10月21日，据公司消息，特斯拉透露其第三代人形机器人Optimus已做好大规模量产准备。业内专家认为，若进入量产爬坡，有望拉动执行器、减速器等零部件需求。特斯拉在10月19日提交的致股东年度报告材料中表示，Optimus第三代已为大规模量产做好准备。董事会主席罗宾·德霍姆在股东信中提到，公司正建设规模化交付所需的基础设施，包括在内华达州建设Semi工厂、搭建Optimus首批产线。特斯拉首席执行官埃隆·马斯克在社交平台表示，Optimus和Cybercab将构成特斯拉价值的绝大部分，并重申公司将把重心放在机器人与人工智能领域。
On October 21, according to company news, Tesla revealed its third-generation humanoid robot Optimus is ready for large-scale mass production. Industry experts believe mass production ramping could drive demand for actuators, reducers, and other components. Tesla stated in annual shareholder report materials filed on October 19 that Optimus Gen 3 is ready for large-scale mass production. Board Chair Robyn Denholm mentioned in the shareholder letter that the company is building infrastructure needed for scaled delivery, including constructing a Semi factory in Nevada and building Optimus's first production lines. Tesla CEO Elon Musk said on social platforms that Optimus and Cybercab will constitute "the vast majority" of Tesla's value, reiterating the company will focus on robotics and AI.
对上游企业来说，潜在增量主要落在执行器、减速器、传感器等机器人零部件。业内人士认为，若Optimus进入量产爬坡，相关零部件需求有望增加。业内专家表示，硬件可靠性、物理AI算法泛化能力、成本控制、行业合规标准及数据闭环建设，仍是人形机器人产业发展的核心约束。
For upstream enterprises, potential increments mainly fall on robot components including actuators, reducers, and sensors. Industry insiders believe if Optimus enters mass production ramping, related component demand is expected to increase. Industry experts said hardware reliability, physical AI algorithm generalization capability, cost control, industry compliance standards, and data loop construction remain core constraints on humanoid robot industry development.
这条消息和前面讨论的特斯拉Optimus周产数据对得上。前面讨论过特斯拉弗里蒙特工厂现在每周能产出数百台Optimus，大约是2026年第二季度产能的十倍，内部目标据称是到年底突破每周1000台，更长远的规划是得州超级工厂投产后达到每周近2万台。当时Automotive Manufacturing Solutions的报告提到，特斯拉已进入规模爬坡阶段，这次致股东年度报告材料的说法是第三代已为大规模量产做好准备，两个信息在时间上衔接，说明特斯拉的量产准备工作已经推进到了产线搭建和基础设施建设的阶段。
This news aligns with the previously discussed Tesla Optimus weekly output data. Earlier we discussed Tesla's Fremont factory now producing several hundred Optimus units weekly, roughly ten times Q2 2026 capacity, with internal targets reportedly exceeding 1,000 units weekly by year-end and longer-term plans near 20,000 weekly once Giga Texas comes online. At the time, Automotive Manufacturing Solutions reported Tesla had entered scaled ramping; this annual shareholder report material saying Gen 3 is ready for large-scale mass production connects in time, indicating Tesla's mass production preparation has advanced to production line construction and infrastructure building stages.
产业链预期升温这个说法，和前面讨论的物理AI概念领涨、减速器板块涨幅居前是同一个脉络。前面讨论过人形机器人产业链里，执行器、减速器、传感器这些核心零部件占整机成本超过60%，且具备刚需属性和确定性订单，业绩兑现最快。也讨论过雷赛智能已进入80%以上人形机器人整机厂家供应体系，法奥机器人做工业级人形力控臂，芯际无限做RISC-V芯片。这些公司分处产业链不同环节，但它们的共同点是，整机厂放量之前，零部件订单会先体现。特斯拉第三代Optimus做好量产准备，对上游零部件企业来说，意味着潜在订单的增加。
The "supply chain expectations warming" formulation follows the same thread as the previously discussed physical AI concept leading gains and reducer sector ranking among top gainers. Earlier we discussed that in the humanoid robot supply chain, core components including actuators, reducers, and sensors account for over 60% of complete machine cost, with rigid demand attributes and certain orders, realizing performance fastest. We also discussed Leisai Intelligence entering over 80% of humanoid robot complete machine manufacturers' supply systems, Fa'ao Robotics making industrial-grade humanoid force-controlled arms, and Xinji Wuxian making RISC-V chips. These companies sit at different supply chain links, but their common point is that component orders appear before complete machine volume. Tesla's Gen 3 Optimus ready for mass production means potential order increases for upstream component enterprises.
不过，业内专家提到的几个核心约束也值得放在同一张账上看。硬件可靠性、物理AI算法泛化能力、成本控制、行业合规标准及数据闭环建设，这五条约束和前面讨论的具身智能落地难是同一组问题。前面讨论过宇树科技创始人王兴兴说在充分标定的固定场景内任务成功率可以接近100%，但一旦环境或物品发生微小改变，执行效果便会断崖式下滑。也讨论过机器人数据采集员的工作状态，机器人要真正像人一样无须调试、适应各种环境，还需要三到五年。特斯拉Optimus做好量产准备，是硬件层面的进展，但物理AI算法泛化能力、成本控制、数据闭环这些软件和运营层面的问题，还需要在真实量产和真实使用中逐步解决。量产准备和量产兑现之间，还隔着交付、验收、客户复购这几道关。
However, the core constraints mentioned by industry experts also deserve viewing on the same ledger. Hardware reliability, physical AI algorithm generalization capability, cost control, industry compliance standards, and data loop construction—these five constraints are the same set of problems as the previously discussed embodied intelligence deployment difficulty. Earlier we discussed Unitree founder Wang Xingxing saying task success rates can approach 100% in fully calibrated fixed scenarios, but once the environment or objects change slightly, execution effects plummet. We also discussed robot data collectors' working conditions—for robots to truly operate without debugging and adapt to various environments like humans, another three to five years are needed. Tesla Optimus ready for mass production is hardware-level progress, but physical AI algorithm generalization capability, cost control, and data loops—software and operational-level issues—still need gradual resolution in real mass production and real use. Between mass production preparation and mass production realization lie delivery, acceptance, and customer repeat purchase hurdles.
