证据支持潜在的“硬件 + 软件 + 解决方案”路径,但每个变现节点仍需私下确认。
仅为定性桥接;没有公开来源披露定价、收入确认或利润率。
[CI001, CI002, CI009, CI012, CI028, CI030]4.2 单位经济性、成本结构与可比公司申报
Simplexity 没有披露通常支撑财务模型的单位经济性输入:机器人 BOM、制造良率、软件附加率、现场服务人力、质保率、支持强度、毛利率或回本期。因此,最好的公开证据只能是间接的。据报资金用途集中在基础模型训练、机器人本体研发与迭代、数据采集和核心算法,这指向一套横跨算力、机器人工程、数据运营、原型制造和现场验证的成本基础。这个画像与纯软件公司根本不同,不适合套用没有支撑的 SaaS 利润率假设。 可比自动驾驶和人形机器人公司的公开申报说明了为什么必须谨慎。WeRide 2025 年收入为 RMB684.6 million,但毛利只有 RMB4.9 million,净亏损 RMB1.25 billion。UBTECH 2024 年收入超过 RMB1.3 billion,但仍处亏损。Horizon 2024 年收入超过 RMB2.38 billion,并录得大额 non-IFRS 净亏损。Pony AI 2025 年收入扩至 US$90 million,但申报记录仍显示,从技术商业化到持久盈利还有很长路径。这些申报不是 Simplexity 的事实;它们是护栏,说明机器人收入可以与重度亏损、客户集中和营运资金压力并存。[CI006, CI015, CI016, CI017, CI018, CI019]
| 指标 | 公开值 | 置信度 | 为何重要 | 尽调要求 |
|---|---|---|---|---|
| 机器人 BOM | null | 低 | 决定硬件毛利率下限 | 提供 BOM 历史、供应商条款和各批次良率 |
| 制造良率 / 返工 | null | 低 | 决定扩产成本和保修风险 | 提供试点制造良率、报废和返工工时 |
| 软件附加率 | null | 低 | 决定模型 IP 能否在硬件之外变现 | 按客户分群提供附加率和已确认软件收入 |
| 每次部署服务人工 | null | 低 | 机器人部署中,服务人工可能吃掉硬件毛利 | 提供实施小时数、支持工单和 SLA 成本 |
| CAC / 回本周期 | null | 低 | 检验企业销售打法是否资本效率高 | 提供销售周期、管线转化率和毛利回本周期 |
| 毛利率 | null | 低 | 机器人部署能否经济性扩张的核心证据 | 按产品线提供经审计或月度毛利率 |
| 客户使用率 / 续约 | null | 低 | 区分演示和可重复商业使用 | 提供使用日志、续约条款和客户流失率 |
所有值均为空值,因为留存的公开来源没有披露 Simplexity 单位经济性;可比公司披露只能定义该问什么。
[CI010, CI029, CI032, CI035, CI037, CI039]| 可比公司 | 披露指标 | 纳入原因 | 对 Simplexity 的启示 | 局限 |
|---|---|---|---|---|
| Pony AI | 2025 年收入 US$90.0M,同比增长 20.0% | 自动驾驶商业化申报可比公司 | 说明技术型自动驾驶收入可以逐步放大 | 产品、上市公司阶段和车队结构不同 |
| WeRide | 2025 年收入 RMB684.6M,毛利 RMB4.9M,净亏损 RMB1.25B | 机器人出租车 / 机器人巴士 / 自动驾驶申报可比公司 | 说明收入可以与接近零的毛利和大额亏损并存 | 不是人形机器人公司 |
| Mobileye | 2025 年中国出货目的地占比 23%;净亏损 US$392M | 汽车自动驾驶供应商可比公司 | 说明地域和 OEM 集中度可能很重要 | 成熟上市供应商,不是创业公司 |
| UBTECH | 2024 年收入约 RMB1.305B,净亏损约 RMB1.160B | 上市人形 / 机器人可比公司 | 说明人形机器人收入不保证盈利 | 产品组合和上市阶段不同 |
| Horizon Robotics | 2024 年收入 RMB2.384B;non-IFRS 净亏损 RMB1.681B;最大客户占比 31.5% | 中国 AI / 自动驾驶芯片与软件可比公司 | 用来框定客户集中度和研发亏损风险 | 芯片 / 平台组合与 Simplexity 不同 |
| TrendForce Unitree 基准 | Unitree / Agibot 预期产出份额较高;Unitree 毛利率被引述接近 60% | 正向行业基准 | 说明头部机器人公司可以做到正毛利 | 不能证明 Simplexity 也有类似毛利 |
这些只是申报或分析师可比项;不作为 Simplexity 收入、毛利或消耗事实。
[CI015, CI016, CI017, CI018, CI019, CI020]公开模型有清晰的成本驱动项,但未披露单位经济模型的具体数字。
节点来自已报道的资金用途和机器人部署逻辑中的成本类别;未估算任何数字成本。
[CI003, CI006, CI010, CI027, CI029, CI032]4.3 资本充足性、跑道与下一轮风险
最强的已披露财务事实是融资,不是经营表现。独立来源一致称累计融资约 CNY2.0 billion、估值超过 USD1.0 billion,投资财团包括 Tencent、Alibaba、HongShan、Legend、CAS Star、Gaorong、BlueRun/Lanchi 等。对一家 2025 年 7 月成立的公司来说,这笔资金异常庞大,也给 Simplexity 在模型训练、机器人迭代、数据获取和早期部署上留下有意义的资本垫。可是,它没有揭示账上现金,也没有说明公司消耗资本的速度。 因此,资本充足性仍是开放尽调问题。没有公开来源披露月度烧钱、跑道月数、债务、租赁、客户预付款、项目融资义务或下一轮融资触发条件。可能的触发条件不是标准 SaaS ARR 里程碑;对这家公司来说,更可能是证明软硬件系统能从 PoC 走向可重复部署,同时保持可接受的毛利率和服务负载。负面背景也重要,因为独立媒体报道了中国官方对人形机器人泡沫风险的担忧。如果市场情绪在 Simplexity 拿出收入质量和单位经济性的私有证据之前收紧,其据报独角兽估值可能更难守住。[CI003, CI004, CI005, CI006, CI013, CI014]
| 资本项 | 公开值 / 状态 | 来源依据 | 投资判断 | 尽调要求 |
|---|---|---|---|---|
| 披露融资总额 | ~CNY2.0B / ~USD289M | 独立融资报道 | 对一家 2025 年成立的公司来说,账面融资垫很厚 | 确认一级发行还是二级转让、实收现金以及投后股权结构表 |
| 报道估值 | >USD1.0B | 独立融资报道 | 有独角兽估值,但没有公开收入分母 | 提供估值条款、清算优先权和最新股价 |
| 账面现金 | null | 未公开披露 | 无法计算资金可支撑期 | 提供最新现金、受限现金和月度结账包 |
| 月度消耗 | null | 未公开披露 | 无法检验资本是否充足 | 提供经营性消耗、资本开支、算力支出和库存采购 |
| 资金可支撑月数 | null | 未公开披露 | 只有融资金额,算不出资金可支撑期 | 提供基准 / 下行情形资金可支撑期和董事会批准的支出计划 |
| 债务 / 项目融资 | 未识别到公开偿付义务 | 已留存公开来源 | 没有证据,不等于没有债务 | 提供银行债务、租赁、供应商信用、客户预付款和担保 |
| 下一轮触发点 | 可能取决于部署和毛利证明 | 由资金用途和机器人可比公司推断 | 更偏硬件 / 研发,而不是 SaaS ARR | 明确下一轮融资前需要打到的里程碑 |
资本充足性需要把已报道融资事实,与没有支撑的现金、消耗、资金可支撑期和偿付义务假设拆开看。
[CI003, CI004, CI005, CI006, CI013, CI014]Simplexity 已披露融资覆盖数条耗资本工作流,但现金是否足够无法从公开信息验证。
由于没有公开现金流量表,矩阵条目均为定性判断。
[CI003, CI004, CI006, CI007, CI014, CI025]4.4 公开牵引与私有指标缺口
公开牵引故事可信但不完整。报道称公司很快完成小批量机器人本体推出和 PoC 验证,并描述其在杭州、北京、上海和苏州的布局。这些是重要运营信号,因为它们显示公司不只是纸面融资载体。可是,保留来源没有任何一个点名付费客户、披露已部署机器人数量、区分试用与付费合同、发布使用率,或展示来自 PoC 的已确认收入。这意味着数据无法支持活跃客户数、收入 run rate、ARR、NRR、毛利率或销售效率判断。 尽调姿态应当明确:公开牵引可以支持继续研究,但承销必须等私有证据。下一个 data room 应包括月度管理账、按产品和客户拆分的已确认收入、递延收入、签署订单积压、部署 cohort、机器人级 COGS、服务工单、质保准备金政策、算力和数据采集支出,以及所有类债务义务。在这些材料可用之前,财务模型中的每个收入、利润率、烧钱和跑道单元都应留为空值,而不是用行业平均数填充。[CI007, CI008, CI010, CI011, CI012, CI029]
| 缺失指标 | 当前公开状态 | 影响 | 精确尽调路径 |
|---|---|---|---|
| 收入 / ARR | 未披露 | 无法测算牵引力或估值倍数 | 要求提供月度确认收入、如适用的 ARR 桥接表,以及按客户 / 产品拆分的收入 |
| 定价和合同条款 | 未披露 | 无法评估收入质量或折扣 | 审阅主服务协议、报价单、PO 和收入确认备忘录 |
| 毛利率和 COGS | 未披露 | 无法判断可规模化程度 | 要求提供产品线 P&L、BOM、人工、质保、算力和支持成本 |
| 消耗和资金可支撑期 | 未披露 | 即便融资金额很大,也无法判断资本是否充足 | 审阅最新现金、13 周现金流、预算和董事会计划 |
| 付费客户数和集中度 | 未披露 | 无法评估 GTM 质量或交易对手风险 | 要求提供客户名单、已签 ARR / 订单积压、队列转化和前 10 大客户集中度 |
| 类债务义务 | 未披露 | 无法识别现金或资产上的下行索取权 | 审阅债务、租赁、供应商融资、客户预收款、补助和担保 |
每个缺口都对应一个公开空值;管理层提供非公开证据前,模型中的这些空值应继续保留。
[CI010, CI011, CI014, CI028, CI029, CI033]4.5 财务结论与尽调优先级
财务结论是继续研究。Simplexity 拥有异常强的融资动能、据报顶级投资人阵容,以及契合 2026 年具身 AI 需求的产品叙事。这些都是真正的正面因素。但公司太年轻、太私有,公开财务模型难以可靠。干净的尽调备忘录必须把已披露 CNY2.0 billion 融资和 USD1.0 billion-plus 估值,与没有支撑的收入、ARR、毛利率、烧钱、跑道或单位经济性主张分开。 承销风险不是公司没有财务价值,而是公开证据尚不能衡量收入质量或资本效率。可比申报显示,机器人和自动驾驶企业可以在收入增长的同时持续大额亏损、承受客户集中,并为重研发和部署成本融资。因此,下一轮尽调应优先拿到 5 类证据包:管理层财务;合同和管线明细;BOM 与制造良率历史;部署服务成本 cohort;以及董事会批准的、覆盖至下一次融资事件的资本计划。没有这些,估值立场应保持未知,而不是有吸引力或昂贵。[CI003, CI004, CI014, CI017, CI022, CI023]
只有融资和估值有公开边界值;运营指标仍不可得,也不作估算。
图中仅绘制融资和估值项;运营指标没有公开披露,因此排除,而不是按零展示。
[CI003, CI004, CI010, CI014, CI028, CI037]4.6 图表
05产品与技术
5.1 产品定义与成熟度
Zhijian Power 应被视为一家具身智能机器人公司,试图把自动驾驶方法转化为面向实体作业的机器人。官网称公司成立于 2025 年 7 月下旬,正从真实场景出发,把高上限统一模型、高效数据闭环和可靠机器人硬件结合起来打造产品;它对外呈现为「全场景具身机器人」,而不是一款具名商业 SKU。第三方报道补上了成熟度边界:Yicai 报道称第一代机器人已小批量生产,概念验证测试已经启动;36Kr EU 称面向 B 端和 C 端用户的两代本体已经开发完成,已实现小批量生产,PoC 验证已全面启动。这是有意义的原型证据,但不等于已认证、可重复、被客户接受的部署。因此,最稳妥的表述是:产品仍在开发中,并在封闭环境里开展 PoC 活动,而不是已经规模化的商业 product-market fit。 [CE001, CE002, CE003, CE004, CE005, CE006]
| 模块 / 资产 | 用户或运营方问题 | 公开状态 / 成熟度 | 差异化信号 | 尽调缺口 |
|---|---|---|---|---|
| 全场景具身机器人本体 | 买方需要能进入真实物理流程的机器人,而不只是演示样机。 | 官方宣布即将推出;媒体报道称,第一代机器人已开始小批量产出和 PoC。 | 公开论点是模型定义通用本体,并做软硬件协同设计。 | 没有公开载荷、移动能力、执行器、续航、传感器、安全或成本规格。 |
| 世界模型 + VLA 基座模型 | 机器人执行任务时,需要理解语言、视觉、空间和状态。 | 官方称为自研,并通过一个 Transformer 统一。 | 单一架构旨在减少手工设计模块,并提升扩展性。 | 没有公开生产基准、模型规模、数据规模、延迟或安全边界。 |
| 端侧学习 / 数据闭环 | 新场景需要适应能力,不能依赖缓慢的集中式再训练周期。 | 官方描述为端侧部署、实时推理 / 训练和高效在线学习。 | 自动驾驶影子模式类比,指向车队数据闭环纪律。 | 没有公开证据证明在线更新安全、可回滚、隐私控制或边缘算力上限。 |
| LaST0 基座模型研究 | VLA 操作既需要快速反应,也需要较慢的物理推理。 | arXiv / 项目页证据;共同署名机构包括 Simplexity Robotics。 | 潜在时空 CoT 与快 / 慢 MoT 专家,用来应对延迟和动态变化。 | 论文实验与商业机器人技术栈之间的关系未披露。 |
| ManualVLA 长时程任务模型 | 长时程装配 / 重排需要程序化规划和精确控制。 | arXiv / PDF / 项目证据;论文中 Peng Jia 隶属于 Simplexity Robotics。 | 规划专家生成多模态手册,再用这些手册约束动作执行。 | 没有公开客户工作流显示 ManualVLA 已进入 Zhijian 机器人。 |
| TwinRL 后训练框架 | 真实世界机器人 RL 慢、成本高,还受安全约束。 | arXiv / PDF / GitHub 证据,且有公开仓库和数据集指南。 | 数字孪生扩大探索范围,并指导人类在环的真实机器人试运行。 | 公开代码 / 数据只能提供信号,不能证明生产部署或车队可靠性。 |
模块状态只基于已审阅的公开来源;硬件规格和认证声明未披露,因此刻意留作缺口。
[CE001, CE002, CE003, CE007, CE011, CE015]| 日期 / 阶段 | 功能或里程碑 | 状态 | 含义 | 来源 |
|---|---|---|---|---|
| 2025-07 | 公司成立 | 官网和媒体称公司成立于 2025 年 7 月下旬 / 7 月。 | 公司非常年轻;成熟度声明需要强验证。 | 官网;Yicai;36Kr EU |
| 2025-12 | ManualVLA arXiv 提交 | 论文于 2025-12-01 提交,Peng Jia 的署名机构为 Simplexity Robotics。 | 长时程规划研究已经公开,但仍处于商业化前。 | arXiv / ManualVLA PDF 技术来源 |
| 2026-01 至 2026-06 | LaST0 arXiv 版本 | arXiv 列出截至 2026-06-12 的第 4 版提交;项目页列出 Simplexity 署名。 | 基座模型研究活跃,且近期仍在更新。 | arXiv / LaST0 项目页 |
| 2026-02 至 2026-05 | TwinRL arXiv 版本和公开仓库 | arXiv 列出截至 2026-05-19 的多个版本;GitHub 托管项目材料。 | 后训练和数字孪生数据闭环已有可见开发者入口。 | arXiv / GitHub / 项目页 |
| 2026-03 | 小批量机器人和 PoC 报道 | Yicai 和 36Kr EU 报道第一代小批量产出、两代本体以及 PoC 启动。 | 支撑原型成熟度,不等于生产就绪。 | Yicai;36Kr EU |
| 2026 年报告日缺口 | 认证、客户部署、SDK/API、车队可靠性 | 已审阅公开来源中未找到。 | 这些是企业尽调的门槛项。 | 本章证据缺口 |
日期来自公开来源日期或来源发布日期;路线图行不代表引用材料之外的内部发布承诺。
[CE002, CE003, CE004, CE005, CE006, CE011]公开证据对研究产物支撑最强,对客户验证过的部署控制支撑最弱。
评分是基于所审阅公开档案给出的定性证据强度标签,不是内部绩效评级。
[CE001, CE002, CE030, CE035, CE038, CE041]5.2 自动驾驶 / VLA 架构
公司的技术栈明确以自动驾驶和 VLA 世界观来框定。官网称其自研一个具身基础模型,用统一 Transformer 融合世界模型与 VLA,联合建模语言逻辑、视觉语义、3D 空间结构和机器人状态,用于理解、生成和预测。媒体报道把这与创始团队的 Li Auto 背景相连:贾鹏被描述为曾负责 Li Auto 智能驾驶研发,并经历 BEV 感知、AD Max 3.0 和 VLA 模型迭代;王佳佳则被描述为带来端到端驾驶模型的量产交付经验。研究成果也契合这个世界观。LaST0 用 latent spatio-temporal chain-of-thought 和 Mixture-of-Transformers 双系统设计处理机器人快慢思考;ManualVLA 把长程操作中的规划 / 手册生成与动作执行拆开;TwinRL 用数字孪生和真实世界 RL 扩大探索并指导机器人本体学习。这些论文支撑了架构词汇,但本身不能证明公司已把它们融合成客户可用的机器人系统。 [CE007, CE008, CE009, CE010, CE011, CE012]
| 层级 / 组件 | 在技术栈中的角色 | 公开证据 | 依赖项 | 产品风险 |
|---|---|---|---|---|
| 统一 Transformer 模型 | 联合建模语言、视觉、3D 空间和机器人状态。 | 官网称世界模型和 VLA 通过统一 Transformer 集成。 | 训练数据广度、算力和机器人状态采集。 | 公开来源未披露规模、延迟或安全行为。 |
| 快 / 慢推理 | 把低频推理和高频动作拆开。 | LaST0 论文 / 项目描述了一个带潜在时空 CoT 的双系统 MoT。 | MoT 协同、异构运行频率和稳健动作条件化。 | 研究基准提升未必能扛住真实客户边界案例。 |
| 规划 / 手册专家 | 把目标状态转成多模态手册和子目标。 | ManualVLA 论文描述了规划专家、ManualCoT 和动作专家。 | 数字孪生手册数据、VLM 微调和下游任务轨迹。 | 没有公开生产编排证据显示如何处理错误恢复或用户纠正。 |
| 数字孪生后训练 | 扩大探索范围,并指导真实世界 RL。 | TwinRL 论文、项目页、GitHub 仓库和数据集指南。 | 精准重建、物理 / 几何保真度、人类在环试运行安全。 | 仿真到现实的差距和物理安全需要独立验证。 |
| 端侧推理 / 训练 | 让机器人边缘端具备低延迟适应和数据采集能力。 | 官网描述端侧部署和实时推理 / 训练。 | 边缘算力、热预算、更新治理和数据权利。 | 硬件和网络安全实现未披露。 |
| 通用机器人本体 | 通过收窄本体差异,提升数据通用性和复用。 | 官网称模型定义本体,并采用一个通用本体。 | 机械设计、执行器、传感器、制造质量。 | 没有公开 BOM、可靠性或可维护性指标。 |
| 系统化汽车执行 | 把 Li Auto 的工程、数据闭环和量产方法迁移过来。 | 36Kr EU、Gasgoo 和 QQ/Auto-First 描述了创始人的自动驾驶背景。 | 人才连续性、供应链执行和验证纪律。 | 汽车方法未必能顺滑映射到灵巧机器人部署。 |
这张架构表把直接披露的组件,与推断出的运营依赖拆开;依赖项是尽调问题,不是已确认的实施细节。
[CE007, CE008, CE009, CE011, CE012, CE013]公开架构呈现为一套统一的具身机器人技术栈,从 VLA / 世界模型,延伸到数据闭环、本体和部署场景。
层名综合已披露公开组件;不意味着生产架构已完整披露。
[CE007, CE011, CE015, CE018, CE024, CE027]5.3 工作流、部署与数据闭环
目标客户工作流是在受控实体环境中渐进式部署机器人。Yicai 称公司计划从工厂车间、超市、物流等封闭场景切入,再从封闭走向半开放和完全开放环境。36Kr EU 给出更技术化解释:公司被描述为试图把智能驾驶中的「shadow mode」模式迁移到机器人,通过部署额外边缘算力,让机器人在本体上采集、训练、测试和验证。官网中的「on device」「body」「hour」表述与这个框架一致:端侧部署、实时推理 / 训练、模型定义的通用本体、可复用数据和快速在线学习。尽调问题在于,这些仍是架构意图和早期 PoC 信号。公开来源没有披露边缘算力 BOM、延迟预算、车队遥测、故障分类、远程运维工具、服务流程、客户验收标准,或机器人学习中安全发生在端侧而非离线的比例。 [CE022, CE023, CE024, CE025, CE026, CE027]
| 工作流 / 用例 | 当前工作流问题 | 公司解决方案信号 | 已声明或暗示的可衡量收益 | 局限 |
|---|---|---|---|---|
| 工厂车间封闭场景 | 结构化但会变化的物理任务,需要稳定操作和正常运行时间。 | Yicai 和 36Kr EU 都把工厂车间列为早期目标。 | PoC 路径可以先在开放环境前验证重复性任务。 | 没有公开具名工厂客户、任务清单、SLA 或验收指标。 |
| 超市 / 商业环境 | 半结构化零售空间需要物品处理、导航和人与机器人近距离安全控制。 | 媒体报道把超市列为初始封闭场景目标之一。 | 零售场景可以生成多样化操作数据。 | 安全、感知边界案例和客户劳动力 ROI 未披露。 |
| 物流工作流 | 仓储 / 物流任务要在物体种类多的情况下保持处理速度和鲁棒性。 | 媒体来源把物流列为早期部署方向之一。 | 封闭环境适合分阶段部署策略。 | 没有与 WMS / 仓储系统打通的吞吐量、错误率或集成证据。 |
| 长时程目标状态操作 | 机器人很难把最终目标状态转成可执行流程。 | ManualVLA 生成多模态手册,并把手册输入动作专家。 | 论文报告平均成功率比分层基线高 32%。 | 证据来自研究任务表现,不是客户部署。 |
| 数字孪生引导的 RL 适应 | 物理 RL 探索成本高、速度慢,还受安全约束。 | TwinRL 重建由智能手机拍摄的数字孪生,并用孪生环境试跑指导真实世界 RL。 | 论文报告四项任务接近 100% 成功率,收敛速度快 30% 以上。 | 四项论文任务之外的迁移,以及进入商业机器人后的表现仍未验证。 |
用例混合了公司声称的目标场景和研究任务工作流;不能把任何一项解读为已验证的创收部署。
[CE005, CE006, CE014, CE015, CE016, CE017]支持的工作流是分阶段的封闭场景 PoC 闭环,包含数据采集、模型适配和验证关口。
流程综合了公司 / 媒体表述和 TwinRL 式学习闭环;公开来源未披露真实客户部署 SOP。
[CE022, CE023, CE024, CE025, CE026, CE018]5.4 差异化、开发者信号与风险
最强的差异化信号不是已披露硬件规格,而是团队背景、全栈野心,以及围绕 VLA 后训练的可见研究 / 开发者成果组合。官网列出硬件、嵌入式系统、机器人控制、世界模型、云端模型、大数据、数据运营、感知、模型部署、标定 / SLAM、强化学习、基础模型、VLA 算法、调度框架、平台软件和仿真等岗位。TwinRL 的 GitHub 仓库和数据集指南提供了开发者信号代理:围绕数字孪生资产和孪生生成轨迹,有公开代码 / 文档;从业者整理仓库也把 TwinRL 列入 RL-VLA 生态。话虽如此,与商业开发者平台相比,开发者信号仍然很薄:没有公开 SDK、API、发布说明流、开放机器人操作栈、包下载历史或由 issue 驱动的客户社区。信任与合规证据更薄。公开资料没有披露安全认证、网络安全控制、生产质量体系、可靠性指标、事故历史或具名客户背书。对承销来说,这些缺口很重要,因为机器人问题同样是验证和运营问题,不只是模型架构问题。 [CE030, CE031, CE032, CE033, CE034, CE035]
| 控制 / 质量领域 | 公开状态 | 已展示范围 | 重要性 | 待补缺口 |
|---|---|---|---|---|
| 机器人安全认证 | 已审阅来源中未见公开披露。 | 未发现认证名称、测试实验室或标准映射。 | 工厂、零售和物流中的实体机器人存在人身安全风险。 | 要求提供安全案例、标准映射、测试报告和事故流程。 |
| 网络安全和隐私控制 | 已审阅来源中未见公开披露。 | 官方材料讨论数据闭环,但未讨论数据治理控制。 | 端侧学习和车队数据采集可能带来隐私 / 安全暴露。 | 要求提供安全架构、更新签名、数据留存和访问控制证据。 |
| 可靠性 / 正常运行时间指标 | 已审阅来源中未见公开披露。 | PoC 和小批量状态已有报道,但正常运行时间没有披露。 | 客户 ROI 取决于持续自主运行。 | 要求提供 MTBF、干预率、恢复流程和现场日志。 |
| 制造质量体系 | 已审阅来源中未见公开披露。 | 媒体报道提到小批量生产和两代本体。 | 机器人规模化需要可重复的零部件质量和可服务性。 | 要求提供供应商名单、质量门、良率数据、质保流程和服务计划。 |
| 开发者 / 社区证据 | TwinRL 的 GitHub / 项目页以及策展列表提供了部分公开信号。 | 研究代码 / 数据和从业者仓库显示出技术关注度。 | 开发者信号支撑可信度,但不能证明企业支持已准备好。 | 要求提供 SDK/API 路线图、发布节奏、文档和支持渠道。 |
这张表刻意记录缺失控制项,因为公开材料不能支撑认证、可靠性、安全或生产质量声明。
[CE030, CE032, CE033, CE034, CE035, CE036]商业就绪度取决于数据、算力、仿真保真度、机器人硬件、验证能力,以及客户 PoC 入口。
这些依赖是尽调关键项:公开来源能支撑架构方向,却支撑不了验证边界和部署边界。
[CE015, CE018, CE023, CE024, CE029, CE030]5.5 图表
06客户
6.1 客户分层与目标需求
公开客户档案指向目标中的 B 端封闭环境买家,而不是已验证的付费用户名单。Zhijian Power 官网把公司定位为从真实场景出发打造高用户价值的具身智能产品,中国报道则反复把第一批商业目标收窄到工厂车间、超市或零售货架运营,以及物流或仓储分拣。因此,可能的经济买家是掌握现场生产力预算的运营、自动化、制造工程或创新负责人;日常用户是产线、仓库或门店运营团队;付款方是运营该场地的企业。公开记录中的地域也以中国为中心:北京、上海和苏州被列为战略地点,苏州全球创新中心被定位为连接研发、制造和应用的实体桥梁。这个分层有用,但不等于客户证明。已审阅公司、媒体或合作伙伴来源都没有披露具名付费客户、活跃场地数量、已安装机器人数量、价格,或按分部拆分的收入贡献。[CU001, CU002, CU003, CU004, CU006, CU014]
| 细分市场 | 可能的买方 / 付款方 / 用户 | 用例 | 公开规模证据 | 战略价值 | 缺口 |
|---|---|---|---|---|---|
| 工厂车间 | 运营 / 制造工程预算;产线团队使用机器人 | 搬运、分拣、检测、封闭工位任务 | 目标细分已被点名;没有客户数量或站点数量 | 环境可控,因此近期匹配度最高 | 需要具名工厂和付费部署状态 |
| 超市 / 零售货架 | 零售运营或创新预算;门店员工与机器人互动 | 半结构化通道中的货架整理或补货 | 目标细分只在二手报道中被点名 | 测试从封闭环境走向半开放公共环境的过渡 | 需要门店名称、安全验收和劳动力 KPI |
| 物流 / 仓库分拣 | 仓库自动化付款方;拣选 / 包装 / 分拣团队使用机器人 | 分拣、物料移动、仓库辅助 | 报道提到未具名合作;没有具名物流客户 | 自动化预算大,吞吐量 KPI 可衡量 | 需要客户、SKU / 任务范围和生产状态 |
| 苏州产业生态 | 合作伙伴和产业化渠道,而不是客户付款方 | 借创新中心打通研发、制造和应用 | 已点名 Leaderdrive 合作和苏州中心 | 提升供应链可信度,也打开本地部署入口 | 还需证明生态能带来终端客户 |
| 未来 C 端或开放场景 | 尚无消费者付费方证据 | 封闭场景成熟后,可能切入家庭 / 服务用途 | 只有泛泛的 B/C 端表述,没有 C 端客户证据 | 长期可选项 | 安全和支持模型跑通前,不应纳入投资假设 |
各行区分公司瞄准的细分场景与已观察到的客户证据;规模为 null 表示截至 runDate 没有公开分母。
[CU001, CU006, CU014, CU015, CU036, CU039]可被支撑的旅程从受控演示走到 PoC,再进入生产;公开记录尚未走到具名生产部署。
[CU005, CU007, CU009, CU014, CU029, CU030]6.2 采用轨迹与具名客户证明
采用轨迹最多只可见到 PoC 和未具名合作。Sohu 及同类中文报道提到两代 B 端和 C 端机器人本体、小批量推出和 PoC 验证。Suzhou/Leaderdrive 文章增加了一个具体伙伴事件:全球创新中心签约、与谐波减速器供应商达成战略合作,以及首款工业双臂机器人进入样机测试和产品化。Sina 更进一步称,截至 2026 年 3 月,Zhijian 已与多家制造、商超和物流企业达成合作。可是,客户尽调的核心答案仍是否定的:这些企业没有被点名,生产状态没有披露,也没有给出结果。Leaderdrive 是有价值的具名生态伙伴,不是客户背书。因此,本章把客户证明视为缺口,而不是隐藏的正面因素。公开证据支持一条从技术可信度到 PoC 和未具名合作的漏斗;它尚不支持付费生产采用、重复使用或客户 ROI。[CU005, CU007, CU008, CU009, CU010, CU011]
| 阶段 / 信号 | 证据价值 | 日期 / 批次 | 置信度 | 含义 | 缺失分母 |
|---|---|---|---|---|---|
| 公司发布和官方露出 | 定位真实场景产品,但没有客户标识 | 2025-07 起 / 当前官网 | 中 | 客户叙事从产品假设起步 | 客户数量和官网案例 |
| 本体开发 | 报称已有两代 B 端 / C 端本体,并推进小批量 | 2026-03 | 中 | 硬件可能已可进入受控试点 | 产量和交付客户的数量 |
| PoC 验证 | 多家中文媒体公开报道了 PoC 验证 | 2026-03 | 中 | 采用至少有概念验证层面的表述 | 付费 / 免费 PoC、现场名称 |
| 苏州中心与 Leaderdrive 合作 | 有具名战略伙伴,并涉及样机测试 / 产品化 | 2026-02 | 中 | 产业化路径比单纯实验室叙事更具体 | 与该中心绑定的终端客户 |
| 未具名垂直场景合作 | 报称涉及多家制造、商超和物流企业 | 2026-03 | 低至中 | 指向目标垂直场景里的管线或试点沟通 | 名称、采购订单、部署数量、结果 |
| 生产部署 / 留存 | 未发现公开证据 | 2026-07 本轮审阅 | 中 | 生产和留存应视为未解 | NRR、流失、续约、利用率、安全验收 |
该轨迹记录公开证据层级,而非内部转化数据;未具名合作不计作具名客户证据。
[CU001, CU005, CU007, CU009, CU011, CU012]| 客户 / 交易对手 | 细分市场 | 部署 / 用例 | 生产还是试点 | 结果证据 | 限制 |
|---|---|---|---|---|---|
| 未找到 Zhijian 具名付费客户 | 所有目标细分市场 | 公开信息未具名 | 未证明 | None | 核心缺口:已审阅来源没有点名付费生产客户 |
| Leaderdrive / 绿的谐波 | 工业部件生态 | 战略合作,以及苏州创新中心的产业化支持 | 伙伴,不是客户 | 伙伴可信度和样机测试背景 | 不能证明采购、使用、留存或客户 ROI |
| 未具名制造企业 | 工厂车间 | 报称围绕工厂车间场景合作 | 不清楚;很可能是试点或管线 | 没有吞吐、可用时间、安全或单元指标 | 缺少名称和付费状态 |
| 未具名商超企业 | 零售 / 门店运营 | 报称围绕理货或零售场景合作 | 不清楚;很可能是试点或管线 | 没有门店、人力或服务指标 | 没有具名零售商或部署地点 |
| 未具名物流企业 | 仓储 / 分拣 | 报称围绕物流分拣场景合作 | 不清楚;很可能是试点或管线 | 没有订单、站点或生产率指标 | 没有具名物流客户或生产范围 |
该枚举刻意保持部分且保守:列出已找到的所有具名或类别级客户证据候选,同时保留客户、伙伴、未具名试点 / 合作之间的区别。
[CU007, CU008, CU009, CU010, CU011, CU012]证据从宽泛市场野心一路收窄,到最后没有具名生产客户证明。
计数代表公开证明层级,不是内部转化指标;数值 0 表示未发现公开证明,不代表没有真实客户。
[CU005, CU008, CU011, CU012, CU016, CU019]Zhijian 有伙伴和试点信号,但缺少同行基准中可见的具名客户和结果证据。
[CU001, CU011, CU012, CU021, CU024, CU025]6.3 留存、扩张与集中风险
公开档案完全没有披露留存。已审阅来源没有报告 NRR、GRR、流失、续约率、复购、合同期限、满意度、客户数量、活跃部署数量或账户级扩张。这不代表早期客户不存在;它意味着公开档案无法区分免费演示、付费试点、战略共研和生产部署。land-and-expand 逻辑合理,因为封闭工业场地往往从一个受限任务开始,在安全验收和使用率证明后再扩展到工位或场地。但对 Zhijian 来说,这仍是假设,不是已观察行为。集中风险也无法界定:如果公司只有少数试点,任何一个战略账户或伙伴都可能主导学习数据、路线图优先级或未来收入。基准来源显示了更强证明该是什么样:JD Logistics 描述了双 11 期间的运营使用;Boots/Locus 把机器人与峰值订单处理相连;UBTECH 披露具名汽车交易方、人形机器人收入和出货量。Zhijian 尚未达到这个证据标准。[CU016, CU017, CU018, CU020, CU021, CU022]
| 指标 | 公开值 | 细分市场 | 置信度 | 尽调问题 |
|---|---|---|---|---|
| 客户数量 | 所有细分市场 | 中 | 按月索取活跃付费账户、活跃试点和停用试点 | |
| 生产部署 | 工厂 / 零售 / 物流 | 中 | 索取客户名称、站点位置、已部署机器人数量和验收日期 | |
| NRR / GRR / 流失 | 所有付费客户 | 中 | 按队列索取客户数留存、总收入留存和扩张收入 | |
| 复购 / 多站点扩张 | 企业账户 | 中 | 索取转为生产的试点,以及每个账户的站点数 | |
| 满意度 / 推荐人 | 具名用户 | 中 | 要求与运营负责人和安全经理做客户证明访谈 |
null 值表示已审阅公开文件未披露该指标;它们不是零值估计。
[CU016, CU017, CU018, CU037, CU040]| 扩张驱动因素 | 集中度风险 | 影响 | 尽调路径 |
|---|---|---|---|
| 封闭场景任务可重复性 | 少数试点可能主导路线图优先级 | 若缺少广泛付费客户基础,则风险高 | 审阅客户名单、试点阶段和按账户收入 |
| 苏州工业生态与 Leaderdrive 伙伴关系 | 伙伴证据可能被误当成客户证据 | 中:有助供应链,但不能验证需求 | 把供应商 / 伙伴合同与客户订单分开核查 |
| 战略投资方 Tencent 和 Alibaba | 战略投资方未必会转为客户或渠道 | 中:潜在分发上行仍未证明 | 询问任一投资方是否签署部署或渠道协议 |
| 安全和集成验收 | 安全验收失败会卡住 PoC 向生产转化 | 在人类工作空间里的实体机器人,风险高 | 审阅安全论证、事故日志、ISO/OSHA 合规映射 |
| 行业资本热潮 | 泡沫 / 产能过剩会挤压定价和客户信任 | 在拥挤的中国市场,风险中至高 | 比较胜率、订单积压质量和客户付费利用率 |
风险来自公开证据缺口和行业反向资料,而非内部客户数据。
[CU018, CU020, CU029, CU030, CU031, CU032]真实的 Zhijian 留存队列未公开,因此图中按生命周期阶段给证据可见度打分。
数值是分析师对证据可见度的打分,不是客户留存率;Zhijian 实际留存未披露。
[CU016, CU018, CU021, CU024, CU025, CU026]6.4 采购摩擦与负面背景
负面情形不是 Zhijian 缺少雄心,而是在人形和具身机器人领域,商业化主张很容易在客户签约、部署和续约前被夸大。OSHA 指出,机器人事故常发生在编程、维护、测试、安装或调整期间;ISO 10218-1:2025 也强调机器级安全要求的必要性。这些约束对工厂、超市和物流场地很重要,因为买方必须验证人机交互、与既有工作流集成、数据采集授权、服务响应和责任承担,才会越过 PoC。更广的中国市场评论也给出警示:France 24 和 The Business Times 报道了官方对泡沫风险、技术和商业化不成熟,以及超过 150 家人形机器人制造商的警告。在这种背景下,未具名合作应保守承销。下一步尽调是客户证据包:签约客户、付费与免费状态、部署地点、在线率与安全验收、使用率、试点转化、复购和收入集中度。[CU027, CU028, CU029, CU030, CU031, CU032]
6.5 图表
07风险
7.1 按严重性排序的风险视图
风险结构的核心,是 2026 年大额融资叙事与机器人能否在真实客户现场安全、稳定、可商业化运行之间的证据缺口。公开资料能支撑 Simplexity 技术野心大、成立时间短、资金充足,并由前 Li Auto 自动驾驶高管带队;但还不能证明公司已有具名量产客户、留存、安全认证、事故记录或单位经济模型。残余商业化和烧钱风险因此很高:公司或许有足够资本推进模型、本体和数据闭环开发,但估值要可投,仍需要转化证据。最重要的法律和运营风险也不能分仓看待,而应联动监控:安全事故会引发产品责任索赔;数据采集会触发隐私或网络安全义务;算力管制会拖慢模型迭代;客户证据缺失则会直接传导为下轮降价融资或过桥融资压力。[CR001, CR003, CR005, CR007, CR008, CR021]
| 风险 | 可监控触发项 | 阈值 / 事件 | 行动含义 |
|---|---|---|---|
| 安全 / 产品责任 | 机器人伤人、险些事故、客户停用,或缺失安全论证 | 任何严重事故,或试点扩张前没有安全论证 | 暂停投资,或要求里程碑托管 |
| 客户证明缺口 | PoC 转量产 | 到下一融资里程碑仍没有具名付费量产部署 | 降级为跟踪 / 继续研究 |
| 芯片 / 出口管制依赖 | BOM 或云审查发现受管制组件且无替代方案 | 没有经过测试的备用架构 | 视为规模化计划的论点破裂点 |
| AI / 数据监管合规 | 律师备忘录发现备案、PIPL、数据出境或重要数据缺口 | 上线需要重设方案,或备案未解决 | 推迟交割,直至拿到整改计划 |
| 烧钱与估值 | 月度烧钱对照验证里程碑 | 客户转化前账上现金可支撑时间被压缩 | 重新定价、分期融资,或回避 |
| 伙伴集中度 | 战略投资人或客户条款限制市场进入 | 排他性条款,或冲突过重的条款 | 要求豁免或下行保护 |
| 创始人 / 关键人风险 | 创始人离职或接班未定 | 任何关键创始人在量产证明前退出 | 重新审视管理层假设 |
| 制造可靠性 | 现场 MTBF、缺陷和服务指标不可得 | 没有量产质量看板 | 不要为大规模部署背书 |
| 法律清理 | 征信、法院、监管或 IP 记录未清 | 重大处罚、纠纷,或缺少清理结论 | 投资条款清单前升级法律尽调 |
否决标准把公开资料缺口转成尽调动作;拿到私有尽调数据后,应替换这些阈值。
[CR037, CR038, CR039, CR040, CR041, CR042]客户证明、安全、芯片获取和烧钱的剩余严重性最高。
序数热力图来自公开证据;私有尽调可能大幅改写可能性判断。
[CR021, CR022, CR036, CR037, CR038, CR039]7.2 监管与法律敞口
中国目前还没有一部单一的具身 AI 机器人法规,能一次性回答所有部署问题。Simplexity 需要在生成式 AI、深度合成、算法、个人信息、数据安全、网络安全审查、产品质量、侵权和出口管制等制度之间穿行。只要公司面向公众提供生成式 AI 服务、通过摄像头或麦克风处理个人数据、处理重要运营数据,或把联网机器人部署到敏感环境,最确定的义务就会落下。即使某条规则尚未被直接触发,它也应进入尽调包;机器人既是实体产品,也是 AI / 数据系统。法律风险不只是罚款,还包括发布延期、被迫重做模型或数据流程、客户采购阻力、安全停机,以及关键部件获取受限。[CR010, CR011, CR012, CR013, CR014, CR015]
| 风险 / 规则 | 司法辖区 | 状态 | 可能性 | 严重性 | 缓释成熟度 | 剩余敞口 | 尽调路径 |
|---|---|---|---|---|---|---|---|
| 生成式 AI 服务义务 | 中国 | 若提供公开模型输出或 API,即会触发 | 中 | 高 | 早期 / 未见证据 | 上线延迟、安全评估、备案、内容和数据控制 | 梳理所有面向用户的 AI 功能,并取得律师备忘录 |
| PIPL、《数据安全法》、《网络安全法》 | 中国 | 很可能适用于机器人遥测和人类数据 | 高 | 高 | 公开未见证据 | 同意、最小化、本地化、重要数据和网络安全控制 | 审阅数据地图、DPIA、跨境传输和留存设计 |
| 深度合成和算法备案 | 中国 | 取决于是否有生成内容或推荐功能 | 中 | 中 | 公开未见证据 | 标识、备案、服务调整或暂停上线 | 分类模型输出和推荐触点 |
| 产品质量和侵权责任 | 中国 / 部署市场 | 实体机器人始终适用 | 高 | 高 | 公开未见证据 | 人身伤害、财产损失、召回、保险和客户索赔 | 审阅安全论证、验收测试、警示、保险和事故日志 |
| 出口管制 / 高级算力获取 | 美国 / 中国 / 全球供应链 | 若需要受控芯片或设计服务,则影响重大 | 中 | 高 | Unknown | 模型迭代、边缘算力成本、采购延迟 | 筛查 BOM、云、芯片、供应商和备用架构 |
| 联网 / 自主系统供应链政策 | 美国及盟友市场 | 若出口到受覆盖的出行领域则直接适用;作为政策信号则间接影响 | 低至中 | 中 | Unknown | 海外市场排除或采购摩擦 | 评估目标地区以及车辆 / 机器人通信模块 |
| 企业信用、法院和行政核查 | 中国 | 公开负面信息核查不完整 | 中 | 中 | 弱 | 隐性处罚、纠纷、执法记录 | 调取 GSXT、Credit China、法院、SAIC 和地方市场监管记录 |
部分列举公开法律 / 监管风险中最值得关注的项;状态反映公开证据,而非律师意见。
[CR010, CR011, CR012, CR013, CR014, CR015]7.3 运营、制造与安全风险
运营假设要求公司并行推进硬件、嵌入式系统、感知、控制、模型优化、数据运营和端侧学习。执行协同,这很有力;一旦产品成熟度落后于融资预期,它也会放大失败。小批量本体和 PoC 验证本身无法证明工厂级可靠性、可维护性或安全人机交互。职场机器人指南和产品责任原则把尽调问题落到具体清单:投资人应索取危害分析、现场验收测试、急停设计、维护流程、事故日志、网络加固证据和保险。当前最强缓释项,是团队有自动驾驶规模化经验;残余风险在于,自动驾驶数据闭环不能自动迁移到仓库、工厂、超市或家庭,后者的操作任务和人机近距接触都更不结构化。[CR002, CR006, CR008, CR015, CR018, CR019]
| 失效模式 | 可能性 | 严重性 | 缓释成熟度 | 剩余敞口 | 未解缺口 |
|---|---|---|---|---|---|
| 机器人致伤或不安全人机互动 | 中 | 高 | 公开未见证据 | 产品责任、部署暂停、品牌受损 | 安全认证、危险分析和事故日志未公开 |
| 小批量转生产时质量漏检 | 高 | 高 | 早期 / PoC 阶段 | 保修成本、召回、客户流失 | 缺少良率、MTBF、现场故障和供应商质量数据 |
| 端侧训练或推理不稳定 | 中 | 高 | 有技术主张,但没有现场证据 | 行为不可预测,客户安全验收失败 | 缺少验收测试和运行时监控 |
| 机器人集群或数据闭环遭网络入侵 | 中 | 高 | 公开未见证据 | 客户现场被攻破、监管审查、停机 | 缺少渗透测试、安全更新和访问控制证据 |
| 数据闭环污染或隐私泄露 | 中 | 中高 | 公开未见证据 | 模型退化、PIPL / 数据安全敞口 | 缺少数据血缘、同意和删除流程 |
| 大额融资后制造产能过度拉伸 | 中 | 中高 | Unknown | 可靠性证明前,烧钱速度加快 | 缺少资本开支计划、供应商、工装和产能里程碑 |
各行结合公司架构主张、公开 PoC 阶段证据和机器人安全可比案例;可能性为定性判断。
[CR002, CR006, CR008, CR015, CR018, CR019]核心风险通过安全、客户、融资和估值传导。
定性因果图;边权未用公开数据量化。
[CR002, CR017, CR020, CR033, CR038, CR039]7.4 芯片、伙伴与客户依赖
依赖图有四个关键节点:前沿或边缘算力、战略投资人与生态伙伴、部署客户、监管方。Simplexity 走端侧训练和 VLA / 世界模型路线,算力和硬件供应因此格外重要;美国出口管制与智能网联汽车供应链措施也说明,自动系统的硬件和软件都可能成为政策目标。战略投资人能降低商业落地和基础设施摩擦,但未披露的商业条款也会带来集中度或利益冲突风险。客户依赖眼下是另一类问题:公开证据太薄,无法判断公司究竟有没有具名量产部署。尽调目标应当是把演示、PoC、付费试点、复购和量产铺开分开,再映射出每个阶段可能被哪类伙伴、供应商或监管方卡住。[CR017, CR024, CR029, CR030, CR033, CR038]
| 依赖项 | 交易对手 / 节点 | 角色 | 集中度 | 失效情景 | 严重性 | 缓释措施 | 剩余敞口 |
|---|---|---|---|---|---|---|---|
| 高级 AI 加速器和边缘算力 | 芯片供应商、云、国产替代 | 模型训练和端侧推理 | 未知 / 可能重要 | 出口管制或短缺压低性能或成本目标 | 高 | BOM 筛查和替代算力栈 | 测试前为高 |
| 战略投资方 | Tencent、Alibaba 相关资本、其他生态伙伴 | 资本、云、渠道、数据或客户入口 | Unknown | 排他性、渠道冲突或生态支持撤回 | 中高 | 披露商业条款和独立权利 | 中 |
| 制造供应商 | 执行器、传感器、电池、嵌入式板卡 | 质量和成本曲线 | Unknown | 供应商质量漏检或单一来源瓶颈 | 高 | 供应商认证和冗余 | 审计前为高 |
| 部署客户 | 工厂、物流、商超,之后是消费者用户 | 收入证明和数据生成 | 公开未见证据 | PoC 未能转为付费生产部署 | 高 | 具名客户证据和分阶段放量 | 高 |
| 监管机构和标准组织 | CAC、市场监管、工作场所安全、出口管制机构 | 上线许可和采购信任 | 碎片化 | 备案、问询、安全事故或出口阻断延迟上线 | 高 | 监管矩阵和律师签字 | 中高 |
依赖严重性是公开缓释措施之后的剩余严重性;未披露合同可能显著改善或恶化各行。
[CR004, CR017, CR024, CR029, CR033, CR039]Simplexity 依赖监管方、算力、供应商、战略资本和客户现场。
依赖图基于公开来源和标准尽调类别,不是已披露合同。
[CR004, CR016, CR017, CR024, CR029, CR041]7.5 创始人与执行集中度
创始人质量既是强项,也是风险。公开报道反复把投资逻辑押在前 Li Auto 领导层,以及自动驾驶方法迁移到机器人这件事上。这支撑了系统集成和数据闭环的可信度,但也让论证集中在少数个人身上,并押注“汽车到通用机器人”这一仍未验证的类比。公司招聘页面也确认,建设范围横跨大量稀缺职能,从工业设计、机械工程,到 VLA 算法、SLAM、数据运营和嵌入式系统。尽调因此应要求继任计划、角色边界、独立安全负责人、制造领导层深度,以及决策没有卡在少数创始人手里的证据。创始人离开,或无法招到资深生产和安全负责人,都应视为投资逻辑破裂事件。[CR009, CR027, CR032, CR040, CR045]
| 角色 / 职能 | 依赖或缺口 | 可能性 | 严重性 | 缓释措施 | 尽调路径 |
|---|---|---|---|---|---|
| 创始人 / 前 Li Auto 负责人 | 论点取决于自动驾驶系统经验能否迁移 | 中 | 高 | 公开履历强 | 访谈创始人;核验角色分工、接班安排与留任情况 |
| 制造负责人 | 需要把原型 / 小批量爬坡到可重复生产 | 高 | 高 | 仅有招聘信号 | 审阅组织架构、制造副总裁背景、供应商与质量体系 |
| 安全与合规负责人 | 公开资料未证明其拥有独立权限 | 中 | 高 | Unknown | 要求明确负责安全的具名负责人及董事会汇报线 |
| AI / 数据运营负责人 | 端侧学习需要数据治理和模型监控 | 中 | 中高 | 仅有招聘信号 | 审阅数据运营 SOP、标注 QA、隐私负责人和模型事故响应机制 |
| 商业负责人 | 需要把 PoC 转成可重复的量产合同 | 中 | 高 | Unknown | 审阅管线负责人、客户背书、定价和部署打法 |
人员风险行基于公开创始人报道和公司招聘信息;无法取得私有组织数据。
[CR009, CR027, CR032, CR040, CR045]7.6 监控计划与终止标准
监控计划应把不确定性转成明确触发器。监管尽调需确认是否有任何产品界面构成面向公众的生成式 AI 服务、算法备案义务、个人信息处理、重要数据问题、出口管制问题或安全认证要求。商业尽调应要求客户管线列明名称、合同状态、部署阶段、收入、安全签核、续约意向,以及按交易对手划分的集中度。运营尽调应要求 BOM、算力冗余、供应商认证、质量逃逸、事故响应和现场可靠性指标。财务尽调应把烧钱速度与里程碑证据绑定,而不是与头条融资额绑定。如果公司拿不出这些领域的可信数据,正确动作是把项目标记为继续研究或跟踪,而不是仅靠私人资本动能去承销估值。[CR034, CR035, CR037, CR038, CR039, CR040]
7.7 附录
08估值
8.1 结论:价格昂贵,继续研究
公开记录支撑这是一家严肃公司,但不足以把媒体报道的投后估值超过 US$1 billion 视为可以买入的入场价。正面事实是真实的:多家独立报道称,Zhijian Power / Simplexity Robotics 在大约六个月内完成五轮融资,合计 RMB2.0 billion,吸引顶级财务和战略投资人,并瞄准具备企业场景的具身 AI 机器人。卡住交易的不是叙事质量,而是公开收入、具名客户、单位成本、毛利率、股权结构表和优先权证据缺失。因此本章建议为继续研究,置信度为中低,风险评级为高,估值立场为昂贵。买方不应编造 ARR、回报或客户经济性来合理化媒体标出的估值;它应要求私下证据,或调整价格和交易结构。[CV001, CV002, CV004, CV007, CV008, CV031]
| 维度 | 本章结论 | 证据基础 | 决策含义 |
|---|---|---|---|
| 投资建议 | 继续研究,而非买入 | 融资动能已被核验,但收入、客户、毛利率和条款未公开 | 仅推进私有尽调,或要求结构化价格重置 |
| 置信度 | 中低 | 多家独立来源相互印证融资;决定性的经营指标缺失 | 不要用对赛道的信心替代公司经济性 |
| 风险评级 | 高 | 赛道拥挤警示、商业化早期、单位经济性不透明,以及媒体口径独角兽估值 | 要求否决触发项和下行条款 |
| 估值立场 | 偏贵 / 公开证据不支撑 | 报称投后估值超过 US$1B,但公开收入与客户证明缺失 | 把该估值视为待验证主张,而非成交清算价 |
| 决策含义 | 等待证据 | 强投资人和市场顺风让该标的仍值得跟踪 | 不作脱离价格的买入建议 |
判断表结合已被印证的公开事实和投委会解读;不插入私有回报或收入估算。
[CV001, CV002, CV007, CV008, CV031, CV034]融资和市场证据先经过披露和反向风险门槛过滤,最后才导出 research-more 建议。
流程代表决策逻辑,不是定量加权模型。
[CV001, CV002, CV007, CV011, CV028, CV034]8.2 投资逻辑与反论
投资逻辑是,Simplexity 在一个长期需求可信的赛道里,早期背书异常强。公司似乎已经融到足够资本,可在更广泛的消费者或开放环境使用前,先训练基础模型、开发硬件和算法,并推进可控的企业部署。市场数据也支持中国出货预期上行、商业化加速。反论同样有力:公开材料没有量化收入、客户集中度、已签部署、利润率、烧钱速度或轮次条款;负面报道还指出,人形机器人市场拥挤,容易过热。综合看,证据支持把公司留在尽调池里,而不是接受当前价格。若管理层拿出已签客户证据、付费部署转化、毛利率轨迹,以及可执行的投资人条款,解释为什么它相较披露更充分的上市可比公司应享有溢价,观点才会改变。[CV003, CV004, CV005, CV006, CV011, CV012]
| 论点 | 支持证据 | 什么会改变判断 |
|---|---|---|
| 顶级背书 | 多篇报道列出的投资人包括战略投资人与财务投资人 | 确认轮次文件、所有权和投资人权利 |
| 赛道时点 | 分析师与市场数据来源指向中国人形机器人商业化和产出快速增长 | 核验 Simplexity 是否参与付费部署,而不只是演示 |
| 团队可信度 | 报称创始人是前 Li Auto 高管 | 确认高级技术与制造负责人仍在岗 |
| 募资用途匹配 | 报称融资用于模型、算法、数据和应用扩张 | 审阅研发路线图、烧钱速度和里程碑预算 |
| 披露缺口 | 已留存来源均未量化收入或客户数量 | 在 NDA 下获取收入台账和客户背书 |
| 过热风险 | NDRC / 泡沫风险报道警示入局者拥挤且同质化 | 展示差异化数据、硬件可靠性和自有部署闭环 |
| 价格纪律 | 公开可比公司披露收入 / 亏损,Simplexity 未披露 | 围绕已验证里程碑重新定价或分期投资 |
各行配对正向与反向证据;若干单元格是尽调条件,而非既成事实。
[CV003, CV004, CV005, CV007, CV008, CV011]方向性敏感性显示,哪些尽调结果最会改变估值立场。
数值是 IC 敏感性分数,不是估值金额或回报预测。
[CV006, CV007, CV008, CV011, CV014, CV039]8.3 不编造私有指标的牛市、基准与熊市情形
情景分析看显性证据条件,不编造收入或回报测算。牛市情形下,私下尽调证明企业试点已转为经常性付费部署,制造良率在改善,数据和模型优势守得住,股权结构表条款没有把过多下行风险转嫁给新增普通股敞口。基准情形下,公开记录仍是证据集:它证明融资势能和赛道热度,但无法支撑价格。熊市情形下,监管顾虑、供给拥挤、产品同质化或 PoC 转化失败,在收入质量可见之前,就把公司推向降价融资、内部过桥或战略出售。这个区间因此不是预测回报区间,而是“可支撑性”区间,显示媒体报道的估值要被视为公允,还需要多少私下证据。[CV010, CV011, CV012, CV013, CV032, CV033]
| 情景 | 明确假设 | 估值支撑逻辑 | 概率信号 / 下行触发项 |
|---|---|---|---|
| 牛市 | 企业付费部署、可验证的毛利率路径、自有模型 / 数据优势、干净条款 | 只有私有证据证明规模和防御性,报称的 >US$1B 估值才可能站得住 | 信号:具名付费客户和复购订单;触发项:未验证 PoC |
| 基准 | 公开记录仍限于融资、投资人、市场顺风和笼统的企业客户描述 | 不要为媒体口径估值背书;维持继续研究 / 偏贵判断 | 信号:可信的 NDA 材料包;触发项:管理层无法披露指标 |
| 熊市 | 赛道资本降温、产品同质化、试点无法转化,或公开可比公司估值下修 | 下轮降价融资、内部过桥或战略出售风险明显上升 | 信号:部署延迟;触发项:没有付费转化或条款恶化 |
情景表是条件式推演,刻意不编造收入、毛利率、退出估值或回报倍数。
[CV010, CV011, CV012, CV013, CV032, CV033]只有私有证据改善后,报道价格才会从缺少支撑转为有条件可支撑。
区间使用 1–10 的证据支撑分数,因为公开来源未披露收入、利润率、稀释或退出输入。
[CV010, CV032, CV033, CV042, CV044]8.4 可比证据:市场上行,但披露标准更严
可比公司证据要求纪律。UBTECH、Baidu、Pony AI、WeRide 和 Mobileye 都不是完美匹配,但它们各自说明,可投的公开证据长什么样:审计或交易所备案财务、明确收入或运营指标、风险因素和持续披露。UBTECH 表明,人形机器人上市可比公司即便有 RMB1.3 billion 收入,一年也可能亏损超过 RMB1.1 billion。Baidu 和 Mobileye 显示真实运营规模和收入,同时提醒投资人,预期变化时估值标记会重置。Pony AI 和 WeRide 的备案文件显示,AV 技术栈公司在进入公开市场前如何披露风险。与这些参照相比,Simplexity 报道中超过 US$1 billion 的价格,除非私下尽调补齐缺失的经济性,否则只是私人市场叙事标记。[CV018, CV019, CV020, CV021, CV022, CV023]
| 可比对象 / 来源 | 指标或状态 | 与 Simplexity 的相关性 | 局限 |
|---|---|---|---|
| UBTECH Robotics | HKEX 文件披露 2024 年收入 RMB1.3054B、亏损 RMB1.1599B | 最接近的公开人形机器人 / 机器人披露基准 | 成熟度、产品组合和上市公司披露不同 |
| UBTECH 招股书 / 年报 | 可查阅风险因素和治理披露 | 显示公开市场估值前要求的披露门槛 | 不是直接的私募轮次可比对象 |
| Baidu Apollo Go | 2025 Q4 完全无人驾驶出行 3.4M 次,2026 Q1 AI 业务收入增长 | 展示公开投资人可核验的 AV 平台规模指标 | Apollo 属于一家多元化上市公司 |
| Pony AI | F-1 和 20-F 申报轨迹;AV 牌照 / 应用市场数据 | 可作为 AV 技术栈商业化和风险披露的相关代理 | 自动驾驶不同于人形机器人硬件 |
| WeRide | 已公告 2025 年 20-F 文件;Nasdaq 与 HKEX 上市 | 另一项公开 AV 技术栈披露基准 | 具体财务数据需在本章之外审阅申报文件 |
| Mobileye | 2026 Q1 收入 US$558M、指引更新、商誉减值 | 说明公开市场奖励收入,也会重置估值预期 | ADAS / AV 芯片模式不同于具身机器人 OEM |
| IFR World Robotics | 机器人统计与趋势来源 | 独立机器人市场基线 | 不是公司特定估值证据 |
| Morgan Stanley 经 CNBC | 上调中国人形机器人出货观点 | 支撑市场时点的牛市论点 | 市场增长不能证明 Simplexity 经济性 |
| TrendForce / Axis / Robozaps | 2026 年出货、融资、定价和公司跟踪信号 | 提供市场数据交叉验证 | 数据来源方法不同,可能夹杂炒作周期噪音 |
枚举为部分可比对象集合,围绕人形机器人、具身 AI 和 AV 技术栈公开披露的估值相关性筛选。
[CV014, CV015, CV017, CV018, CV019, CV020]IC KPI 分数在品类和投资人支持上最强,在估值证据和经济性透明度上最弱。
分数是对有来源证据的判断性总结,并刻意避免编造私有指标。
[CV003, CV004, CV007, CV008, CV011, CV018]8.5 决策条件与尽调要求
投资委员会应把当前估值当作一个尽调假设。从继续研究转向买入,需要同时具备商业和财务证据:按产品或部署拆分的收入、付费客户证明、订单积压和续约条款、BOM 与服务成本轨迹、安全和可靠性指标、部署利用率、股权结构表中的优先权,以及融资后资金跑道。投资逻辑破裂触发器很具体:没有具名付费客户、试点卡在未付费 PoC 状态、毛利率轨迹为负、轮次结构让新投资人排在更高级优先权之后,或板块融资压缩。若管理层无法在 NDA 下提供这些数据,合适动作是放弃或等待更低价格。若能够提供,下一版备忘录应基于已验证里程碑重估公司,而不是泛泛的具身 AI 热情。[CV009, CV030, CV033, CV039, CV040, CV041]
| 触发项 | 阈值 / 事件 | 对论点的传导 | 行动含义 |
|---|---|---|---|
| 没有付费客户证明 | 管理层无法在 NDA 下提供具名付费客户背书 | 商业证明停留在叙事或 PoC 层面 | 放弃或等待 |
| 收入不透明持续 | 没有按产品、部署或客户群组拆分的收入 | 估值无法锚定变现 | 不要为 >US$1B 背书 |
| 毛利为负或未知 | 没有 BOM、服务成本或良率证据 | 规模化可能吃掉现金,而不是创造价值 | 要求里程碑分期 |
| 拥挤市场重估 | 可比轮次或公开可比公司估值下修 | 下轮降价融资风险上升 | 要求更低进入价格 |
| 部署转化弱 | PoC 无法转成重复付费订单 | 企业采用的牛市论点破裂 | 除非价格重置,否则停止 |
| 不利优先权 | 新资金排在沉重清算优先权之后 | 媒体口径估值误导普通股经济性 | 重谈条款 |
| 安全 / 可靠性缺口 | 机器人达不到企业正常运行时间或事故阈值 | 客户扩张和监管接受度停滞 | 等待现场数据后再推进 |
触发项是尽调阈值和监控标准,不是事件已经发生的主张。
[CV009, CV011, CV012, CV025, CV033, CV039]| 主题 | 缺失证据 | 为什么重要 | 尽调路径 |
|---|---|---|---|
| 收入质量 | 按产品、付费部署、经常性 / 项目制组合拆分的收入 | 从叙事估值切到倍数或里程碑估值时需要 | CFO 数据室和银行流水 |
| 客户证明 | 具名客户、合同条款、续约或扩张数据 | 验证试点之外的企业需求 | 客户访谈和合同审阅 |
| 部署漏斗 | PoC 到付费转化、部署数量、利用率 | 检验商业化主张能否规模化 | 销售运营导出数据和现场走访 |
| 毛利率 | BOM、服务人工、质保、良率和制造成本曲线 | 决定规模化能否盈利 | 销货成本模型和供应商发票 |
| 股权结构和优先权 | 持股、清算优先权、反稀释、老股转让所得 | 媒体口径投后估值可能不等于普通股价值 | 由律师主导审阅融资文件 |
| 技术防御性 | 模型 / 数据护城河、硬件可靠性、安全事故、正常运行时间 | 需要用来解释为何能在拥挤赛道拿溢价 | 架构审阅和现场日志 |
| 现金可支撑期和烧钱 | 月度烧钱、已承诺资本开支、下一轮计划 | 防止融资速度叙事掩盖融资风险 | 董事会材料和现金预测 |
| 监管 / 安全 | 认证、产品责任、事故响应 | 影响客户采用和下行风险 | 安全档案和保险审阅 |
| 可比对象桥接 | 为什么 Simplexity 相比 UBTECH、Pony、WeRide、Mobileye 代理对象值得溢价 | 倒逼明确估值逻辑 | 用已验证指标搭建投委会可比模型 |
| 退出准备度 | 审计准备度、治理、汇报节奏 | 决定 IPO 或战略出售可选性 | 审计师、法律和董事会流程审阅 |
每一项都关系到价格、风险或退出;不能只因本轮超募就豁免。
[CV028, CV039, CV040, CV041, CV045]8.6 附录
免责声明
本报告是基于公开证据的尽调快照,不构成投资建议。关键财务、法律、技术和合同事实仍未公开;任何投资决策前,都应直接向管理层和一手文件核验。
证据索引
| 编号 | 陈述 | 可信度 | 来源 |
|---|---|---|---|
| CO001 | The company publicly uses the Simplexity Robotics / 至简动力 brand for a full-scenario embodied-robotics business. | 中 | SO001 |
| CO002 | The company website says Simplexity Robotics was founded in late July 2025. | 高 | SO001, SO007 |
| CO003 | AiQicha lists Hangzhou Zhijian Power Technology Co., Ltd. as established on 2025-07-31. | 高 | SO018, SO016 |
| CO004 | AiQicha lists the company's address in Xianlin Street, Yuhang District, Hangzhou, Zhejiang. | 高 | SO018, SO016 |
| CO005 | AiQicha identifies Jia Peng as the legal representative and as a company executive/manager. | 高 | SO018, SO004 |
| CO006 | AiQicha shows the company as currently open/operating. | 中 | SO018 |
| CO007 | The registered business scope includes AI software, intelligent robots, industrial robots, service robots, data processing, system integration, and related hardware/manufacturing activities. | 中 | SO018 |
| CO008 | Public company and media sources frame the business primarily around embodied intelligence and robotics. | 中 | SO001, SO002, SO005, SO006 |
| CO009 | 36Kr explicitly frames the strategy as moving VLA/autonomous-driving technical concepts from cars into robots. | 中 | SO008 |
| CO010 | The company claims a soft-hardware integrated path built around model, data-loop, and robot hardware co-development. | 中 | SO001, SO002 |
| CO011 | The official site describes a self-developed embodied foundation model integrating world-model and VLA capabilities. | 中 | SO001 |
| CO012 | The official site claims on-device deployment with real-time inference/training and low latency. | 中 | SO001 |
| CO013 | The official site lists hiring across hardware, algorithms, software, and functional roles. | 中 | SO001 |
| CO014 | Multiple sources report that the company completed five financing rounds totaling RMB 2 billion. | 高 | SO002, SO004, SO005, SO006, SO009, SO023 |
| CO015 | The reported five financing rounds occurred within less than roughly six months of the company's founding or launch. | 中 | SO005, SO006, SO009 |
| CO016 | 36Kr and Baidu Baike report that the company's valuation crossed USD 1 billion. | 中 | SO006, SO008, SO016 |
| CO017 | Reported financial investors include Yuanjing/Vision Capital, Lanchi/BlueRun, Sequoia China/HongShan, Legend Capital, CAS Star, and Gaorong. | 中 | SO002, SO004, SO005, SO006, SO013 |
| CO018 | Tencent and Alibaba Group are repeatedly reported as strategic investors. | 中 | SO002, SO004, SO005, SO006, SO026 |
| CO019 | Tencent News and 36Kr report Light Source / Lighthouse Capital as financial advisor for the latest round. | 中 | SO004, SO006 |
| CO020 | Yicai reports that the financing will support foundation-model training, robot R&D/iteration, data collection, and core algorithm development. | 中 | SO005 |
| CO021 | Public financing coverage frames the company as a private embodied-intelligence unicorn rather than a public company. | 中 | SO006, SO007, SO011, SO016 |
| CO022 | 36Kr reports that the first self-developed robot body emerged in under 45 days from first employee arrival. | 中 | SO006, SO008 |
| CO023 | 36Kr reports that the company completed two generations of robot bodies for B-end and C-end users. | 中 | SO006, SO008 |
| CO024 | 36Kr and Sina report small-batch robot-body production and PoC verification. | 中 | SO002, SO006, SO007, SO008 |
| CO025 | Public reports name factories, supermarkets, and logistics as initial closed or B-end application scenes. | 中 | SO002, SO016 |
| CO026 | Sina and 36Kr report strategic layouts in Beijing, Shanghai, and Suzhou beyond the Hangzhou registered base. | 中 | SO002, SO006, SO008 |
| CO027 | 36Kr reports joint laboratories with top universities and a global innovation center in Suzhou. | 中 | SO006, SO007 |
| CO028 | Multiple sources report that Jia Peng, Wang Kai, and Wang Jiajia all came from Li Auto. | 中 | SO004, SO005, SO008, SO009 |
| CO029 | Jia Peng is reported as CEO and former Li Auto intelligent-driving R&D leader with IBM and Nvidia experience. | 中 | SO002, SO004 |
| CO030 | Wang Kai is reported as chairman, former Li Auto CTO, and earlier Visteon autonomous-driving leader/chief architect. | 中 | SO002, SO004 |
| CO031 | Wang Jiajia is reported as COO and former Li Auto intelligent-driving mass-production leader. | 中 | SO002, SO004, SO008 |
| CO032 | AiQicha lists a board/director roster beyond the three core founders, including Liu Yiran, Cao Wei, Qi Na, Wang Jiajia, Zheng Qingsheng, Wang Lei, and Yao Yao. | 中 | SO018 |
| CO033 | Baidu Baike reports founder shareholding percentages for Jia Peng, Wang Kai, and Wang Jiajia, but those figures are not supported by signed cap-table documents in reviewed sources. | 低 | SO016, SO019, SO020 |
| CO034 | Baidu Baike reports a mix of founder, partnership-vehicle, and institutional shareholders. | 低 | SO016, SO019, SO020 |
| CO035 | AiQicha says the company has 44 registered trademarks and one brand project. | 中 | SO018 |
| CO036 | No reviewed source disclosed revenue, ARR, revenue run-rate, or gross margin for the company. | 中 | SO001, SO002, SO005, SO006, SO018 |
| CO037 | No reviewed source named signed customers or disclosed customer count; available coverage cites only target scenarios and PoC verification. | 中 | SO002, SO006, SO016 |
| CO038 | No reviewed source disclosed company headcount, although the official site lists multiple open hiring categories. | 中 | SO001 |
| CO039 | Static fetches of national registry and CreditChina portals did not yield a conclusive company-specific official adverse-record result. | 中 | SO021, SO022 |
| CO040 | NBD describes the 2026 embodied-intelligence financing market as early, path-uncertain, and partly FOMO-driven. | 中 | SO027 |
| CO041 | NetEase sector commentary flags technical bottlenecks, pseudo-demand, and profitability concerns across the broader embodied/humanoid robotics boom. | 中 | SO028 |
| CO042 | Gasgoo reports that auto executives are moving into embodied intelligence because smart cars and robots share algorithms, compute needs, and hardware logic. | 中 | SO010 |
| CO043 | The ManualVLA project page lists Peng Jia with Simplexity Robotics affiliation on a VLA/robotic-manipulation research artifact. | 中 | SO030 |
| CO044 | The TwinRL arXiv abstract says VLA models are constrained by expert-demonstration costs and limited real-world interaction. | 高 | SO031, SO033 |
| CO045 | The TwinRL GitHub repository says real-world RL training code is still coming soon. | 中 | SO032 |
| CO046 | The official website does not disclose named customers, pricing, revenue, headcount, or detailed governance terms. | 中 | SO001 |
| CO047 | Gasgoo notes sector consolidation pressure, including more than 20 robotics companies facing bankruptcy or layoffs in the past year while top-tier players secured funding. | 中 | SO009 |
| CO048 | Baidu Baike reports that the company was recognized as a Zhejiang/Hangzhou unicorn at an April 24, 2026 Hangzhou event and included in a Zhejiang embodied-intelligence map. | 低 | SO016 |
| CM001 | The appropriate market boundary for Simplexity is embodied-intelligence robots for controlled industrial, logistics, retail, and service workflows, not the entire AI, EV, or consumer humanoid market. | 中 | SM017, SM016 |
| CM002 | Simplexity publicly describes a progressive deployment path from closed scenarios to semi-open and fully open scenarios, with factory workshops, supermarkets, and logistics as the initial closed-scenario wedge. | 中 | SM017, SM016 |
| CM003 | Simplexity’s first-generation robot had entered small-batch production and proof-of-concept testing, so current market sizing should be anchored on pilot-to-deployment conversion rather than proven high-volume revenue. | 中 | SM017 |
| CM004 | The founders’ Li Auto intelligent-driving background explains why external coverage frames the company through autonomous-driving stack migration into robots. | 中 | SM015, SM016, SM017 |
| CM005 | Modern smart-car capabilities overlap with robotics in algorithms, compute requirements, navigation, controllers, batteries, and system-integration methods, but robots face harder manipulation and unstructured-environment demands. | 中 | SM015, SM016 |
| CM006 | The core included spend for Simplexity should include embodied-AI robot systems, fleet software, integration, deployment services, maintenance, and data/skill iteration for controlled facilities. | 中 | SM002, SM009, SM010, SM012 |
| CM007 | Excluded spend should include generic model training unrelated to robots, consumer chatbots, broad EV autonomy, standalone warehouse management software, and commodity industrial automation that does not embody mobile or manipulative autonomy. | 中 | SM004, SM009, SM012 |
| CM008 | China’s 2026 real-scene training action explicitly targets industrial, service, and special scenarios and includes production, inspection, maintenance, warehousing logistics, catering retail, healthcare, safety, emergency, and disaster-response applications. | 高 | SM002, SM027 |
| CM009 | The 2026 action sets a policy goal for representative-scenario validation, normal deployment, more than 100 high-value application scenarios, and ten-thousand-unit scale landing capability by the end of 2026. | 高 | SM002, SM027 |
| CM010 | The Robot+ application plan provides a broader policy backdrop by directing multiple ministries to deepen robot adoption across manufacturing, logistics, healthcare, agriculture, energy, and other application fields. | 中 | SM003 |
| CM011 | China’s first national standard system for humanoid robots and embodied AI covers common foundations, intelligent computing, limbs, whole machines, applications, safety, and ethics across the industrial chain and lifecycle. | 高 | SM001, SM025, SM026 |
| CM012 | The standard system was developed under MIIT organization with more than 120 research institutions, enterprises, and industry users, which supports buyer confidence but also formalizes compliance expectations. | 高 | SM001, SM026 |
| CM013 | CAC’s generative-AI interim measures make model-data governance, security, and personal-information compliance part of the adoption environment for embodied-AI vendors using foundation models. | 中 | SM004 |
| CM014 | World Robotics is a recognized market-data source spanning industrial robots, service robots, and mobile robots, but public page access does not isolate a Simplexity-specific SAM. | 中 | SM006 |
| CM015 | Fortune Business Insights estimates the global warehouse robotics market at USD 6.51 billion in 2025, USD 7.35 billion in 2026, and USD 25.41 billion by 2034, with 16.8% CAGR. | 中 | SM007 |
| CM016 | Fortune Business Insights reports Asia-Pacific held 51.7% of the global warehouse robotics market in 2025 and projects China’s warehouse robotics market at USD 2.94 billion in 2026. | 中 | SM007 |
| CM017 | MarkNtel estimates China warehouse automation at USD 3.02 billion in 2025 and USD 9.17 billion by 2032, implying a 17.2% CAGR from 2026 to 2032. | 中 | SM008 |
| CM018 | MarkNtel says retail and e-commerce represent about 36% of China warehouse automation demand, East China about 33% of the market, and hardware about 46% of component value during the forecast period. | 中 | SM008 |
| CM019 | Global Market Insights estimates the global logistics robots market at USD 20.7 billion in 2026 and USD 91.4 billion by 2035, with 17.9% CAGR. | 中 | SM009 |
| CM020 | Global Market Insights lists e-commerce growth, labor shortages, AI/computer vision/navigation advances, RaaS models, efficiency, and resilience as logistics-robot drivers. | 中 | SM009 |
| CM021 | Global Market Insights identifies high upfront capex, uncertain ROI, and integration complexity with legacy warehouse systems as logistics-robot market challenges. | 中 | SM009 |
| CM022 | JD Logistics’ Zhilang system provides direct buyer proof that Chinese logistics operators deploy AGVs, lifting robots, shelves, workstations, algorithms, navigation, and verification for goods-to-person warehousing. | 中 | SM010 |
| CM023 | JD claims Zhilang improves picking efficiency by more than three times, raises storage density to 2.5 times industry average for facilities up to 10 meters high, and shortens payback by 30% versus comparable solutions. | 中 | SM010 |
| CM024 | JD reported nearly 100 Zhilang AGVs and lifting robots operating in a Beijing intelligent logistics park and handling nearly a million items. | 中 | SM010 |
| CM025 | JD says its warehouse-control, execution, robotics-management, 3D SCADA, and logistics-park software are used in more than 1,600 JD-operated warehouses and client facilities. | 中 | SM010, SM011 |
| CM026 | Geekplus targets e-commerce, 3PL, apparel, healthcare, groceries, auto manufacturing, and temperature-controlled storage, which maps the main buyer segments for industrial-logistics robotics. | 中 | SM012 |
| CM027 | Geekplus claims its warehouse automation can integrate into traditional warehouses with minimal adjustments and boost picking efficiency by up to 200%, creating a benchmark for buyer ROI narratives. | 中 | SM012 |
| CM028 | Geekplus cites Interact Analysis for seven consecutive years of No. 1 global AMR market share and projects order-fulfillment deployment sites rising from 5,500 in 2024 to 18,000 by 2030. | 中 | SM013, SM020 |
| CM029 | Geekplus says order fulfillment is one of the fastest-growing warehouse automation segments and reports 23% global market share plus 48.5% shelf-to-person share in that segment. | 中 | SM013 |
| CM030 | RoboticsTomorrow reports Geekplus 2025 revenue of RMB 3.171 billion, 31.6% year-on-year growth, positive adjusted net profit, and more than 72,000 robots delivered to about 950 end customers. | 中 | SM020 |
| CM031 | China’s warehouse robotics landscape is crowded with AMR, case-handling, vision-sorting, and intralogistics companies such as Geekplus, Hai Robotics, Hikrobot, Quicktron, and Youi Robotics. | 中 | SM021, SM013 |
| CM032 | The same crowding that validates demand also constrains Simplexity’s SOM because incumbents already sell deployed warehouse automation and have customer references. | 中 | SM010, SM013, SM020, SM021 |
| CM033 | Simplexity reportedly raised about CNY 2 billion, or USD 289.3 million, across five financings in less than half a year, with Tencent, Alibaba, Vision Capital, Lanchi, HongShan, Legend Capital, CAS Star, and Gaorong among investors. | 高 | SM017, SM016, SM022, SM023 |
| CM034 | Simplexity’s reported valuation exceeded USD 1 billion after the latest funding, making the market narrative dependent on rapid commercial application rather than disclosed revenue. | 中 | SM017, SM022, SM023 |
| CM035 | Public reports use Simplexity Robotics, Zhijian Power, and Zhijian Dynamics labels around the same Li Auto-linked embodied-intelligence story, creating a naming-normalization diligence risk. | 中 | SM015, SM016, SM017, SM018 |
| CM036 | The autonomous-driving-to-robotics migration is commercially plausible because VLA/world-model, mass-production, data-loop, and systems-engineering skills can transfer into controlled robot deployments. | 中 | SM015, SM016, SM017 |
| CM037 | The transfer is not one-for-one because robot operation data in factories, supermarkets, and logistics environments is more complex and random than road-driving data, and shadow-mode style validation remains unproven for robots. | 中 | SM016, SM015 |
| CM038 | Embodied robot companies face high model-training difficulty, high compute consumption, and large data needs, making training data access and compute economics material market constraints. | 中 | SM016, SM019 |
| CM039 | BIS 2026 guidance says advanced-computing exports to entities headquartered in Country Group D:5 or Macau require licenses, so China-based embodied-AI vendors can face external compute-supply constraints. | 中 | SM019 |
| CM040 | The 2026 MIIT action explicitly encourages leasing, pay-for-utility, and humanoid-robot-as-a-service models to lower user investment thresholds and accelerate market promotion. | 中 | SM002 |
| CM041 | The same MIIT action requires real-scene verification of success rate, efficiency improvement, safety reliability, and economic feasibility before normal deployment, which turns ROI evidence into a gating adoption step. | 中 | SM002 |
| CM042 | The embodied-AI funding environment is hot: one public source claims more than 200 domestic embodied-AI financings in Q1 2026, but that figure should be treated as low-confidence market-temperature evidence. | 低 | SM024 |
| CM043 | No public source reviewed discloses Simplexity’s paid customer count, contract value, unit price, gross margin, or repeat-deployment rate. | 中 | SM017, SM016, SM022, SM023 |
| CM044 | A practical TAM/SAM/SOM lens should use global logistics robots as a ceiling, China warehouse automation and warehouse robotics as SAM proxies, and closed-scenario pilots in factories, supermarkets, and logistics as the initial SOM wedge. | 中 | SM007, SM008, SM009, SM017 |
| CM045 | The public sizing sources are directionally consistent on high growth but not directly comparable because they use different units and boundaries: warehouse automation, warehouse robotics, and logistics robots. | 中 | SM007, SM008, SM009, SM006 |
| CM046 | Buyers, users, and payers split by segment: operations leaders fund throughput, warehouse teams use robots daily, IT/automation teams integrate systems, and finance validates payback. | 中 | SM009, SM010, SM012, SM013 |
| CM047 | Government support improves top-down scenario access, standards, and financing channels, but it does not remove the need for customer-level proof of uptime, safety, payback, and integration feasibility. | 中 | SM002, SM001, SM009, SM015 |
| CP001 | Simplexity Robotics is a newly formed embodied-robotics company founded in late July 2025, with Gasgoo reporting a $50 million angel round tied to Li Auto executives. | 高 | SP001, SP032 |
| CP002 | Simplexity claims a unified world-model and VLA foundation model, on-device real-time inference/training, and model-defined robot hardware. | 中 | SP001 |
| CP003 | The competitive landscape splits into direct embodied-robot peers, AV-stack adjacencies, substitutes/status quo, internal build, and likely entrants from auto/industrial AI. | 中 | SP001, SP032, SP033, SP034 |
| CP004 | AgiBot’s English site presents A2, G1, X2, and X1 robot lines and describes AGIBOT World as an embodied-AI development platform. | 中 | SP002 |
| CP005 | AgiBot’s Chinese about page positions the company as a global embodied-intelligence general AI robot company with full-stack robot body, algorithm, and open-platform pillars. | 中 | SP004 |
| CP006 | AgiBot’s Chinese home page states that the company’s 15,000th general embodied robot has rolled off the production line. | 中 | SP003 |
| CP007 | AgiBot’s 15,000th robot milestone is corroborated by its Chinese official page and a PR Newswire release describing the milestone unit as the industrial-grade G2. | 高 | SP003, SP005 |
| CP008 | AgiBot declared 2026 “Deployment Year One” and said it plans to invest more than RMB 2 billion over five years to expand its ecosystem. | 中 | SP006 |
| CP009 | Unitree’s official site presents a broad robot portfolio spanning consumer/education, industry, quadrupeds, humanoids, arms, perception, and components. | 中 | SP007 |
| CP010 | Unitree’s official G1 page and shop support a $13,500 to $13.5K public price signal for G1. | 高 | SP008, SP011 |
| CP011 | Unitree publishes a $1,600 Go2 price, while the H1 page/shop signals contact-us or conflicting public price information that should not be treated as a clean list price. | 中 | SP009, SP010, SP011 |
| CP012 | UBTECH’s official pages show humanoid service robot application scenarios, Walker-related product navigation, and investor/financial report access. | 中 | SP012, SP013 |
| CP013 | Horizon’s official page claims 10 million-plus Journey-series shipments, 400-plus design wins, 300-plus vehicles in mass production, and 40-plus partner OEM brands. | 高 | SP014, SP015, SP022 |
| CP014 | Horizon’s 2024 annual report disclosed RMB2.384 billion of revenue from contracts with customers and an operating loss of RMB2.144 billion. | 中 | SP015 |
| CP015 | Galbot has an official website presence, but the fetched official home page exposed very limited readable detail beyond the site title. | 中 | SP016 |
| CP016 | Galbot’s 2026 PR Newswire release states that it raised more than $300 million, reached $800 million total funding, and was valued at $3 billion. | 中 | SP017, SP018 |
| CP017 | TechNode independently reported Galbot’s RMB2.5 billion 2026 round and listed state/industrial investors including China Integrated Circuit Industry Investment Fund, Sinopec, CITIC-linked entities, and SAIC Motor’s financial arm. | 中 | SP018 |
| CP018 | Robotics & Automation News reported Galbot’s prior $151 million round led by CATL and Puquan and a Bosch investment-arm joint venture for global commercialization. | 中 | SP019 |
| CP019 | Galbot’s PR source claims thousands of unit orders, autonomous retail in more than 30 cities, and warehouse operations running continuously for over a year. | 中 | SP017 |
| CP020 | Pony.ai’s IR page describes PonyWorld and Virtual Driver powering Robotaxi, Robotruck, licensing, and applications businesses across multiple regions. | 高 | SP025, SP026 |
| CP021 | Pony.ai’s F-1 warns that it is still in a nascent stage of commercialization and faces highly competitive autonomous-driving markets. | 中 | SP026 |
| CP022 | WeRide’s IR and official pages claim tested or operated vehicles in over 40 cities across 12 countries, permits in eight markets, and a five-product portfolio including Robotaxi. | 高 | SP027, SP028 |
| CP023 | WeRide’s official page says it offers a WeRide One platform and five products, including Robotaxi, while being listed on Nasdaq and HKEX. | 中 | SP028 |
| CP024 | Mobileye’s official page markets driver-assist and autonomous-driving technology and highlights an intent to establish a vertically integrated robotaxi business. | 中 | SP029 |
| CP025 | Mobileye’s IR page shows ongoing public-company disclosure and a first-quarter 2026 results/share-repurchase announcement. | 中 | SP030 |
| CP026 | Mobileye describes SuperVision as a bridge from ADAS to consumer AVs, reinforcing its edge-safety and automaker-channel relevance. | 中 | SP031 |
| CP027 | Baidu Apollo’s official page states that Baidu began autonomous-driving work in 2013 and launched Apollo as an open autonomous-driving platform in 2017. | 高 | SP023, SP024 |
| CP028 | Baidu’s IR financial-report page indicates Baidu can provide audited annual reports to stakeholders and ADS holders, giving Apollo an incumbent public-company disclosure backdrop. | 中 | SP024 |
| CP029 | Gasgoo reports that auto executives are moving into embodied intelligence, that Li Auto leaders founded Zhijian Power, and that NEV price wars are pushing interest toward humanoid robotics. | 中 | SP032 |
| CP030 | TrendForce expects China humanoid robot output to grow up to 94% in 2026 and projects Unitree plus AgiBot at nearly 80% of total shipments. | 中 | SP033 |
| CP031 | DirectIndustry cautions that many robot media examples are demonstrations and reports 2025 production figures of 5,100 robots for AgiBot and 4,200 for Unitree. | 中 | SP034 |
| CP032 | Scale leaders and low-priced hardware create commoditization pressure for a new entrant whose public proof is still model thesis rather than deployments. | 中 | SP001, SP008, SP010, SP033, SP034 |
| CP033 | AgiBot, Unitree, UBTECH, and Galbot are the most relevant direct embodied-robotics peers because each has public robot product, scale, funding, or deployment evidence absent from Simplexity’s current public site. | 中 | SP001, SP003, SP007, SP012, SP017 |
| CP034 | AV-stack companies are adjacent rather than direct humanoid peers, but they can pressure Simplexity through world models, fleet data, safety/regulatory experience, OEM channels, and physical-AI talent. | 中 | SP020, SP021, SP022, SP025, SP027, SP029, SP032 |
| CP035 | Only Unitree provided clear reviewed list-price signals; Simplexity, AgiBot, Galbot, UBTECH humanoid enterprise products, and most AV adjacencies should remain unknown or quote-required. | 中 | SP001, SP008, SP010, SP011, SP017, SP012 |
| CP036 | Simplexity’s model-defined body and on-device training claims are promising but not yet durable moats without public benchmarks, customer uptime, or proprietary deployment data. | 中 | SP001, SP005, SP017, SP033, SP034 |
| CP037 | Public sources do not yet show strong switching costs or lock-in for Simplexity because no customer contracts, data-exclusivity terms, partner APIs, or installed-base metrics were found. | 低 | |
| CP038 | Distribution power appears stronger at AgiBot, Unitree, UBTECH, Galbot, and AV incumbents than at Simplexity because those peers expose production, shop, investor, OEM, or public-market channels. | 中 | SP003, SP007, SP011, SP013, SP017, SP022, SP025, SP027 |
| CP039 | Internal build is a credible status quo for automotive and industrial buyers because Gasgoo shows auto leaders entering robotics and AV-stack firms already own relevant perception, planning, and safety assets. | 中 | SP020, SP022, SP025, SP027, SP032 |
| CP040 | The provided AutoX official URL returned a 404 body, so AutoX should remain an unscored adjacent competitor until a current official source is retrieved. | 中 | SP035 |
| CP041 | Unitree’s shop context lists related robot products such as R1 and H2 around low-to-mid five-figure prices, reinforcing visible price segmentation in robot hardware. | 中 | SP011 |
| CP042 | Unitree H1 pricing should be treated as quote-required because the shop title says contact us for real price while related listings show a numeric price. | 中 | SP009, SP011 |
| CP043 | DirectIndustry’s market overview indicates logistics and manufacturing are first key humanoid deployment areas, supporting factory/industrial use as a near-term battlefield. | 中 | SP034 |
| CI001 | Simplexity Robotics says on its official website that it was founded in late July 2025 to build embodied-intelligence products from real-world scenarios. | 中 | SI001 |
| CI002 | The official website describes LaST-series models, ManualVLA, and robotics hardware as core technology surfaces, but it does not publish revenue, pricing, margin, cash, or runway metrics. | 中 | SI001 |
| CI003 | Multiple independent reports state that Simplexity Robotics raised about CNY2.0 billion, or roughly USD289 million, across five financings within less than six months. | 高 | SI002, SI003, SI004, SI005, SI006, SI007 |
| CI004 | Independent coverage reports that Simplexity Robotics crossed a post-money valuation above USD1.0 billion after the latest 2026 funding activity. | 高 | SI002, SI005, SI007, SI008 |
| CI005 | Reported investors include Tencent, Alibaba, Vision Plus/Yuanjing, BlueRun/Lanchi, HongShan, Legend Capital, CAS Star, Gaorong, and other financial or strategic investors. | 中 | SI002, SI003, SI004, SI005, SI009, SI012 |
| CI006 | Reported use of funds centers on foundation-model training, robot body research and iteration, data collection, and core algorithm development rather than sales expansion alone. | 中 | SI002, SI003, SI004, SI006, SI010 |
| CI007 | 36Kr reports that Simplexity moved from first employee arrival to first-generation self-developed robot body in under 45 days and had started small-batch body rollout and PoC validation. | 中 | SI003, SI011 |
| CI008 | Sina Finance reports that Simplexity is headquartered in Hangzhou and has Beijing, Shanghai, and Suzhou R&D or business layouts. | 中 | SI004 |
| CI009 | Preqin describes Simplexity revenue as coming from direct sales of embodied-AI robotic hardware, supporting software models, and autonomous-driving-system solutions to enterprise clients. | 中 | SI006 |
| CI010 | No retained source discloses Simplexity revenue, ARR, gross margin, contribution margin, CAC, payback, monthly burn, cash balance, or runway. | 中 | |
| CI011 | Public reports describe enterprise PoC validation and initial closed-scenario targets, but they do not disclose named customers, contract values, paid conversion, utilization, or renewal terms. | 中 | |
| CI012 | Simplexity reportedly prioritizes closed environments such as factory workshops, supermarkets, and logistics before moving into more open scenarios. | 中 | SI002, SI003, SI004 |
| CI013 | Because Simplexity is private, very young, and pre-audited in the public record, its valuation cannot be translated into a reliable revenue multiple from public evidence. | 中 | SI001, SI002, SI003, SI010 |
| CI014 | The disclosed CNY2.0 billion financing gives Simplexity a large headline capital buffer, but public sources do not show cash on hand, monthly burn, runway months, debt, or project-finance obligations. | 中 | SI002, SI003, SI004, SI006 |
| CI015 | Pony AI disclosed in its F-1 that total revenues increased from US$68.4 million in 2022 to US$71.9 million in 2023 and that first-half 2024 revenue was US$24.7 million. | 高 | SI017, SI018 |
| CI016 | Pony AI disclosed in its 2025 Form 20-F that total revenues increased 20.0% from US$75.0 million in 2024 to US$90.0 million in 2025. | 高 | SI017, SI019, SI030 |
| CI017 | WeRide disclosed 2025 total revenue of RMB684.6 million, gross profit of RMB4.9 million, and net loss of RMB1.25 billion, illustrating low gross profit and high losses in an autonomous-robotics comparable. | 高 | SI020, SI021, SI029 |
| CI018 | WeRide disclosed that its largest customer represented 11.4% of 2025 revenue, down from 55.3% in 2023 and 24.4% in 2024. | 高 | SI020, SI021 |
| CI019 | Mobileye disclosed that China, Germany, and South Korea accounted for 23%, 16%, and 10% of 2025 revenue by shipment destination, showing geographic concentration can be material in autonomy suppliers. | 高 | SI022, SI023 |
| CI020 | Mobileye reported net losses of US$392 million in 2025 and US$3.09 billion in 2024, with the 2024 loss primarily reflecting a non-cash goodwill impairment. | 高 | SI022, SI023 |
| CI021 | Baidu disclosed 2025 total revenue of RMB129.1 billion, down 3% from 2024, with cloud-services growth partly offsetting lower online-marketing revenue. | 高 | SI024, SI025, SI031 |
| CI022 | UBTECH public materials and HKEX filings show 2024 revenue of about RMB1.305 billion and a 2024 net loss of about RMB1.160 billion. | 高 | SI026, SI027 |
| CI023 | Horizon Robotics reported 2024 total revenue of RMB2.384 billion, non-IFRS net loss of RMB1.681 billion, and a largest-customer share of 31.5% for 2024. | 中 | SI028 |
| CI024 | TrendForce reports Unitree and AgiBot are expected to capture nearly 80% of China humanoid output in 2026, and cites Unitree segment gross margin near 60% as a positive benchmark. | 中 | SI015 |
| CI025 | The Business Times reported an NDRC warning that humanoid robotics faces bubble risk, with more than 150 makers operating in China. | 中 | SI013 |
| CI026 | France 24 separately reported the same Chinese official warning that speed and bubble risk must be balanced in humanoid robotics. | 中 | SI014 |
| CI027 | The comparable filings show that autonomy and robotics companies can report meaningful revenue while still consuming large operating capital through R&D, support, fleet, manufacturing, or impairment costs. | 中 | SI018, SI019, SI021, SI023, SI027, SI028 |
| CI028 | Simplexity has no public list pricing or realized contract-pricing evidence, so any hardware ASP, software license fee, service fee, or autonomous-driving solution price would be unsupported. | 中 | |
| CI029 | Simplexity unit economics cannot be calculated publicly because the numerator and denominator for robot cost, service labor, software attach rate, utilization, warranty, and data-collection cost are all undisclosed. | 中 | |
| CI030 | The most defensible public revenue model is a hybrid enterprise model spanning robot hardware, supporting AI software, and solution delivery, but only Preqin explicitly describes it as a revenue model. | 中 | SI001, SI006, SI012 |
| CI031 | Closed-scenario PoCs are a traction signal but should not be treated as recurring revenue until management provides signed contract values, paid conversion, deployment counts, and renewal economics. | 中 | SI003, SI004, SI006 |
| CI032 | Full-stack robot development and foundation-model training imply capital needs across model compute, robot bill of materials, testing, field support, and working capital. | 中 | SI001, SI002, SI003, SI006 |
| CI033 | Public evidence does not identify any Simplexity debt, credit facility, lease obligation, customer prepayment, or project-finance obligation. | 中 | |
| CI034 | Customer concentration is a diligence risk for Simplexity because public comparables such as WeRide and Horizon disclose material dependence on major customers, while Simplexity has not disclosed its customer base. | 中 | SI021, SI028 |
| CI035 | Gross-margin path for Simplexity depends on robot production yield, component costs, field-service load, software attach, and deployment utilization, none of which are public. | 中 | |
| CI036 | The sector bubble warning is financially adverse because it can tighten next-round terms for young robotics companies whose revenue quality and unit economics are not yet visible. | 中 | SI013, SI014 |
| CI037 | A reasonable diligence stance is to treat the disclosed funding and valuation as real reported facts while treating revenue, ARR, margins, burn, runway, and unit economics as unknown until management provides private data. | 中 | SI002, SI003, SI004, SI006, SI013, SI021, SI027, SI028 |
| CI038 | Simplexity’s planned use of funds is R&D-heavy, so the next-round trigger is more likely to be productized deployments and gross-margin proof than a conventional SaaS ARR threshold. | 中 | SI002, SI003, SI006, SI015 |
| CI039 | No retained source supports a claim that Simplexity is profitable, cash-flow positive, gross-margin positive, or revenue-generating at scale. | 中 | |
| CI040 | The chapter’s financial recommendation is research-more rather than avoid because the financing syndicate is strong, but the private financial statements needed to underwrite revenue quality and runway are absent. | 中 | SI002, SI003, SI005, SI013, SI014, SI021, SI027, SI028 |
| CE001 | Zhijian Power's official site defines the company as building embodied-intelligence products from real scenarios through a unified model, data closed loop, and robot hardware. | 中 | SE001 |
| CE002 | The official site says Simplexity Robotics was founded in late July 2025. | 中 | SE001 |
| CE003 | Yicai reported that Simplexity Robotics was set up in July 2025 by Jia Peng, Wang Kai, and Wang Jiajia. | 中 | SE003 |
| CE004 | Yicai reported that Simplexity's first-generation robot had been produced in small batches and that PoC testing had started. | 中 | SE003 |
| CE005 | 36Kr EU reported that Zhijian had developed two generations of robot bodies for B-end and C-end users, achieved small-batch production, and fully launched PoC verification. | 中 | SE002 |
| CE006 | Yicai reported that Simplexity plans to start with closed scenarios including factory workshops, supermarkets, and logistics. | 中 | SE003 |
| CE007 | The official site says Zhijian has built a world-model-and-VLA integrated model through a unified Transformer. | 中 | SE001 |
| CE008 | The official site says the unified model jointly models language logic, visual semantics, 3D spatial structure, and robot state for understanding, generation, and prediction. | 中 | SE001 |
| CE009 | 36Kr EU characterized Zhijian's approach as moving a VLA model from cars into robot bodies. | 中 | SE002 |
| CE010 | 36Kr EU reported that Jia Peng led Li Auto intelligent-driving R&D and worked on VLA-related technology before founding Zhijian. | 中 | SE002 |
| CE011 | The LaST0 arXiv abstract describes a latent spatio-temporal chain-of-thought approach that captures visual dynamics, 3D structure, and robot proprioceptive states. | 中 | SE005, SE006 |
| CE012 | LaST0 uses a Mixture-of-Transformers dual-system design with a low-frequency reasoning expert and high-frequency acting expert. | 中 | SE005, SE008 |
| CE013 | The LaST0 paper reports mean success-rate improvements of 13%, 14%, and 14% over prior VLA methods across tabletop, mobile, and dexterous real-world tasks. | 中 | SE005, SE006 |
| CE014 | The ManualVLA arXiv abstract says the framework uses a Mixture-of-Transformers architecture with planning and action execution. | 中 | SE009, SE010 |
| CE015 | ManualVLA equips a planning expert to generate intermediate manuals of images, position prompts, and textual instructions, then feeds them into an action expert through ManualCoT. | 中 | SE009, SE010, SE011 |
| CE016 | The ManualVLA PDF says its digital-twin toolkit uses 3D Gaussian Splatting to automatically generate manual data for planning-expert training. | 中 | SE010 |
| CE017 | The ManualVLA paper reports an average success rate 32% higher than the previous hierarchical SOTA baseline on LEGO assembly and object rearrangement tasks. | 中 | SE009, SE010 |
| CE018 | TwinRL is described as a digital twin-real-world collaborative post-training framework for VLA models. | 中 | SE012, SE013, SE016 |
| CE019 | TwinRL reconstructs high-fidelity digital twins from smartphone-captured scenes and uses twin rollouts to guide real-world reinforcement learning. | 中 | SE012, SE013, SE017 |
| CE020 | The TwinRL paper reports near-100% success across four tasks, over 30% faster convergence, and about 20 minutes of on-robot interaction. | 中 | SE012, SE013, SE022 |
| CE021 | Yicai reported that Simplexity had secured CNY2 billion for foundation models, robot R&D and iteration, data collection, and core algorithms. | 中 | SE003 |
| CE022 | Yicai reported that Simplexity follows a progressive path from closed to semi-open and fully open scenarios. | 中 | SE003 |
| CE023 | 36Kr EU described Zhijian's learning paradigm as collecting human operation data, using human demonstrations for downstream exploration, and using real-time manual guidance for online learning. | 中 | SE002 |
| CE024 | The official site describes on-device deployment, real-time inference and training, low latency, high reliability, and fast learning as technical highlights. | 中 | SE001 |
| CE025 | The official site says its model-defined body and one general body are intended to improve data generality and reuse. | 中 | SE001 |
| CE026 | The official site says the company aims to build an efficient data closed-loop system and accelerate data-collection efficiency. | 中 | SE001 |
| CE027 | CSDN's industry compilation summarizes Zhijian's four-O system as one model, on device, one body, and one hour. | 低 | SE025 |
| CE028 | CSDN reports that Zhijian proposes a Human data is all you need paradigm with an edge shadow-mode data loop. | 低 | SE025 |
| CE029 | Public sources do not disclose edge-compute specifications, latency budgets, data rights, or rollback controls for Zhijian's on-device learning claims. | 中 | SE001, SE002, SE003, SE025 |
| CE030 | The official site lists hiring needs across hardware, embedded, robotic control, world model, cloud model, big data, perception, model deployment, calibration/SLAM, reinforcement learning, foundation model, VLA algorithm, scheduling, platform software, application interaction, and simulation roles. | 中 | SE001 |
| CE031 | Gasgoo reported that autonomous-driving entrepreneurs are entering embodied intelligence and that automotive and robotics share similarities in algorithms, computing power needs, and hardware. | 中 | SE004 |
| CE032 | Tencent News / Auto-First listed Wang Kai, Jia Peng, and Wang Jiajia as Li Auto alumni who moved to Zhijian Power roles. | 中 | SE026 |
| CE033 | The TwinRL GitHub repository publicly describes Twin-RL as a digital twin-real-world collaborative RL framework for VLA models. | 中 | SE014, SE015 |
| CE034 | The TwinRL dataset guide says the repository provides high-fidelity digital twin assets and twin-generated trajectories. | 中 | SE017, SE018 |
| CE035 | Independent GitHub curation repositories list VLA, RL-VLA, and TwinRL-related resources, showing practitioner-community visibility around the technical area. | 低 | SE019, SE020, SE021 |
| CE036 | alphaXiv pages mirror the LaST0, ManualVLA, and TwinRL paper texts, creating an additional practitioner review surface for the research artifacts. | 低 | SE022, SE023, SE024 |
| CE037 | Reviewed public sources did not reveal a Zhijian SDK, public API, package registry, release notes, or customer-support developer forum. | 中 | SE001, SE014, SE015, SE017, SE019, SE020, SE021 |
| CE038 | The public maturity signal is prototype/PoC-level because media reports describe small batches and PoC tests but do not name scaled customers or production KPIs. | 中 | SE002, SE003 |
| CE039 | The relationship between LaST0, ManualVLA, and TwinRL research artifacts and a shipped Zhijian customer robot is not disclosed in public sources. | 中 | SE001, SE005, SE009, SE012, SE014 |
| CE040 | Public sources do not disclose integration details such as WMS/warehouse systems, retail systems, remote operations, or customer acceptance tests. | 中 | SE002, SE003, SE025 |
| CE041 | Reviewed public sources did not disclose robot safety certifications, cybersecurity controls, privacy controls, uptime, MTBF, intervention rate, or incident history. | 中 | SE001, SE002, SE003, SE014, SE015, SE017 |
| CE042 | Public sources do not disclose manufacturing quality systems, supplier lists, yield data, warranty process, or service plans for Zhijian robots. | 中 | SE001, SE002, SE003, SE025 |
| CU001 | Zhijian Power's official site positions the company as starting from real scenarios to build high-user-value embodied-intelligence products, but it does not name any customer, case study, or deployment logo. | 中 | SU001 |
| CU002 | The official site describes technical pillars such as one model, on-device deployment, one body, and one-hour fast learning, which support a customer-value narrative without proving adoption. | 中 | SU001 |
| CU003 | 36Kr reports that Zhijian Power was founded by former Li Auto leaders and financed rapidly, making the public story more team-and-capital-led than customer-led. | 中 | SU002 |
| CU004 | Yicai reports that the CNY2 billion financing is intended to accelerate large-scale application of embodied-intelligence technologies across multiple scenarios. | 中 | SU003 |
| CU005 | Sohu reports that Zhijian Power had completed two generations of B-end and C-end robot bodies, achieved small-batch body rollout, and opened PoC verification. | 中 | SU004 |
| CU006 | Sohu identifies factory workshops, supermarkets, and logistics as the company's priority closed scenarios, but the article does not name end customers in those verticals. | 中 | SU004 |
| CU007 | Toutiao/Yinlibo reports that Zhijian Power signed a Suzhou global innovation center and strategic cooperation with Leaderdrive on 2026-02-05. | 中 | SU005 |
| CU008 | The Leaderdrive relationship is best classified as strategic partner and component ecosystem proof, not a paying-customer deployment, because the public evidence describes algorithm/manufacturing complementarity rather than a customer purchase. | 中 | SU005, SU006, SU007 |
| CU009 | The Suzhou signing article says a first industrial dual-arm robot was in sample testing and productization, which indicates technical progress before production customer proof. | 中 | SU005 |
| CU010 | Leaderdrive's official site and Xinhua profile support its credibility as a precision harmonic-drive supplier to the robotics industry. | 中 | SU006, SU007 |
| CU011 | Sina Finance reports that by March 2026 Zhijian had cooperated with multiple manufacturing, supermarket, and logistics enterprises around factory, shelf-tidying, and sorting scenes. | 中 | SU008 |
| CU012 | The Sina cooperation claim remains unnamed and outcome-free, so it should be treated as pipeline or pilot/cooperation evidence rather than named customer-proof. | 中 | SU008, SU004, SU005 |
| CU013 | Multiple Chinese secondary articles repeat the small-batch and PoC narrative, but none of the reviewed articles names a paying production customer. | 中 | SU002, SU004, SU008, SU009, SU010, SU011 |
| CU014 | The most supportable customer segmentation is B-end closed environments: industrial manufacturing, supermarket/retail shelf operations, and logistics/warehouse sorting. | 中 | SU004, SU008, SU009, SU011 |
| CU015 | The public geography footprint relevant to customer access is China-centered, with Beijing, Shanghai, and Suzhou cited in coverage and Suzhou positioned as the innovation and industrialization center. | 中 | SU004, SU005 |
| CU016 | No reviewed source discloses Zhijian Power customer count, active sites, robots deployed at customer sites, contract length, NRR, GRR, churn, renewal rate, or satisfaction score. | 中 | SU001, SU002, SU003, SU004, SU005, SU008 |
| CU017 | No reviewed Zhijian-specific source discloses customer outcomes such as throughput improvement, labor savings, uptime, safety incidents, or error-rate reduction. | 中 | SU001, SU002, SU003, SU004, SU005, SU008 |
| CU018 | Because named customers and revenue mix are absent, top-customer concentration and channel-dependence risk cannot be bounded from public data. | 中 | SU013, SU016, SU024, SU025 |
| CU019 | Zhijian's public adoption funnel is best represented as funding and team credibility, then prototype/small-batch output, then PoC/unnamed cooperation, with named production deployment still unproven. | 中 | SU002, SU003, SU004, SU005, SU008 |
| CU020 | Tencent and Alibaba are strategic investors according to multiple reports, but no reviewed source shows either acting as a customer, channel customer, or deployment reference. | 中 | SU002, SU003, SU004, SU011 |
| CU021 | JD Logistics' Zhilang article is a benchmark customer-proof source because it names JD Logistics operations, describes Goods-to-Person warehousing, says the system supports Singles Day operations, and quantifies picking efficiency and storage density. | 中 | SU014 |
| CU022 | Geek+ states its warehouse automation targets eCommerce, 3PL, apparel, healthcare, grocery, auto manufacturing, and temperature-controlled storage, providing a peer segmentation benchmark for warehouse robotics demand. | 中 | SU013 |
| CU023 | Geek+ states its robots can boost picking efficiency by up to 200 percent, a level of quantified outcome proof that Zhijian-specific public sources do not yet provide. | 中 | SU013 |
| CU024 | Locus Robotics' Boots UK customer proof says Boots more than doubled online volume during the pandemic and that LocusBots enabled it to manage peak volumes. | 中 | SU016 |
| CU025 | UBTECH's 2024 annual report names automotive and logistics counterparties including BYD, Geely Automobile, FAW-Volkswagen Qingdao Branch, Audi FAW, Dongfeng Liuzhou Motor, Beijing Automotive New Energy, Foxconn, and SF for humanoid robot training or cooperation. | 中 | SU021 |
| CU026 | UBTECH's 2025 annual report reports RMB820 million in full-size embodied-intelligent humanoid robot products and services revenue and sales volume of 1,079 units in 2025. | 中 | SU020 |
| CU027 | IFR reports 4,281,585 operational factory robots worldwide in 2023 and 276,288 installations in China, equal to 51 percent of global installations. | 中 | SU017 |
| CU028 | IFR's July 2026 leadership update describes the global robotics industry as at an important inflection point driven by AI and automation. | 中 | SU018 |
| CU029 | OSHA describes industrial robots as used for unsafe, hazardous, repetitive, and unpleasant tasks, but warns accidents often occur during programming, maintenance, testing, setup, or adjustment. | 中 | SU022 |
| CU030 | ISO 10218-1:2025 is relevant to industrial robot safety as machine-level safety requirements, reinforcing that deployment requires integration and safety diligence beyond technical demos. | 中 | SU023 |
| CU031 | France 24 reports official Chinese concern that humanoid robotics faces bubble risk and is not yet mature in technology, commercialization, or use. | 中 | SU024 |
| CU032 | The Business Times reports China's NDRC warned that more than 150 humanoid robot makers and highly similar models could create bubble and overcapacity risk. | 中 | SU025 |
| CU033 | Adverse industry warnings make it unsafe to underwrite Zhijian's unnamed cooperation claims as durable customer demand without purchase orders, utilization data, or renewal evidence. | 中 | SU024, SU025, SU008 |
| CU034 | The absence of named customer-proof is a core finding rather than a data-cleaning issue: reviewed direct, news, partner, and benchmark sources support pilots and ecosystem readiness but not named paid production adoption for Zhijian. | 中 | SU001, SU002, SU003, SU004, SU005, SU008, SU009 |
| CU035 | A likely early customer journey starts with innovation-center demonstration, proceeds to controlled PoC in closed scenes, then safety/integration validation, small production rollout, and only later multi-site expansion. | 中 | SU004, SU005, SU008, SU022, SU023 |
| CU036 | The likely buyer is an operations, automation, or innovation leader at manufacturing, retail, or logistics companies; the likely users are line workers and site operations teams; and the payer is probably the enterprise operating site. | 中 | SU004, SU008, SU022 |
| CU037 | Public evidence supports a land-and-expand hypothesis only as a mechanism, not as observed behavior, because no account-level pilot-to-production or multi-site expansion record is disclosed. | 中 | SU008, SU016, SU021 |
| CU038 | Zhijian's customer proof quality is currently materially below peer benchmarks from JD Logistics, Boots/Locus, and UBTECH, all of which provide named deployment or operating details. | 中 | SU014, SU016, SU020, SU021, SU001, SU008 |
| CU039 | The strongest named external relationship in the reviewed Zhijian file is Leaderdrive, but Leaderdrive evidence validates supply-chain and industrial ecosystem access rather than customer retention. | 中 | SU005, SU006, SU007 |
| CU040 | The diligence ask should focus on signed customer list, pilot-to-production conversion, paid robots deployed by site, utilization, safety acceptance, renewal/expansion terms, and revenue concentration. | 中 | SU016, SU021, SU022, SU023, SU024, SU025 |
| CR001 | Simplexity Robotics publicly describes itself as founded in late July 2025 to build embodied-intelligence products using a unified model, data loop, and reliable robot hardware. | 高 | SR001, SR008 |
| CR002 | The official site markets a world-model and VLA integrated embodied foundation model plus on-device deployment and online training, which makes telemetry, model safety, and compute availability central diligence topics. | 高 | SR001, SR002 |
| CR003 | Multiple 2026 reports state Simplexity completed five financing rounds totaling about RMB 2 billion within less than half a year. | 高 | SR002, SR003, SR004, SR005, SR006 |
| CR004 | Public reports name top-tier financial and strategic backers including HongShan or Sequoia China, Legend Capital, BlueRun, Yuanjing, Tencent, and Alibaba-related strategic capital. | 高 | SR003, SR004, SR006, SR007 |
| CR005 | The same financing reports imply a valuation around or above US$1 billion for a company founded in 2025, increasing sensitivity to proof of production deployments and future financing conditions. | 高 | SR002, SR003, SR004 |
| CR006 | Simplexity has public evidence of technical ambition and hiring needs across industrial design, mechanical, embedded, controls, world model, data operations, perception, SLAM, and VLA roles. | 中 | SR001 |
| CR007 | Public company-facing evidence reviewed for this chapter does not identify signed production customers, paid deployments, customer revenue, retention, safety certifications, or incident history. | 中 | SR001, SR002, SR003, SR004 |
| CR008 | Several reports describe small-batch body production and PoC validation, but that wording is materially weaker than evidence of recurring production revenue. | 中 | SR002, SR006, SR007 |
| CR009 | The founding team is repeatedly tied to former Li Auto autonomous-driving leaders, concentrating the thesis around transfer of autonomous-driving methodology into robotics. | 中 | SR002, SR005, SR008 |
| CR010 | China generative-AI measures require providers to manage training data legality, IP, personal information, safety, service stability, complaints, and reporting for qualifying public generative-AI services. | 高 | SR011, SR016 |
| CR011 | The generative-AI measures require safety assessment and algorithm filing for services with public-opinion or social-mobilization attributes, making consumer-facing AI functions a potential launch gate. | 高 | SR011, SR017 |
| CR012 | Deep-synthesis rules create labeling and provider-responsibility obligations that could apply if robots generate or transform synthetic images, audio, video, or text for users. | 高 | SR013, SR017 |
| CR013 | Robots that learn from cameras, microphones, human demonstrations, or workplace telemetry can implicate personal-information, data-security, and cybersecurity duties even if the robot is not a public chatbot. | 高 | SR011, SR014, SR015, SR016 |
| CR014 | Cybersecurity review rules can become relevant when network products, services, critical information infrastructure, important data, or national-security concerns enter a deployment. | 高 | SR012, SR016 |
| CR015 | Product-quality and tort-liability sources support treating injury, property damage, defective design, warning defects, and inadequate maintenance as core residual legal risks for autonomous robots. | 高 | SR018, SR019, SR028 |
| CR016 | BIS connected-vehicle rules show the United States is willing to restrict Chinese-linked autonomous-system hardware and software supply chains on national-security grounds. | 高 | SR021, SR022 |
| CR017 | Advanced-computing export controls and entity-list restrictions create a material medium-confidence risk for any China robotics company that depends on frontier GPUs, accelerators, or restricted design services. | 中 | SR020, SR023, SR024, SR037 |
| CR018 | The NHTSA autonomous-vehicle safety page is a useful regulatory analogy because it frames automated physical-world systems around safety, testing, and public-road risk rather than software performance alone. | 中 | SR027 |
| CR019 | OSHA’s robotics guidance identifies workplace robot hazards, so factory and logistics deployments should require customer-site risk assessment, guarding, procedures, and incident reporting. | 中 | SR028 |
| CR020 | IFR reports demonstrate that China and global factories already have large installed robot bases, increasing both opportunity and benchmark pressure for reliable industrial-grade deployment. | 高 | SR029, SR030 |
| CR021 | Chinese state-linked warnings reported by Business Times and France 24 describe bubble risk in humanoid robotics, which is adverse evidence for valuation and capital-cycle assumptions. | 高 | SR031, SR032 |
| CR022 | The Robot Report’s discussion of humanoid difficulty supports a medium-confidence view that manipulation, reliability, cost, and safety remain hard commercialization barriers. | 中 | SR033 |
| CR023 | TrendForce’s 2026 humanoid market coverage supports continued investor attention but does not remove execution or unit-economics risk for pre-scale entrants. | 中 | SR034 |
| CR024 | Comparable customer-proof sources from JD Logistics and Locus Robotics show that mature warehouse-robot vendors publish named deployment evidence, unlike the public Simplexity record reviewed here. | 中 | SR035, SR036, SR001 |
| CR025 | Pony AI’s F-1 shows China autonomous-system issuers disclose risks around PRC regulation, data, product liability, commercialization, and dependence on permits or partners. | 中 | SR025 |
| CR026 | UBTECH’s public filings are a relevant humanoid-robotics comparable for commercialization, losses, supply chain, and product-development risk in embodied robotics. | 中 | SR026 |
| CR027 | The strongest current mitigant for Simplexity is founder experience in autonomous-driving commercialization, but that also creates key-person and transferability risk. | 中 | SR002, SR005, SR008 |
| CR028 | A second mitigant is substantial 2026 financing, but the same capital intensity raises burn and milestone pressure if production customers do not convert quickly. | 中 | SR003, SR004, SR005 |
| CR029 | Strategic investors Tencent and Alibaba can help with cloud, channels, data, or ecosystem access, but dependency or exclusivity terms are not public and require diligence. | 中 | SR003, SR004 |
| CR030 | A general-purpose body and one-model strategy can improve data reuse, but it also concentrates product, data, and safety risk in one architecture. | 中 | SR001, SR002 |
| CR031 | The public record reviewed here does not establish whether Simplexity has required safety certifications, factory acceptance tests, cyber audits, recall procedures, or insurance for robot deployments. | 中 | SR001, SR002, SR028 |
| CR032 | The company’s official hiring list spans both hardware and software roles, indicating that execution requires simultaneous scaling across mechanical, embedded, AI, data, and product functions. | 中 | SR001 |
| CR033 | Because robots operate in physical workplaces and collect multimodal data, a single customer incident could transmit into product liability, deployment pauses, regulatory scrutiny, and financing risk. | 中 | SR015, SR018, SR019, SR028 |
| CR034 | No public source reviewed provides a complete China administrative-penalty, credit, court-enforcement, or litigation clearance for the company as of the run date. | 中 | SR008, SR009, SR010 |
| CR035 | Credit China and court-enforcement portals were considered diligence paths, but public access limitations mean negative assurance cannot be drawn from this chapter alone. | 中 | SR010 |
| CR036 | The highest residual risks are customer-proof gap, safety/product liability, chip/export-control dependence, burn after a large private round, and key-person execution concentration. | 中 | SR001, SR003, SR011, SR018, SR020, SR021, SR031, SR033 |
| CR037 | A thesis-break trigger should be any serious robot injury, safety shutdown, regulatory inquiry, or inability to document site-level safety controls before pilots expand. | 中 | SR018, SR019, SR028 |
| CR038 | A thesis-break trigger should be failure to convert PoCs into named paying production deployments within the next financing cycle. | 中 | SR002, SR003, SR024 |
| CR039 | A thesis-break trigger should be loss of access to required AI accelerators or edge-compute components without a tested domestic or lower-spec alternative. | 中 | SR020, SR023, SR024, SR037 |
| CR040 | A thesis-break trigger should be founder departure, unresolved governance concentration, or inability to recruit senior manufacturing and safety leadership. | 中 | SR001, SR002, SR008 |
| CR041 | A monitorable mitigation is a formal regulatory matrix covering AI service scope, algorithm filing, PIPL/data-security workflow, product safety, and export-control screening. | 中 | SR011, SR012, SR013, SR014, SR015, SR018, SR020 |
| CR042 | A monitorable mitigation is an auditable safety case for each deployment, including hazard analysis, acceptance tests, remote stop, maintenance, incident logs, and customer sign-off. | 中 | SR018, SR019, SR028 |
| CR043 | A monitorable mitigation is named-customer evidence that separates demos, PoCs, paid pilots, production deployments, repeat orders, and retention. | 中 | SR024, SR035, SR036 |
| CR044 | A monitorable mitigation is supplier and compute redundancy, with bill-of-materials screening against export controls and entity-list exposure. | 中 | SR020, SR023, SR024, SR037 |
| CR045 | A monitorable mitigation is founder succession and operating-governance evidence, including independent safety ownership and documented decision rights for incident response. | 中 | SR001, SR008 |
| CV001 | Multiple business-media sources report that Simplexity Robotics raised RMB2.0 billion across five financing rounds in roughly six months. | 高 | SV001, SV002, SV003, SV005 |
| CV002 | Yicai and other funding coverage report that Simplexity Robotics exceeded a post-money valuation of US$1.0 billion after the latest round. | 高 | SV002, SV004, SV007 |
| CV003 | Public profiles identify the company as founded in 2025 by former Li Auto executives including Jia Peng, Wang Kai, and Wang Jiajia. | 中 | SV002, SV007 |
| CV004 | Public funding coverage names major financial and strategic investors including Tencent- and Alibaba-linked capital, HongShan or Sequoia China, Legend Capital, CAS Star, Lanchi, and Gaorong. | 中 | SV001, SV003, SV004, SV005 |
| CV005 | Reported use of proceeds centers on robotics foundation models, core algorithms, data collection, platform iteration, and large-scale application rather than disclosed near-term revenue. | 中 | SV002, SV004, SV007 |
| CV006 | Preqin describes Simplexity as serving enterprise clients in manufacturing and smart-vehicle sectors with integrated embodied-AI software and hardware. | 中 | SV008 |
| CV007 | The retained public sources do not provide a quantified revenue run rate for Simplexity Robotics. | 中 | SV001, SV002, SV003, SV004, SV005, SV008 |
| CV008 | The retained public sources do not provide a named customer list or customer-count disclosure for Simplexity Robotics. | 中 | SV001, SV002, SV003, SV004, SV006, SV008 |
| CV009 | No retained public source discloses Simplexity round preferences, liquidation stack, secondary mix, or common-equity price. | 中 | SV001, SV002, SV003, SV004, SV005 |
| CV010 | A reported valuation above US$1.0 billion cannot be tested with standard revenue-multiple math from public materials because revenue, margin, backlog, and customer concentration are undisclosed. | 中 | SV001, SV002, SV007, SV008 |
| CV011 | The Business Times reported that China’s economic-planning agency warned about bubble risk in humanoid robotics. | 高 | SV009, SV011, SV012 |
| CV012 | Adverse coverage frames China’s humanoid-robotics market as crowded, with more than 150 companies and limited proven deployments in factories or homes. | 中 | SV011, SV012, SV013 |
| CV013 | Firstpost reported that humanoid-robotics sector valuation multiples had run ahead of the broader Chinese industrial-equipment market. | 中 | SV010 |
| CV014 | Morgan Stanley’s 2026 forecast upgrade indicates faster commercialization momentum in China’s humanoid-robotics market. | 中 | SV014 |
| CV015 | TrendForce expects China’s humanoid-robot output to surge in 2026 and highlights Unitree and AgiBot as leading share capturers. | 中 | SV015 |
| CV016 | Grand View Research and other analyst-market-data sources support the view that humanoid robotics is a growing global category. | 中 | SV016, SV018, SV033 |
| CV017 | The International Federation of Robotics frames World Robotics as a source for robotics statistics, trends, and analyses across industrial, service, and mobile robots. | 中 | SV017 |
| CV018 | UBTECH’s 2024 annual results reported revenue of RMB1.3054 billion and a loss for the year of RMB1.1599 billion. | 高 | SV020, SV019 |
| CV019 | UBTECH’s filings show that a public humanoid-robotics comparable can have meaningful revenue while still reporting large operating losses. | 中 | SV020, SV021, SV022 |
| CV020 | Pony AI’s F-1 cites Frost & Sullivan estimates for autonomous-vehicle licensing and applications growth from US$12.3 billion in 2023 to US$64.7 billion in 2030. | 中 | SV028 |
| CV021 | Pony AI’s F-1 includes substantial net-loss disclosures, underscoring that AV-stack commercialization can remain loss-making even with large market forecasts. | 中 | SV028 |
| CV022 | Baidu reported Apollo Go delivered 3.4 million fully driverless operational rides in Q4 2025 with weekly rides peaking above 300,000. | 高 | SV025, SV023 |
| CV023 | Baidu reported Q1 2026 Baidu Core AI-powered Business revenue above RMB13.6 billion, up 49% year over year. | 高 | SV026, SV023 |
| CV024 | Mobileye reported Q1 2026 revenue of US$558 million, up 27% year over year, and raised the midpoint of 2026 revenue guidance. | 中 | SV032 |
| CV025 | Mobileye’s Q1 2026 release also announced a goodwill impairment, showing that public AV/robotics-adjacent valuations can reset even with real revenue. | 中 | SV032 |
| CV026 | WeRide announced that it filed a 2025 annual report on Form 20-F with the SEC and is listed on Nasdaq and HKEX. | 高 | SV030, SV031 |
| CV027 | Pony AI announced that it filed a 2025 annual report on Form 20-F and is listed on Nasdaq and HKEX. | 高 | SV027, SV029 |
| CV028 | Public AV-stack and humanoid comps provide audited or exchange-filed revenue and risk disclosures that Simplexity does not yet provide publicly. | 中 | SV020, SV024, SV028, SV029, SV031, SV032 |
| CV029 | Axis Intelligence reports that humanoid-robot sector VC funding exceeded US$9.8 billion by the end of 2025 and that 2026 Q1 funding rose sharply year over year. | 中 | SV033 |
| CV030 | Robozaps says its 2026 humanoid-robot industry report tracks 26 robots across seven countries and more than US$5 billion of industry investment including acquisitions. | 中 | SV034 |
| CV031 | The base-case valuation stance is expensive because the public evidence proves financing momentum but not revenue quality, customer depth, or margin path. | 中 | SV001, SV002, SV007, SV008, SV009, SV020 |
| CV032 | The bull case would require private proof of repeatable paid deployments, signed enterprise contracts, credible gross-margin trajectory, and defensible model or data advantage. | 中 | SV005, SV006, SV008, SV014, SV015 |
| CV033 | The bear case is a down-round or bridge-financing outcome if pilots do not convert, if humanoid-sector capital cools, or if public comps continue repricing. | 中 | SV009, SV010, SV011, SV012, SV025, SV032 |
| CV034 | The warranted public-evidence recommendation is research-more rather than buy because the reported unicorn price lacks public revenue, customer, and terms support. | 中 | SV001, SV002, SV007, SV008, SV009, SV020 |
| CV035 | Recommendation confidence is medium-low: the financing and market facts are well sourced, but the decisive private metrics are absent. | 中 | SV001, SV002, SV014, SV020, SV028, SV032 |
| CV036 | Risk rating should be high because the company combines early-stage commercialization, opaque economics, crowded-sector bubble warnings, and a headline valuation above US$1.0 billion. | 中 | SV002, SV007, SV009, SV012, SV013 |
| CV037 | The core thesis is that elite investors, rapid financing, and a large embodied-AI market could make Simplexity an important China robotics platform. | 中 | SV001, SV004, SV005, SV014, SV015 |
| CV038 | The core anti-thesis is that public evidence does not yet bridge from technology narrative to paid deployments, durable economics, or a supportable billion-dollar entry price. | 中 | SV007, SV008, SV009, SV011, SV020 |
| CV039 | Entry discipline should require either a materially lower price, milestone-based tranche, or strong investor protections until revenue and customer evidence are verified. | 中 | SV002, SV009, SV020, SV028, SV032 |
| CV040 | The most important diligence asks are audited or bank-supported revenue, named customer references, deployment conversion data, gross-margin or bill-of-materials evidence, cap-table terms, and safety or reliability metrics. | 中 | SV007, SV008, SV020, SV021, SV028 |
| CV041 | Exit readiness is not publicly underwriteable because Simplexity lacks the public reporting history, audited metrics, and scaled revenue disclosures visible in listed comps. | 中 | SV020, SV024, SV029, SV031, SV032 |
| CV042 | A public-evidence bull case can keep the company on an IC watchlist, but it cannot justify a buy decision without private proof that manufacturing and enterprise PoCs are converting into paid scale. | 中 | SV006, SV008, SV014, SV015, SV020 |
| CV043 | Public comps suggest that valuation should be tied to verified revenue, losses, and commercialization milestones rather than a generic robotics TAM multiple. | 中 | SV020, SV025, SV028, SV032, SV033 |
| CV044 | Market forecasts are necessary but insufficient support for the price because the same sources coexist with adverse warnings about overinvestment and homogeneous entrants. | 中 | SV014, SV015, SV016, SV009, SV012 |
| CV045 | Final diligence should focus on evidence that changes valuation stance, not cosmetic validation of an already popular financing narrative. | 中 | SV001, SV002, SV007, SV009, SV020, SV032 |
| 编号 | 出版方 | 标题 | 引文 |
|---|---|---|---|
| SO001 | Simplexity Robotics | Simplexity Robotics - 至简动力 | 至简动力Simplexity Robotics于2025年7月底成立,是一家从真实场景出发...打造具备高用户价值的具身智能产品。 |
| SO002 | 新浪财经 | 杭州至简动力半年完成5轮融资吸金20亿元,理想前CTO王凯与贾鹏联手创业 | 在不到半年时间里连续完成5轮融资,累计融资金额达20亿元。 |
| SO003 | 新浪财经 | 迅猛!理想前高管打造的企业半年融资20亿! | 理想前高管打造的企业半年融资20亿。 |
| SO004 | 腾讯新闻 | 至简动力宣布在半年内完成20亿元规模融资 核心创始团队来自理想汽车 | 至简动力CEO为贾鹏...董事长王凯...COO王佳佳。 |
| SO005 | Yicai Global | China's Simplexity Robotics Raises USD289.3 Million in Six Months to Speed Large-Scale Application of Embodied AI | Simplexity Robotics said the Chinese company has secured CNY2 billion (USD289.3 million) in less than half a year. |
| SO006 | 36Kr | 6个月5轮融资,累计20亿人民币,至简动力成具身智能赛道最快独角兽 | 2025年7月底注册成立的至简动力,迎来首次正式亮相并官宣融资。 |
| SO007 | 36Kr Europe | Zhijian Dynamics Becomes Fastest Unicorn in Embodied Intelligence Track with 5 Rounds of Financing Totaling 2 Billion RMB | Simplexity Robotics, founded in late July 2025, made its first official appearance and announced its financing. |
| SO008 | 36Kr | 前理想系把造车的那套技术观搬进机器人,半年融了20亿 | 至简动力现在做的,就是把VLA模型从车上拆下来,装进机器人身体里。 |
| SO009 | Gasgoo Auto News | Simplexity Robotics Secures RMB 2 Billion in Cumulative Funding | Less than six months elapsed between the first and fifth rounds. The total raised was 2 billion yuan. |
| SO010 | Gasgoo Auto News | Auto Executives Flock to Embodied Intelligence | In July 2025, Jia Peng...teamed up with ex-CTO Wang Kai to found Zhijian Power. |
| SO011 | Pandaily | Simplexity Robotics Raises RMB 2 Billion in Just Six Months, Becoming the Fastest Unicorn in Embodied AI | Simplexity Robotics Raises RMB 2 Billion in Just Six Months, Becoming the Fastest Unicorn in Embodied AI. |
| SO012 | Pandaily | Embodied AI Startup Founded by Former Li Auto Executives Nears Unicorn Valuation | Embodied AI Startup Founded by Former Li Auto Executives Nears Unicorn Valuation. |
| SO013 | Taibo | Zhijian Power has completed 5 consecutive rounds of financing, accumulating a total of 2 billion RMB | completed five consecutive rounds of financing within six months, with a total financing amount of 2 billion RMB. |
| SO014 | C114 Pro | 中科创星投资具身智能企业至简动力 | 中科创星投资具身智能企业至简动力。 |
| SO015 | Redplanx | Simplexity Robotics funding news | Simplexity Robotics funding coverage from Redplanx. |
| SO016 | Baidu Baike | 杭州至简动力科技有限公司 | 杭州至简动力科技有限公司是一家科技公司,专注于具身智能技术的研发。 |
| SO017 | Baidu Baike | Hangzhou Zhijian Power Technology Co., Ltd. | Hangzhou Zhijian Power Technology Co., Ltd. |
| SO018 | Baidu AiQicha | 杭州至简动力科技有限公司 - 爱企查 | 杭州至简动力科技有限公司成立于2025年07月31日,位于浙江省杭州市余杭区闲林街道嘉企路11号1号楼394室,目前处于开业状态。 |
| SO019 | Qichacha | 杭州至简动力科技有限公司 | 杭州至简动力科技有限公司。 |
| SO020 | 企查猫 | 杭州至简动力科技有限公司工商信息 | 杭州至简动力科技有限公司工商信息。 |
| SO021 | 国家企业信用信息公示系统 | 国家企业信用信息公示系统 | 国家企业信用信息公示系统。 |
| SO022 | 信用中国 | 信息公示 | Fetch returned HTTP 412; live company-specific credit search remains an evidence gap. |
| SO023 | 投资界 | 6个月连融五轮,至简动力累计融资20亿 | 6个月连融五轮,至简动力累计融资20亿。 |
| SO024 | 每日商报 | 至简动力半年完成5轮融资累计20亿元 | 至简动力半年完成5轮融资累计20亿元。 |
| SO025 | 同花顺财经 | 至简动力半年完成5轮融资累计20亿元 | 至简动力半年完成5轮融资累计20亿元。 |
| SO026 | 网易订阅 | 阿里腾讯一起投!理想前高管组团做机器人,半年融了20亿 | 阿里腾讯一起投!理想前高管组团做机器人,半年融了20亿。 |
| SO027 | 每日经济新闻 | 3个月超30笔融资!具身智能的估值狂飙与万亿赌局:没有共识,为信仰下注 | 行业还在起点,但估值已经在半山腰,甚至更高。 |
| SO028 | 网易订阅 | 2026年具身智能融资超370亿,出货量暴增508%!风口还是泡沫? | 技术瓶颈、伪需求疑云、普遍亏损的现实,也让‘泡沫’二字如影随形。 |
| SO029 | 搜狐 | 2026具身智能投资价值重估:头部企业融资量产提速 | 2026年,具身智能产业进入商业化落地与资本化提速的关键期。 |
| SO030 | ManualVLA project page | ManualVLA: A Unified VLA Model for Chain-of-Thought Manual Generation and Robotic Manipulation | Peng Jia³... 3Simplexity Robotics. |
| SO031 | arXiv | TwinRL: Digital Twin-Driven Reinforcement Learning for Real-World Robotic Manipulation | VLA models remain constrained by the high cost of expert demonstrations and limited real-world interaction. |
| SO032 | GitHub | TwinRL official repository | Release real-world RL training code (coming soon). |
| SO033 | TwinRL project page | TwinRL: Digital Twin–Driven Reinforcement Learning for Real-World Robotic Manipulation | TwinRL... designed to scale and guide exploration for VLA models. |
| SM001 | State Council Information Office / Xinhua | China releases national standard system for humanoid robotics and embodied AI | China released its first national standard system covering the entire industrial chain and lifecycle of humanoid robots and embodied artificial intelligence. |
| SM002 | Ministry of Industry and Information Technology | 工业和信息化部办公厅 国务院国资委办公厅关于联合开展2026年度人形机器人与具身智能实景实训专项行动的通知 | 到2026年底,人形机器人等重点产品在一批代表性场景中率先完成应用验证和常态部署,开启“作业模式”。 |
| SM003 | The State Council of the PRC | 工业和信息化部等十七部门关于印发“机器人+”应用行动实施方案的通知 | 文件下载:“机器人+”应用行动实施方案 |
| SM004 | Cyberspace Administration of China | 生成式人工智能服务管理暂行办法 | 生成式人工智能服务管理暂行办法 |
| SM005 | World Artificial Intelligence Conference | 世界人工智能大会 | 世界人工智能大会 |
| SM006 | International Federation of Robotics | World Robotics | The annually published World Robotics report provides comprehensive and up-to-date information on the global robotics market. |
| SM007 | Fortune Business Insights | Warehouse Robotics Market Size, Share Report | 2026-2034 | The global warehouse robotics market size was valued at USD 6.51 billion in 2025 and is projected to grow from USD 7.35 billion in 2026 to USD 25.41 billion by 2034. |
| SM008 | MarkNtel Advisors | China Warehouse Automation Market Size, Trends & Forecast | The China Warehouse Automation Market size was valued at around USD3.02 billion in 2025 and is projected to reach USD9.17 billion by 2032. |
| SM009 | Global Market Insights | Logistics Robots Market Size, Forecast Report 2026-2035 | The global logistics robots market was estimated at USD 17.8 billion in 2025 and is expected to grow from USD 20.7 billion in 2026 to USD 91.4 billion in 2035. |
| SM010 | JD Corporate Blog | JD Logistics Introduces “Zhilang” Intelligent Warehousing Solution at CeMAT Asia 2024 | Zhilang improves picking efficiency by over three times compared to traditional methods. |
| SM011 | JD.com Investor Relations | Annual Reports | JD.Com, Inc. | 2024 Annual Report 5.8 MB |
| SM012 | Geekplus | Geek+ | Robotics Solutions for Warehouse & Logistics Automation | Geekplus caters to eCommerce, 3PL, apparel, healthcare, groceries, auto manufacturing, and temperature-controlled storage. |
| SM013 | Geekplus | Global Warehouse Robotics Leader Geekplus Maintains the Largest AMR Market Share for the 7th Consecutive Year in a Growing Market | Order-fulfilment deployment sites are projected to increase from 5,500 in 2024 to 18,000 by 2030. |
| SM014 | UBTECH Robotics | Financial Reports | UBTECH Robotics | Financial Reports | UBTECH Robotics |
| SM015 | Gasgoo | Auto Executives Flock to Embodied Intelligence | The embodied intelligence sector remains in the early days; most companies have yet to turn a profit. |
| SM016 | 36Kr Global | Former Li Auto Execs Apply Car-Making Technology Philosophy to Robots, Raise 2 Billion Yuan in Half a Year | The founding team includes Wang Kai, Jia Peng, and Wang Jiajia, nearly half of Li Auto’s core intelligent driving team. |
| SM017 | Yicai Global | China's Simplexity Robotics Raises USD289.3 Million in Six Months to Speed Large-Scale Application of Embodied AI | Simplexity Robotics secured CNY2 billion in less than half a year and valuation exceeded USD1 billion. |
| SM018 | 36Kr Global | Zhijian Dynamics Becomes Fastest Unicorn in Embodied Intelligence Track with 5 Rounds of Financing Totaling 2 Billion RMB in Six Months | Zhijian Dynamics became a unicorn in the embodied intelligence track with five rounds totaling 2 billion RMB. |
| SM019 | Bureau of Industry and Security | Guidance Regarding Enforcement of License Requirements for Advanced Computing Items | A license is required to export advanced computing items to entities headquartered in Country Group D:5 or Macau. |
| SM020 | RoboticsTomorrow | Geekplus Hits Profitability Milestone with 31.6% YoY Revenue Growth, Fueled by Embodied Intelligence-Driven Tech Innovation | Geekplus reported 31.6% year-on-year revenue growth, reaching 3.171 billion RMB. |
| SM021 | Robotics & Automation News | Top 20 Chinese warehouse robotics companies driving logistics automation | The article lists 20 leading Chinese companies developing AMRs and advanced automation systems for warehouses and factories. |
| SM022 | Pandaily | Simplexity Robotics Raises RMB 2 Billion in Just Six Months, Becoming the Fastest Unicorn in Embodied AI | Simplexity Robotics Raises RMB 2 Billion in Just Six Months. |
| SM023 | EqualOcean | Five Rounds, RMB 2B: Simplexity Robotics Emerges as Fastest Unicorn in Embodied AI | Five Rounds, RMB 2B: Simplexity Robotics Emerges as Fastest Unicorn in Embodied AI. |
| SM024 | Internet Info Agency | Domestic Embodied AI Sector Secures Over 200 Funding Rounds in Q1 2026, Averaging $460M Daily | Domestic Embodied AI Sector Secures Over 200 Funding Rounds in Q1 2026. |
| SM025 | TechNode | China releases first national standard framework for humanoid robots and embodied AI | China releases first national standard framework for humanoid robots and embodied AI. |
| SM026 | Seconded European Standardization Expert in China | China’s First Standards System for Humanoid Robots and Embodied Intelligence | China’s First Standards System for Humanoid Robots and Embodied Intelligence. |
| SM027 | Ministry of Industry and Information Technology | 经济日报:开启人形机器人作业模式 | 经济日报:开启人形机器人作业模式 |
| SM028 | MIIT | 工业和信息化部关于印发《人形机器人创新发展指导意见》的通知 | 现将《人形机器人创新发展指导意见》印发给你们,请结合实际,认真贯彻落实。 |
| SP001 | Simplexity Robotics | Simplexity Robotics - 至简动力 | 至简动力Simplexity Robotics于2025年7月底成立,是一家从真实场景出发...打造具备高用户价值的具身智能产品。 |
| SP002 | AGIBOT | AGIBOT Innovation (Shanghai) Technology Co., Ltd. | AGIBOT A2 Full-Size Humanoid Robot; AGIBOT G1 Universal Embodied Intelligent Robot. |
| SP003 | Zhiyuan Robotics / AgiBot | 以智能机器创造无限生产力-智元创新(上海)科技股份有限公司 | 智元量产下线15000台; 智元第15000台通用具身机器人量产下线。 |
| SP004 | Zhiyuan Robotics / AgiBot | 关于我们-智元创新(上海)科技股份有限公司 | 我们是一家全球领先的以具身智能(Embodied Intelligence)为核心技术方向的通用AI机器人公司。 |
| SP005 | PR Newswire / AGIBOT | AGIBOT's 15,000th Robot Rolls Off the Production Line, Marking a New Milestone in Embodied AI Deployment | AGIBOT...announced that its 15,000th robot has officially rolled off the production line. |
| SP006 | PR Newswire / AGIBOT | AGIBOT Declares 2026 "Deployment Year One" at APC 2026 | Over the next five years, AGIBOT plans to invest more than RMB 2 billion to expand its ecosystem. |
| SP007 | Unitree Robotics | Unitree Robotics | Robot Dog_Quadruped_Humanoid Robotics Company | Robots; Robot - Consumer/Education; Robot - Industry; Human Humanoid Robot. |
| SP008 | Unitree Robotics | Humanoid robot G1_Humanoid Robot Functions_Humanoid Robot Price | Unitree G1 Humanoid agent AI avatar Price from $13.5K. |
| SP009 | Unitree Robotics | Universal humanoid robot H1_Bipedal Robot_Humanoid Intelligent Robot Company | Unitree H1 / H1-2 Unitree's first universal humanoid robot. |
| SP010 | Unitree Robotics | Robot Dog Go2_Quadruped_Robot Dog Company | Unitree Go2 Infinite Revolution Price from $1600. |
| SP011 | UnitreeRobotics Shop | Unitree G1 – UnitreeRobotics | Unitree G1 $13,500.00 USD; Unitree H1 Contact us for the real price. |
| SP012 | UBTECH Robotics | UBTECH: Humanoid Robot | AI Education Robot | Commercial Robot Solutions | Humanoid Service Robot Application Scenarios; AI Education Solution. |
| SP013 | UBTECH Robotics | Financial Reports | UBTECH Robotics | Investor Relations; Financial Reports; Humanoid; Walker Tienkung. |
| SP014 | Hong Kong Exchanges and Clearing | Horizon Robotics Global Offering Prospectus | Horizon Robotics Global Offering; maximum offer price HK$3.99 per offer share. |
| SP015 | Hong Kong Exchanges and Clearing | Horizon Robotics Annual Report 2024 | Revenue from contracts with customers 2,383,554 RMB thousands in 2024. |
| SP016 | Galbot | Galbot-银河通用机器人官方网站 | Galbot-银河通用机器人官方网站. |
| SP017 | PR Newswire / Galbot | Galbot Secures Over $300 Million in New Funding, Breaking Records with $3 Billion Valuation | The company's valuation has reached $3 billion; total funding to $800 million. |
| SP018 | TechNode | Humanoid robot maker Galbot raises RMB 2.5 billion | Galbot...completed a new funding round totaling RMB 2.5 billion (about $350 million). |
| SP019 | Robotics & Automation News | Galbot raises $151 million to scale embodied AI humanoid robots, partners with Bosch investment arm | Galbot...raised $151 million...and announced a strategic joint venture with Boyuan Capital, the investment arm of Bosch Group. |
| SP020 | Momenta | Momenta | The World's Leading Physical AI Company | Momenta | The World's Leading Physical AI Company. |
| SP021 | DeepRoute.ai | DeepRoute.Ai | DeepRoute.Ai. |
| SP022 | Horizon Robotics | Smart Driving Technology_Smart Driving Solutions_Smart Driving Vehicles | Shipments of Journey Series 10million+; Design Wins 400+; Vehicles in Mass Production 300+; Partner OEMs & Brands 40+. |
| SP023 | Baidu Apollo | 百度Apollo-自动驾驶、智能汽车解决方案 | 百度2013年开始布局自动驾驶,2017年推出全球首个自动驾驶开放平台Apollo。 |
| SP024 | Baidu Inc. | Financial Reports | Baidu Inc | We will provide a hard copy of our annual report containing our audited consolidated financial statements. |
| SP025 | Pony AI Inc. | Investor Relations | Pony AI Inc. | Pony.ai...is a global leader in achieving large-scale mass production and commercialization of autonomous driving technology. |
| SP026 | Securities and Exchange Commission | Pony AI Inc. Form F-1 Registration Statement | The autonomous driving industry is highly competitive...and we are still in a nascent stage of commercializing our technologies. |
| SP027 | WeRide Inc. | Investors | WeRide Inc. | Our autonomous vehicles have been tested or operated in over 40 cities across 12 countries. |
| SP028 | WeRide Inc. | To Transform Urban Living with Autonomous Driving | WeRide...offers a portfolio of five core products, including Robotaxi. |
| SP029 | Mobileye | Mobileye | Driver Assist and Autonomous Driving Technologies | Mobileye to establish vertically integrated robotaxi business. |
| SP030 | Mobileye | Mobileye | Investor Relations | Mobileye Releases First Quarter 2026 Results...and Announces a $250 Million Share Repurchase Program. |
| SP031 | Mobileye | Mobileye SuperVision™ | The Bridge from ADAS to Consumer AVs | Mobileye SuperVision; The Bridge from ADAS to Consumer AVs. |
| SP032 | Gasgoo Auto News | Auto Executives Flock to Embodied Intelligence | The new energy vehicle market is locked in fierce competition, with relentless price wars; Jia Peng...founded Zhijian Power. |
| SP033 | TrendForce | China’s Humanoid Robot Output to Surge 94% in 2026; Unitree and AgiBot to Capture Nearly 80% Market Share | China...annual output growth up to 94% in 2026; Unitree Robotics and AgiBot...nearly 80% of total shipments. |
| SP034 | DirectIndustry e-Magazine | A Deep Look Into China's Humanoid Robot Market | Many examples people see in the media are essentially demonstrations...market leader AgiBot only produced 5,100 robots in 2025 and Unitree made 4,200. |
| SP035 | AutoX | AutoX official website fetch result | The requested URL /index.html was not found on this server. |
| SI001 | Simplexity Robotics | Simplexity Robotics - 至简动力 official website | 至简动力Simplexity Robotics于2025年7月底成立,是一家从真实场景出发...打造具备高用户价值的具身智能产品。 |
| SI002 | Yicai Global | China's Simplexity Robotics Raises USD289.3 Million in Six Months to Speed Large-Scale Application of Embodied AI | Simplexity Robotics said the Chinese company has secured CNY2 billion (USD289.3 million) in less than half a year... |
| SI003 | 36Kr | 6 个月 5 轮融资,累计 20 亿人民币,至简动力成具身智能赛道最快 “独角兽” | 在半年不到的时间内连续完成 5 轮融资,总融资额达 20 亿人民币... |
| SI004 | Sina Finance | 杭州至简动力半年完成5轮融资吸金20亿元,理想前CTO王凯与贾鹏联手创业 | 公司总部位于杭州,目前已经在北京、上海、苏州三地布局研发与业务团队。 |
| SI005 | Pandaily | Simplexity Robotics Raises RMB 2 Billion in Just Six Months, Becoming the Fastest “Unicorn” in Embodied AI | Simplexity Robotics Raises RMB 2 Billion in Just Six Months, Becoming the Fastest “Unicorn” in Embodied AI. |
| SI006 | Preqin | Simplexity Robotics Asset Profile | The company generates revenue from the direct sale of embodied artificial intelligence robotic hardware, supporting software models, and autonomous driving system solutions to enterprise clients. |
| SI007 | ChinaVenture | 独家|半年融资20亿,最年轻具身独角兽诞生 | 至简动力在不到半年的时间内,连续完成5轮融资,总融资金额达20亿元人民币。 |
| SI008 | EqualOcean | Five Rounds, RMB 2B: Simplexity Robotics Emerges as Fastest Unicorn in Embodied AI | Five Rounds, RMB 2B: Simplexity Robotics Emerges as Fastest Unicorn in Embodied AI. |
| SI009 | Tencent News | 腾讯阿里联手加注!理想智驾班底创业,至简动力5轮融资20亿 | 腾讯阿里联手加注...至简动力5轮融资20亿。 |
| SI010 | Gasgoo Auto News | Simplexity Robotics secures RMB 2 billion in cumulative funding | Simplexity Robotics secures RMB 2 billion in cumulative funding. |
| SI011 | 36Kr | 半年融资20亿,最年轻具身独角兽诞生 | 2026年开年...披露融资的总额接近150亿元人民币...至简动力...总融资金额达20亿元人民币。 |
| SI012 | Sohu | 腾讯阿里联手加注!理想智驾班底创业,至简动力5轮融资20亿 | 理想智驾班底创业,至简动力5轮融资20亿。 |
| SI013 | The Business Times | China warns of bubble risk in booming humanoid robotics industry | China’s top economic-planning agency has warned over the risk of a bubble forming in humanoid robotics. |
| SI014 | France 24 | China says humanoid robot buzz carries bubble risk | Speed and bubble have always been issues that need grasping and balance in the development of frontier industries. |
| SI015 | TrendForce | China’s Humanoid Robot Output to Surge 94% in 2026; Unitree and AgiBot to Capture Nearly 80% Market Share | Combined gross margin across both segments reached 60%, which challenges the perception that robotics is a purely cash-burning industry. |
| SI016 | SEC | Company tickers JSON | SEC company tickers JSON identifies public-reporting issuers used as comparable filing sources. |
| SI017 | SEC | Pony AI Inc. submissions JSON | Pony AI Inc. filings include 2026 and 2025 annual reports and its 2024 F-1 registration statement. |
| SI018 | SEC | Pony AI Inc. Form F-1 registration statement | Our limited operating history makes it difficult to predict our future prospects, business and financial performance. |
| SI019 | SEC | Pony AI Inc. 2025 Form 20-F | Total revenues increased by 20.0% from US$75.0 million in 2024 to US$90.0 million in 2025. |
| SI020 | SEC | WeRide Inc. submissions JSON | WeRide Inc. recent filings include a 2026 Form 20-F annual report. |
| SI021 | SEC | WeRide Inc. 2025 Form 20-F | Total revenue 401,844 361,134 684,587 ... Gross profit ... 4,912 ... Net loss ... (1,246,739). |
| SI022 | SEC | Mobileye Global Inc. submissions JSON | Mobileye Global Inc. recent filings include a 2026 Form 10-K annual report. |
| SI023 | SEC | Mobileye Global Inc. 2025 Form 10-K | China, Germany, and South Korea made up 23%, 16%, and 10% of total revenue respectively. |
| SI024 | SEC | Baidu, Inc. submissions JSON | Baidu, Inc. recent filings include a 2026 Form 20-F annual report. |
| SI025 | SEC | Baidu, Inc. 2025 Form 20-F | Total revenue in 2025 were RMB129.1 billion (US$18.5 billion), decreasing by 3% from 2024. |
| SI026 | UBTECH Robotics | Financial Reports | Financial Reports: Annual Report 2025, Interim Report 2025, Annual Report 2024, Interim Report 2024. |
| SI027 | HKEXnews | UBTECH Robotics 2024 annual results / annual report PDF | Gross profit increased from RMB332.8 million for the year ended December 31, 2023 to RMB374.0 million for the year ended December 31, 2024. |
| SI028 | HKEXnews | Horizon Robotics 2024 annual report PDF | Total revenues 2,383,554 1,551,607 ... Customer A 31.50%. |
| SI029 | WeRide Investor Relations | Annual Reports | WeRide Inc. | 2025 Annual Report 3.5 MB; 2024 Annual Report 4.4 MB. |
| SI030 | Pony AI Investor Relations | Investor Relations | Pony AI Inc. | Investor Relations | Pony AI Inc. |
| SI031 | Baidu Investor Relations | Financial Reports | Baidu Inc | Financial Reports | Baidu Inc. |
| SE001 | Simplexity Robotics | Simplexity Robotics - 至简动力 | 至简动力Simplexity Robotics于2025年7月底成立,致力于通过高上限的大一统模型、高效的数据闭环、高可靠的机器人硬件打造具身智能产品。 |
| SE002 | 36Kr | 至简动力:三个理想人创业,做成最像造车新势力的机器人公司 | 文章称至简动力已开发两代面向B端和C端用户的本体、小批量生产并全面启动PoC验证。 |
| SE003 | Yicai Global | China's Simplexity Robotics Raises USD289.3 Million in Six Months to Speed Large-Scale Application of Embodied AI | Simplexity Robotics's first-generation robot has been produced in small batches, with a proof-of-concept test initiated. |
| SE004 | Gasgoo | Auto executives flock to embodied intelligence | Gasgoo reports Jia Peng and Wang Kai founded Zhijian Power and frames automotive talent as moving into embodied intelligence. |
| SE005 | arXiv | LaST0: Latent Spatio-Temporal Chain-of-Thought for Robotic Vision-Language-Action Model | LaST0 adopts a dual-system architecture implemented via a Mixture-of-Transformers design. |
| SE006 | arXiv | LaST0 PDF | The PDF lists Simplexity Robotics-PKU affiliation and reports real-world manipulation task results for LaST0. |
| SE007 | Google Sites project page | LaST0 project page | The project page says LaST0 uses latent spatio-temporal chain-of-thought for robotic VLA. |
| SE008 | LaST0 project page | LaST0 GitHub Pages project page | The project page lists Simplexity Robotics affiliation and describes demonstrations on tabletop, mobile, and dexterous manipulation. |
| SE009 | arXiv | ManualVLA: A Unified VLA Model for Chain-of-Thought Manual Generation and Robotic Manipulation | ManualVLA is described as a unified VLA framework built upon a Mixture-of-Transformers architecture. |
| SE010 | arXiv | ManualVLA PDF | The PDF lists Simplexity Robotics as an affiliation and reports a 32% higher average success rate than a hierarchical baseline. |
| SE011 | Google Sites project page | ManualVLA project page | The project page summarizes ManualVLA as generating manuals and executing robotic manipulation. |
| SE012 | arXiv | TwinRL: Digital Twin-Driven Reinforcement Learning for Real-World Robotic Manipulation | TwinRL is a digital twin-real-world collaborative post-training framework for VLA models. |
| SE013 | arXiv | TwinRL PDF | The PDF lists Simplexity Robotics affiliation and reports near-100% success with about 20 minutes of on-robot interaction across four tasks. |
| SE014 | GitHub | zhourui9813/TwinRL repository | The repository describes Twin-RL as a digital twin-real-world collaborative RL framework for VLA models. |
| SE015 | GitHub | zhourui9813/TwinRL README | The README presents repository structure, environment setup, and offline training sections for Twin-RL. |
| SE016 | Google Sites project page | TwinRL project page | The project page describes TwinRL as digital twin-driven reinforcement learning for real-world robotic manipulation. |
| SE017 | GitHub | Twin Assets and Dataset Guide | The guide says the repository provides digital twin assets and demonstration datasets used in the TwinRL project. |
| SE018 | Raw GitHubusercontent | Digital Twin Assets and Twin Dataset raw guide | The raw guide describes high-fidelity digital twin assets and twin-generated trajectories for the TwinRL project. |
| SE019 | GitHub | Awesome Reliable Robotics | The curation lists TwinRL-VLA among 2026 robotics reliability/RL works and summarizes its digital-twin mechanism. |
| SE020 | GitHub | Awesome RL-VLA | The repository frames online RL-VLA as interactive policy learning through continuous environment interaction. |
| SE021 | GitHub | Awesome VLA | The repository describes itself as a curated list of papers on Vision-Language-Action Models for Embodied AI. |
| SE022 | alphaXiv | TwinRL alphaXiv page | The alphaXiv page mirrors the TwinRL abstract and paper text for practitioner review. |
| SE023 | alphaXiv | ManualVLA alphaXiv page | The alphaXiv page mirrors ManualVLA paper details and abstract for practitioner review. |
| SE024 | alphaXiv | LaST0 alphaXiv page | The alphaXiv page mirrors LaST0 paper details and abstract for practitioner review. |
| SE025 | CSDN | 2026年3月国内具身智能机器人企业融资情况汇总 | 汇总称至简动力以“四个O”为核心技术体系,提出Human data is all you need,并已完成两代本体研发和PoC验证。 |
| SE026 | Tencent News / 汽势Auto-First | 风口换了,车圈留不住人了 | 文章列出王凯、贾鹏、王佳佳转向至简动力,并称自动驾驶积累的工程红利正在转化为机器人产业优势。 |
| SU001 | Simplexity Robotics | Simplexity Robotics - 至简动力 | 至简动力Simplexity Robotics于2025年7月底成立,是一家从真实场景出发...打造具备高用户价值的具身智能产品。 |
| SU002 | 36Kr | 前理想系把造车的那套“技术观”搬进机器人,半年融了20亿 | 2026年3月,一家成立刚满八个月的公司,用五轮融资、20亿人民币、十亿美金估值... |
| SU003 | Yicai Global | China's Simplexity Robotics Raises USD289.3 Million in Six Months to Speed Large-Scale Application of Embodied AI | Simplexity Robotics said the Chinese company has secured CNY2 billion... accelerating the large-scale application... across multiple scenarios. |
| SU004 | Sohu | 6 个月 20 亿元!至简动力成具身智能赛道最快独角兽 | 先后完成两代面向 B 端及 C 端的本体研发、实现本体小批量下线并全面开启 PoC 验证。 |
| SU005 | Toutiao / 引力播新闻 | 具身智能新锐牵手零部件“冠军”,至简动力全球创新中心落户苏州 | 至简动力科技有限公司全球创新中心签约暨至简动力与绿的谐波战略合作签约仪式在苏州市吴中区机器人产业园举行。 |
| SU006 | Leaderdrive | leaderdrive - 绿的谐波 | 苏州绿的谐波传动科技股份有限公司从事精密传动装置研发、设计和生产。 |
| SU007 | Xinhua Jiangsu | 苏州吴中新型工业化“雁阵”∣绿的谐波:突破国外技术封锁,争做谐波减速器引领者 | 成功打破国外品牌垄断。 |
| SU008 | Sina Finance | 杭州,横空出世一家具身智能独角兽! | 2026年3月,已与多家制造业、商超、物流企业达成合作,聚焦工厂车间、商超理货、物流分拣等封闭场景率先落地。 |
| SU009 | Baidu Baijiahao / VCA创投社 | 半年狂飙20亿,至简动力如何撬动具身智能新引擎? | 小批量下线并启动PoC验证。下一步,工厂车间、商超货架、物流仓储将成为它的“练兵场”。 |
| SU010 | Baidu Baijiahao / 猎云网 | 6个月20亿元!至简动力成具身智能赛道最快独角兽 | 实现本体小批量下线并全面开启PoC验证。 |
| SU011 | Baidu Baijiahao / 财通社 | 阿里腾讯一起投!理想前高管组团做机器人,半年融了20亿 | 该人士表示,至简动力计划2026年实现量产,主要面向B端场景。 |
| SU012 | 广州日报新花城 / 机器人参考 | 2026 Q1中国具身智能融资盘点:资本疯狂押注“百亿俱乐部” | 国内具身智能赛道披露融资事件超50起,获投企业超30家,累计融资额约200亿元。 |
| SU013 | Geek+ | Geek+ | Robotics Solutions for Warehouse & Logistics Automation | Geekplus caters to a wide range of industries, including eCommerce, 3PL, apparel, healthcare, groceries, auto manufacturing... |
| SU014 | JD Corporate Blog | JD Logistics Introduces “Zhilang” Intelligent Warehousing Solution at CeMAT Asia 2024 | This advanced system is already supporting operations during the Singles Day Grand Promotion. |
| SU015 | Geek+ | Case Studies | TÜV-certified safety across robots, workstations, and maintenance zones. |
| SU016 | Locus Robotics | Boots UK: Scaling Efficiency with LocusBots | Watch Video | Without the robots, there’s no way we’d have gotten through the volumes. |
| SU017 | International Federation of Robotics | Record of 4 Million Robots in Factories Worldwide | China is by far the world´s largest market. The 276,288 industrial robots installed in 2023 represent 51% of the global installations. |
| SU018 | International Federation of Robotics | Jane Heffner is New President of International Federation of Robotics | The global robotics industry is at an important inflection point, driven by rapid advancements in artificial intelligence and automation. |
| SU019 | UBTECH Robotics | Financial Reports | UBTECH Robotics | Financial Reports |
| SU020 | UBTECH Robotics | Annual Report 2025 | In 2025, full-size embodied intelligent humanoid robot products and services... achieved sales volume of 1,079 units. |
| SU021 | UBTECH Robotics | Annual Report 2024 | cooperated with many automobile enterprises such as Dongfeng Liuzhou Motor, Geely Automobile, FAW-Volkswagen Qingdao Branch, Audi FAW, BYD... |
| SU022 | Occupational Safety and Health Administration | Robotics - Overview | Studies indicate that many robot accidents occur during non-routine operating conditions, such as programming, maintenance, testing, setup, or adjustment. |
| SU023 | International Organization for Standardization | ISO 10218-1:2025 | ISO 10218-1 provides the safety requirements for the robot as a machine itself. |
| SU024 | France 24 | China says humanoid robot buzz carries bubble risk | the sector is not yet mature in terms of technology, commercialisation or use |
| SU025 | The Business Times | China warns of bubble risk in booming humanoid robotics industry | More than 150 makers of humanoid robots are operating in China and their number is still rising. |
| SR001 | Simplexity Robotics | Simplexity Robotics - 至简动力 company site | 至简动力Simplexity Robotics于2025年7月底成立 |
| SR002 | 36Kr | 至简动力融资与技术路线 coverage | 最终要看能不能在真实的工厂里稳定干活,能不能让客户买单 |
| SR003 | Yicai Global | China’s Simplexity Robotics Raises USD289.3 Million in Six Months | |
| SR004 | Pandaily | Simplexity Robotics Raises RMB 2 Billion in Just Six Months | |
| SR005 | Gasgoo | Simplexity Robotics secures RMB 2 billion in cumulative funding | |
| SR006 | Sina Finance | 具身智能赛道再迎快马:至简动力融资报道 | |
| SR007 | Tencent News | 至简动力融资与创始团队 coverage | |
| SR008 | Baidu Baike | Hangzhou Zhijian Power Technology Co., Ltd. profile | |
| SR009 | Aiqicha | 杭州至简动力科技有限公司 company profile | |
| SR010 | National Enterprise Credit Information Publicity System | National enterprise credit information query portal | |
| SR011 | Cyberspace Administration of China | Interim Measures for the Management of Generative AI Services | 提供者应当依法开展预训练、优化训练等训练数据处理活动 |
| SR012 | Cyberspace Administration of China | Cybersecurity Review Measures | |
| SR013 | Cyberspace Administration of China | Provisions on Deep Synthesis Internet Information Services | |
| SR014 | China Law Translate | Personal Information Protection Law translation and legal reference | |
| SR015 | China Law Translate | Data Security Law translation and legal reference | |
| SR016 | China Law Translate | Cybersecurity Law translation and legal reference | |
| SR017 | China Law Translate | Algorithmic recommendation provisions legal translation | |
| SR018 | China Law Translate | Product Quality Law legal translation | |
| SR019 | China Law Translate | Civil Code tort liability legal reference | |
| SR020 | China Law Translate | Export Control Law legal translation | |
| SR021 | Bureau of Industry and Security | Connected Vehicles supply-chain rule page | |
| SR022 | Bureau of Industry and Security | Commerce finalizes rule to secure connected vehicle supply chains | |
| SR023 | Bureau of Industry and Security | Entity List / EAR part 744 reference | |
| SR024 | Federal Register | Implementation of Additional Export Controls: Advanced Computing Items | |
| SR025 | SEC | Pony AI Inc. Form F-1 registration statement | |
| SR026 | HKEX | UBTECH Robotics annual report 2024 | |
| SR027 | NHTSA | Automated Vehicles for Safety | |
| SR028 | OSHA | Robotics - Overview | |
| SR029 | International Federation of Robotics | Record of 4 Million Robots in Factories Worldwide | |
| SR030 | International Federation of Robotics | Robots: China breaks historic records in automation | |
| SR031 | The Business Times | China warns of bubble risk in booming humanoid robotics industry | China warns of bubble risk in booming humanoid robotics industry |
| SR032 | France 24 | China says humanoid robot buzz carries bubble risk | |
| SR033 | The Robot Report | Humanoid robots: Why are they so difficult? | |
| SR034 | TrendForce | TrendForce humanoid robot market outlook | |
| SR035 | JD Corporate Blog | JD Logistics introduces Zhilang intelligent warehousing solution | |
| SR036 | Locus Robotics | Boots case study | |
| SR037 | DLA Piper | US expands export controls on advanced computing and semiconductor manufacturing items | |
| SV001 | Gasgoo Auto News | Simplexity Robotics Secures RMB 2 Billion in Cumulative Funding | The startup closed five rounds in six months, raising a total of 2 billion yuan. |
| SV002 | Yicai Global | China's Simplexity Robotics Raises USD289.3 Million in Six Months to Speed Large-Scale Application of Embodied AI | The firm's valuation exceeded USD1 billion after the latest funding round. |
| SV003 | 36Kr | 6 个月 5 轮融资,累计 20 亿人民币,至简动力成具身智能赛道最快 “独角兽” | |
| SV004 | The AI Insider | China's Simplexity Robotics Raises $289M USD to Develop and Deploy Embodied AI Tech | The startup said its valuation has surpassed $1 billion. |
| SV005 | EqualOcean | Five Rounds, RMB 2B: Simplexity Robotics Emerges as Fastest Unicorn in Embodied AI | |
| SV006 | Tencent News | 6个月20亿元!至简动力成具身智能赛道最快独角兽 | |
| SV007 | Baidu Baike | Hangzhou Zhijian Power Technology Co., Ltd. | The funds raised from this round will be used for training foundational models, platform research and development and iteration, data collection, and core algorithm development. |
| SV008 | Preqin | Simplexity Robotics Asset Profile | The company works with enterprise clients in the manufacturing and smart vehicle sectors. |
| SV009 | The Business Times | China warns of bubble risk in booming humanoid robotics industry | China's economic-planning agency has warned over the risk of a bubble forming in humanoid robotics. |
| SV010 | Firstpost | Are humanoid robots spooking the Chinese government? | The industry's price-to-earnings ratio was reported at roughly 58 times forward earnings. |
| SV011 | Interesting Engineering | Beijing flags humanoid robotics bubble risk as hype intensifies | Investment has poured into the sector despite limited proven use cases in factories or homes. |
| SV012 | Unite.AI | China Warns of Bubble Risk as 150 Companies Flood Humanoid Robot Market | More than 150 companies are flooding the market with nearly identical products. |
| SV013 | RobotToday | Bloomberg Warns of Bubble Risk as China’s Humanoid Robotics Boom Overheats | The humanoid-robot boom is real—so is the bubble risk. |
| SV014 | CNBC | Morgan Stanley doubles China humanoid robot shipment forecast as commercialization accelerates | Morgan Stanley has sharply raised its outlook for China’s humanoid robotics market. |
| SV015 | TrendForce | China’s Humanoid Robot Output to Surge 94% in 2026; Unitree and AgiBot to Capture Nearly 80% Market Share | China’s Humanoid Robot Output to Surge 94% in 2026. |
| SV016 | Grand View Research | Humanoid Robot Market Size & Share | Industry Report, 2030 | |
| SV017 | International Federation of Robotics | World Robotics | The publication covers various aspects of robotics, such as industrial robots including cobots, service robots as well as mobile robots. |
| SV018 | RoboticsTomorrow | Humanoid.guide Publishes Landmark 2025/2026 Humanoid Robot Market Report | |
| SV019 | UBTECH Robotics | Financial Reports | UBTECH Robotics | |
| SV020 | Hong Kong Exchanges and Clearing | UBTECH Robotics 2024 Annual Results Announcement | Our revenue increased from RMB1,055.7 million for the year ended December 31, 2023 to RMB1,305.4 million for the year ended December 31, 2024. |
| SV021 | Hong Kong Exchanges and Clearing | UBTECH Robotics Prospectus | Please refer to the section headed “Risk Factors”. |
| SV022 | Hong Kong Exchanges and Clearing | UBTECH Robotics 2024 Annual Report | |
| SV023 | Baidu Inc. | Financial Reports | Baidu Inc | We will provide a hard copy of our annual report containing our audited consolidated financial statements, free of charge. |
| SV024 | PR Newswire | Baidu, Inc. Files Its Annual Report on Form 20-F | Baidu announced it filed its annual report on Form 20-F for the fiscal year ended December 31, 2025 with the SEC on March 17, 2026. |
| SV025 | PR Newswire | Baidu Announces Fourth Quarter and Fiscal Year 2025 Results | Apollo Go delivered 3.4 million fully driverless operational rides with weekly rides peaking at over 300,000 during the quarter. |
| SV026 | Nasdaq | Baidu Announces First Quarter 2026 Results | Revenue from Baidu Core AI-powered Business exceeded RMB 13.6 billion, up 49% year over year. |
| SV027 | Pony AI Inc. | Investor Relations | Pony AI Inc. | |
| SV028 | U.S. Securities and Exchange Commission | Pony AI Inc. Registration Statement on Form F-1 | The global licensing and applications market was valued at US$12.3 billion, by revenue and is projected to reach US$25.2 billion, US$64.7 billion, and US$88.0 billion by 2025, 2030, and 2035, respectively, according to Frost & Sullivan. |
| SV029 | Nasdaq | PONY AI Inc. Files Its Annual Report on Form 20-F and Publishes Inaugural ESG Report | Pony AI announced that it filed its annual report on Form 20-F for the fiscal year ended December 31, 2025. |
| SV030 | WeRide Inc. | Annual Reports | WeRide Inc. | |
| SV031 | GlobeNewswire / FinancialContent | WeRide Inc. Files Its 2025 Annual Report on Form 20-F | WeRide announced that it filed its annual report on Form 20-F for the fiscal year ended December 31, 2025 with the SEC on April 23, 2026. |
| SV032 | Mobileye Global Inc. | Mobileye Releases First Quarter 2026 Results, Updates Full-Year Outlook, and Announces a Goodwill Impairment | Revenue of $558 million in the first quarter increased 27% year over year compared to Q1 2025. |
| SV033 | Axis Intelligence | Humanoid Robot Statistics 2026: Market Size, Deployments, and Who's Actually Winning | The industry shipped an estimated 16,000–18,000 units globally in 2025. |
| SV034 | Robozaps | Humanoid Robot Industry Report 2026 | Updated monthly, this report tracks 26 humanoid robots across 7 countries. |