发布会·中文精读

Tesla Q3 2026 财报及问答网络直播

Tesla Q3 2026 Financial Results and Q&A Webcast

中文精读由 AI 基于原始视频与文稿改写,非官方翻译,可能存在偏差或错误;表述以原视频为准。说明

特斯拉 2026 年第三季度财报电话会上,马斯克给出的最有信息量的数字是 10 倍。

奥斯汀的 Cybercab 车队从上线到现在大约一个月,规模涨了 10 倍。他把每周累计自动驾驶里程的曲线形容成火箭入轨:基数不大,斜率吓人。

同一次电话会上,他还押了一个可以事后翻账的预测:只要一个国家的政府不拦着,特斯拉最终会承担这个国家的大部分载客里程。他特意补了一句,你可以回来查这句话。

这场电话会的信息密度不低。FSD 的全球化、Optimus 的量产节奏、Terafab 的资本开支、与 SpaceX 的合并传闻,每条都值得单独拆开看。下面按重要性依次说。

需求跑在产能前面,原因只有一个

三季度很强。马斯克开场先谢了整个团队,然后给了一个在汽车行业不太常见的判断:需求增长明显超过产量。

一家车企说需求大于供给,通常意味着要降价换量,或者产能爬坡卡住了。特斯拉这次的原因反过来了。马斯克把需求的第一驱动力归给了 FSD(Full Self-Driving,完全自动驾驶):

今天你花钱能买到的车里,FSD 没有对手。(马斯克)

这句话可以拆成两层。软件层面,FSD 是唯一在量产车上大规模跑起来的端到端自动驾驶系统。商业层面,它把一辆车的价值从一次性硬件销售,变成了硬件加持续订阅。后者对需求的拉动是复利的,用户体验过之后很难退回去。

产能这边,特斯拉在尽可能多造车、多交付。同时推进 FSD 在更多市场获批。欧洲已经批了好几个国家,接下来想要的是欧盟范围的统一批准。马斯克给的时间是几个月内,最晚不超过明年一季度。他自己也承认,这个时间不由特斯拉控制。

奥斯汀一个月 10 倍

Robotaxi 是这场电话会里数据最硬的一块。

完全自动驾驶里程在指数级增长,周环比速度极快。马斯克的原话是「很难夸大这一点」。奥斯汀的 Cybercab 车队规模比上线时扩大了 10 倍,累计里程的周曲线在他的描述里像火箭入轨。

他还给了个很具体的验证方式:去奥斯汀,街上就能看到成规模的 Cybercab 在跑。前不久的奥斯汀城市极限音乐节(Austin City Limits)期间,很多人已经看到了。

这里有个容易被忽略的结构性差别。Waymo 的模式是自己买车、自己改装、自己运营,扩张速度受制于车辆改装和单城运营成本。特斯拉的 Cybercab 是从产线直接下来的专用车型,规模化复制的边际成本低得多。马斯克说目前世界上没有第二家公司能在自动驾驶车辆的规模上跟特斯拉比,这句话放在一年前还值得争,放在现在,论据是那条周里程曲线。

他还明确了 Cybercab 的定位:主要做运输服务(transport as a service),绝大多数由特斯拉自己持有和运营。想买来自己开的,还是 Model 3、Model Y 和 Roadster 这些车型。这条线划得很清楚,特斯拉不打算把 Cybercab 当成消费品卖。

Semi、Roadster 和 500 万辆

三季度还有两件事:Tesla Semi 卡车开始生产,新 Roadster 做了演示。

马斯克对 Roadster 演示的评价是「可以说是有史以来最酷的产品演示」,并把它归功于特斯拉和 SpaceX 两个团队的协作。这个说法有背景,Roadster 二代项目拖了很多年,一直跳票,这次终于有了可看的东西。

产能目标上,他给出的数字是 2030 年前后年产 500 万辆。原话是「有不错的机会实现」。考虑到 2025 年特斯拉全球交付量在 180 万辆上下,五年翻近三倍,这个目标不算保守。支撑它的会是多条产线同时爬坡,包括 Optimus 那条。

卖光的 Megapack

能源业务这次给出的信号比往常更强。

Megapack 2027 年的产能几乎售罄,特斯拉在尽可能快地扩产。马斯克预期这块的毛利率会显著提升。

储能业务过去几年一直是特斯拉财报里被低估的一块。它的特点是订单可见度高、客户是公用事业公司和大型数据中心、合同周期长。2027 年卖光意味着未来两年的收入基本锁定,剩下的变量只有产能和毛利。

马斯克对整体状态的总结是:熬了很多年,业务一直比较平,现在开始进入有意思的增长期。这句话放在特斯拉身上,上一次出现类似语气还是 2019 年 Model 3 产能地狱结束的时候。

Optimus 的一万件事

关于人形机器人,这场电话会最重要的信息不在进展,在减速带。

Fremont 的高产量产线已经进入早期生产。但马斯克花了大段篇幅解释为什么不能更快。他把 Optimus 的成长曲线定义成一条 S 曲线,前半段的斜率很平,还会颠簸,原因是产品太新。

有一万件事必须同时做对,机器人的产出速度取决于这一万件事里最慢、最不可能的那一件。(马斯克)

年产 100 万台的产线,绝大部分设备已经落地 Fremont。但从设备到位到跑出年化 100 万台的节奏,也就是每周 2 万台,他认为至少还要一年,可能更久。

为什么?因为设计还在改。

他特别强调这一点,原因是外界经常问:既然零件都有了,为什么不直接造一大堆发出去。答案是要从零建起一整条全球供应链,还要有一座巨型工厂,每一环都得跑通。同时机器人得是安全的、可靠的、寿命够长的,不能一个月就磨坏。AI 这一侧也还有大量工作,要让它真正能处理开放式任务。

他给的时间点是明年年底。到那时候,机电设计、可靠性、使用寿命、AI 软件这几件事预计会同时收敛,也是 Optimus 开始向特斯拉以外的公司交付的时点。他同时给了一句预防针:

我不想压低预期,我们还有很多东西要走完,大概还要一年左右。(马斯克)

关于竞争,他的表述很强硬:就通用人形机器人而言,目前没有任何一家公司接近特斯拉。他还说,等到开始向客户交付的时候,如果谁家的机器人能跟 Optimus 相提并论,他会非常震惊。

机器人要先学会玩

训练数据是 Optimus 最难的一关,而且跟汽车完全不同。

特斯拉有数百万辆在路上跑的车,天然就是一个数据采集车队。Optimus 没有这个东西。马斯克说得很直接:要让 Optimus 拿到需要的训练数据,得先造出机器人本身。数量级是一万到两万台,让它们互相练习(self-play),同时建一个「Optimus 学院」,让机器人在里面练各种各样的任务,产生的数据再回流,形成一个越转越快的飞轮。

他用小孩打比方。孩子刚生下来摇摇晃晃什么都不会,慢慢学会走路,学会自己擦屁股(养过孩子的人都知道这是个大里程碑),然后学会读书,学会做各种各样的事。整个过程靠的是玩。

Optimus 也一样,必须在规模上「玩」,才能在真实世界里把任务做好。这也解释了为什么他反复强调要先造出一两万台:那批机器人不是产品,是训练基础设施。

负责 AI 的 Ashok 补充了一个发布计划:大约一个月后可能会放出一段 Optimus 视频。他特意说明这段视频的标准:不是罐头演示(canned demo)。市面上很多人形机器人的演示看着惊艳,实际是预先编程的确定性动作,真放到现实环境里基本跑不动。他要展示的是没有预编程的行为,是「你看到什么就能信什么」的那种视频。

最难的是那只手

被问到机电层面最大的瓶颈,马斯克的答案很具体:手。

难点在于人手会自我修复。我们的身体会自己长好,机器人不会。人的手指是软的,皮肤里布满传感器,既能捏起易碎的东西,又能承受反复摩擦。要造出一只既耐磨、又足够柔软、传感器在成千上万次重复动作后还不损坏的手,非常难。

做到最后你会发现,人的手真是个了不起的东西。(马斯克)

只有真正下场做硬件的人才说得出这种话。机器人行业的公开叙事里,大家谈的都是 AI 模型和自由度数量,很少有人谈材料和耐久性。而这两样恰恰决定了机器人能不能连续工作几千小时不出故障,也直接决定了它有没有商业价值。

除了手,还有 AI5 芯片的上车。他给的时间是明年 4 月前后。

AI5 换脑,明年四月

AI5 相对 AI4 是一个数量级的算力提升。马斯克说这对 Optimus 帮助很大,因为人形机器人要处理的自由度远高于汽车。

他的比较方式是:人形机器人要做的事,大概是汽车的 1000 倍。这个倍数很粗,但方向没错。汽车在受约束的道路环境里做有限决策,人形机器人要在开放环境里完成没有边界的任务。同样的芯片架构,后者的算力需求高出一个量级完全合理。

所以 Optimus 的能力收敛,要等几块拼图同时到位:机电灵巧度的改进、长期运行不磨损的可靠性、AI 软件的架构、AI5 芯片的集成。四件事都在并行推进,他因此把「能交付给客户」的时间定在明年年底。

第一批买家是 24/7 的危险活

商业化的路线上,马斯克给了一个明确排序。

首发客户是工业场景,具体是危险、重复、没人愿意干、而且需要 24 小时不间断的活。这类场景的价值密度最高,因为机器人不需要休息,不需要社保,也不会因为重复劳动受伤。他把重复性劳损(repetitive stress injuries)单独点了出来,在制造业和物流业,这是真实存在的成本项。

等产线规模起来、成本降下去,才会进入家庭市场。这个顺序跟当年电脑的路径有点像:先进入企业和机构,再进家庭。人形机器人的家庭场景对安全性、泛化能力、价格的要求都远高于工厂,先啃工厂是唯一理性的选择。

他还提到一件正在发生的事:特斯拉现在造的就是「手还没做到完美」的机器人。同时推进的是 Optimus 的训练、供应链打通、生产系统调试、设计迭代、AI 软件开发。全部并行,速度已经是人力所及的极限。

三样东西相乘,然后是递归

这场电话会里最有结构性的一个观点,是他给出的价值公式。

人形机器人的有用程度,大致等于机电灵巧度、AI 软件、AI 芯片硬件三者的乘积。他特意用了「相乘」这个词,因为这三样都在指数级改进。相加的话增长是线性的,相乘的话,三条指数曲线叠在一起,结果是超指数。

然后是递归。Optimus 会制造未来的 Optimus。

当你有递归、有乘法、有指数,增长会非常疯狂。起点可能很小,但会涨到离谱的水平。(马斯克)

他的推论是:某个时点,地球上的机器人数量会超过人类数量。

这个推论听起来像科幻,拆开看只是算术。全球人口大约 80 亿,汽车保有量大约 15 亿辆,人形机器人的潜在应用场景覆盖了所有体力劳动、部分服务劳动、以及大量家庭劳动。只要单台成本降到几万美元、可靠性达到工业级,需求的天花板会远远越过 80 亿台这条线,落在「买得起的人都想买几台」那里。剩下的是产能和时间问题。

一栋可能排进国家前二十的楼

Terafab 是这场电话会里最容易被低估的部分。

奥斯汀的研究性晶圆厂进展很快。明年上半年开始设备调试,5 月或 6 月尝试做出第一颗测试芯片。马斯克的原话很朴素:就是想看看我们到底能不能做出任何一颗芯片。

真正有意义的节点在 2028 年三季度,那时候的目标是量产某种有用的芯片。真正的规模化生产要等到 2028 年底。他补了一句:按晶圆厂行业的标准,这个速度是闪电级的。这话不算夸张,一座先进制程晶圆厂从动工到量产,行业惯例是三到五年。

资本开支的量级:到 2028 年初,设备投入不少于数百亿美元。长期看,可能超过 1000 亿美元。SpaceX 和特斯拉之间怎么分摊还没定。

他对这栋楼的估值判断是:Terafab 会成为世界上最有价值的建筑,其经济价值可以排进国家的前二十名。这个说法听着离谱,逻辑链却清楚:如果 Optimus 真的做到年产千万台,半导体供应就是唯一的硬约束,内存、封装、逻辑芯片,任何一环卡住都会限速。自己有产能,等于把命脉握在手里。

他还提到,CapEx 实际执行下来比预期略低一些。明年开始会更有意义,2028 年会到数百亿美元的级别。

政府是唯一的变量

关于 Robotaxi 的终局,马斯克给了一个可以被检验的预测。

除非被政府拦住,比如他们要保护本国产业,否则在任何允许特斯拉自由运营的国家,特斯拉最终会承担这个国家的大部分载客里程。你可以回来查这句话,我有信心。(马斯克)

他主动把政府列为唯一变量,这个表态值得留意。自动驾驶的扩张速度取决于监管批准,而监管批准在不同国家的政治考量完全不同。欧洲的 FSD 批准已经拖了很久,中国市场对外资自动驾驶的开放节奏也是变量。技术不是瓶颈,许可才是。

关于与 SpaceX 的潜在合并,他只说了一句:昨天该说的都说了,踩着舞步穿过雷区。没有更多信息。这类问题在电话会上通常不会被正面回答,能给出这句话已经算是表态。

时间表

把散落在各处的节点排一下,节奏感就出来了。

  • 未来几个月:欧盟范围的 FSD 批准,马斯克预期不超过明年一季度
  • 约一个月后:Optimus 非预编程演示视频
  • 明年上半年:Terafab 开始设备调试
  • 明年 4 月:AI5 芯片集成进 Optimus
  • 明年 5 月或 6 月:Terafab 第一颗测试芯片
  • 明年二季度:某项新业务的财务影响开始显现
  • 明年年底:Optimus 解决可靠性、寿命、AI 三项问题,开始向外部客户交付
  • 明年三、四季度:Robotaxi 业务在财务上成为大事
  • 2028 年三季度:Terafab 量产有用芯片
  • 2028 年底:Terafab 真正规模化生产,CapEx 达到数百亿美元级
  • 2030 年前后:年产 500 万辆

被问到某项业务的财务贡献时,管理层的回答是:本季度不会有明显影响,明年二季度开始能感觉到,明年三、四季度变成大事。另一处被追问时,回答是还处在早期,不清楚会怎么展开,明年一季度末或二季度会有更清楚的判断,潜力很大,但现在很难量化。

这种「有大潜力但说不清」的表态,在特斯拉的财报会上出现过很多次,事后回看,有些兑现了,有些拖了很久。区别在于有没有可观测的中间指标。

接下来盯什么

这场电话会给出的信息里,有些可以直接验证,有些需要等。给几条具体的观察建议:

  • 盯奥斯汀 Cybercab 的车队规模和每周自动驾驶里程。整场电话会里最硬的领先指标就是它,比交付量和毛利率都更早反映趋势。如果三个月后这个数字没有继续翻倍,Robotaxi 的扩张速度就要重新评估。
  • 把「欧盟范围 FSD 批准不超过明年一季度」记下来。马斯克自己说了这个时间不由他控制,它因此成了一个纯粹的外部验证点。批了,欧洲需求会有一波跳升;没批,就要看他怎么解释。
  • Optimus 视频发布后,重点看机器人做的是不是预编程动作。判断方法很简单:看它有没有在失败后自我纠正,看它面对没见过的物体时的反应。有纠错、有迟疑,才是真的端到端控制。
  • 记下明年 4 月 AI5 上车和明年年底对外交付这两个节点。前者决定 Optimus 的能力上限,后者决定它有没有商业收入。两个都推迟,人形机器人的时间表就要整体往后挪一年。
  • Megapack 2027 年售罄这件事,去财报里核对储能板块的毛利率变化。订单锁定之后,毛利率是唯一还能改善的变量,也是这块业务被低估最久的地方。
  • Terafab 的资本开支分摊方案值得等。SpaceX 和特斯拉谁出多少钱,直接决定特斯拉未来三年的自由现金流曲线,也决定这座「世界上最有价值的建筑」会不会变成资产负债表上的负担。
  • 把「特斯拉会在任何允许自由运营的国家承担大部分载客里程」这句话存下来。这条预测有明确的边界条件、明确的时间窗口,可以被证伪。两三年后回看,它要么是马斯克最准的一次判断,要么是他最脱离现实的一次。

最后提一句他自己的措辞。被问到这些目标有多大把握时,他说:

我们当然是在拉斯维加斯押注未来会来。(马斯克)

这句话既是自嘲,也是实话。特斯拉过去十年的每一次大跳,都是在结果还没出现的时候先押上去。区别只在于,这次押的筹码比以往都大。

第三方整理文稿,仅保留马斯克本人发言。已按语句合并成段落,并去掉 um、uh 等口头填充词。 文稿来源:The Motley Fool (transcription)。

Thank you. So the third quarter was very strong, and I'd like to give a big congratulations to the whole Tesla team for working hard to make it happen. Demand is growing rapidly for our vehicles. It's exceeding our production by a significant margin. And obviously, we're going to be ramping and we are ramping production meaningfully. The biggest driver for demand is full self-driving. And today, there's really no competitor for FSD in any car that you can buy. And to be clear, we are producing and delivering as many cars as we can, and we're pushing very hard for FSD approval in additional markets. We're now approved in many countries in Europe, and we're hoping for an EU-wide approval.

It's hard to predict, but maybe within a few months, but I would be surprised if it takes longer than Q1. But obviously, that's not a time frame we control, but we do anticipate being fully approved in throughout Europe and many other countries. We also began production of the Tesla Semi truck in Q3 and did the new Roadster demo, which was, I think, one of the most exciting events. I mean, arguably, the coolest demo of any product ever, and that's a testament to amazing work by the Tesla and SpaceX teams. So we've also been like laying the groundwork for a lot more manufacturing capacity. We're, I think, aiming for 5 million vehicles a year by approximately 2030.

And that's -- I think we've got a good shot of achieving that goal. And yes, so energy demand is very strong. We're actually almost sold out of all Megapacks for 2027, and we're ramping production as quickly as possible to meet demand. We also expect gross margins on that to increase meaningfully. So in a nutshell, after many years of sort of grind and things being fairly flat, I think we're now headed for a lot of fun and significant growth. Regarding Robotaxi, we're seeing the Robotaxi miles driven -- fully autonomous miles driven scale exponentially, meaning scaling very rapidly. And it would be hard to overstate this point.

It really is an extremely fast week-over-week growth in autonomous miles. And I don't think there's any company in the world that can match our scale with autonomous vehicles as we continue to ramp. The Cybercab fleet in Austin, for example, has grown by a factor of 10 since launch about a month ago. And the cumulative mileage is really -- when I look at the weekly chart, it's wow. I mean it looks like a rocket going to orbit basically is growing very fast. And I recommend if you want to try it out, come into Austin and you'll see Cybercabs in use at scale in Austin, as I think a lot of people saw with the Austin City Limits concert recently. So this is really going to a very good place.

Regarding Optimus, we've started like what I'd call early production on our high-volume line in Fremont. I want to elaborate on this point because -- and I've said this before that the growth of Optimus, it's an S-curve, but the initial part of the S-curve is -- has a shallow slope and is somewhat bumpy because it is such a radically new product. So the way to think about it is there's like 10,000 things that have to go right for production of the robot, and it will move as fast as the slowest least likely of the 10,000. But we do have the vast majority of a 1 million unit per line production line landed in Fremont right now.

But I would expect that it takes us probably at least a year, maybe a little longer to get to that annualized rate to get to 20,000 units a week. So this is just fundamentally because there's still parts of the design that are changing. Like a lot of people ask me about Optimus. That's why I think I've got to kind of sort of belabor this point because if people aren't -- don't sort of understand like new technology that well or volume production of complex new hardware that well, it sort of seems like, well, can't you just make a whole bunch of these things immediately and send them out.

And it's like, no, you've got to basically build up an entire gigantic global supply chain and have domestically a very, very big factory and everything has got to work. And there's still some things we need to tweak about the design because you need to deliver a robot that is safe and reliable and has long duration. So it can't, like, wear out after a month. And there's still quite a lot of work to do with regard to the AI and the robot for it to be truly useful for open-ended tasks.

But like I said, if -- I would expect that the AI, the electromechanical design and the reliability and duration -- usage duration, like the usage life of the robot, I expect -- I do expect these issues to be solved around the end of next year. So that's the time to expect when we're delivering Optimus to companies outside of Tesla. And I remain convinced that Optimus will be the biggest product on Earth and humanoid robots in general will be the biggest product on Earth. So Optimus is a very big deal, but I don't want to temper expectations that we still have a lot to get through, and it's about a year or so that will take us to get through all that stuff.

But as to the best of my knowledge, there is no one that is even close to Tesla with respect to a general purpose humanoid robot. Regarding the Terafab, the Austin-based research fab is progressing quickly. We'll begin commissioning machinery in the first half of next year. And hopefully, if things go well, we'll make our first test chip just to see if we can make any chip at all around May or June. So that's, I think, pretty exciting. And then we're -- our aspiration is to be making some kind of useful chip at volume around Q3 of 2028.

So, and as I said publicly, the Terafab, I think, will be the most valuable building in the world and will probably have an economic value that will place it in the top 20 of countries. So it's an epic project. And this will be an incredibly important asset, if not a necessary asset in order to scale, especially Optimus as we can address bottlenecks in semiconductors, whether it's memory, packaging, logic or whatever the case may be. With regard to a potential merger with SpaceX, I think I said everything that should be said yesterday, tap-dancing through a minefield there. And yes, that's it. In conclusion, we're super excited about our autonomy and robotics road map.

There's a lot of incredibly awesome stuff coming. And I'd like to once again thank the Tesla team for their strong execution and all of our supporters for joining us on this journey.

I mostly talked about this already, but it's not crowdsourcing is not the issue for Optimus. There's a lot of things we have to get right about the AI architecture that we're still working on. We're making good progress, but that needs to be sorted out. And really, in order for Optimus to get the training data we need, we actually need to make a lot of Optimi. So we need to make, I think, on the order of 10,000 or 20,000 bots and then have them do self-play and we're building up an Optimus Academy where Optimus can practice doing a very wide range of tasks. And then we can use that data in sort of a flywheel to make it better and better at these various tasks.

A big difference with Optimus versus the cars is we don't have this vast fleet of cars from which we can gather training data. So with Optimus, we have to bootstrap it by making the robots and having them do self-play, which is really how a human learns to navigate the world. That's -- you can see this when you have a toddler and they grow up and they -- initially, they're sort of wobbly and they don't know how to do things and eventually, they learn to wipe their own ass, which is a big milestone. As anyone's who has had kids knows. And then they learn to read and do all sorts of things and be able to do these things. And they learn this by playing.

And we actually need Optimus to play at scale in order to have it be really good at tasks in the real world. And Ashok, you may want to add to that.

Yes. I think we may be ready to release like an Optimus video in a month or so. I want to make sure that like when we release this video that it's not like a canned demo that -- like when you see these robot, humanoid robot demonstrations, some of them look very impressive, but they're not really using any kind of advanced AI or SI. I mean they're like preprogrammed deterministic tasks. And if you actually try to use that humanoid robot in the real world, you would have a lot of trouble. It wouldn't work basically. So I want to release a video that's where it's Optimus doing things that are not preprogrammed.

And I think that would be -- it's maybe a month or 2 for that to have a demo that you can believe what you see type of thing. And then there's another milestone that's, I think, important we want to get to, which is to integrate our -- the Tesla AI5 chip into Optimus, which if things go according to plan, will be in approximately April of next year. And that's a giant improvement in capability with an order of magnitude more sort of more operations per second. Like it's an order of magnitude more compute, AI5 versus AI4. And that will help Optimus a lot because the degrees of freedom that a humanoid robot has to deal with are far greater than what a car has to deal with.

So it is a bit of a struggle to squeeze the -- everything needed to operate with all the degrees of freedom of Optimus and the wide range of tasks. If you can think of a humanoid robot is going to do like 1,000x more things than a car would do. That's some very big number basically. So I think the additional compute of AI5 will make a big difference, and that comes in, hopefully in volume around April of next year. And that's why I think like all the pieces like the refinements in electromechanical dexterity, making sure Optimus can operate for long periods of time without wearing out, getting the AI software right and the AI5 chip integrated. All of these things kind of come together.

That's why I think about a year from now is when it's something that would be -- that you could ship to customers. And it would also be reaching scale in production. So that's probably the right timing to think about. In terms of the initial customers for Optimus, I think it probably would be things that are industrial applications at first, things that are like dangerous repetitive tasks, things that people really don't want to do. So -- and that you want done basically 24/7. So like around-the-clock stuff. That's going to be the highest value for Optimus in the beginning. As we scale production, of course, it would be something that people can buy for their home.

But the initial high-value use will be -- the highest value use, I think, will be kind of heavy industry, dangerous tasks or things that may cause repetitive stress injuries and that you want done 24/7. So that's the way to think about it.

I just answered that one basically.

No. I mean Cybercab is primarily going to be something that is offered as like transport as a service whereas our other cars would be more the cars that you'd want to own like the 3 and the Y and whatnot, and Roadster, obviously. But the vast majority of the Cybercab stuff will be essentially Tesla-owned and transport as a service.

Yes.

Super hard problem.

Yes. I wouldn't say it's not like going to move the needle this quarter. But if you say like maybe -- I think it starts to become noticeable probably in Q2 and then probably a big deal by Q3 or Q4 of next year.

I think it's still early days on that front. We don't know the full extent of how -- what that will -- how that will unfold. I think we'll have a better sense for that probably in maybe end of Q1 or Q2. But it does have a lot of potential. It's just hard to say how much potential right now.

Well, it's going to be some pretty significant capital. It's certainly in the tens of billions for Terafab. Although we've got to figure out like what the sort of relative split is between SpaceX and Tesla in terms of who pays for what at Terafab. But it's not going to be less than tens of billions in capital equipment up through early 2028. And then it will be -- it will probably be north of $100 billion long term.

The research fab.

Yes. That's where we're just figuring out how to work the equipment type of thing. But real production would be late '28, which is lightning fast by fab standards.

Yes. It's mostly the hand and just kind of the fine-tuning of the hand design and the fingers, especially just -- it's a hard problem because humans have self-repair, like our hands repair -- our body repairs itself, but robots don't. So you have to -- and our fingers are kind of squishy and have lots of sensors in our skin. It's hard to make a hand that is -- that can handle lots of wear and tear, but it's still squishy enough to pick up delicate objects and has all the sensors and doesn't damage the sensors when you do repeated actions. And there's a lot of little things that you realize actually what an incredible thing the human hand is at the end of the day.

So it's mostly from an electromechanical standpoint, the hands. There are other things that are improving as well throughout the robot, but that's the main thing. And then as I mentioned, upgrading the intelligence -- the onboard intelligence of the robot with AI5, which is a big step change in intelligence that will help a lot with the capabilities of Optimus. So I've said before that like humanoid robots, biggest product on Earth and its usefulness will be -- it's a rough approximation multiple of the electromechanical dexterity, the AI software and the AI chip hardware. And all 3 of those things are improving exponentially.

I think we used that word a lot, but -- and the value is roughly those 3 things multiplied by each other. And then the Optimus robots will build the future Optimus robots. So you have a recursive function. So when you have a recursive multiplicative exponential, things will really grow extremely fast. They may start from a small base, but they're going to grow to crazy levels. And I think at some point, there will be more humanoid robots on Earth than there are humans.

I mean I think we -- unless we are stopped by governments, which is certainly a possibility, like if they're trying to protect the domestic industry, I think Tesla will end up doing a majority of all passenger miles on Earth in any country where it is allowed to operate freely.

And you can come back and look at this prediction, but I feel confident about this prediction.

Well, that's what we're doing is we're building robots that do not have perfected hands, and we are making progress with Optimus training and ironing out all aspects of the supply chain and the production system. So that's -- we're doing all these things in parallel, like this giant production line in -- at our Fremont factory, massive supply chain and then fine-tuning the design and engineering simultaneously with the AI software. So all these things are happening simultaneously. And we're growing as fast as humanly possible here.

I don't think anyone is going to have -- and in fact, I'd be stunned if anyone has a robot that is remotely competitive with Optimus when we go -- when we start delivering it to customers, which is likely to be towards the end of next year.

Yes. We're definitely, as I say, in Vegas betting on the come.

It's not a massive impact right now. I think CapEx is actually -- it's actually ending up being a little lower than we anticipated. But it will start to become more meaningful next year and then noticeable like in the tens of billions in '28.

从现场发言里挑出的 8 句原话,附中文译文(AI 翻译,以英文原话为准)。点时间戳可跳到视频对应位置。

我们,我想,目标是到大约2030年达到每年500万辆车。而且——我认为我们很有机会实现这个目标。

We're, I think, aiming for 5 million vehicles a year by approximately 2030. And that's -- I think we've got a good shot of achieving that goal.

所以我们需要制造,我想,大约1万到2万台机器人,然后让它们进行自我对弈;我们正在建立一个Optimus学院,让Optimus练习执行非常广泛的任务。

So we need to make, I think, on the order of 10,000 or 20,000 bots and then have them do self-play and we're building up an Optimus Academy where Optimus can practice doing a very wide range of tasks.

就Optimus的初始客户而言,我认为一开始很可能是工业应用,比如危险、重复性的任务,那些人们真的不想做的事。

In terms of the initial customers for Optimus, I think it probably would be things that are industrial applications at first, things that are like dangerous repetitive tasks, things that people really don't want to do.

但Cybercab的绝大部分业务本质上将由Tesla拥有,并作为运输即服务来提供。

But the vast majority of the Cybercab stuff will be essentially Tesla-owned and transport as a service.

但真正的生产要到28年末,按晶圆厂的标准这已经是快得惊人。

But real production would be late '28, which is lightning fast by fab standards.

我认为到了某个时候,地球上的人形机器人数量将超过人类数量。

I think at some point, there will be more humanoid robots on Earth than there are humans.

我认为,在任何允许Tesla自由运营的国家,Tesla最终将承担地球上大部分客运里程。

I think Tesla will end up doing a majority of all passenger miles on Earth in any country where it is allowed to operate freely.

如果在我们开始向客户交付Optimus时——那很可能是在明年年底——有人拥有一个能与Optimus有丝毫竞争力的机器人,我会感到震惊。

I'd be stunned if anyone has a robot that is remotely competitive with Optimus when we go -- when we start delivering it to customers, which is likely to be towards the end of next year.

资源与出处

相关视频

同频道、同主题或相邻年份