一辆车,一周开 10 到 12 个小时。剩下的 150 多个小时,它停在车位上,占着一块地,慢慢贬值。
这是今天绝大多数私家车的真实状态。马斯克在和罗恩·巴伦(Ron Baron)的这场对话里,把这个数字翻了个面:如果车能自己开出去跑,一周 50 到 60 个小时,同样的制造成本,用处是原来的五倍。
巴伦资本创始人罗恩·巴伦是特斯拉的长期大股东,这场对话发生在 2022 年。两个人聊的范围很宽:两万美元的车、删零件、睡工厂地板、垂直整合、挖矿、70 倍市盈率。把这些散落的点串起来,会发现它们指向同一件事:特斯拉在重新定义一辆车到底值多少钱。
两万美元的车,先从成本倒推
罗恩·巴伦在对话里提到,马斯克想造两万美元一辆的车。
马斯克的回应比较克制。他说下一代车型平台还没有正式发布,具体细节不能讲,但确认公司预期会造出比现在的 Model 3 和 Model Y 更便宜的车。
我们确实预期会造出比 Model 3 和 Model Y 更便宜的车,但下一代车型项目还没有正式公布,所以我不能讲太多。(马斯克)
这个"更便宜"到底便宜到什么程度,可以用现有数据往回倒推。
Model 3 在上海的售价大约 3.8 万美元。有分析师算过这款车的毛利率在 32% 左右,扣掉之后,制造成本大约 2.5 万美元。而眼下这台 Model 3,在特斯拉自己的产品线里已经算是相对"老"的技术:它没有前车身一体压铸,连后压铸也没有,还没有自研的结构化电池包。
一台没用上最新制造工艺的车,成本 2.5 万美元。下一代平台把这些工艺全部用上,成本压到 2 万美元以下,有依据。解说者自己的估算更激进,认为接近 1.5 万美元。
所以两万美元更可能是一个成本目标,落到终端售价上,3 万美元左右更合理。马斯克早先公开提过的数字是 2.5 万美元。
这里有个容易被忽略的细节:马斯克说的是"造出更便宜的车",没有说"卖给消费者"。它也可能是 Robotaxi(无人驾驶出租车)。但从整场对话的上下文看,面向消费者销售的可能性更大。
再往下想一层。如果新平台真能把成本压到 1.5 万到 2 万美元,售价定在 3 万美元,这个价位的车在全球市场对应的是一个完全不同的客户群。它不是从 Model 3 的用户里抢人,而是把一批原本买不起电动车的人拉进来。这也是为什么"两万美元"这个数字在对话里被反复提起,它代表的是市场边界往外推了一大截。
车价降四成,等效门槛降八成
车价便宜下来,需求会怎么变,这件事比直觉复杂。
先看第一层。卖车之外,特斯拉还有 FSD(Full Self-Driving,完全自动驾驶)的订阅收入。一辆 3 万美元的车,假设毛利率 50%,卖出去赚 1.5 万美元。之后每年还有大约车价 10% 的 FSD 订阅费流回特斯拉。这笔订阅收入的利润率,比整车毛利高得多。
再看第二层,这一层才是真正的杠杆。
燃油车的使用成本里,油费和保养占了固定的大头。电动车把这块基本砍掉了。所以比较电动车和油车,不能只比标价,要比总拥有成本。
一个粗略的换算:
- 一辆 5 万美元的特斯拉,总拥有成本大约相当于一辆 3 万美元的燃油车
- 一辆 3 万美元的特斯拉,总拥有成本大约相当于一辆 5000 美元的燃油车
车价从 5 万降到 3 万,降幅 40%。但对消费者来说,等效的购车门槛从 3 万美元掉到了 5000 美元,降幅 83%。
能掏 5000 美元买车的人,比能掏 3 万美元买车的人多多少,不用算也知道。这就是为什么电动车降价带来的需求增长呈指数级,而非线性。
很多人在看特斯拉的销量预测时,会不自觉地把汽车当成普通消费品来建模:价格降一点,销量涨一点,画一条平缓的斜线。但电动车的成本结构和油车不一样,它的使用成本占比高、固定成本低,价格每往下走一步,撬动的都是另一个数量级的客户群。这条曲线是陡的。
删零件才是真护城河
对话里有一段,马斯克讲制造,讲得比讲电池还细。
他说特斯拉可能是最擅长汽车制造的公司。罗恩·巴伦在旁边补了一句:可能是有史以来最擅长的。马斯克也认了。
特斯拉可能是在汽车制造上做得最好的公司。(马斯克)
这句话听着像自夸,背后是一套方法。
马斯克的说法是,要把制造的要求和规格变得"不那么蠢",然后想办法删掉零件、删掉工序。他们在车里找到过一些零件和流程,存在的理由只是"以防万一",有些甚至是被意外加进去的。现在这些都在被一个个清掉。
从车上删掉的零件,数量已经非常可观。竞争对手很难做同样的事。
原因在节奏上。特斯拉一款车的设计能用十年甚至更久,这十年里团队一直在找办法减少步骤、减少零件,制造工艺被反复打磨。对手大约每五年就换一代车型,每次换代都要重新走一遍流程,没有足够时间把一个工序磨到极致。
所以那些说特斯拉"设计老旧"的评论,看错了地方。同一套外观用了十年,换来的是一套别人追不上的制造流程。零件越少,成本越低,装配越简单,产线速度越快。这是复利。
这里面有个反常识的地方:消费者看车,看的是外观和内饰,觉得没变化就是没进步。但制造业的进步往往藏在看不见的地方。一辆车少了二十个零件,买车的人感受不到,可产线上的节拍快了,良率高了,维修率低了,成本就下来了。积累十年,这个差距会大到对手没法用一次改款追平。
马斯克在整场对话里反复回到"删零件"这种听起来很无聊的细节,原因就在这里。它不性感,但它是所有价格的源头。
每周四万辆,和算错账的分析师
罗恩·巴伦说,特斯拉现在每周造 4 万辆车。
马斯克确认,公司公开说过的目标是年底达到每周 4 万辆。这个数字换成简单算术:4 万乘以 52 周,年化 208 万辆,约等于 210 万辆。
2022 年的产量在 140 万辆左右。从 140 万到 210 万,增幅正好 50%。
反过来推更有意思:就算特斯拉在 2023 年完全不扩建、不爬产,维持每周 4 万辆的速度,全年也会自动实现 50% 的增长。这是算术,并非预测。
但市场上有些分析师和基金经理,包括把特斯拉当作最大持仓的人,说特斯拉 2023 年能做 200 万辆。这个数字低于"不扩张也能达到"的水平。
解说者对这件事的评价很直接:指出这些人的数学问题,目的是降低他们在社区里的可信度,而非嘲笑。因为这些人的错误预测会扭曲投资者对特斯拉的预期,进而扭曲市场给出的估值。
接着罗恩·巴伦说,每周 4 万辆,相当于每 15 秒造一辆车。
马斯克点了点头,像是在心里快速验算,然后表示同意。
这道题是:7 天 × 24 小时 × 60 分钟 × 60 秒 ÷ 40000,答案是 15.12 秒。
他当时还在倒时差,状态并不好。这个细节说明的事很清楚:只要不出现黑天鹅事件或者重大灾难,特斯拉 2023 年超过 50% 的增长几乎没有悬念。
而市场当时的定价,连这 50% 都没算进去,更没算经营杠杆。解说者的判断是股价仍然明显低估。下一阶段的增长,主要取决于 4680 电池的出货。
这段对话的价值在于它把一个抽象的判断变成了小学算术。看产能、看周产量、乘一下,就知道增速的下限在哪。市场情绪每天都在变,但产线不会骗人。一个工厂每周能下线多少台车,是硬约束,也是硬承诺。
船不够用,交付摊平到整季
规模上来之后,新的瓶颈出现在物流。
特斯拉的量已经大到船和汽车运输车都不够用的程度。解法是把交付摊平到整个季度,而不是集中在季末冲量。这样反而能降成本,因为季末集中交付要付出额外的运输和人力溢价。
对话里也提到了自动驾驶卡车。2024 年有 5 万辆 Semi 的规划,其中一部分可能会用来在全国范围运送特斯拉自己。
马斯克还提到一个场景:如果你的位置在车辆交付路线的范围内,车可以自己开到你那里。
如果你正好在附近,车会自己开到你那儿。(马斯克)
这个想法听上去像科幻,但它背后的逻辑很朴素。一辆刚下线的新车,本来就要从工厂运到客户手里。如果它自己会开,这段运输就不需要司机、不需要板车、不需要装卸。交付成本被压掉一大块,而这部分成本在传统车企的账上一直是个负担。
被供应商逼出来的垂直整合
有一段历史,现在回看是特斯拉的关键转折。
美国的汽车公司生存概率本来就很低。特斯拉早期去谈供应商,对方不太把他们当回事,很多零部件根本拿不到。没办法,只能自己做。
这种被迫的垂直整合,当时是无奈,现在成了资产。
因为必须自己搞懂整条供应链,特斯拉最终设计出的车集成度更高:零件更少,成本更低,重量更轻,性能反而更好。这几件事通常是互相矛盾的,特斯拉把它们同时做到了。
对竞争对手来说,这种优势很难追。他们习惯了从供应商那里采购成熟模块,供应链是分散的,改一个设计要协调十几家厂商。协调成本本身就是一道墙。
这段历史还有个侧面值得说。供应商不认真对待一家新公司,是行业常态,因为他们的产能有限,优先服务大客户。特斯拉当年被拒绝,逼着自己建能力,结果反而绕开了所有传统车企都陷在里面的路径依赖。很多能力是被逼出来的,不是规划出来的。
睡工厂地板,和采矿这件事
马斯克解释过为什么那段时间睡在工厂地板上。
一部分原因很朴素:激励。工人看到 CEO 愿意睡在车间,会相信这件事真的重要,然后多使一把劲。在冲产能最艰难的那段时间,这种示范作用是实打实的。
解说者在这里加了一句调侃:你能想象玛丽·博拉(Mary Barra,通用汽车 CEO)睡在工厂地板上吗?
关于上游,马斯克这次的语气变了。
过去他谈到采矿业,说法是"如果需要,特斯拉会去做"。这次他说的是,特斯拉可能真的需要进入采矿业务,而且这种可能性正在变大。
解说者对这个变化的解读值得注意:这或许是个好信号。需要更多矿物,意味着公司的规模会比原先设想的更大。
精炼这块,特斯拉本来就在做。德州的锂精炼厂很快就会开工。
从"如果需要"到"可能需要",措辞上的这一点位移,反映的是产量规划的变化。电池产能往上走,锂、镍这些原材料的需求就跟着往上走,靠市场采购能不能稳住价格和供应,越来越没把握。自己下场,等于把成本曲线的一部分握在手里。
70 倍市盈率,看用哪个基数算
现场有观众提问:特斯拉的市盈率是 70 倍,奔驰只有 6 倍,而奔驰也卖电动车,凭什么?
马斯克的回答先撇清一件事:他不劝任何人买特斯拉的股票。他过去公开说过股价太高,也说过如果你承受不了波动,就别碰。
他还讲了一个反常识的点:股价高对特斯拉本身没什么好处。如果公司要回购股票,股价低反而更划算。
至于估值差距,他的解释是特斯拉在自动驾驶上领先太远,没有对手接近。
解说者接过这个话头,做了一层更具体的拆解。
第一,市盈率要看你用哪个盈利基数去算。70 倍是按当时的利润算的。如果把盈利拉到每周 4 万辆、年化 210 万辆的水平,市盈率其实只有 25 倍左右。而奔驰过去四个季度的盈利,跟特斯拉可能不相上下。
第二,两家的电动车生意质量完全不同。奔驰的电动车不赚钱,特斯拉的电动车非常赚钱。
第三,特斯拉还有一堆别的优势。解说者的感慨是,很多人到现在还没意识到,把电动车造出来并且造得赚钱,到底有多难。
这个对比值得多想一会儿。传统车企做电动车,往往是拿燃油车的利润去补贴电动车的亏损,为的是满足排放法规和市场份额。特斯拉从一开始就只做电动车,成本结构里没有燃油车业务可以腾挪,只能靠制造效率活下来。活下来之后,它的每一分利润都来自电动车本身,没有水分。
同一辆车,五倍价值
马斯克在对话里说,一辆自动驾驶车的制造成本和普通车一样,但它能创造的价值是五倍。
解说者顺着这个思路算了一笔账。
假设一辆特斯拉卖 5 万美元,成本 3.5 万美元,毛利 1.5 万美元。
现在这辆车因为能自动驾驶,价值变成 5 倍,也就是 25 万美元。成本还是 3.5 万美元。
毛利从 1.5 万美元变成 21.5 万美元。
21.5 除以 1.5,大约是 14 倍。
一家毛利率 25% 到 30% 的公司,卖的东西没变,成本没变,价值突然翻五倍,利润翻十四倍。这就是自动驾驶对估值的意义。
这件事还有个前提:所有已经卖出去的特斯拉,硬件都已经装好了。不需要重新造车,一次 OTA 升级就能解锁。
再往上叠加 Optimus(人形机器人),想象空间还要再放大一层。
解说者的结论是,70 倍市盈率并不离谱,因为它定价的是这个还没发生的未来。
这里可以换个角度看。传统车企卖车,是一次性交易,卖完就结束。特斯拉卖车,是把一台可以持续升级的设备交付给用户,后续每次软件更新都可能带来新的收入。这两种商业模式对应的估值方法本来就不一样。用制造业的市盈率去衡量一家软件和服务占比在上升的公司,得出的结论会系统性偏低。
路灯充电与僵尸末日
最后有个观众的提问挺有意思:能不能用路灯作为充电基础设施?
马斯克说他知道这个思路,英国已经有实际案例。
更常规的路径是超级充电站做成自给自足:顶上铺太阳能板,配电池储能,脱开电网也能工作。解说者在这里开了个玩笑,说这套配置在僵尸末日里也能用。
这场对话真正的信息量在哪
把上面这些拼起来,能看到一条完整的逻辑链。
制造成本往下走,靠的是删零件和垂直整合。车价往下走,靠的是成本下降。需求往上走,靠的是等效拥有成本比车价降得更快。产能往上走,靠的是同一套设计被反复优化十年。而这一切的上限,被自动驾驶重新定义了一次。
所以马斯克在这场对话里反复回到"删零件"这种听起来很无聊的细节,并不奇怪。一家公司的估值,最终由它把东西造得多便宜、多快决定。自动驾驶只是把这个便宜和快,乘上了一个更大的系数。
至于 2023 年能不能做到 50% 的增长,答案写在那道 15 秒的算术题里。
- 看特斯拉财报时,自己拿周产能乘以 52 算年化,别直接引用分析师给的整数,很多数字对不上算术
- 把 4680 电池的爬产进度当作下一阶段增长的先行指标,不要只盯交付总量
- 关注 FSD 订阅收入在总收入里的占比变化,这块的利润率远高于整车销售
- 盯住下一代低价平台的正式发布节点和定价口径,两万美元到底是成本还是售价,差别很大
- 把锂精炼厂和采矿业务的进展放进观察清单,这是成本曲线的上游变量
- 判断电动车需求时,用等效总拥有成本和油车对比,别只比标价
- 如果考虑持有特斯拉,先想清楚自己能不能扛住波动,马斯克本人都不建议承受不了的人买入
- 读制造业公司的材料时,多留意"删掉了什么",而不是"加了什么",成本和速度往往藏在减法里
0:00hi everyone and welcome to the test economist I thought it would be remiss not to discuss this interview with Elon and Ron Baron it's a great interview I would recommend watching it but I'm sure you already have I'll discuss the elements that are relevant to Tesla Ron mentions that Elon wants to make cars for twenty thousand dollars a piece mainly referring to the new platform of vehicles that Elon mentioned on the latest earnings call that would cost half the price of a model 3 to build for the record my mat has this closer to fifteen thousand dollars cost rather than 20. Elon says they have not formally announced their next car program so he can't go into specifics
0:37but he says they do expect to make cars that are more affordable than the current model 3 and model y to give you some reference currently the model 3 in Shanghai is for sale at around 38 000 some analysts are saying this even has a 32 margin implying it costs around twenty five thousand dollars to build currently and this is for lack a better word at least relative to how fast test the moves old technology not only does it not have a front casting but not even a rear casting or in-house structural battery packs which gives you some indication as to why I think the next platform will cost less than twenty thousand dollars but if we're already at
1:19thirty eight thousand dollars for an Antiquated model 3 then a thirty thousand dollar price tag for this new platform is probably reasonable bear in mind we were initially told twenty five thousand dollars I mean if we want to be specific he does say we expect to make cars that are more affordable than the current model 3 or why he doesn't officially say for sale to Consumers so it may just be a robo taxi but I would infer from the context that he is talking about for sale to Consumers remember the FSD profit will be significantly higher again than any gross margin on the sale of the car and that is what we've been talking about lately a part of Tesla's unlimited
2:00Demand right there a 30 000 car with perhaps a 50 margin and then probably another 10 of that going to Tesla every year on FSD subscription and like I've explained in the past that the demand for a Tesla or any electric car actually reducing in price becomes exponentially more elastic due to fixed costs of the competition when it comes to gas and servicing just as a rough example a 50 000 Tesla May compare to a similar cost of ownership of a thirty thousand dollar Ice car but a thirty thousand dollar Tesla compares to more like a five thousand dollar Ice car now think how many more people can afford a five thousand dollar Ice car than a thirty
2:43thousand dollar Ice car yet Tesla only dropped their price by forty percent yet the cost of equivalent ownership has dropped 83 percent Elon then goes on to talk about how big a factory autonomy is which he has mentioned many times before because it's such a big deal Elon says instead of a car driving about 10 to 12 hours a week they could instead drive for 50 to 60 hours again the five-fold increase in utility that he said many times before and that we need fewer parking lots and fewer cars in general so it's better for the environment Elon says that Tesla is probably the best at auto manufacturing which no one was expecting Elon sounds like he is
3:24being modest here when he says this and as a result Ron interjects and says probably in the history of the world right presumably including all manufacturing ever and Elon again modestly agrees his modesty seems sincere as in he doesn't want to brag about it but equally he's not going to lie about it either and admits that Tesla are that good then he talks about how you need to make the requirements and specifications of manufacturing less dumb and to try and delete parts and processes Elon talks about how they found parts and processes that were in there for just in case purposes or even by accident and now they're going through and removing them
4:04all he also gives some examples and they have deleted so many parts from the car I don't think the competition of doing this this is also one of the major advantages of a Tesla car now I get a lot of comments about how Tesla need a new design or their design looks old or dated but think what happens when you have the same design for a decade or even decades and how you can refine the manufacturing process over that time especially when you are continually looking for ways to improve it and reduce steps and parts compared to other manufacturers who will make a new version of this car every five years or so it's not feasible for them to continually refine the steps oh and for
4:43the record testers do not look dated they are Timeless design obviously the more parts and processes test to remove the costs reduce the manufacturing becomes simpler and the production speed increases Elon talks about why he was sleeping on the floor in the factory a lot of it was due to Simply motivating employees so the workers could see that the CEO was so motivated and hard-working that he'll sleep on the factory floor then they'll simply work that much harder and this is through the really tough times getting to volume production then the workers give it their all it was good leadership could you imagine Mary Barra sleeping on the factory floor
5:18all right lots of great conversation goes on about Twitter and SpaceX and I might do a video on Twitter sometime too but let's focus on Tesla Ron talks about logistics but starts by saying Tesla is making 40 000 cars a week Elon then confirms that he thinks he has said publicly that their aspiration is to reach 40 000 cars a week by the end of the year which I'm not entirely sure he has literally said but perhaps in other ways either way that is some Easy Math it equates to 2.1 million a year run rate well guess what people if we hit 1.4 million this year then 2.1 million is 50 more in other words if Tesla do no expansion or ramping and remain at their current rate
6:03the whole of 2023 then they'll still hit 50 production growth yet there are analysts and fund managers out there even some of which Tesla is their largest position who are saying that Tesla will do 2 million next year I don't say this to point out how bad these people's meth are I just like to point out and reduce the credibility from the authoritative people in the community as they distort investors and the Market's valuation on Tesla from the True Value and expectations Ron then says that 40 000 a week is producing a car every 15 seconds Elon then nods his head as if he is quickly working out if what Ron said was true and then agrees possibly Elon just
6:43worked out the equation of seven days times 24 hours times 60 Minutes times 60 seconds divided by 40 000 in his head and yes the answer was indeed 15 seconds and this is while Elon is also very tired and jet lagged in the interview so in case you didn't already work it out assuming no Black Swan events or any major catastrophes Tesla will certainly beat 50 growth in 2023. the market is not even priced in 50 on top of that the market has not priced in Tesla's operating leverage the stock still seems very undervalued but the next stage of growth is mainly dependent on 4680 sales in case I haven't said that lately then we talk about logistics and that
7:24Tesla are simply getting too big and there are not enough ships so motor transport available so instead they will smooth out deliveries throughout the quarter which will actually reduce cost everyone talks about autonomous trucks for deliveries and that the idea and Elon agrees right remember those 50 000 semi-tracks for 2024 well I think how many of them might be used to deliver Teslas around the country then Elon says if you're even in the area then the car will just drive to you something I've said in the past too they talk about the fact that a US car company has a very small chance of survival and that Tesla struggled to get suppliers as a result and ended up
8:01having to do a lot of these processes themselves and we just simply forced into vertical integration as a result suppliers wouldn't take them seriously this has now become an asset as Tesla had to understand the entire supply chain and as a result Tesla designed a more integrated vehicle that needed far fewer parts cost less weighs less has increased performance again going to be so tough for the competition Elon now says the Tesla may actually need to get into the mining business and it is looking more likely that this will be the case in the past Elon has only said Tesla will if they need to well it's starting to look like they need to perhaps this is a good sign and means
8:43they are going to be bigger than they thought and will thus need more minerals he also says Tesla will get into refining 2 of which we knew already and going to start lithium refinery in Texas soon we then get some audience questions and one of the questions is why should people invest in Tesla with a 70 times earnings ratio compared to Mercedes at six times earnings which also sells EVS Elon says he's not telling people to invest in Tesla and has said in the past that the price is too high or not to invest in Tesla if you can't handle the volatility at this stage a higher share price is of little consequence to Tesla in fact lower is better if they're doing
9:20a buyback Elon thinks that this is due to the fact that Tesla is so far ahead in autonomy and no one is even close yep that's what I've said Tesla will have this Monopoly for probably at least a decade but let's be realistic here Mercedes last four quarters likely have similar earnings whereas Tesla despite being 70 times earnings is actually only about 25 earnings when you compare it to the level they are today at around 40 000 cars a week Mercedes do not make profitable EVS test to make very profitable ones and all the other advantages Tesla have over EVS compared to anyone else it seems people still don't quite understand yet just how difficult EVS are to make
10:04but the mind-blowing part of elon's answer is when he talks about a self-driving car costing the same to build but becoming five times more useful then imagine what would happen if a company was doing 25 to 30 percent gross margins and suddenly that same thing was five times more valuable what
10:25would that do to the value of Tesla and the car now there are a few ways to interpret this and do math around this but let's just use a simple way if a Tesla is worth fifty thousand dollars and then it's worth five times as much I.E 250 000 but still only costs thirty five thousand dollars to build in both scenarios the profit or value goes from fifteen thousand dollars gross profit to essentially two hundred and fifteen thousand dollars or fourteen times as much profit or value I think this is roughly the mind-blowing aspect of what Elon talks about the value of autonomy also remember all Teslas already have this necessary hardware for this and then there's
11:14Optimus on top of that which is also hard to comprehend I.E a PE ratio of 70 is Justified due to the unbelievable potential there's another question then Elon is asked about the street lights being a potential source of charging due to the infrastructure I was happy to hear that Elon did already know about this and I'd seen articles of this being used in the UK it's good to see Elon is aware of everything going on in the industry but basically the supercharger stations will be self-sustained with solar and batteries and will even work in a zombie apocalypse Elon has a great sense of humor anyway it's always great to hear Elon share some of his thoughts thanks
11:53for listening please hit the Thumbs Up And subscribe you can follow me on Twitter and talk to me on patreon
资源与出处
- 视频在 YouTube 收看
- 档案silicon-valley.tech 档案页本页视频与文稿数据来自 Elon Musk Archive
- 精读本篇精读 Markdown 版