Sunday
消费级物理 AI:拥有差异化的真实家庭数据闭环,但单位经济学不透明
Sunday 把差异化的真实家庭数据闭环、顶尖机器人团队履历和一线机构融资结合起来,但公开记录仍缺少评估 $1.15 billion 估值所需的价格、客户验证和单位经济性。
覆盖范围与披露说明
本次刷新报告基于截至 2026-10-14 的公开披露和第三方报道;定价、付费客户转化、现金消耗和硬件毛利率仍未披露。
封面要素
公司概况
Sunday 是一家私营消费机器人公司,正在研发 Memo——一款带轮家庭机器人,利用专有的 Skill Capture Glove 数据闭环和模仿学习模型, 自动完成叠衣服、收拾餐桌、装洗碗机等家务。公司位于 Mountain View,累计融资 $200 million,并在 2026 年 3 月达到 $1.15 billion 投后估值,但仍处于尚未产生收入的阶段,也未披露客户定价、付费部署或单位经济学。
- 成立时间
- 2024-04-01
- 创始人
- Tony Zhao, Cheng Chi
- 创立地点
- Mountain View, California, USA
- 总部
- Mountain View, California, USA
- 产品
- Memo 是一款带轮家用机器人,配有伸缩脊柱和柔顺双夹爪,专为家庭家务设计;其背后是 ACT-1、ACT-2 模仿学习技术栈和 Skill Capture Glove 数据采集系统。
- 客户
- 富裕家庭和愿意容忍 Beta 阶段机器人缺陷的早期采用者
- 商业模式
- 消费硬件销售,未来可能获得持续性软件 / 服务收入;目前仍未产生收入
- 阶段
- Series B / late-stage private
- 融资情况
- Seed 轮和 Series B 轮累计融资 $200 million,其中包括 2026 年 3 月以 $1.15 billion 投后估值完成的 $165 million Series B
执行摘要
主要优势
- 顶尖机器人创始团队,兼具扎实的学术与产业履历
- 真实家庭数据采集闭环,降低对昂贵远程操作设备的依赖
- 一线机构融资,以及清晰差异化的消费级家用机器人产品逻辑
主要风险
- 未披露客户定价、付费部署或已验证的收入转化
- 原型硬件的单位经济性仍远高于目标零售价
- 达到商业规模后,家庭安全、隐私和可靠性风险仍未解决
未决问题
- 备忘录中的价格、融资计划、(如有)订金,以及从测试版转为付费的预期。
- 计划部署规模下的硬件 BOM、毛利率、支持成本、烧钱速度和现金跑道。
- 测试家庭的使用数据、人工介入或远程操作率、现场可靠性和更换指标。
- Series B 及此前轮次的条款堆叠,包括清算优先权、治理权和反稀释保护。
- 机器人专属的数据留存、家庭内感知政策、安全测试和认证路线图。
目录
01公司概览
1.1 公司身份与创立使命
Sunday 是一家人工智能和消费机器人初创公司,由机器人研究员 Tony Zhao 和 Cheng Chi 于 2024 年 4 月在加州 Mountain View 创立。 公司最初在没有大学或企业实验室背景的情况下,于公寓里制作原型,并将使命定为用低成本物理 AI 解决家庭家务。2025 年 11 月 19 日, 公司结束隐身状态,公开推出 Memo——一款面向日常家务的自主家庭机器人,可收拾餐桌、装洗碗机、叠衣服和制作浓缩咖啡。 与追求双足行走、面向工业场景的当代人形机器人初创公司不同,Sunday 有意把 Memo 设计成带轮底盘的移动躯干,并配备可调高度、 最大升至 2.1 米的垂直升降机构。这种架构差异消除了跌倒风险,断电时仍能保持被动机械稳定,也让 Memo 能在机载端运行 Sunday 专有的 ACT-1 基础模型。公司不到两年就达到 $1.15 billion 估值,但仍是一家没有公开商业收入或经审计单位经济学数据的未产生收入企业。[CO001, CO002, CO009, CO010, CO015]
| 指标 | 数值 / 状态 | 截至日期 | 置信度 | 尽调缺口 / 备注 |
|---|---|---|---|---|
| 估值(投后) | $1,150M | 2026-03-12 | 高 | 二级交易或清算优先权条款尚未确认 |
| 累计融资额 | $200M | 2026-03-12 | 高 | 包含 $35M 种子轮和 $165M B 轮 |
| B 轮领投投资人 | Coatue Management | 2026-03-12 | 高 | 已确认 Thomas Laffont 担任董事 |
| 收入运行率 | 尚无收入 | 2026-10-14 | 中 | 商业销售尚未开始 |
| 年度经常性收入(ARR) | null | 2026-10-14 | 中 | 未披露经常性软件订阅 |
| 单位制造成本(原型) | ~$20,000 | 2026-03-24 | 中 | 目标是在规模化生产时实现低于 $10,000 的消费者零售价 |
| 核心团队人数 | 30-50 名全职员工 | 2026-03-24 | 中 | 不包括分散的合同 Memory Developers |
| 计划中的 beta 发布 | 2026 年下半年 | 2026-03-12 | 高 | 目标覆盖 50 个创始家庭 |
所有 KPI 指标均汇总自 B 轮新闻稿和技术报道;对于尚无收入的业务,财务运行率仍未披露。
[CO001, CO002, CO005, CO007, CO012, CO013, CO014, CO015]1.2 管理层、学术背景与治理
Sunday 的技术可信度和快速融资,直接来自创始团队顶尖的研究背景。联合创始人兼 CEO Tony Zhao 先在 UC Berkeley 获得电气工程与计算机科学学位, 随后进入 Stanford University CS 博士项目,师从机器人学习领域的 Sergey Levine 和 Chelsea Finn。Zhao 作为开源低成本 ALOHA 远程操作系统的 核心架构师,在机器人圈广受认可;之后曾在 Google DeepMind 和 Tesla Autopilot 任研究职务,最终从 Stanford 退学创办 Sunday。 联合创始人兼 CTO Cheng Chi 在 Shuran Song 指导下于 Columbia University 完成计算机科学博士论文,并发明 Diffusion Policy——现代操作机器人中 应用最广泛的模仿学习方法之一,曾多次获得 Best Paper。公司治理也纳入了机构风投代表:Sunday 于 2026 年 3 月完成 Series B 融资后, Coatue Management 联合创始人 Thomas Laffont 正式加入董事会,负责后期规模化监督。[CO001, CO003, CO004, CO006]
| 人员 | 职务 | 技术背景 | 创始人与市场匹配度及职能覆盖 | 关键人物依赖 |
|---|---|---|---|---|
| Tony Zhao | 联合创始人兼 CEO | UC Berkeley EECS、Stanford CS 博士(辍学);ALOHA 负责人、Google DeepMind、Tesla Autopilot | 机器人模仿学习先驱;负责愿景、数据采集策略和外部融资 | 高;低成本数据飞轮的核心架构师及公司对外代表 |
| Cheng Chi | 联合创始人兼 CTO | Columbia CS 博士,师从 Shuran Song;密歇根大学 BS;Diffusion Policy 发明者 | 世界知名生成式策略专家;负责 ACT 基础模型和算法栈 | 高;通用化算法的核心技术发明者 |
| Thomas Laffont | 董事(Coatue) | Coatue Management 联合创始人;资深成长科技和深科技投资人 | 成长资本策略、后期治理和科技生态连接 | 中;负责资本联合体治理和规模化监督 |
| Eric Vishria | 种子轮领投 / 合伙人(Benchmark) | Benchmark 普通合伙人;早期消费和企业软件投资人 | 高速产品规模化和消费品类发布指导 | 低;投资人治理和早期种子轮顾问 |
管理层和正式董事会代表均根据公司公告及 B 轮披露确定。
[CO003, CO004, CO006, CO007]1.3 融资历程与股东格局
Sunday 在成立后的二十三个月内,通过两轮机构融资累计筹集约 $200 million 股权资金。公司先完成约 $35 million 的 Seed 轮,由 Benchmark 和 Conviction 领投,用于硬件原型、手套制造和隐身期数据运营。2026 年 3 月 12 日,Sunday 宣布完成超额认购的 $165 million Series B, 投后估值 $1.15 billion,以消费机器人行业极快的速度跻身独角兽。Series B 由 Coatue Management 领投,Bain Capital Ventures、 Fidelity Management & Research Company、Tiger Global、Benchmark、Conviction 和 Xtal Ventures 等跨界及成长型机构参与。 该银团把 Benchmark 的一线消费风投支持,与 Coatue 和 Fidelity 的深科技成长资本结合起来。不过,完整股东权利、二级交易和清算优先权 仍属私密信息;Sunday 也未披露商业订阅收入或付费客户定金来支撑这一估值。[CO005, CO006, CO007, CO008, CO015]
| 利益相关方 | 角色与轮次 | 经济与战略重要性 | 已知治理权利 | 尽调要求 |
|---|---|---|---|---|
| Coatue Management | B 轮领投($165M) | 主要成长资本合作方;$1.15B 估值的锚定方 | 董事席位(Thomas Laffont) | 确认保护性条款和清算优先权层级 |
| Benchmark | 种子轮领投($35M)及 B 轮参与方 | 对早期消费规模化的信心和品牌背书 | 主要投资人权利;早期董事会观察员 | 审阅种子轮股东协议和信息权 |
| Conviction | 种子轮联合领投及 B 轮参与方 | Sarah Guo 提供的 AI 专业风险投资支持 | 按比例参与权 | 查阅附函和技术顾问角色 |
| Bain Capital Ventures | B 轮参与方 | 多阶段机构验证和企业网络 | B 轮联合投资体成员 | 确认共同投资条款和持股比例 |
| Fidelity Management & Research | B 轮参与方 | 在最终走向公开市场前提供跨阶段资产管理验证 | B 轮联合投资体成员 | 核实二级转让限制和注册权 |
| Tiger Global | B 轮参与方 | 高速成长资本合作方 | B 轮联合投资体成员 | 评估后续资本承诺和按比例参与权 |
所列机构均参与了 Sunday 于 2026 年 3 月 12 日宣布的 B 轮融资,或此前的种子轮融资。
[CO005, CO007, CO008]1.4 运营规模、里程碑与风险画像
Sunday 的核心竞争资产不是传统远程操作机队,而是一套运营级数据采集装置。为在没有模拟或合成数据的情况下训练 ACT 基础模型,Sunday 向美国各地 数百名远程 Memory Developers 部署了超过 2,000 只取得专利的 Skill Capture Gloves,每小时支付 $30 至 $60 记录真实家庭日常。 这条运营流水线已在 500 个家庭中积累超过 10 million 个真实世界交互片段。全职员工从最初住在公寓里的两名创始人,扩展到约 30 至 50 名 专业机器人和机器学习工程师;承包商招聘名单显示,分布式运营规模更大。Sunday 计划在 2026 年底将 Founding Family Beta 部署到 50 个家庭, 并计划在 2026 年感恩节前后开始全面商业发货。然而,外部批评者指出了多重障碍:原型制造成本目前约 $20,000,却对应低于 $10,000 的零售价目标; 真实家庭 Beta 部署还带来尚未解决的安全、财产责任和分布外模型泛化风险。[CO002, CO011, CO012, CO013, CO014]
Sunday 从公寓创立一路发展到估值 $1.15 billion 的独角兽,并推进 beta 部署。
里程碑日期来自公司公开发布和创始人访谈。
[CO001, CO002, CO005, CO009, CO010, CO012]1.5 展示项
02市场分析
2.1 市场边界、工作范围与现有替代方案
Sunday 将旗舰产品 Memo 明确定位在尚处早期的家庭辅助机器人领域,并刻意与工业自动化、仓储物流区分开来。它不去争夺工厂物料搬运合同, 因为那里的流程结构化、且有专门的安全区域;Sunday 瞄准的是无序的单户住宅环境。在这一家庭边界内,Sunday 聚焦摩擦最重的日常家务: 收拾餐桌、刮除食物残渣、装洗碗机、折叠衣物和整理放错位置的个人物品。Memo 优先处理脆弱的家庭物件,而不是追求工业级举重能力, 因而用柔顺性和家庭安全绕开传统工业机器人指标。Sunday 的现状替代方案不是其他自主移动机器人,而是传统洗碗机、洗衣机等单一用途白色家电, 加上外包的专业清洁劳动和无偿的家庭劳动。要从家庭预算中分得一杯羹,Memo 必须与这些旧有家电协同,而不是要求改造住宅结构。[CM001, CM006, CM007]
| 细分市场 / 类别 | 纳入支出 | 排除支出 | 买方 / 付款方 | 尽调相关性 |
|---|---|---|---|---|
| 自主家庭辅助 | 轮式多任务机器人硬件、技能更新和持续车队维护 | 工业地面清洁机和仓储 AGV | 富裕郊区房主和早期科技采用者 | Sunday 的主要收入引擎;界定可服务的硬件与服务支出。 |
| 专用家电自动化 | 单一用途清洁家电(洗碗机、洗衣机、扫地机器人) | 通用操作软件或人工任务辅助 | 普通住宅消费者家庭 | 奠定价格锚点;Memo 必须与现有家电整合,而不是取代它们。 |
| 外包家务劳动 | 付费住宅清洁、家政服务和专业洗衣护理 | 资本设备采购和硬件维护 | 双收入家庭和高净值家庭 | 关键预算替代项;为每日节省时间提供经济基准。 |
| 工业 / 物流机器人 | 重载物料搬运、工厂自动化和仓库分拣 | 无结构家庭环境和住宅生活空间 | 企业供应链和制造业采购部门 | 排除的相邻市场;凸显 Sunday 有意聚焦家庭环境,而非工厂地面。 |
工作量范围源自 Sunday 的产品披露和分析师画像;不包括工业工厂机器人。
[CM001, CM006, CM007]2.2 可服务市场规模、成本现实与经济约束
评估通用家庭机器人的市场,必须把关于数万亿美元通用人工智能的投机叙事,过滤进物理硬件成本结构。2026 年 3 月,风投基于 Sunday 解锁消费级物理 AI 的潜力,将其投后估值定为 $1.15 billion;但眼下可服务市场受到原型经济学的强力约束。Memo 手工装配的单台制造成本 约 $20,000,原因包括定制双夹爪工具、伸缩脊柱执行器、传感器阵列和小批量零部件采购。管理层明确表示,大规模制造后目标零售价低于 $10,000。 按 $10,000 的资本性价格分摊 60 个月计算,相当于每月约 $167,直接与高端奢侈家电或第二辆车的融资支出竞争。若没有普及的消费金融基础设施 或补贴型硬件订阅合同,早期可服务可获得市场仍局限于富裕郊区家庭和早期科技采用者;这些人能承受高额前期硬件支出,却没有确定的二手流动性。[CM002, CM004, CM005, CM010, CM011, CM012, CM014]
| 发布方 / 视角 | 年份 | 地理范围 | 市场价值 | CAGR / 增长 | 方法论 | 置信度 | 局限 |
|---|---|---|---|---|---|---|---|
| 风险投资估值视角 | 2026 | 全球 / 美国 | $1.15B 投后估值 | 估值提升超过 5 倍 | Coatue 领投的 B 轮投资定价,基于平台潜力 | 高 | 反映风险投资预期,而非当前经营收入。 |
| 规模化前生产成本视角 | 2026 | 美国 | $20,000 / 台 | N/A(量产前) | 分析师和公司工程团队对低批量原型制造的估算 | 中 | 手工组装原型成本;不包括规模化制造收益。 |
| 目标规模零售视角 | 2026 | 美国 | 单台低于 $10,000 | 目标降本 50% | 管理层目标定价,对标高端家电 | 中 | 依赖未经验证的合同制造和供应链规模折扣。 |
| 消费家电融资视角 | 2026 | 美国 | $167 / 月(5 年) | 可比汽车贷款 | 60 个月内摊销硬件付款,不含融资利润 | 中 | 假设消费者接受为未经验证的机器人承担多年债务。 |
规模视角综合了风险投资定价、原型成本和消费者融资模型;尚未报告商业收入。
[CM002, CM004, CM005, CM010, CM011]逐层收窄市场规模,从全球消费机器人机会过滤至 Memo 可行的早期部署。
各层宽度仅表示层级,不代表数值比例。
各层反映从广义行业 TAM 到最初 50 个家庭 beta 群体的连续商业过滤。
[CM003, CM004, CM005, CM008, CM012, CM014]2.3 买家分层、工作流摩擦与早期采用者经济学
自主家庭机器人的商业采用,会沿着明显的多阶段客户分层路径推进。Sunday 的首批客户群由 Founding Family Beta 计划定义:计划在 2026 年底向 约 50 个入选家庭部署带独立编号的 Memo。这个切入群体是自我选择的科技爱好者家庭,愿意承受商业化前的运营摩擦、偶尔的远程操作介入和诊断软件更新, 以换取优先使用家庭自动化的机会。Beta 之外,主要目标买家是因日常家务而严重缺时间的双收入职业家庭。在这一群体中,经济决策者和日常用户住在 同一屋檐下,采购关系比企业软件更简单。但工作流摩擦依然很大:家庭厨房台面高度各异,酒杯易碎,地面杂乱,家具摆放不可预测。 能否触发采用,很大程度取决于 Memo 能否在装洗碗机等边界明确的家务上接近完美地工作,同时不需要人类预先分类或持续看护。[CM003, CM008, CM010, CM013]
| 细分市场 | 买方 | 用户 | 付款方 | 工作流 | 预算负责人 | 采用触发因素 |
|---|---|---|---|---|---|---|
| 创始家庭 beta 群体 | 富裕的早期科技采用者 | 家庭成员和家政人员 | 家庭负责人(自费) | 部署 50 台 beta 设备,用于清理餐桌、装载洗碗机和叠衣 | 家庭负责人 | 希望抢先使用新技术,并能直接向产品团队反馈。 |
| 富裕科技爱好者 | 高净值郊区家庭 | 在职父母和家庭居住者 | 个人可支配收入 | 清理餐桌、装载洗碗机和分类干净衣物 | 主要房主 | 为高支付意愿的双收入职业人士夺回时间。 |
| 主流独栋住宅 | 中等至中高收入家庭 | 多代家庭成员 | 消费信贷或家电融资 | 日常多房间整理、每日餐具管理和地面收纳 | 家庭共同预算 | 零售价低于 $10,000,同时证明五年可靠性并具备安全记录。 |
| 老年居住与辅助护理 | 成年子女或辅助生活机构 | 需要任务支持的老年居民 | 私人付费或长期护理计划 | 轻量家务辅助、拾取地面物品和整理居住空间 | 护理人员 / 机构负责人 | 安全认证、轻柔的实体互动和被动稳定性保障。 |
细分反映 Sunday 从最初 50 户家庭 beta 到更广泛消费部署的商业化路线图。
[CM003, CM008, CM010, CM013]2.4 增长驱动、商业瓶颈与尽调风险
推动投资者关注 Sunday 的首要增长动力,是消费者对自动化减轻家务负担的巨大潜在需求;公司结束隐身后不久,约 1,000 人加入了等待名单。 Sunday 大幅扩充工程、研究和数据运营团队,为进入家庭部署做准备。但商业化瓶颈仍然突出。最直接的风险是硬件成本鸿沟:在通过合同制造的规模折扣 释放之前,$20,000 的原型成本根本没有正毛利空间。更严重的是,在亲密的居住空间里运行重型电动机械臂,会带来严重的产品责任和儿童安全风险; 一旦硬件失控,可能造成灾难性的声誉损害。最后,Sunday 未披露客户定金指标或确定的定价计划,商业转化率和客户生命周期价值等基本问题, 在接近首次 Beta 交付里程碑时仍没有答案。[CM009, CM012, CM013, CM014]
2.5 展示项
03竞争对手
3.1 家庭机器人格局与形态分化
2026 年的机器人行业已经分裂为两种截然不同的通用操作架构。一边是 Tesla Optimus、Figure AI 和 1X Technologies 等资金雄厚的人形通用机器人, 主张用双足拟人形态在为人类行动建造的环境中移动;另一边是 Sunday,它有意拒绝人形外观,转向以家庭为先、实用主义的 Memo。Memo 高 1.7 米、重 170 磅,不用双足行走,而是靠稳定的带轮底盘和伸缩垂直 Z 轴脊柱移动,脊柱可从地面升到 2.1 米。Sunday 放弃跨越多层楼梯的能力,换来单次充电 4 小时续航、较低制造复杂度,以及断电时可防止灾难性跌倒的被动物理稳定性,从而应对真实家庭中严苛的安全和责任约束。面向工厂的竞争者 接受笨重的液压或高扭矩电动双足机身,因为工业环境能容纳安全笼和严格的隔离带。住宅则需要柔顺控制,让机器一接触人就立即让开; 因此 Sunday 的形态取舍,构成了对工业人形既有企业的运营护城河。[CP001, CP002, CP003, CP004]
| 竞争者 | 类别 | 形态 | 规模与融资 | 目标细分市场 | 关键差异化 | 核心局限 |
|---|---|---|---|---|---|---|
| Sunday (Memo) | 直接家庭机器人同行 | 带伸缩脊柱和双臂的轮式移动底座 | $200M 总额($165M B 轮,$1.15B 估值) | 消费者家庭(洗衣、餐具、整理) | 通过 $200 手套实现真实家庭数据闭环;全自主架构 | 无法爬楼;零售价和量产计划尚未公布 |
| Weave Robotics (Isaac 1) | 直接家庭机器人同行 | 带伸缩脊柱和双臂的轮式移动底座 | 种子轮 / 早期风险投资支持;车队运行超过 2,000 小时 | 单层住宅和轻型商业场所 | $7,999 标价($449/月);活跃车队每周叠衣 1,000+ 磅 | AI 遇到失败时依赖人工远程操作 |
| 1X Technologies (NEO) | 双足人形机器人同行 | 带柔性触感外壳的轻型双足人形机器人 | 已筹集超过 $125M 风险资本(OpenAI、EQT 支持) | 消费者家庭和企业设施 | 安静的腱驱动双足运动;类人的形态 | 目标成本高($20,000 或 $499/月);双足机器人在家庭中有跌倒风险 |
| Tesla (Optimus) | 现有工业 / 通用机器人 | 配备灵巧双手的全尺寸双足人形机器人 | 由 Tesla 资产负债表和内部 AI 基础设施提供资金 | 先面向工业制造;未来面向家庭消费者 | 巨大的汽车制造规模和 Dojo AI 算力 | 优先工业工厂;家庭安全和入户测试尚远 |
| Figure AI (Figure 02) | 现有工业 / 通用机器人 | 配备 16 自由度双手的全尺寸双足人形机器人 | 融资超过 $675M,估值超过 $2.6B(OpenAI、Nvidia、Bezos) | 商业物流和工业制造 | 商业工厂试点(BMW);高负载操作 | 聚焦企业 / 工业;占地和成本不适合家庭 |
主要数据截至 2026 年,汇总自公司披露、新闻稿和技术报告。Sunday 原型成本为 $20,000;零售价尚未公布。
[CP001, CP002, CP003, CP004, CP007]3.2 能力矩阵与自主性分歧
物理人工智能长期以来的核心瓶颈,是在无序、杂乱环境中缺少高保真操作数据。Sunday 不用传统的 $20,000 远程操作台或模拟环境,而是向数百名远程 Memory Developers 部署轻量、单只 $200 的 Skill Capture Gloves,在真实家庭布局中收集超过 ten million 个真实家务片段。这个数据引擎驱动 ACT-1 和 ACT-2 模仿学习模型;据报道,在 9 种不同衣物、785 次叠衣试验中,无需针对每个家庭微调,就达到了 99.1% 成功率。 这一技术架构与 Sunday 最接近的带轮直接对手 Weave Robotics 形成鲜明对比。Weave 已部署 Isaac 机队,用主动远程人工操作折叠商业洗衣;一旦自动策略卡住, 就由人介入。Sunday 则明确排除在客户家中进行人工远程操作和监控直播。Sunday 的纯自主理念有利于建立消费隐私形象,但也让公司暴露在真实世界 边缘案例中:无辅助的操作失败需要用户亲自介入,而不是无缝远程恢复。[CP005, CP006, CP007, CP008]
| 购买标准与能力 | Sunday Memo | Weave Robotics Isaac 1 | 1X NEO | Tesla Optimus / Figure 02 | 现状(家电与人工劳动) |
|---|---|---|---|---|---|
| 运动与稳定性 | 轮式底座;被动稳定;柔顺控制;无法爬楼 | 轮式底座;无法爬楼 | 双足行走;可走楼梯;跌倒风险更高 | 双足行走;需要主动平衡;冲击风险高 | 固定式家电;人工在楼梯间移动 |
| 自主架构 | 端到端学习型自主系统(ACT-2);客户家中零远程操作 | 混合自主系统,人工远程操作介入 | 受监督的自主系统,配合远程操作支持 | 端到端神经网络操作(工业试点) | 100% 人工操作;零算法自主性 |
| 叠衣可靠性 | 9 种衣物类型、785 次试验中据报道达到 99.1%(未经第三方验证) | 通过操作员辅助,商业车队每周叠衣 1,000+ 磅 | 在受控实验室测试中演示;客户侧数据未知 | 受控演示能力;在住宅中未经验证 | 100% 人工灵巧度和速度;零家电自动化 |
| 洗碗机与餐桌清理 | 已演示从餐桌到洗碗机的流程(33 次灵巧交互) | 公开生产披露中未支持 | 已演示操作杯子和轻型物品 | 已在工厂演示容器分拣和零件搬运 | 人工装载;家电自动完成洗涤周期 |
| 家庭数据隐私政策 | 手套数据由合同 Memory Developers 采集;客户家中没有训练摄像头 | 客户家中的摄像头画面会流向远程操作员,以便介入 | 远程操作会传输视频,供操作员控制 | 工厂 / 企业摄像头记录 | 绝对隐私;摄像头和网络连接均为零 |
评估截至 2026 年,依据公开演示基准、技术报告和厂商披露。未获支持的能力标记为未知或缺失。
[CP005, CP006, CP007, CP008, CP013]按家庭杂务专门化程度和自主架构对家庭机器人替代方案进行定位,展示 Sunday 作为自主轮式专家相对于远程操作和工业同业的位置。
| 条目 | 家庭杂务专门化(0 = 工厂 / 通用型,10 = 专用家庭杂务) (1–8.5) | 自主架构(0 = 远程操作 / 手动,10 = 完全学习式自主) (1–8) | 说明 |
|---|---|---|---|
| 1. Sunday (Memo) | 8.5 | 8 | |
| 2. Weave Robotics (Isaac 1) | 8 | 3.5 | |
| 3. 1X Technologies (NEO) | 6.5 | 5 | |
| 4. Tesla (Optimus) | 3.5 | 6 | |
| 5. Figure AI (Figure 02) | 2.5 | 6.5 | |
| 6. 现状(家电与人工劳动) | 1 | 1 |
分数依据截至 2026 年的公开技术报告、产品架构和商业披露,代表有证据支持的序数评估。
[CP001, CP005, CP007, CP014]3.3 定价、产品包装与历史失败先例
家庭操作硬件商业化,必须跨过高昂的物料清单门槛,也要面对惨痛的历史失败先例。Memo 原型目前手工装配,每台约花费 Sunday $20,000;随着制造规模化, 管理层目标是降价 50% 至 $10,000。相比之下,Weave Robotics 通过 Isaac 1 建立了明确的商业价格上限:首付 $7,999,或每月 $449、定金 $250; 1X NEO 的目标售价为 $20,000,或每月订阅 $499。然而,家庭洗衣和家务机器人行业有着令人警醒的商业破产史。日本初创公司 Seven Dreamers Laboratories 从包括 Panasonic 在内的企业支持者处筹得约 10 billion yen,用于打造 Laundroid 叠衣柜;但由于无法解决叠衣速度慢(每件衬衫三至十分钟)、 零售价高($16,000)和深色衣物传感器失灵等问题,最终于 2019 年申请破产并背负 2.2 billion yen 债务。FoldiMate 也在 2021 年停止运营, 未能交付可行的低于 $1,000 的消费级叠衣设备。Sunday 必须在即将到来的 2026 年秋季 Founding Family Beta 中证明,Memo 的多任务效用超过 过去市场参与者致命的单任务单位经济学。[CP009, CP010, CP011, CP012]
| 解决方案 / 平台 | 硬件前期价格 | 月度订阅 / 租赁 | 押金要求 | 部署状态 | 关键经济与实用影响 |
|---|---|---|---|---|---|
| Sunday(备忘录) | 未公布(原型制造成本约 $20,000;目标约 $10,000) | 未公布(Beta 计划免费且仅限受邀者) | 无需公开押金(Founding Family Beta 优先) | Founding Family Beta 计划于 fall 2026 启动;shipments 2027-2028 | 初始制造成本高;需扩大规模才能达到消费者可接受的经济性 |
| Weave Robotics(Isaac 1) | 标价 $7,999 | $449 / 月 | $250 可退押金 | 2026 年秋季在加州发货;2027 年全国铺开 | 为早期消费者家务自动化设定了已知价格上限;持续承担远程操作人工成本 |
| 1X Technologies(NEO) | 预计标价 $20,000 | $499 / 月 | 预约候补名单已开放 | 目标是在 2026 年内交付首批家庭 Beta | 高端定价将初始可服务市场限制在高净值早期采用者 |
| Seven Dreamers(Laundroid)[历史] | 约 $16,000(计划零售价 1.85M JPY) | 无(一次性购买) | 破产前接受预订 | 2019 年破产(提交时负债 2.2B JPY;出货量为零) | 证明高价格、折叠缓慢(每件衬衫 3–10 分钟)和单一任务范围难以实现商业化 |
| FoldiMate [历史] | 计划目标价约 $980 | 无(一次性购买) | 接受候补登记 / 预订 | 2021 年停止运营,未实现大规模商业发布 | 表明即使低于 $1,000 的硬件,在柔性纺织品上的机械可靠性也很难达标 |
Laundroid 和 FoldiMate 的历史定价由历史记录换算而来。Sunday 的零售价尚未正式公布;原型制造成本为 $20,000。
[CP009, CP010, CP011, CP012]3.4 数据护城河持久性、转换成本与替代风险
Sunday 在 2026 年 3 月由 Coatue 领投、以 $1.15 billion 估值完成超额认购的 $165 million Series B 后,将工程和运营团队扩展到 100 多名专业人士, 并从 Tesla Optimus 项目招募有经验的基础设施和系统负责人。Sunday 认为,从手套硬件、Skill Transform 具身翻译、基础策略预训练到机队部署的垂直整合, 打造了难以攻破的数据飞轮。然而,长期护城河仍面临明显的替代威胁。家庭家电的消费者转换成本很低,家庭可以同时使用传统洗碗机、洗衣机和扫地机器人, 几乎没有固有的供应商锁定。此外,Physical Intelligence(Pi0)等第三方研究机构快速推进基础操作模型,可能把通用操作策略商品化,让资金雄厚的硬件 OEM 无需复制 Sunday 的专有手套机队,也能达到类似任务成功率。Sunday 的竞争持久性最终取决于:它的全自主架构能否比竞争者扩张多任务远程操作更快赢得家庭信任。[CP013, CP014]
3.5 展示项
04财务
4.1 收入模式、定价架构与变现路径
Sunday 为 Memo 家庭机器人采用消费家电式变现框架,而不是双足人形竞争者常见的企业自动化或工业租赁合同。公司的长期消费零售价目标低于 $10,000; 大规模制造后,市场公开预期集中在 $5,000 至 $10,000 价格区间。但 Sunday 尚未公布商业订阅费、付费软件分层或经常性维护计划, 购买后的变现结构仍未得到验证。管理层把 Memo 定位为高端消费电子产品,但公司如今完全没有收入,也未收取商业客户定金或敲定零售融资方案。 商业收入要等首次家庭安全和可靠性试验完成后才会产生。[CI001, CI002, CI003, CI004]
| 收入来源 | 变现机制 | 单位 / 指标 | 当前价值 / 状态 | 收入质量 | 尽调问题 |
|---|---|---|---|---|---|
| 硬件单元销售 | 一次性消费者零售硬件购买 | 每售出一台机器人 | 目标零售价低于 $10,000(尚未产生收入) | 依赖资本开支、非经常性 | 确认最终 MSRP、经销商扣点和客户押金政策 |
| 软件与智能层 | 为模型更新和云服务提供的经常性订阅 | 每台活跃设备的月费或年费 | 未公布 / 可选层 | 若强制订阅,利润率可能较高 | 弄清基础模型更新(ACT-2)是否要求有效订阅 |
| 替换部件与耗材 | 直接销售替换夹爪、电池和易损件 | 每个订购部件 | 商业化前 / 未说明 | 辅助利润来源 | 获取部件磨损测试曲线和预期年度维护支出 |
| 上门服务与安装 | 一站式安装、建图和技术支持 | 每次事件或服务合同 | 为 50 个 Beta 家庭免费提供 | 除非变现,否则属于成本中心 | 审查现场技术支持人员配置模型和保修准备金计提规则 |
截至 2026 年末,Sunday 仍完全处于收入前阶段;收入机制代表管理层的变现目标和计划中的商业化结构。
[CI001, CI002, CI003, CI011]| 变现组成 | 价格 / 单位 / 合同 | 标价与实际实现价格 | 折扣 / 未知事项 | 来源 / 参考 |
|---|---|---|---|---|
| Founding Family Beta 硬件 | 免费(前期硬件费用 $0) | 前 50 台的标价和实际实现价格均为 $0 | 仅限受邀的研究群体;完全未变现 | Humanoids Daily / Sunday 官方公告 |
| 商业零售单价 | 目标低于 $10,000(预计 $5,000 至 $10,000) | 目标标价;尚无商业收入实现 | 商业批量折扣和融资条款尚未公布 | MLQ.ai / AI2.work 报道 |
| Skill Capture Glove 硬件 | 制造成本约 $200 | 提供给 Memory Developers;不向消费者销售 | Sunday 将其计入数据收集资本开支 | 来源:Humanoids Daily / Sunday.ai |
| Memory Developer 任务报酬 | 每个获批准演示小时 $30 至 $60 | 通过 Stripe 支付的实际承包商费用 | 费率随任务复杂度和提交质量变化 | Simplify Jobs 招聘信息 |
商业零售定价反映公开的管理层目标;截至 2026 年 10 月,不存在付费客户交易或具有约束力的购买承诺。
[CI003, CI004, CI009, CI011]展示 Sunday 如何从众包数据采集和免费 beta 试用,走向规模化零售销售,并验证毛利率能否成立。
| 节点 / 连接 | 节点 / 起点 | 终点 | 说明 |
|---|---|---|---|
| 节点 1 | 数据采集与手套部署 [n1] | 向时薪 $30-$60 的承包商分发 $200 Skill Capture Gloves,以低于 $20,000 远程操作设备的资本开支采集 10M+ 段数据。 | |
| 节点 2 | 未变现的 Beta 测试 [n2] | 2026 年末向 Founding Family 家庭部署 50 台免费且各有编号的 Memo,用于现实世界可靠性和安全测试。 | |
| 节点 3 | 消费者候补名单转化 [n3] | 安全基准和制造准备度达标后,将候补名单上 1,000 多名自然流量申请者转化为付费零售订单。 | |
| 节点 4 | 硬件商业化 [n4] | 目标消费零售价低于 $10,000($5,000 - $10,000 区间),将 Memo 定位为高端家电而非工业机器人。 | |
| 节点 5 | BOM 成本压缩挑战 [n5] | 靠规模化合同制造,把手工组装原型成本从约 $20,000 降至 $10,000 以下,消除负单位毛利。 | |
| 节点 6 | 软件与机群变现 [n6] | 从一次性硬件销售扩展到经常性智能更新(ACT-2)和机群服务,带来可持续的毛利。 |
流程图展示管理层规划的商业化顺序;产量、零售上市时间和经常性服务定价尚未最终确定。
[CI001, CI003, CI005, CI007, CI011, CI012]4.2 单位经济学、硬件成本结构与毛利路径
Memo 当前的底层单位经济学存在严重毛利率缺口,只有通过规模化制造才能解决。手工装配的原型每台成本约 $20,000,原因是专用双夹爪、定制聚合物外壳和小批量执行器采购。Sunday 预计规模化后至少降本 50%,把生产成本压到 $10,000 以下;但行业观察者强调,若供应链没有大幅压缩,按目标零售价附近 销售硬件将带来负毛利或微薄毛利。在数据采集端,Sunday 用单只约 $200 的 Skill Capture Gloves 替代 $20,000 的远程操作台,资本效率很高; 众包 Memory Developers 每个获批小时的报酬为 $30 至 $60。不过,多步骤家务中不断叠加的错误率,会带来尚未建模的保修和服务责任, 进而在零售部署后侵蚀单位经济学。[CI005, CI006, CI007, CI008, CI009]
| 单位指标 | 当前价值 | 置信度 | 战略意义 | 尽调问题 |
|---|---|---|---|---|
| 原型单元制造成本(BOM + 人工) | 每台约 $20,000 | 中 | 超过目标零售价;按当前制造规模会产生负毛利率 | 要求提供逐项 BOM 拆解,涵盖执行器、计算、传感器和结构底盘 |
| 规模化制造目标成本 | 每台低于 $10,000(降幅 50%+) | 低 | 在目标零售 MSRP 下实现硬件正毛利率所需的门槛 | 审查合同制造报价、最小订单量和模具资本开支 |
| Skill Capture Glove 生产成本 | 每台约 $200 | 中 | 以相当于 $20,000 远程操作设备 1% 的成本,实现众包人工数据收集 | 检查手套传感器寿命、校准频率和更换率 |
| 数据收集人工费率 | 每小时 $30.00 至 $60.00 | 高 | 每小时获批准上门家庭演示数据的直接变动人工成本 | 审计承包商总拨付额,以及每项已部署技能对应的演示小时数 |
| 客户保修与服务准备金 | 未披露 / 私下 | 低 | 若机械故障或上门事故需要现场支持,可能严重拖累利润率 | 检查客户支持升级流程、保修政策和准备金建模 |
原型制造成本反映低产量手工组装;成熟的单位经济取决于尚未验证的供应链规模和合同制造协议。
[CI005, CI006, CI007, CI008, CI009]4.3 商业牵引、Beta 部署与 GTM 动能代理指标
Sunday 的商业 go-to-market 仍处于商业化前演示和研究阶段。公司商业收入为零,ARR 为零,也没有任何付费企业或零售客户。为完成 2026 年底的部署里程碑, Sunday 将约 50 台独立编号的 Memo 免费发给入选的 Founding Family Beta 参与者,而不是进行商业硬件交付。超过 1,000 个报名的自然等待名单, 证明消费者兴趣很高。但免费 Beta 部署不能证明付费意愿或获客成本效率;分析师警告,商业承保判断开始前,入户部署就会让公司立刻暴露于 安全事故、法律责任和公关风险。[CI010, CI011, CI012, CI013]
4.4 资本充足性、烧钱估计与尽调承保结论
Sunday 的远期资产负债表由累计 $200 million 风投资金支撑;2026 年 3 月的 Series B 还让 Coatue 联合创始人 Thomas Laffont 加入董事会, 同时扩大了部署资本池。公司积极扩大运营支出:工程团队扩大 3 倍,研究团队扩大 4 倍,数据运营扩大 5 倍,并在 Silicon Valley 招聘资深人才。 与汽车或大批量硬件发布相比,50 台免费 Beta 所需资本开支很少,因此 Sunday 拥有足够流动性支撑运营到 2027 年。不过,从投资判断角度看, Sunday 仍是一项纯研发资产:没有经审计的现金消耗、已实现的物料清单成本或经过验证的消费者付费意愿,财务判断无法在不做深度私有尽调的情况下 证明 $1.15 billion 估值合理。[CI014, CI015, CI016]
4.5 展示项
05产品与技术
5.1 产品定义与家庭家务工作流
Sunday 将 Memo 定位为自主家庭设备,旨在把被重复、灵巧家务占用的家庭时间还回来。Memo 不处理宽泛的工业任务或结构化工厂装配, 而是围绕住宅空间凌乱且不可预测的几何环境设计,主要聚焦叠衣服、整理厨房和管理洗碗机。家庭机器人能否落地,关键标准是无需针对特定场所的人类介入 或远程操作,也能执行多步骤物理流程。在演示试验中,Memo 执行了一个 33 步的 Table-to-Dishwasher 工作流,涉及 21 件物体上的 68 次灵巧交互, 并完成 130 英尺的室内导航,处理陶瓷餐具、金属餐具、食物残渣和易碎的酒杯。 家庭家务自动化与工业自动化有根本差异,因为家庭物品高度多变、会变形,还常常部分被遮挡。衣服会不可预测地变形,盘子直径各不相同, 洗碗机篮架的空间几何也不规则。Sunday 已展示抽浓缩咖啡、整理鞋子等基础能力,但近期产品重点是把叠衣服做成首项生产级能力。 然而,长时段的厨房清理任务仍然更容易失败;在自主进入厨房部署前,还需要大量感知推理和恢复策略,才能达到叠衣服已经展示的稳定性。 [CE001, CE006, CE007, CE011, CE012]
| 家务 | 当前人工流程 | Memo 机器人方案 | 已演示基准 | 当前运行限制 |
|---|---|---|---|---|
| 折叠衣物 | 人工分类和折叠 | ACT-2 自主折叠并堆叠 | 在陌生家庭的 785 次试验中成功率 99.1% | 轻薄衬衫的准确率较低,为 94.7% |
| 餐桌到洗碗机 | 人工清空、刮除、装架 | ACT-1 33 步自主清理流程 | 试点中对 21 件物品完成 68 次灵巧动作 | 杂乱水槽中的精致透明玻璃器皿 |
| 浓缩咖啡制作 | 人工压粉并锁定手柄 | 双臂柔顺性与扭矩控制 | 已演示从咖啡豆到成杯的 crema 萃取 | 无法适应不同的研磨机 / 咖啡机布局 |
| 地面鞋子整理 | 人工整理入口区域 | 单臂空间排列 | 首个学会的任务(2024 年 12 月原型) | 仅限低摩擦平整地面 |
任务表现基准来自公司在固定模型检查点下报告的评估试验;真实家庭干预率仍需在 Beta 期间确定。
[CE006, CE007, CE011, CE012]5.2 机械架构、移动能力与物理安全
Memo 明确偏离 Figure、Tesla 等人形竞争者偏好的拟人化双足设计,采用带轮移动底座和垂直升降躯干。机身高 1.7 米,约重 77 千克(170 磅), 配有伸缩 Z 轴脊柱,可将垂直触达范围从地面延伸至 2.1 米(6 英尺 10 英寸)。这种机械布局让 Memo 无需消耗能量维持双足动态平衡, 就能从地毯捡起丢弃物并够到高处的厨房搁板。带轮架构消除了灾难性跌倒风险,以 1 小时快速充电换来 4 小时续航,并大幅减少机械执行器数量。 住宅环境的安全,建立在柔顺执行和被动结构稳定性之上。Memo 的多自由度手臂使用柔顺控制回路,住户或儿童可以把手臂推到一旁,而不会遇到刚性阻力。 突然断电时,Memo 的低重心和稳定轮距能防止机器人倾倒或因自重坍塌。Memo 没有采用人类式五指手,而是使用定制的三指柔顺夹爪, 模仿人类手指的功能性抓取几何。这样既降低了电机复杂度和故障点,又保留了抓取易碎玻璃器皿和操作柔软纺织品所需的触觉灵巧性。 手工制造的原型目前每台约 $20,000,制造规模化仍是重大挑战。 [CE002, CE003, CE010, CE013]
| 模块 / 资产 | 目标用户 / 角色 | 成熟度与状态 | 技术差异化 | 尽调缺口 |
|---|---|---|---|---|
| Memo Base & Spine | 家庭家务移动 | 现场测试中的原型(1.7m,77kg) | 以 2.1m 伸缩升降结构实现轮式被动稳定 | 长期传动系磨损和门槛跨越 |
| 双机械臂与夹爪 | 双手物品操作 | 已在实验室和试点中验证 | 具备柔顺性的机械臂,采用三指欠驱动几何结构 | 沉重湿厨具的动态负载上限 |
| Skill Capture Glove | Memory Developers(众包) | 量产硬件(已发货 >2,000 件) | $200 可穿戴形态,复刻 Memo 手部运动学 | 手套传感器随时间的漂移校准和耐用性 |
| ACT 基础模型(ACT-1 / ACT-2) | 自主策略执行 | ACT-2 于 2026 年 7 月预览 | 无机器人预训练,单次演示任务适配 | 在复杂湿润 / 反光表面上的泛化能力 |
成熟度状态和规格来自已披露的原型指标及 2026 年 7 月技术披露;商业化量产公差仍未验证。
[CE001, CE002, CE003, CE005]5.3 物理 AI 基础模型与家庭数据飞轮
Sunday 的核心技术差异,来自创始人将机器人模仿学习的学术研究融入产品。CEO Tony Zhao 在 Stanford 共同开发 ALOHA、Mobile ALOHA 和 Action Chunking with Transformers(ACT),建立了让机器人学习数秒动作轨迹、而非孤立电机步骤的方法。CTO Cheng Chi 则在 Columbia 和 Stanford 开创 Diffusion Policy 与 Universal Manipulation Interface(UMI),建立了生成式扩散技术,用于多模态动作生成,并把演示采集与实体机器人硬件解耦。 Sunday 将这些学术突破变成商业物理 AI 技术栈,核心是专有的 Skill Capture Glove。 传统机器人学习依赖每台成本超过 $20,000 的远程操作台,形成严重的资本瓶颈,限制训练数据集规模。Sunday 用一只成本约 $200 的低成本传感手套替代, 让手套模仿 Memo 的手部运动学和传感器布局。公司把数千只手套发给美国各地家庭中的 500 多名远程众包 Memory Developers,让他们在自家记录真实日常家务。 记录下来的多模态交互经由 Skill Transform 处理;这是一个专有软件流水线,可重定向人类运动学,并以 90% 的转换产出率补偿视觉具身差距。 这条流水线为 ACT-1 提供了超过 10 million 个真实世界片段,使基础模型能在没有任何实体机器人远程操作数据的情况下训练。 [CE004, CE005, CE009, CE014, CE015]
| 架构层 | 主要组件 | 运营角色 | 关键依赖 | 关键技术或运营风险 |
|---|---|---|---|---|
| 数据采集硬件 | Skill Capture Glove | 捕捉手部运动学和接触力 | 覆盖 >500 个美国家庭的分布式网络 | 人类演示噪声和物理柔顺性保真度 |
| 数据转换管线 | Skill Transform | 将人体运动学映射到 Memo 动作空间 | 视觉域适配和运动学重定向 | 手套到机器人轨迹转换的失败率为 10% |
| 基础 AI 模型 | ACT-1 与 ACT-2 | 生成多模态动作块轨迹 | 在 10M+ 常规情景上大规模预训练 | 对未覆盖的边缘物品,模型性能趋于平台期 |
| 机器人控制与安全 | 柔顺电机驱动 | 确保人机安全的物理接触 | 低延迟传感运动控制回路 | 连续多小时负载下电机热节流 |
| 家庭移动与电源 | 轮式底座与 4h 电池 | 在单层家庭平面内导航 | 平整地面与障碍物清除 | 无法爬楼或适应多层住宅布局 |
架构组件汇编自公司技术披露和创始人访谈;专有模型权重规模及推理硬件规格未披露。
[CE004, CE005, CE009, CE010, CE013]多层物理 AI 栈,展示从手套数据采集、基础模型推理到机器人柔顺执行的完整链路。
- Tier 1 - 家庭数据采集
- 条目
- 专有 Skill Capture Glove($200 BOM)
- 超过 500 个 Memory Developer 家庭组成的分布式网络
- 可穿戴动作、力和光学采集传感器
- 模块
- 硬件数据记录器
- 演示质量验证器
- 输出
- 人类原始多模态运动学演示流
- Tier 2 - 数据转换与领域适配
- 条目
- Skill Transform 软件管线
- 将 20 多个 DoF 的运动学重定向至三指手
- 视觉手臂遮罩与领域合成
- 模块
- 运动学转换器
- 视觉合成器
- 输出
- 机器人兼容轨迹片段(90% 转换良率)
- Tier 3 - 预训练与基础 AI
- 条目
- ACT-1 与 ACT-2 基础模型
- 使用 Transformers 架构进行动作分块
- Diffusion Policy 多模态轨迹生成
- 模块
- 动作 Transformer 核心
- 扩散轨迹规划器
- 输出
- 通用操作策略权重
- Tier 4 - 微调与求解校准
- 条目
- 单次演示监督微调
- 现实世界故障恢复闭环
- 求解泛化基准协议
- 模块
- 快速 SFT 适配器
- 故障采集器
- 输出
- 校准后的任务策略(例如 99.1% 叠衣)
- Tier 5 - 边缘推理与传感运动执行
- 条目
- 机器人端神经推理引擎
- 3D 地图条件室内导航
- 双臂轨迹跟踪控制器
- 模块
- 地图定位引擎
- 轨迹控制器
- 输出
- 协同双臂毫米级操作指令
- 第 6 层——物理硬件与安全子系统
- 条目
- 覆硅胶底盘与轮式移动底座
- 伸缩式 Z 轴脊柱(伸展范围 0m 至 2.1m)
- 柔顺执行器、被动稳定与电源安全
- 模块
- 柔顺电机控制器
- 断电装置与被动阻尼器
- 输出
- 安全完成家庭实体家务
分层拆解综合了公开技术论文(ACT、Diffusion Policy、UMI)与公司披露;边缘推理芯片的具体细节仍属机密。
[CE004, CE005, CE009, CE010, CE014, CE015]5.4 泛化基准、Solve 标准与失败恢复
2026 年 7 月,Sunday 发布 ACT-2,这是一款针对机器人领域长期泛化缺口的先进基础模型:实验室环境训练的模型,部署到陌生家庭时往往会失败。 Sunday 在未参与训练的真实家庭中,针对 9 个不同衣物类别进行了 785 次自主叠衣试验。Memo 成功叠好 778 次,综合成功率 99.1%(标准误差 ±0.3%); 按严格的缺陷扣分标准,平均外观质量为 4.72 / 5 星。短裤和 Polo 衫成功率达到 100%,T 恤为 99.0%,轻薄女式衬衫为 94.7%, 后者因材料柔软而更难操作。 更关键的是,Sunday 证明,扩大多样化预训练数据能系统性压缩分布外泛化差距。内部消融实验显示,没有预训练数据的模型在实验室同域成功率为 96%, 但在陌生环境中骤降至 14%(泛化损失 82 个百分点)。将预训练数据扩展到 100% 后,在测试条件下这一差距完全消失。 此外,ACT-2 还展示了单次示范监督微调能力:仅凭一个人的示范,就能学习新的折叠方式,并应用到未见过的衣物。为建立严格的行业问责, Sunday 提出了 Solve 框架,主张用未见环境中的经验证自主率衡量机器人能力,而不是靠精心剪辑的亮点视频。 [CE001, CE006, CE007, CE012]
5.5 部署路径、生产经济学与尽调风险
2026 年 3 月,Sunday 以 $1.15 billion 估值完成超额认购的 $165 million Series B 后,开始把资源从研究演示转向商业部署。 Sunday 计划在 2026 年底通过 Founding Family Beta 向早期采用者部署约 50 台带编号的 Memo,目标是在 2026 年感恩节前后开始交付。 这次 Beta 部署既是经验验证实验室,用来评估未编排的家庭边缘案例(包括儿童、宠物、液体洒落、环境光变化和 Wi-Fi 延迟),也用于在扩大消费硬件生产前 测试长期机械可靠性。 尽管技术有所进展,承保判断所需的重大商业和运营问题仍未解决。定制机加工原型的单台制造成本目前接近 $20,000;Sunday 预计转向注塑规模化工具后, 可将生产成本降至 $10,000 以下,但工具排期和供应商合同尚未披露。把配备视觉系统的机器人放进私人住宅,也会带来显著隐私敏感性。 Sunday 强调 Memo 运行自主策略,不会接入实时人工远程操作视频;但正式安全认证、设备端数据脱敏协议和云端传输边界尚未接受公开审计。 此外,公开专利缺失表明,Sunday 的知识产权很大程度依赖手套硬件和数据转换流水线中的商业秘密。 [CE002, CE008, CE010, CE013]
5.6 展示项
06客户
6.1 客户分层与目标家庭画像
Sunday 瞄准住宅消费者家庭,用于自动化家庭家务,有意绕开商业酒店、医疗和工业仓储物流。公司的 go-to-market 设计集中在高收入的独栋住宅业主和双职工家庭, 他们每天都受叠衣服、收拾餐桌、整理厨房等家务的持续时间约束。消费机器人硬件本身部署复杂,Sunday 因而意识到,早期客户会以科技爱好者为主: 他们愿意与早期自主系统共处,也愿意容忍偶尔的运行故障。与企业软件或 B2B 硬件不同,Sunday 没有商业合同结构、企业采购渠道或 SLA 承诺, 只专注于建立直接的消费者关系。[CU001, CU002, CU006, CU007]
| 细分群体 | 买方 / 用户 / 付费方 | 主要使用场景 | 目标规模 | 战略 / 收入价值 | 尽调缺口 |
|---|---|---|---|---|---|
| 科技早期采用者 | 崇尚科技的房主与机器人爱好者 | 自动折叠衣物和清理厨房 | 初始约 50 个 Beta 家庭,扩展至数百个 | 初期参与度高;可为用户体验和边缘情况提供定性反馈 | 愿意从免费 Beta 转为付费硬件 / 订阅 |
| 高收入家庭 | 希望节省家务时间的双收入父母 | 日常地面整理、装载洗碗机和洗衣流程 | 郊区独栋住宅中的广泛大众富裕型住宅 TAM | 长期经常性收入和机队密度的主要驱动 | 能容忍原型恢复干预、杂物和宠物干扰 |
| Memory Developers | 收集家务演示的有偿家庭承包商 | 使用 Skill Capture Glove 记录高保真操作流程 | 已向 500+ 个家庭发货 2,000+ 副手套 | 驱动 ACT 基础模型的数据获取引擎 | 承包商流失、数据质量一致性和地域多样性 |
| 商业 / B2B 酒店业 | 酒店运营商和商业设施经理 | 标准化床品折叠和托盘清理 | 尚未覆盖;非核心探索性细分 | SLA 合同推动企业机队部署的潜力 | 没有商业 B2B 销售动作、现场支持或企业认证 |
细分反映目标运营画像;由于 Sunday 目前尚未产生收入,商业定价和收入贡献仍未公布。
[CU001, CU002, CU006, CU007]6.2 采用路径与商业化前需求
自 2025 年底隐身亮相以来,公众对家庭机器人自动化的兴趣,为 Sunday 带来了大量漏斗顶部需求。Memo 原型制作浓缩咖啡和收拾玻璃器皿的演示走红后, 这家初创公司收到数千个 Beta 计划咨询,并建立了超过 1,000 个家庭的初始等待名单。公司围绕计划在 2026 年底启动、设有门槛的 Founding Family Beta 设计了首条客户管道,时间点接近感恩节。Sunday 希望借此从实验室评估转向持续的家庭数据采集,利用客户反馈扩大 Memo 的家务范围, 并验证全机队模型更新。[CU003, CU004, CU005, CU008]
| 指标 | 数值 | 日期 / 时间范围 | 来源 | 置信度 | 含义 | 缺失分母 |
|---|---|---|---|---|---|---|
| Beta 候补名单申请 | 数千(初始 1,000+) | March 2026 | Sunday B 轮发布与科技报道 | 高 | 病毒式演示视频带来强劲的消费者漏斗顶端兴趣 | 申请人资格标准总数及地域分布 |
| 计划中的 Beta 机队规模 | 约 50 台编号设备 | Fall 2026 | GlobeNewswire 与 Note.com 技术分析 | 中 | 密切监控受控初始群体,并提供全程工程支持 | 已确认首批秋季群体中的活跃家庭具体数量 |
| 商业客户数 | 0 个付费账户 | Current (October 2026) | 独立分析师研究 | 高 | 公司完全处于收入前阶段,尚无商业部署 | 从免费 Beta 申请转化为付费商业客户的转化率 |
| 数据收集家庭 | 500+ 个家庭 | November 2025 – March 2026 | 公司披露与硬件追踪 | 高 | 在部署客户之前已积累广泛的预训练数据基础 | 手套接收者的每周活跃数据收集留存率 |
采用指标反映商业化前的管线规模;从免费 Beta 兴趣到付费部署的转化尚未测试。
[CU003, CU004, CU005, CU008]从排队申请到自主执行家庭任务,再到长期扩展的端到端客户旅程。
[CU003, CU004, CU005]6.3 客户证明、试点现实与运营限制
尽管市场热情高涨、融资达到独角兽水平,Sunday 尚未将任何一台商业产品放入客户家中,仍处于没有收入、没有付费账户的阶段。 当前客户证明仅限于对计划中的 50 台 Founding Family Beta 的潜在兴趣,以及付费 Memory Developer 承包商佩戴 Skill Capture Gloves 进行的数据采集工作流。 此外,Sunday 的运营理念与 Weave Robotics 等家庭机器人同行截然不同:Weave 依靠远程人工操作员实时介入,Sunday 则严格承诺自主执行物理 AI, 不允许入户实时远程操作。这种纯自主立场抬高了运营门槛,Memo 必须在无序家庭环境中成功工作,不能依赖人类参与的后备方案。[CU008, CU009, CU010, CU011]
| 客户 / 群体 | 细分群体 | 部署 / 使用场景 | 量产与试点 | 观察结果 | 已知限制 |
|---|---|---|---|---|---|
| Founding Family Beta Cohort | 住宅早期采用者 | 家庭家务(折叠衣物、清理餐桌、装载洗碗机) | 尚未发布的试点(2026 年秋季) | 正在筛选;优先获得编号 Memo 设备 | 仅限受邀且免费参与;硬件尚未部署到客户家中 |
| Memory Developer 网络 | 付费家庭数据贡献者 | 使用 Skill Capture Glove 录制家庭杂务演示 | 活跃运营管线 | 已在 500 多个家庭采集超过 1,000 万段动作片段 | 数据采集承包商,而非商业付费机器人客户 |
| 内部员工部署 | Sunday 工程团队 | 在员工住所和租赁 Airbnb 中验证生活空间 | 内部测试 | ACT-2 零样本洗衣模型经过数月迁移验证 | 受监督的工程环境,开发人员可立即介入 |
| 商业企业试点 | 酒店及设施管理 | 潜在的商业洗衣与清洁流程 | 尚未覆盖的管线 | 未宣布任何企业试点或商业合同 | 公司仍只专注于独栋住宅部署 |
商业化前部署群体,依据公开宣布的 Founding Family Beta 和 Memory Developer 训练网络。
[CU008, CU009, CU010, CU011]6.4 留存持久性、流失风险与单位经济学
Sunday 的核心尽调难题不是漏斗顶部的消费者好奇心,而是客户留存能否持久以及单位经济学是否成立。在机器人每天必须连续执行几十个物理动作的家庭环境中, 即使任务可靠性达到令人印象深刻的 99.1%,按数学推算也意味着约每 100 次动作会失败一次,导致餐具每天掉落、衣物堆积卡住,或机器人在杂乱房间中迷路。 家庭机器人历史凸显了这一持久性挑战:Seven Dreamers Laboratories 和 FoldiMate 等资本雄厚的先驱,没能解决自动化演示与可靠家庭实用性之间的差距, 最终失败。更糟的是,Memo 原型目前每台制造成本约 $20,000,而未公布的目标零售价低于 $10,000,客户获客回本周期和经常性订阅毛利都高度不确定。[CU012, CU013, CU014]
6.5 展示项
07风险
7.1 管理层风险排名与论点破坏层级
Sunday 从学术模仿学习突破走向消费家庭机器人,风险集中在物理执行、单位经济学和消费者信任上。风投支持者已通过超额认购的 Series B 为 Sunday 提供 $200 million 融资,公司估值达到 $1.15 billion;但商业成功要求把一台重 170 磅的电动机器部署进不可预测的私人住宅。 实验室级演示无法覆盖住宅环境的组合式混乱:家具布局、光线变化和儿童互动,都会打断自主任务完成。2026 年底的早期 Beta 部署涉及约 50 个家庭, 一旦出现未受监控的失效模式,公司就会暴露在严重的产品责任和声誉损害之下。Sunday 仍完全没有收入,公开披露的商业 ARR 和客户合同均为零; 如果量产长期无法把零售价压到 $10,000 以下,在大批量发货前就可能触发下轮估值或严重资本约束。[CR001, CR004, CR005, CR012]
7.2 监管合规、产品安全与家庭隐私
将无系留自主操作部署到家庭生活空间,会让消费品安全和消费者监控监管面临前所未有的审视。Memo 采用低重心带轮底盘,通过被动物理稳定性在断电时保持直立, 并结合柔顺关节执行,让住户可以主动改变机器人的方向。然而,公开记录显示,Sunday 没有按照 UL 1740 或 ISO 13482 等机器人标准取得正式第三方产品安全认证, 也未披露覆盖核心操作硬件的已授权公开专利。持续使用摄像头和麦克风感知家庭环境,还会使公司暴露在州级隐私和窃听法规之下。 Sunday 通过拒绝入户人工远程操作、改用付费 Memory Developers 事先采集的演示数据,来缓解隐私反弹;但客户同意框架和家庭边缘数据安全, 仍是尚未验证的监管尽调重点。[CR002, CR010, CR011]
| 风险领域 / 规则 | 司法辖区 | 状态 | 可能性 | 严重程度 | 缓释策略 | 剩余暴露 | 尽调路径 |
|---|---|---|---|---|---|---|---|
| 消费品安全与人身伤害(CPSC / CPSA) | 美国(联邦) | 未认证原型机 | 中 | 严重 | 被动稳定底座、合规控制臂、柔软硅胶外壳 | 高:若夹持检测或制动失效,170 磅移动质量可能造成伤害 | 核实是否按 UL 1740 / ISO 13482 个人护理机器人标准完成 NRTL 测试 |
| 家庭隐私与室内视频录制(州隐私法 / 窃听法) | 美国(加州 / 多州) | 自行施加的政策 | 高 | 高 | 本地执行政策;不在客户家中训练远程操作 | 中:环境视频和麦克风信号可能引发窃听审查 | 审阅用户同意协议、边缘数据留存和隐私条款 |
| 硬件供应链与关键部件关税(301 条款 / EAR) | 美国 / 跨境 | 持续暴露 | 高 | 中等 | Mountain View 国内最终组装;传感器来源多元 | 中:专业电机和执行器易受贸易中断影响 | 审计一级供应商来源及部件海关归类 |
| 知识产权与自由实施 | 美国 | 无公开专利 | 中 | 中等 | 开源基础沿袭(ALOHA、UMI、ACT);专有数据管线 | 低:工业机器人专利持有人声称提出侵权索赔的风险 | 对移动操作领域开展正式自由实施专利检索 |
截至 2026 年 10 月,按严重程度排序;反映商业化前部署姿态。
[CR002, CR010, CR011, CR015]7.3 运营执行、长尾泛化与现场失效模式
运营层面的脆弱性,核心在于长流程家务中任务错误率的复合效应。Sunday 的 ACT-2 基础模型在 785 次自主叠衣尝试中取得了令人印象深刻的 99.1% 成功率,但该基准测试有意排除了袜子、衣物分类和抽屉收纳等复杂边界情况。独立工程评估显示,即使每次操作的错误率只有 1%,每天完成一百项家务动作,平均也会出现一次失败,表现为餐具破损、液体洒落或机构卡死。当前,实体原型仍需约 $20,000 的手工组装成本;要超越低批量原型阶段,还需要工装、精密定制双夹爪和执行器供应链,而这些环节尚未达到成熟的工业良率。如果 2026 年下半年 beta 试验期间无人监督下的现场可靠性下降,Sunday 将承担高昂的维护派遣成本,纯软件创业模式无法消化这笔开支。[CR006, CR007, CR015]
| 失效模式 | 可能性 | 严重程度 | 缓释成熟度 | 剩余暴露 | 未解决的尽调缺口 |
|---|---|---|---|---|---|
| 多步骤任务序列失败(复合长时程错误) | 高 | 严重 | 中等(ACT-2 洗衣任务 99.1%) | 高:需要 20 多步的家务会出现几何级可靠性衰减 | 非结构化家庭中、洗衣任务之外的自主恢复成功率 |
| 硬件磨损、执行器故障与现场维护负担 | 高 | 高 | 早期(手工组装原型) | 高:自定义双夹爪续航 4 小时,易出现机械疲劳 | 平均故障间隔时间(MTBF)及每台现场服务成本 |
| 视觉语言传感器遮挡与边缘场景感知下降 | 中 | 高 | 中等(Diffusion Policy 与 VLM) | 中:透明玻璃器皿、光照变化和蒸汽会干扰感知 | 在杂乱厨房环境中装载洗碗机的独立验证 |
| 家庭障碍物干扰(儿童、宠物、移动杂物) | 高 | 中等 | 先进(柔顺机械臂控制) | 低:被动底座防止倾倒,但机器人会停滞并需要人工清障 | 通道或物体受阻时任务放弃的频率 |
按严重程度排序;评估实体硬件和软件自主性方面的运营脆弱性。
[CR002, CR006, CR007, CR015]评估 Sunday 主要运营、监管和财务风险暴露类别的严重程度及缓释效果。
评估综合了据报道的 ACT-2 基准、原型组装成本以及截至 2026 年 10 月的高管披露。
[CR003, CR005, CR006, CR012, CR013]7.4 资本密集度、合作伙伴依赖与商业可行性
消费机器人要同时承担前沿人工智能的算力开销和精密硬件制造的营运资金压力。Sunday 迅速扩充组织规模,员工已超过 100 人;为赶在量产前加快模型部署,工程团队扩大了三倍,研究团队扩大了四倍。然而,要达到低于 $10,000 的目标零售价,商业化必须把制造成本至少削减 50%,同时还要与资金充足的双足和轮式竞争者 Figure AI、Weave Robotics 竞争。家庭自动化的历史先例——包括 Seven Dreamers Laboratories 和 FoldiMate 破产——说明,机器吞吐速度慢或需要频繁人工介入时,消费者支付意愿会迅速崩塌。如果 Sunday 在 beta 项目期间无法证明自主运行可靠、单位经济性可行,公司在 2027 年寻求量产工装所需的成长资本时,估值将面临严重压缩。[CR005, CR008, CR013, CR014]
| 依赖类别 | 交易对手 / 生态 | 价值链角色 | 集中度 | 失败情景 | 严重程度 | 缓释策略 | 剩余暴露 |
|---|---|---|---|---|---|---|---|
| 合同制造与精密执行器 | 专业机器人制造商 | 部件供应与规模化生产 | 高(单一 / 双来源) | 良率瓶颈使规模化难以超越手工组装 | 严重 | 扩大内部组装;扩充 Mountain View 工程团队 | 高:没有规模制造,原型组装成本仍约为 $20,000 |
| 前沿 AI 与基础视觉模型 | 公共 VLM 与开放世界模型 | 感知和情境场景推理 | 中等(混合生态) | 模型 API 变动或计算成本上涨 | 高 | 垂直模型闭环;专有真实环境物理演示数据集 | 中:每台部署机器人推理计算成本高 |
| 数据采集基础设施 | Memory Developer 家庭网络 | 借助 Gloves 众包采集家务演示 | 高(专有网络) | 采集者流失或演示质量下降 | 中等 | 付费任务激励($30-$60/hr);标准化审核工具 | 低:已向 500 多个家庭分发超过 2,000 副手套 |
| 后期资本联合体 | Coatue、Benchmark、Bain Capital 与 Tiger Global | 成长融资与董事会治理 | 中等(顶级联合体) | 宏观下行导致后期硬件融资关闭 | 高 | $165M B 轮融资提供资金 runway 至 2027 年 | 中:若商业化延迟至 2027 年之后,将需要低价融资轮 |
按严重程度排序;反映硬件、AI 软件和融资方面的集中风险。
[CR001, CR005, CR010, CR013]7.5 展品
08估值
8.1 投资论点、立场与建议
Sunday 提出了一个颇具吸引力的消费级 Physical AI 理论论点:它认为,通过专有 Skill Capture Glove 捕捉可规模化的真实环境操作数据,就能跨过家庭机器人部署的经典障碍。Sunday 将人类示范硬件与轮式 Memo 平台共同设计,并采用 ACT-1 基础模型,于 2026 年 3 月以 $1.15 billion 估值完成超额认购的 $165 million B 轮融资。Coatue、Benchmark、Bain Capital Ventures、Fidelity、Tiger Global 和 Conviction 等顶级风险投资机构都认可这套数据优先架构。 尽管如此,反方论点仍然严峻:过去二十年,家庭机器人已吸收数十亿美元风险投资,却没有诞生一款具备商业可行性的消费产品。Sunday 尚无收入,进入的是一个未经验证的消费市场;Memo 当前的商业化制造成本是手工组装约 $20,000/台,而其期望零售价低于 $10,000。由于客户押金规模、账单转化率、保修负债和单位经济性均未披露,投资建议为继续研究,并附带高风险评级和未知估值立场,而不是给出虚假的精确判断。[CV001, CV002, CV003, CV004, CV005, CV006, CV011, CV012]
| 维度 | 评估 | 支持证据 | 决策含义 |
|---|---|---|---|
| 投资建议 | 继续研究 | 在 2026 年末 beta 启动前仍处于收入前运营;尚无商业账单或转化记录 | 在现场可靠性和定价转化数据出现前,暂不作出新的资本承诺 |
| 信心水平 | 高 | B 轮主要新闻稿、领投方董事席位及官方 beta 申请条款 | 对资本结构事实信心较高,但对商业估值坚持克制,不贸然下注 |
| 风险评级 | 高 | 未经验证的消费硬件需求、$20,000 原型成本及较高的家庭故障责任 | 需要严格的下行保护,并审查清算优先权 |
| 估值立场 | 未知 | $1.15 billion 投后估值,未披露收入运行率或客户毛利率 | 没有单位经济性或经核实的生产成本,不应为私人市场 headline 估值背书 |
| 目标回报 / 行动 | 持有 / 关注 Beta | 50 个家庭的 Founding Family Beta 将于 2026 年末启动,使用编号 Memo 设备 | 跟踪人工介入频率、硬件损坏事故及 beta 转付费转化率 |
由于截至 2026 年 10 月没有公开财务报表、商业收入或披露的单位账单,估值立场标为未知。
[CV001, CV002, CV003, CV004, CV005, CV006, CV011, CV012]逐步推进的尽调决策链,把资本规模、硬件成本、运营层面的验证与最终建议串联起来。
| 节点 / 连接 | 节点 / 起点 | 终点 | 说明 |
|---|---|---|---|
| 节点 1 | 融资:$165M B 轮,估值 $1.15B [capital-scale] | ||
| 节点 2 | 原型 BOM:单台组装成本约 $20,000 [hardware-cost] | ||
| 节点 3 | 部署验证:50 个家庭参与测试 [beta-deployment] | ||
| 节点 4 | 商业现实:尚未产生收入,也没有定价方案 [commercial-status] | ||
| 节点 5 | 风险因素:消费硬件行业二十年来的高失败率 [downside-risk] | ||
| 节点 6 | 决策:继续研究 / 立场未知 [diligence-conclusion] | ||
| 连接 1 | 融资:$165M B 轮,估值 $1.15B [capital-scale] | 原型 BOM:单台组装成本约 $20,000 [hardware-cost] | |
| 连接 2 | 原型 BOM:单台组装成本约 $20,000 [hardware-cost] | 部署验证:50 个家庭参与测试 [beta-deployment] | |
| 连接 3 | 部署验证:50 个家庭参与测试 [beta-deployment] | 商业现实:尚未产生收入,也没有定价方案 [commercial-status] | |
| 连接 4 | 商业现实:尚未产生收入,也没有定价方案 [commercial-status] | 风险因素:消费硬件行业二十年来的高失败率 [downside-risk] | |
| 连接 5 | 风险因素:消费硬件行业二十年来的高失败率 [downside-risk] | 决策:继续研究 / 立场未知 [diligence-conclusion] |
该顺序流程图概括了决定尽调建议的定性关卡。
[CV001, CV003, CV004, CV005, CV006, CV011]8.2 论点与反方论点的驱动因素
评估 Sunday 的长期商业价值,需要把核心技术差异化优势与执行脆弱性放在一起比较。正向投资逻辑建立在数据护城河和资本高效的数据采集模式上:向居家 Memory Developers 发放可穿戴手套,由他们收集日常家务示范,绕开成本高达数万美元的工业远程操作设备。Sunday 打通硬件、模型训练和持续车队评估的全栈能力,因此获得了蓝筹机构支持并完成超额认购融资。 相反,反方逻辑聚焦于消费机器人行业的历史高死亡率、未经验证的消费者支付意愿门槛,以及数据运营和专业工程人才持续消耗现金。Sunday 以按任务计酬的方式雇用 Memory Developers,每个获批数据小时支付 $30 至 $60,同时维持超过 100 名员工,但尚未披露 beta 到付费用户的转化计划或量产制造协议。在不受约束的家庭环境中证明自主车队的可靠性和家庭留存之前,私人估值标记无法由经营现金流支撑。[CV003, CV004, CV005, CV007, CV008, CV009, CV010, CV012]
| 支柱 | 论点 | 反论点脆弱性 | 哪些变化会改变判断 |
|---|---|---|---|
| 数据采集护城河 | 低成本 Skill Capture Glove 可在真实家庭中大规模采集多样化的家庭动作 | 人类演示数据不保证机器人在未测绘家庭中具备长时程自主可靠性 | 独立验证零样本泛化能覆盖 100 多个不受约束的客户家庭 |
| 硬件形态 | 带轮底盘配合被动稳定,避开双足跌倒风险并降低机械复杂度 | 无法爬楼限制了可服务的多层住宅总市场空间 | 证据表明单层家庭消费者愿意支付的价格超过硬件摊销成本 |
| 单位经济性与定价 | 目标零售价低于 $10,000,使 Memo 成为可触达的家庭奢侈电器 | 当前手工组装原型成本为 $20,000,没有巨大规模就会产生负毛利率 | 经审计的物料清单和合同制造协议将生产成本降至 $5,000 以下 |
| 运营现金消耗 | 超过 $165M 的 B 轮资本提供 runway,可停止演示并执行初始部署 | 远程 Memory Developer 每小时 $30-$60,加上 100 多名员工,会迅速加剧收入前现金消耗 | 披露商业订阅或预付定价模式,并实现正贡献毛利 |
| 市场可行性 | 1,000 多名候补名单申请者显示出对家庭杂务自动化的早期爱好者需求强劲 | 历史上的消费机器人烧掉数十亿美元,却未产生商业市场赢家 | 将候补申请者转化为不可退款的付费订单,并保持每日使用 |
依据截至 2026 年 10 月的公开公告、技术发布和招聘信息,评估论点与反论点支柱。
[CV001, CV003, CV004, CV005, CV007, CV008, CV009, CV010, CV012]8.3 情景分析、可比公司与尽调要求
鉴于 Sunday 尚无收入,DCF 模型缺乏经验输入,因此,情景分析和基于里程碑的持有估值才是合适的估值框架。在乐观、基准和悲观路径下,估值最敏感的变量是:Memo 能否从仅邀请 50 户家庭参与的 beta,转向商业可行的合同制造,而无需接受惩罚性下轮融资。Sunday 的目标零售价低于 $10,000,而初始手工组装成本为 $20,000,因此必须把硬件供应链规模大幅做大。 从可比市场看,消费机器人历史上充斥着资金充足但最终清算的公司;当代人形机器人同行则筹集了数亿美元,目标是工业工厂部署,而非无结构的消费住宅。打破核心论点的触发因素和眼下的尽调要求,集中在 beta 远程操作介入率、现场可靠性、硬件物料清单的规模化,以及 B 轮优先股条款的完整结构。[CV001, CV003, CV004, CV006, CV007, CV009, CV011, CV012]
| 情景 | 运营与商业假设 | 估值与回报逻辑 | 下行触发器与终止标准 |
|---|---|---|---|
| 乐观情景(20% 概率) | Beta 实现 99% 以上自主任务完成率;BOM 降至 $4,000;2027-2028 年以 $9,999 售出 10,000 台 | 估值扩张至 $3.0B-$5.0B,成为定义品类的消费 AI 硬件平台(3x-4.5x 回报) | 严重的家庭安全事故、财产损坏,或消费者因室内摄像头隐私问题反弹 |
| 基准情景(50% 概率) | Beta 暴露出频繁的边缘场景卡顿;Memo 需要远程操作协助;单价定为 $12,000-$15,000 | 估值维持在 $1.0B-$1.2B;需要追加融资为硬件重新设计提供资金 | 硬件制造报废率超过 20%;90 天内 beta 客户流失超过 40% |
| 悲观情景(30% 概率) | 制造成本无法降至 $18,000 以下;beta 家庭退出;消费者转化率停留在 10% 以下 | 估值压缩至低于 $300M 的清算 / 下轮融资估值(资本减值 75% 以上) | B 轮优先股堆叠的条款触发资本重组,或科技集团低价收购资产 |
这些概率和估值倍数,是基于可用融资基准对尚无收入的深科技消费硬件所作的风险投资综合判断。
[CV001, CV002, CV003, CV004, CV006, CV007, CV009, CV011, CV012]8.4 展品
免责声明
本报告仅用于支持尽调,不构成投资建议。私营公司的事实可能迅速变化,未披露的运营指标仍可能修订。
证据索引
| 编号 | 陈述 | 可信度 | 来源 |
|---|---|---|---|
| CO001 | Sunday was founded in April 2024 in Mountain View, California, by robotics researchers Tony Zhao and Cheng Chi to develop autonomous household robots. | 高 | SO001, SO002 |
| CO002 | Sunday emerged from stealth on November 19, 2025, introducing Memo, a wheeled personal robot designed to automate household chores including dishes, laundry, and tidying. | 高 | SO001, SO007 |
| CO003 | Co-founder and CEO Tony Zhao earned his EECS degree from UC Berkeley, pursued a Stanford CS PhD in robot learning under Sergey Levine and Chelsea Finn, led the ALOHA project, and worked at Google DeepMind and Tesla Autopilot. | 中 | SO002, SO008 |
| CO004 | Co-founder and CTO Cheng Chi completed his computer science PhD at Columbia University under Shuran Song, where he invented Diffusion Policy for robot manipulation and received two Best Paper Awards. | 高 | SO002, SO007 |
| CO005 | On March 12, 2026, Sunday closed an oversubscribed $165 million Series B financing round at a $1.15 billion post-money valuation, led by Coatue Management. | 中 | SO002, SO003 |
| CO006 | Coatue co-founder Thomas Laffont joined Sunday's board of directors in connection with the Series B round. | 中 | SO003, SO004 |
| CO007 | Prior to Series B, Sunday raised an initial seed round of approximately $35 million backed by Benchmark and Conviction, bringing lifetime equity funding to roughly $200 million. | 高 | SO002, SO007 |
| CO008 | Additional participants in Sunday's Series B round included Bain Capital Ventures, Fidelity Management & Research, Tiger Global, Benchmark, Conviction, and Xtal Ventures. | 中 | SO003, SO004 |
| CO009 | Sunday trains its foundation model ACT-1 using human motion data collected from over 2,000 Skill Capture Gloves deployed in real households rather than teleoperated robots or synthetic simulations. | 高 | SO001, SO002 |
| CO010 | Memo uses a wheeled mobile base with an adjustable vertical reach up to 2.1 meters, four-hour battery runtime, and passive mechanical stability when unpowered. | 中 | SO002, SO010 |
| CO011 | Sunday contracts remote Memory Developers across US households at $30 to $60 per hour to record daily routines using the Skill Capture Glove. | 中 | SO006 |
| CO012 | Sunday announced a 50-household Founding Family Beta launching in late 2026, targeting initial consumer shipments by Thanksgiving 2026. | 高 | SO001, SO003 |
| CO013 | Critics highlight that Memo's current prototype build cost of roughly $20,000 blocks consumer pricing, while real-home beta deployments create safety, liability, and generalization risks. | 中 | SO006 |
| CO014 | Sunday reported approximately 30 engineers and researchers in early 2026, with recruitment portals indicating 201-500 total affiliated personnel including remote contractor roles. | 中 | SO005, SO006 |
| CO015 | Public disclosures provide no audited revenue, ARR, consumer deposit totals, hardware gross margins, or unit economics ahead of the planned late 2026 beta. | 中 | SO002, SO008 |
| CM001 | Sunday emerged from stealth in late 2025 to develop Memo, a wheeled personal robot designed for domestic chores rather than industrial automation. | 高 | SM001, SM003 |
| CM002 | Sunday announced a $165 million Series B financing round in March 2026 at a $1.15 billion valuation led by Coatue Management. | 高 | SM001, SM006 |
| CM003 | Sunday had approximately 1,000 individuals on its waitlist following its emergence from stealth and funding announcements. | 中 | SM001, SM005 |
| CM004 | Building a single Memo robot by hand currently costs approximately $20,000 prior to volume manufacturing economies. | 中 | SM002, SM005 |
| CM005 | Sunday targets a long-term retail price under $10,000 per unit once Memo reaches scaled volume manufacturing. | 中 | SM002, SM005 |
| CM006 | Sunday focuses exclusively on everyday household chores such as dish loading, laundry folding, and tidying rather than commercial or industrial logistics. | 中 | SM003, SM004 |
| CM007 | Memo utilizes a wheeled mobile base and vertical telescoping spine rather than a bipedal humanoid form factor to improve domestic stability and reduce mechanical cost. | 中 | SM004, SM007 |
| CM008 | Sunday initiated a Founding Family Beta program accepting applications to deploy Memo to approximately 50 initial households starting in late 2026. | 高 | SM004, SM006 |
| CM009 | Sunday grew its engineering team three-fold, research four-fold, and data operations five-fold ahead of scheduled 2026 household beta deployments. | 高 | SM006, SM008 |
| CM010 | At an upfront price point between $10,000 and $15,000, initial demand will be restricted to affluent households and enthusiasts unless multi-year consumer financing is offered. | 中 | SM009 |
| CM011 | A $10,000 household robot amortized over a five-year period represents roughly $167 per month before financing fees and service margins, comparable to a premium home appliance. | 中 | SM009 |
| CM012 | Memo unit production costs near $20,000 create severe gross margin headwinds and block mass-market pricing until substantial supply chain scale is achieved. | 中 | SM010 |
| CM013 | Deploying autonomous manipulation hardware into unconstrained domestic environments introduces acute safety, liability, and product reliability risks during initial beta testing. | 中 | SM010 |
| CM014 | Sunday has not publicly disclosed paid deposit counts, gross billings, or post-beta pricing structures for Memo. | 中 | SM001, SM002 |
| CP001 | Sunday diverged from the humanoid form factor favored by competitors like Tesla and Figure by building Memo as a wheeled, vertical-lift machine designed specifically for domestic chores. | 中 | SP004 |
| CP002 | Memo stands 1.7 meters tall, weighs approximately 170 lbs, features a vertical reach of 2.1 meters via a telescoping spine, and achieves a 4-hour battery runtime on a wheeled base. | 中 | SP004 |
| CP003 | Memo incorporates compliant control and passive stability so the robot does not fall over when powered down or resist human touch, mitigating injury hazards around children. | 中 | SP003 |
| CP004 | Bipedal humanoid competitors such as Figure 02 and Tesla Optimus prioritize factory and enterprise logistics over domestic households due to high cost and bipedal balance complexities. | 中 | SP003, SP008 |
| CP005 | Sunday developed the $200 Skill Capture Glove to record authentic household motion from Memory Developers across hundreds of homes, bypassing $20,000 teleoperation rigs. | 高 | SP002, SP004, SP008 |
| CP006 | Sunday's ACT-1 and ACT-2 foundation models were trained on millions of real-home routine episodes without robot teleoperation, demonstrating a 99.1% laundry folding reliability in 785 internal trials. | 中 | SP001, SP003 |
| CP007 | Direct competitor Weave Robotics offers the wheeled Isaac 1 robot for $7,999 or $449 per month with a $250 deposit, utilizing human remote teleoperation whenever autonomous execution fails. | 中 | SP003 |
| CP008 | Sunday explicitly excludes customer-home video teleoperation from its product design, distinguishing its autonomous privacy model from Weave Robotics' remote operator streaming. | 中 | SP003 |
| CP009 | Building the prototype Memo unit by hand using custom parts costs Sunday approximately $20,000, with management targeting at least a 50% cost reduction in mass production. | 中 | SP003 |
| CP010 | Previous venture-backed laundry folding robot attempts failed catastrophically, including Seven Dreamers Laboratories which raised 10 billion yen before filing for bankruptcy in 2019, and FoldiMate which shut down in 2021. | 中 | SP003 |
| CP011 | Sunday announced an oversubscribed $165 million Series B round at a $1.15 billion valuation led by Coatue in March 2026 to transition from demos to beta home deployments. | 高 | SP005, SP007, SP009 |
| CP012 | Sunday recruits remote contract Memory Developers at $30 to $60 per hour to record chore video demonstrations while strictly requiring a full household and kitchen setup. | 中 | SP006 |
| CP013 | Sunday expanded its team to over 100 people by summer 2026, recruiting senior engineering talent from Tesla's AI infrastructure and Optimus teams to scale manufacturing and operations. | 中 | SP003 |
| CP014 | Consumer switching costs between specialized appliances and general-purpose home robots remain low, exposing early robotics platforms to commoditization if foundation models generalize across embodiments. | 中 | SP003 |
| CI001 | Sunday intends to commercialize Memo primarily through consumer hardware sales positioned as a high-end home appliance rather than an industrial machine. | 中 | SI002, SI005 |
| CI002 | Sunday has not announced public subscription pricing or recurring software fees for Memo fleet management or model updates. | 中 | SI001, SI003 |
| CI003 | Sunday targets a commercial retail price under $10,000 for Memo once production reaches manufacturing scale, with an expected range of $5,000 to $10,000. | 中 | SI002, SI005 |
| CI004 | Sunday is not yet selling Memo commercially and has not disclosed deposit terms or financing options for retail customers. | 中 | SI003 |
| CI005 | Building a Memo prototype unit by hand currently costs approximately $20,000 due to custom components and lack of economies of scale. | 中 | SI002, SI005 |
| CI006 | Sunday projects that large-scale manufacturing will reduce Memo unit production costs by at least 50 percent. | 中 | SI002 |
| CI007 | Industry analysts warn that current $20,000 per unit prototype costs block viable consumer pricing and gross margins until manufacturing is scaled. | 中 | SI006 |
| CI008 | Multi-step domestic chores compound robotic failure rates, meaning a 99% single-step policy success rate still produces an estimated 81.8% task success rate across a 20-step routine. | 中 | SI008 |
| CI009 | Sunday pays Memory Developers $30 to $60 per approved data hour to collect household manipulation demonstrations using its $200 wearable glove. | 中 | SI006, SI009 |
| CI010 | Sunday is entirely pre-revenue with zero recorded commercial sales or ARR as of late 2026. | 中 | SI003, SI006 |
| CI011 | Sunday is deploying approximately 50 uniquely numbered Memo units free of charge to Founding Family beta households in late 2026 for reliability research. | 中 | SI003, SI009 |
| CI012 | Sunday accumulated an organic waitlist of more than 1,000 applicants prior to making public commercial product announcements. | 中 | SI003, SI005 |
| CI013 | Independent analysts note that unvetted beta household deployments in late 2026 expose robotics startups to safety failures, liability risks, and negative publicity. | 中 | SI006, SI008 |
| CI014 | Coatue co-founder Thomas Laffont joined Sunday's board following the March 2026 Series B financing. | 高 | SI001, SI002 |
| CI015 | Sunday emerged from stealth in November 2025 with $35 million in prior financing from Benchmark and Conviction, bringing total capital raised to $200 million. | 高 | SI001, SI004 |
| CI016 | Sunday has expanded its engineering headcount by 3x, research by 4x, and data operations by 5x to support in-home robot deployment. | 中 | SI001, SI007 |
| CE001 | Sunday unveiled ACT-2 in July 2026, reporting that Memo folded laundry autonomously with a 99.1% success rate across 785 trials in unfamiliar homes without per-home fine-tuning. | 中 | SE001, SE003, SE004 |
| CE002 | Memo prototypes cost approximately $20,000 each to manufacture by hand, with Sunday targeting a production cost reduction below $10,000 through volume injection-molded manufacturing. | 中 | SE002, SE003, SE006 |
| CE003 | Memo is a wheeled domestic robot standing 1.7 meters tall and weighing 77 kilograms, featuring a telescoping spine reaching 2.1 meters, dual compliant arms, and passive stability against tipping upon power loss. | 高 | SE005, SE006, SE009 |
| CE004 | Sunday collects imitation-learning data through the Skill Capture Glove, a $200 wearable sensor device that mirrors Memo's hand geometry and is distributed to a remote crowd of Memory Developers. | 中 | SE001, SE002, SE007, SE009 |
| CE005 | Sunday's proprietary Skill Transform pipeline converts glove demonstration recordings into robot action space trajectories with a reported 90% conversion success rate. | 中 | SE002, SE006, SE009 |
| CE006 | ACT-2's laundry folding evaluation covered nine garment categories, achieving 100% success on shorts and polos, 99.0% on T-shirts, and 94.7% on lightweight blouses with an average quality score of 4.72 out of 5 stars. | 中 | SE003, SE004 |
| CE007 | Memo's demonstrated skill repertoire includes table clearing, espresso preparation, and loading dishwashers, but multi-step tasks across complex kitchen environments remain less reliable than laundry folding. | 中 | SE001, SE002, SE003, SE004 |
| CE008 | Sunday closed an oversubscribed $165 million Series B financing round in March 2026 led by Coatue Management at a $1.15 billion valuation to transition from demonstrations to home deployments. | 高 | SE006, SE008 |
| CE009 | Sunday's ACT-2 experiments demonstrated that increasing pretraining data scale eliminated the out-of-domain generalization gap from 82 percentage points at 0% pretraining to 0 points at 100% pretraining. | 中 | SE004 |
| CE010 | Memo utilizes a wheeled mobile base rather than bipedal legs to eliminate fall hazards, reduce balance compute overhead, extend battery life to 4 hours, and lower structural manufacturing complexity. | 中 | SE002, SE003, SE005, SE009, SE010 |
| CE011 | Sunday plans to initiate an in-home beta deployment for Memo with Founding Family households in late 2026 targeting deliveries around Thanksgiving 2026. | 高 | SE001, SE005, SE006, SE008 |
| CE012 | Sunday proposed the Solve evaluation standard to benchmark robotic capabilities by requiring verified autonomous success rates in unseen environments without site-specific human tuning. | 中 | SE001, SE003, SE004 |
| CE013 | Memo employs compliant control and a low-center-of-gravity rolling chassis to ensure that human contact deflects arms safely and power cutoffs do not cause sudden structural collapse. | 高 | SE003, SE004, SE005, SE006 |
| CE014 | ACT-2 demonstrated single-demonstration supervised fine-tuning, successfully acquiring new folding methods from a single human example and generalizing the technique to unseen garments. | 中 | SE004 |
| CE015 | Co-founders Tony Zhao and Cheng Chi founded Sunday in 2024 by synthesizing Stanford research lineages in ALOHA, Mobile ALOHA, ACT, Diffusion Policy, and the Universal Manipulation Interface. | 高 | SE002, SE003, SE004, SE008 |
| CU001 | Sunday targets residential consumer households seeking automation for routine domestic chores rather than commercial facilities or industrial logistics. | 中 | SU001, SU009 |
| CU002 | Sunday anticipates that early adopters for Memo will predominantly comprise technology enthusiasts willing to tolerate hardware rough edges and actively teach tasks. | 中 | SU002, SU007 |
| CU003 | Sunday accumulated thousands of consumer applications following its stealth exit, including over 1,000 waitlist applicants by March 2026. | 中 | SU004, SU005 |
| CU004 | Sunday launched applications for its Founding Family Beta program in November 2025 with shipments planned to commence in late 2026. | 高 | SU003, SU008 |
| CU005 | Participation in Sunday's Founding Family Beta is invite-only and free for selected participants, featuring white-glove installation and direct engineering support. | 中 | SU006, SU007 |
| CU006 | Sunday distributed more than 2,000 Skill Capture Gloves across over 500 households to collect domestic training episodes before customer deployment. | 中 | SU004, SU006 |
| CU007 | Memory Developers participate as paid remote contractors recording household tasks rather than commercial paying robot customers. | 中 | SU010 |
| CU008 | Sunday has not placed a single commercial robot unit into a customer's home as of October 2026, remaining completely pre-revenue. | 中 | SU006, SU009 |
| CU009 | Sunday's initial Founding Family Beta program is designed around a limited pilot cohort of approximately 50 individually numbered Memo units. | 中 | SU003, SU004 |
| CU010 | Sunday architectures Memo for autonomous in-home execution rather than the remote human teleoperation model adopted by competitor Weave Robotics. | 中 | SU006 |
| CU011 | Competitors such as Weave Robotics have deployed units accumulating over 2,000 operating hours, while Sunday has yet to begin external customer pilot hours. | 中 | SU006 |
| CU012 | Independent robotics analysis indicates that a 1% per-step operation failure rate results in daily physical failures for domestic robots, threatening customer retention. | 中 | SU006 |
| CU013 | Historical domestic robotics startups including Seven Dreamers Laboratories and FoldiMate folded after raising substantial capital without establishing viable customer retention. | 中 | SU006 |
| CU014 | Building a Memo prototype unit currently costs approximately $20,000 by hand, with Sunday targeting a retail price under $10,000 once mass-manufactured. | 中 | SU004, SU006 |
| CR001 | Sunday completed an oversubscribed $165 million Series B financing in March 2026 led by Coatue Management at a $1.15 billion valuation, bringing total disclosed funding to $200 million. | 高 | SR007, SR009 |
| CR002 | Memo uses a wheeled base with passive stability and compliant arm control to prevent tipping during power loss and allow safe physical redirection in homes with children. | 高 | SR001, SR004 |
| CR003 | Early deployment of autonomous robots into residential households creates direct physical safety, property damage, and product liability exposures during unsupervised operation. | 中 | SR006 |
| CR004 | Sunday plans to ship individually numbered Memo beta units to approximately 50 selected founding families in late 2026 to evaluate reliability before commercial launch. | 中 | SR002, SR005 |
| CR005 | Hand-assembled prototype Memo units currently cost approximately $20,000 each, with Sunday targeting production cost reductions of at least 50% to achieve retail pricing below $10,000 at volume. | 中 | SR001, SR010 |
| CR006 | An independent robotics analysis indicates that even an individual operation failure rate of 1% produces daily failures when a domestic robot executes 100 daily household tasks. | 中 | SR008 |
| CR007 | Sunday reported 99.1% autonomous laundry-folding success across 785 attempts with ACT-2 in test homes, but excluded socks, underwear, sorting, and putting away from its declared solve scope. | 中 | SR002, SR008 |
| CR008 | Historical consumer laundry-folding robotics ventures such as Seven Dreamers Laboratories and FoldiMate raised substantial capital but collapsed prior to sustainable commercialization. | 中 | SR008 |
| CR009 | Sunday leadership stated that early adopters should expect rough edges in the 2026 beta as real-world home deployments feed training data into its flywheel. | 中 | SR005 |
| CR010 | Sunday relies on demonstration data collected via Skill Capture Gloves rather than in-home human teleoperation, distinguishing its data privacy architecture from competitors. | 中 | SR006, SR008 |
| CR011 | Public databases indicate no public patents granted to Sunday and no third-party product safety or robotics certifications disclosed for Memo as of late 2026. | 中 | SR003 |
| CR012 | Sunday remains pre-revenue with no publicly disclosed ARR, customer count, or gross margin figures as of October 2026. | 中 | SR003 |
| CR013 | Home robotics requires simultaneous capital deployment across frontier foundation AI model training, specialized physical component sourcing, and warranty support infrastructure. | 中 | SR010 |
| CR014 | Sunday expanded engineering threefold, research fourfold, and data operations fivefold following its Series B round, scaling overall headcount past 100 personnel by summer 2026. | 高 | SR009, SR011 |
| CR015 | Memo relies on complex multimodal Vision-Language Models and custom dual-gripper manipulation hardware that create single-point dependencies during volume ramp. | 中 | SR001, SR002 |
| CV001 | Sunday raised an oversubscribed $165 million Series B funding round at a post-money valuation of $1.15 billion in March 2026. | 高 | SV003, SV008 |
| CV002 | The Series B round was led by Coatue Management, with Thomas Laffont joining Sunday's board of directors. | 高 | SV003, SV008 |
| CV003 | Sunday previously emerged from stealth mode in November 2025 with $35 million in initial backing from Benchmark and Conviction. | 中 | SV002, SV004 |
| CV004 | Sunday's total lifetime private venture capital raised reached $200 million following completion of the Series B round. | 中 | SV002, SV008 |
| CV005 | Sunday remains pre-revenue with zero commercial revenue run-rate, disclosed customer billings, or production contracts as of late 2026. | 中 | SV001, SV003 |
| CV006 | Building an individual prototype Memo robot by hand costs approximately $20,000, compared to Sunday's aspirational target retail price under $10,000 at production scale. | 中 | SV005 |
| CV007 | Venture capital in domestic household robotics has absorbed billions of dollars across twenty years without establishing a single commercially sustainable consumer product. | 中 | SV003 |
| CV008 | Sunday pays distributed Memory Developers between $30 and $60 per approved data hour for home routine video demonstrations. | 中 | SV007 |
| CV009 | Sunday's Founding Family Beta program plans to deploy individually numbered Memo units across 50 early adopter households in late 2026. | 中 | SV001 |
| CV010 | Sunday accumulated a consumer waitlist exceeding 1,000 prospective households following its public Memo reveal. | 中 | SV003 |
| CV011 | The diligence stance on valuation is unknown and the recommendation is research-more due to unverified unit economics and absence of customer revenue. | 中 | SV003, SV005 |
| CV012 | Sunday faces valuation downside risk if pre-revenue operational cash burn exhausts Series B capital before contract manufacturing scales unit costs below consumer pricing. | 中 | SV003, SV005 |
| 编号 | 出版方 | 标题 | 引文 |
|---|---|---|---|
| SO001 | Sunday Robotics | Sunday Robotics | The helpful robotics company | Built for busy households, Memo works 24/7 to make your life lighter. Hand off your repetitive to-do’s, so you can focus on what really matters. |
| SO002 | Biped.News | Sunday Hit $1.15B Valuation Building a Robot for Chores | In April 2024, two robotics researchers moved into a Mountain View apartment and started building a home robot from scratch. Twenty-three months later, Sunday closed a $165M Series B at a $1.15B valuation, becoming one of the fastest robotics startups ever to reach unicorn status. |
| SO003 | GlobeNewswire | Sunday Raises $165M to Launch First Autonomous Robots by Thanksgiving | Sunday, the company building helpful home robots, announced it has raised an oversubscribed $165 million Series B funding round at a $1.15 billion valuation. The round was led by Coatue, with Thomas Laffont joining the board. |
| SO004 | AI2 Work | Sunday Robotics Hits $1.15B Valuation With Home Robot Memo | Sunday, the Mountain View-based robotics startup founded by Stanford PhD researchers Tony Zhao and Cheng Chi, announced on March 12, 2026 that it has raised $165 million in an oversubscribed Series B at a $1.15 billion valuation, officially earning unicorn status. |
| SO005 | SHORTList | Sunday - Curated Startup Jobs, Alerted ASAP | SHORTList | Sunday develops autonomous home robots, like their flagship product Memo, which assists users by performing household chores such as loading dishwashers and folding laundry. |
| SO006 | Simplify Jobs | Memory Developer @ Sunday Robotics | Beta households in late 2026 expose Sunday to safety failures, liability, and bad publicity. Demo-driven robotics collapses if ACT-2 stops generalizing; Sunday becomes an expensive research lab. |
| SO007 | Yahoo Finance | Sunday Launches Memo, the Robot That Actually Learns Your Home | Backed by $35 million from Benchmark and Conviction, Sunday unveils Memo, a robot trained on millions of episodes of everyday household routines, built for safety, stability, and practical home help |
| SO008 | AI Market Watch | Sunday Robotics - AI Startup Profile | AI Market Watch | Skill Capture Glove technology enables training robots on real human interactions at low cost ($200 vs $20,000 for teleoperation), proprietary ACT-1 foundation model trained on millions of household episodes |
| SO009 | Goldsea | Sunday Robotics Wants to Sell Your First Domestic Helper | Sunday Robotics founders Tony Zhao and Cheng Chi think they know why. Building the mechanical body of a domestic robot is hard, but it isn’t the real bottleneck. |
| SO010 | Humanoids Daily | A Wheeled, Domestic Robot That Learns From $200 Gloves | Standing 1.7 meters (5'7") tall and weighing 170 lbs, the robot moves on a wheeled base rather than legs. This design choice sacrifices the ability to climb stairs for increased stability, energy efficiency, and a run time of 4 hours on a single charge. |
| SM001 | TechCrunch | Humanoid robotics maker Sunday reaches $1.15B valuation to build household robots | TechCrunch | Sunday says it has raised $165 million at a $1.15 billion valuation in a Series B round led by Coatue Management. Other investors in the round include Tiger Global, Benchmark, and Bain Capital Ventures. |
| SM002 | Silicon Valley Investclub | Sunday Robotics — $1.15B · Tony Zhao, Cheng Chi | Current Price (Est.)~$20,000/unit (pre-scale); <$10,000 at volume |
| SM003 | AI Market Watch | Sunday Robotics - AI Startup Profile | AI Market Watch | Positioned as a consumer-facing home robotics pioneer, directly competing with emerging humanoid robotics startups but focusing on everyday household tasks rather than industrial or commercial use cases |
| SM004 | Yahoo Finance | Sunday Launches Memo, the Robot That Actually Learns Your Home | Starting November 19th, 2025, Sunday will accept applications for Memo... Founding Family Beta, launching in late 2026. Fifty households will become early adopters |
| SM005 | AI2 Work | Sunday Robotics Hits $1.15B Valuation With Home Robot Memo | Building a single unit by hand costs approximately $20,000 today, but Sunday has publicly stated a target retail price under $10,000 once manufactured at scale — positioning Memo as a premium home appliance rather than an industrial machine. |
| SM006 | GlobeNewswire | Sunday Raises $165M to Launch First Autonomous Robots by Thanksgiving | With this capital, Sunday moves to focus entirely on real-world deployment of its robot, Memo. The robotics industry has long been defined by impressive demos. But deploying autonomous object manipulation in a real-world home environment has never been achieved. |
| SM007 | Humanoids Daily | Sunday Unveils "Memo": A Wheeled, Domestic Robot That Learns From $200 Gloves | Memo is not a bipedal walker but a wheeled, vertical-lift machine designed specifically for the "messy reality" of domestic chores... Standing 1.7 meters (5'7") tall and weighing 170 lbs |
| SM008 | Sunday Robotics | Sunday's Series B: No More Demos | Sunday Robotics | The helpful robotics company | Public deployment starts with our beta program, rolling out to users this Fall. To deliver on this promise with quality and care, we’ve grown Engineering by 3x, Research by 4x, and Data Operations by 5x. |
| SM009 | Goldsea | Sunday Robotics Wants to Sell Your First Domestic Helper | At $10,000 or $15,000 paid upfront, the market would initially be affluent households and enthusiasts. Financed over several years, however, a domestic robot begins looking more like an appliance or automobile payment. |
| SM010 | Simplify | Memory Developer @ Sunday Robotics | Simplify | Memo still costs about $20,000 each, blocking consumer pricing and margins. Beta households in late 2026 expose Sunday to safety failures, liability, and bad publicity. |
| SP001 | AI Magazine / GlobeNewswire | Sunday Launches Memo, the Robot That Actually Learns Your Home | Unlike most in-home robots, Memo’s training comes from approximately 10 million episodes of genuine household routines, representing an unmatched level of data diversity, quality, and in-the-wild volume. |
| SP002 | Bain Capital Ventures | Home Robots Finally Arrive: Inside Sunday Robotics | Teleoperation rigs and robot twins often cost tens of thousands of dollars and require trained operators in controlled environments. Meanwhile, egocentric video and other lightweight data sources do not translate cleanly to robot control because the human and robot embodiments differ. |
| SP003 | Tomo Kagawa / Note | Sunday Robotics 'ACT-2' Full Analysis: Can Home Physical AI Generalize? | Direct competitor Weave Robotics announced that it will offer the wheeled home robot Isaac 1 for $7,999... or $449... per month. It can be reserved with a refundable $250 deposit, with plans to begin shipping from California in the fall of 2026 |
| SP004 | Humanoids Daily | A Wheeled, Domestic Robot That Learns From $200 Gloves | The reveal confirms the company's divergence from the humanoid form factor favored by competitors like Tesla and Figure. Memo is not a bipedal walker but a wheeled, vertical-lift machine designed specifically for the "messy reality" of domestic chores. |
| SP005 | ai2.work | Sunday Robotics Hits $1.15B Valuation With Home Robot Memo | The target: Memo, Sunday's wheeled home robot capable of clearing dinner tables, loading dishwashers, folding laundry, and even pulling espresso shots — autonomously, in real homes it has never seen before. |
| SP006 | Simplify Jobs | Memory Developer @ Sunday Robotics | Record and submit video demonstrations of household tasks such as folding T-shirts and arranging shoes. |
| SP007 | GlobeNewswire | Sunday Raises $165M to Launch First Autonomous Robots by Thanksgiving | The robotics industry has long been defined by impressive demos. But deploying autonomous object manipulation in a real-world home environment has never been achieved. |
| SP008 | AI Market Watch | Sunday Robotics - AI Startup Profile | AI Market Watch | Sunday Robotics differentiates through wheeled design vs humanoid competitors, focus on practical household automation with AI-driven learning, consumer-centric approach with $200 training gloves vs expensive industrial solutions |
| SP009 | Sunday Robotics | Sunday's Series B: No More Demos | We just raised an oversubscribed $165M Series B at a $1.15B valuation to stop demos and bring the first autonomous home robot to households this year. |
| SI001 | Sunday Robotics | Sunday's Series B: No More Demos | Sunday Robotics | The helpful robotics company | We just raised an oversubscribed $165M Series B at a $1.15B valuation to stop demos and bring the first autonomous home robot to households this year. Our round is led by Coatue, with Thomas Laffont joining the board. |
| SI002 | MLQ.ai | Sunday Robotics Raises $165 Million Series B at $1.15 Billion Valuation for Home Robot Launch | Building Memo today costs approximately $20,000 per unit using hand-assembled custom components without economies of scale. The company expects large-scale manufacturing to reduce costs by at least 50 percent, potentially bringing the final retail price to the $5,000 to $10,000 range. |
| SI003 | AI for Automation | Sunday's home robot just hit $1.15B — trained in 500 real homes | Sunday isn't selling Memo yet. The Fall 2026 milestone is 50 uniquely numbered beta units going to selected households for safety and reliability research. A broader commercial rollout timeline hasn't been announced. |
| SI004 | Yahoo Finance | Sunday Launches Memo, the Robot That Actually Learns Your Home | Backed by $35 million from Benchmark and Conviction, Sunday unveils Memo, a robot trained on millions of episodes of everyday household routines, built for safety, stability, and practical home help |
| SI005 | AI2 Work | Sunday Robotics Hits $1.15B Valuation With Home Robot Memo | Building a single unit by hand costs approximately $20,000 today, but Sunday has publicly stated a target retail price under $10,000 once manufactured at scale — positioning Memo as a premium home appliance rather than an industrial machine. |
| SI006 | Simplify Jobs | Memory Developer @ Sunday Robotics | Simplify | Memo still costs about $20,000 each, blocking consumer pricing and margins. Beta households in late 2026 expose Sunday to safety failures, liability, and bad publicity. Demo-driven robotics collapses if ACT-2 stops generalizing; Sunday becomes an expensive research lab. |
| SI007 | SHORTList | Curated Startup Jobs, Alerted ASAP | SHORTList | Sunday raised $165 million in a Series B funding round led by Coatue, achieving a valuation of $1.15 billion. |
| SI008 | Note (Tomoshi Kagawa) | Sunday Robotics 'ACT-2' Full Analysis: Can Home Robots Move from 'Demos' to '99.1% Reliability'? | Tomoshi Kagawa | If a robot performs 100 operations a day, and the failure rate for each independent operation is 1%, it will fail once a day on average... Moreover, for multi-step tasks, the success rates are multiplied. Even if each step is 99%, the probability of succeeding in all 20 steps is, in simplified terms, about 81.8%. |
| SI009 | Humanoids Daily | Sunday Unveils "Memo": A Wheeled, Domestic Robot That Learns From $200 Gloves | Notably, the beta program will be free for selected participants, though it is currently invite-only. This aligns with CEO Tony Zhao’s view that early adopters should be enthusiasts willing to tolerate "rough edges" and potentially help teach the robot new skills. |
| SE001 | Business Insider | Sunday Robotics says its robot can fold clothes it has never seen in unfamiliar homes | On Thursday, the Bay Area-based startup unveiled ACT-2, a new AI model that powers Memo, its wheeled home robot. Sunday said that Memo folded laundry successfully more than 99% of the time when tested in unfamiliar homes, and on garments it had not been specifically trained to handle. |
| SE002 | EmbodyVC | Sunday Robotics Company Profile: Founders, Funding, Consumer Robotics | EmbodyVC | Second, it built a scalable data-collection pipeline centered on a low-cost (reportedly $200–$400) wearable glove system, called the Skill Capture Glove, that mirrors Memo’s own hand and sensor configuration. Rather than have engineers teleoperate robots directly inside customer homes, Sunday recruited a distributed network of U.S.-based “Memory Developers” who wear the gloves to perform ordinary household tasks—loading dishwashers, folding laundry, tidying—in their own homes, generating training data that maps human movement directly onto the robot’s action space. |
| SE003 | Goldsea | Sunday Robotics Wants to Sell Your First Domestic Helper | Memo completed 778 successfully—a 99.1 percent success rate. |
| SE004 | note.com | Sunday Robotics 'ACT-2' Full Analysis: Can Home Robots Move from 'Demos' to '99.1% Reliability'?|香川友志 | Sunday reports that their home robot, Memo, performed 785 autonomous laundry-folding tasks across 9 types of clothing in actual homes not used for evaluation, without any per-home fine-tuning, and succeeded 778 times. The success rate was 99.1% (standard error ±0.3%). |
| SE005 | GlobeNewswire via Yahoo Finance | Sunday Launches Memo, the Robot That Actually Learns Your Home | Memo is built for homes, with safety and stability as priorities. Instead of a bipedal humanoid shape, Memo features a rolling base for greater balance and lower weight. In the rare event of power loss, Memo remains stable, avoiding the risk of falls. Its silicone-clad design is soft, approachable, and built to fit comfortably into family living spaces. |
| SE006 | AI2.work | Sunday Robotics Hits $1.15B Valuation With Home Robot Memo | Physically, Memo is a 1.7-meter, 77 kg torso-style robot on a wheeled base, with dual multi-DoF arms and custom dual grippers designed for fragile items like wine glasses and ceramic plates. Its silicone-clad, corner-free body is family-safe, and its passive stability design means the robot won't topple even if power is suddenly cut. Building a single unit by hand costs approximately $20,000 today, but Sunday has publicly stated a target retail price under $10,000 once manufactured at scale — positioning Memo as a premium home appliance rather than an industrial machine. |
| SE007 | Simplify | Memory Developer @ Sunday Robotics | Simplify | As a Memory Developer, you won’t just collect data—you will shape the “memory” of our robot and play a vital role in training the AI that powers it. Using provided hardware, you will record high-quality demonstrations of household tasks that directly influence how our robot learns to perceive and act in real environments. This is a flexible, remote, part-time role for individuals who are excited to contribute to the future of robotics. |
| SE008 | GlobeNewswire | Sunday Raises $165M to Launch First Autonomous Robots by Thanksgiving | By owning the full stack, from hardware design and manufacturing to the proprietary Skill Capture Glove, Sunday moves from data collection to model training and evaluation at record speeds. This vertical integration enabled Sunday’s most recent demo to achieve industry-leading capabilities in just three months. |
| SE009 | Humanoids Daily | Sunday Unveils "Memo": A Wheeled, Domestic Robot That Learns From $200 Gloves | The robot features a telescoping "Z-axis" spine that allows it to reach from the floor up to 2.1 meters (6'10"), enabling it to handle tasks ranging from picking up objects off the rug to reaching high cabinets. |
| SE010 | AI Market Watch | Sunday Robotics - AI Startup Profile | AI Market Watch | Sunday Robotics differentiates through wheeled design vs humanoid competitors, focus on practical household automation with AI-driven learning, consumer-centric approach with $200 training gloves vs expensive industrial solutions |
| SU001 | SiliconANGLE | Sunday raises $165M at $1.15B valuation to launch Memo household robot | Early deployments are expected to occur through pilot installations in households where the robots operate continuously while collecting additional operational data that informs further model training and system updates. |
| SU002 | WIRED | This Home Robot Clears Tables and Loads the Dishwasher All by Itself | Sunday plans to give Memo to beta testers next year. The pilot program will show how people respond to having a home robot that can do certain chores—albeit slowly, and perhaps not perfectly every time. |
| SU003 | Yahoo Finance | Sunday Launches Memo, the Robot That Actually Learns Your Home | Starting November 19th, 2025, Sunday will accept applications for Memo's Founding Family Beta, launching in late 2026. Fifty households will become early adopters, receiving individually numbered robots along with direct support and the opportunity to guide future capabilities. |
| SU004 | AI2 Work | Sunday Robotics Hits $1.15B Valuation With Home Robot Memo | Applications opened in November 2025, and the company already had 1,000 people on its waitlist as of the funding announcement. |
| SU005 | GlobeNewswire | Sunday Raises $165M to Launch First Autonomous Robots by Thanksgiving | Since emerging from stealth last November, Sunday has seen strong demand, receiving thousands of applications for its upcoming Beta program. |
| SU006 | Tomoshi Kagawa (note.com) | Sunday Robotics 'ACT-2' Full Analysis: Can Home Robots Move from 'Demos' to '99.1% Reliability'? | Sunday has not yet placed a single unit in a customer's home. |
| SU007 | Humanoids Daily | Sunday Unveils "Memo": A Wheeled, Domestic Robot That Learns From $200 Gloves | Notably, the beta program will be free for selected participants, though it is currently invite-only. This aligns with CEO Tony Zhao’s view that early adopters should be enthusiasts willing to tolerate "rough edges" and potentially help teach the robot new skills. |
| SU008 | Sunday Robotics | Sunday's Series B: No More Demos | Public deployment starts with our beta program, rolling out to users this Fall. |
| SU009 | AI Market Watch | Sunday Robotics - AI Startup Profile | Busy households seeking automation for daily chores and home management |
| SU010 | Simplify Jobs | Memory Developer @ Sunday Robotics | Using provided hardware, you will record high-quality demonstrations of household tasks that directly influence how our robot learns to perceive and act in real environments. |
| SR001 | AI Automation Global | Sunday Memo Robot — The $1B AI That Does Your Chores | AI Automation Global | Current hand-assembled units cost approximately $20,000 each. Through large-scale manufacturing, the company projects retail pricing at “at least 50% less” — putting the target price around $10,000 or below. |
| SR002 | Humanoids Daily | Sunday Robotics — company profile, robots and news | The open question is execution: whether beta hardware reaches homes on the stated timeline, and whether laboratory-grade reliability figures survive contact with 50 unsupervised households. |
| SR003 | Silicon Valley Investclub | Sunday Robotics — $1.15B · Tony Zhao, Cheng Chi | No publicly disclosed revenue, ARR, or profitability figures were found in the provided results. |
| SR004 | Yahoo Finance | Sunday Launches Memo, the Robot That Actually Learns Your Home | In the rare event of power loss, Memo remains stable, avoiding the risk of falls. Its silicone-clad design is soft, approachable, and built to fit comfortably into family living spaces. |
| SR005 | AI2 Work | Sunday Robotics Hits $1.15B Valuation With Home Robot Memo | Zhao has been explicit that the beta is not a polished consumer launch — early adopters should expect rough edges — but the data collected from these real-world deployments is central to Sunday's flywheel strategy. |
| SR006 | Simplify | Memory Developer @ Sunday Robotics | Simplify | Beta households in late 2026 expose Sunday to safety failures, liability, and bad publicity. |
| SR007 | GlobeNewswire | Sunday Raises $165M to Launch First Autonomous Robots by Thanksgiving | The round was led by Coatue, with Thomas Laffont joining the board. The team also welcomes Bain Capital Ventures, Fidelity Management & Research Company, Tiger Global, Benchmark, Conviction, Xtal Ventures, and others. |
| SR008 | Note.com | Sunday Robotics 'ACT-2' Full Analysis: Can Home Robots Move from 'Demos' to '99.1% Reliability'?|香川友志 | If a robot performs 100 operations a day, and the failure rate for each independent operation is 1%, it will fail once a day on average. Dropping a dish once. Spilling a drink. Dropping clothes on the floor. Bumping into furniture. |
| SR009 | Sunday Robotics | Sunday's Series B: No More Demos | Sunday Robotics | The helpful robotics company | To deliver on this promise with quality and care, we’ve grown Engineering by 3x, Research by 4x, and Data Operations by 5x. We just raised an oversubscribed $165M Series B at a $1.15B valuation to stop demos and bring the first autonomous home robot to households this year. |
| SR010 | Goldsea | Sunday Robotics Wants to Sell Your First Domestic Helper | Home robotics combines two businesses notorious for consuming capital: frontier AI and consumer hardware. Sunday must train expensive models while simultaneously buying components, designing production tooling, carrying inventory and building support infrastructure. |
| SR011 | SHORTList | Curated Startup Jobs, Alerted ASAP | SHORTList | Sunday raised $165 million in Series B funding, achieving a valuation of $1.15 billion in March 2026. |
| SV001 | Robotics 24/7 | VIDEO: Sunday launches Memo personal robot that ‘actually learns your home’ | In conjunction with the company launch and Memo announcement, Sunday has begun accepting applications for Memo’s Founding Family Beta, launching in late 2026. Sunday said 50 households will become early adopters, receiving individually numbered robots along with direct support and the opportunity to guide future capabilities. |
| SV002 | eWeek | Meet Memo, the Robot That Does Real Chores | Sunday emerged from stealth with $35 million in backing from Benchmark and Conviction, positioning Memo as a practical robot designed for homes, not labs or factories. |
| SV003 | Biped.News | Sunday Robotics Raises $165M, Hits $1.15B Valuation | Biped.News | The home robot dream has burned through billions of dollars in venture capital over the past two decades without producing a single commercially successful product. Sunday's founders believe the missing piece was always data, not hardware. With $165 million in fresh funding and a growing army of glove-wearing data collectors spread across American kitchens, they are about to find out if they are right. |
| SV004 | Yahoo Finance / GlobeNewswire | Sunday Launches Memo, the Robot That Actually Learns Your Home | Backed by $35 million from Benchmark and Conviction, Sunday unveils Memo, a robot trained on millions of episodes of everyday household routines, built for safety, stability, and practical home help |
| SV005 | AI2.work | Sunday Robotics Hits $1.15B Valuation With Home Robot Memo | Building a single unit by hand costs approximately $20,000 today, but Sunday has publicly stated a target retail price under $10,000 once manufactured at scale — positioning Memo as a premium home appliance rather than an industrial machine. |
| SV006 | SHORTList | Curated Startup Jobs, Alerted ASAP | SHORTList | Sunday raised $165 million in Series B funding, achieving a valuation of $1.15 billion in March 2026. |
| SV007 | Simplify Jobs | Memory Developer @ Sunday Robotics | Simplify | Compensation follows our pay-per-task model: $30 per hour of approved data during a 1-week screening period Up to $60 per hour of approved data for more complex tasks after screening |
| SV008 | GlobeNewswire | Sunday Raises $165M to Launch First Autonomous Robots by Thanksgiving | Sunday, the company building helpful home robots, announced it has raised an oversubscribed $165 million Series B funding round at a $1.15 billion valuation. The round was led by Coatue, with Thomas Laffont joining the board. |