科技003：埃隆·马斯克的特斯拉无人出租车、Optimus等，与Cern Basher一起探讨（科技播客）

您正在收听TIP。

You're listening to TIP.

大家好，欢迎来到本周三的《无限科技》节目。今天与我一同参与对话的是塞恩·巴舍尔，他是关于特斯拉、人工智能、机器人技术及比特币未来最富洞见的专家之一。在这次对谈中，我们将深入探讨宏观图景：特斯拉如何从汽车制造商进一步拓展为AI与机器人领域的巨头，机器人出租车和卡车运输背后的经济学原理，以及为何Optimus人形机器人可能重塑劳动力成本结构。

Hey, everyone. Welcome to this Wednesday's release of Infinite Tech. Today, I'm joined by Cern Basher, one of the smartest voices out there on Tesla, AI, robotics, and the future of Bitcoin. In this conversation, we dig into the big picture. We talk about Tesla's further expansion from carmaker into AI and robotics powerhouse, the economics behind the robotaxi and robotrucking, and why Optimus humanoid robots could reshape the cost of labor.

我们还将解析Dojo系统的崛起与停运始末，探讨自动化驱动的通货紧缩如何与作为终极国库资产的比特币产生关联。这期节目精彩不容错过，话不多说，让我们正式开始。

We also covered the rise and the shutdown of Dojo and how automation driven deflation ties back to Bitcoin as the ultimate treasury asset. This is surely an episode you will not wanna miss. So without further ado, let's jump into the show.

您正在收听由普雷斯顿·皮什主持的《投资者播客网络》旗下节目《无限科技》。我们以丰裕思维与健全货币为视角，探索比特币、人工智能、机器人技术、长寿科技等指数级增长领域。跟随我们共同解码塑造未来十年的突破性技术，助您即刻掌握未来趋势。现在有请主持人普雷斯顿·皮什。

You're listening to Infinite Tech via the Investors Podcast Network hosted by Preston Pysh. We explore Bitcoin, AI, robotics, longevity, and other exponential technologies through a lens of abundance and sound money. Join us as we connect the breakthrough shaping the next decade and beyond, empowering you to harness the future today. And now here's your host, Preston Pysch.

各位听众好，欢迎收听节目。能进行这场对话让我无比兴奋，今天我们将探讨一些真正引人入胜的话题——而且有幸请到塞恩·巴舍尔本人。

Hey, everyone. Welcome to the show. I am super pumped to have this conversation. We're gonna get into some really fascinating stuff, folks. And no other than Sern Basher here.

他现任Brilliant Advice公司总裁兼首席投资官，横跨多个领域的顶尖专家。塞恩，欢迎来到节目。

He's the president and CIO at Brilliant Advice, just an expert in all of these different domains. So, Sern, welcome to the show.

你好普雷斯顿，非常高兴能参与这次对话，感谢邀请。

Hi, Preston. It's so good to be here. Thank you for having me.

很高兴你能来。我们就从这里开始吧，因为我们要讨论很多关于埃隆·马斯克的事情，特斯拉在做什么，可能还会涉及一点比特币。但我想先从他为自己建立的这个道场说起，基本上算是与英伟达的竞争。我不确定你是否会将其视为竞争。他在打造自己的芯片，有能力掌控整个技术栈，从用于训练的芯片到整合所有不同数据源，再到构建模型，一应俱全。

Awesome to have you. Let's start off here because we're gonna be talking a lot about Elon Musk, about what Tesla's doing, and maybe a little bit of Bitcoin in there as well. But where I wanna start is he had this dojo for his own, basically, competition. I don't know if you would frame it as competition to NVIDIA. So that he was building his own chips, and he had the ability to kinda own the entire stack, clear down, you know, from the chips that are doing the training to to collect of all the different data sources that he has to building the models and the whole bit.

但在八月份，他宣布放弃道场项目，将其关闭并彻底终止。这在整体布局中究竟会产生什么影响？作为特斯拉股东，这是否令人担忧？还是说，请谈谈你的看法？

But back in August, he announced that he's done with Dojo. He's shutting it down, and he's just banning it. What impact does that really kinda have on the grand scheme of things? Is this concerning as a Tesla shareholder, or just kinda give us your take on it?

是的。乍一看这确实令人担忧，因为他们放弃了一个投入多年的项目。我们曾对这个道场超级计算机的概念感到兴奋，它本可以成为出色的训练计算机，减少对英伟达系统的依赖。对吧？这听起来很棒，不必支付英伟达75%的毛利率。

Yeah. At first blush, it feels like a concern because they're taking something, you know, they've been working on for years, and we all got excited about this idea of this dojo supercomputer that would be this amazing training computer that would reduce the reliance on NVIDIA's systems. Right? That all sounds good, right? Not to have to pay NVIDIA's 75% gross margins.

但现实情况是，首先这证明了英伟达及其开发时间线的卓越。这些大型训练系统，埃隆显然能获得所需资源。黄仁勋似乎对埃隆的能力和快速建立数据中心印象深刻。对英伟达而言，这是个非常优质的客户。

But I think the reality is that, one, it shows how amazing NVIDIA is and their development timeline. Right? So these big training systems, Elon clearly is able to get the allocation that he needs. Jensen seems very impressed with Elon and his capabilities and in standing up these data centers very quickly. That's a really nice customer for NVIDIA.

但更重要的是，我认为特斯拉始终自主设计车载AI推理芯片——从硬件3到AI4，即将推出的AI5，最终到AI6。这些都是特斯拉设计、台积电代工的强大芯片。而现在至少AI5版本，特斯拉正与三星合作开发新型推理芯片。

But I also think that more importantly, it's probably this. That Tesla has always made its own AI inference chips that are essentially in the car. The hardware three, now AI four, soon to be AI five, and then eventually AI six. These are very powerful chips that Tesla has designed, that TSMC historically has made for Tesla. And they're now, for at least AI five, moving with Samsung to develop this new version of this inference chip.

所以埃隆意识到，构建模型是一回事，在车辆或机器人上高效运行是另一回事。嗯。最终，你需要数十亿个推理芯片。嗯。

So Elon recognizes that it's one thing to build the model. It's another thing then to run it efficiently on a vehicle or in a robot. Mhmm. And ultimately, you're going to need billions billions of inference chips. Mhmm.

而且事实证明，或许可以通过堆叠足够多的推理芯片来构建自己的AI超级计算机。因此我认为道场的重心已从最初方向转向推理领域。特斯拉清楚知道他们需要怎样的推理芯片，因为他们正在开发运行其上的软件。而英伟达并不了解特斯拉在这方面的具体需求。

And also, it turns out that maybe you can take enough inference chips and stack them together and build your own AI supercomputer with that. So I think Dojo has kinda shifted from what they were working on to now focusing on the inference side. And Tesla knows exactly what they want from the inference chip because they're building the software that's going to run on the chip. Yeah. Whereas NVIDIA doesn't know what Tesla's needs are in that regard.

是的。长话短说，我对他们转向Dojo的方向其实感到挺兴奋的。我认为我们将看到基于特斯拉自家推理芯片的全新Dojo系统。

Yeah. So to make a long story short, I'm actually kind of excited about the direction that they've pivoted to with Dojo. I think we're going to see a new Dojo system based on Tesla's own inference chips.

另外，Sern，如果我理解有误请纠正。我只是想为听众梳理这个概念，让他们理解‘推理’这个术语。就是说有个芯片接收所有输入数据，快速生成输出结果或判断其效用。比如对于汽车，它实时分析道路图像信息，这些输入数据进入运行着已训练好压缩模型的推理芯片，最终转化为有用输出——比如车辆该如何转向或是否该刹车。

And, Sern, correct me if I'm wrong. I'm just trying to frame this for the listener so that they understand this term inference. So you got a chip that's taking all the inputs, and it's quickly coming up with what the output or the usefulness of, like, what it's discerning. So for, like, a car, you know, it's looking at the imagery that's just going down the road. That input is then going into this inference chip that's running the model that was already generated and compressed and built, but then it goes into this inference chip so that it can become useful outputs of, like, how the car should be steered or whatever, whether it should be braking.

对吧？没错。那些来自数月或数年前训练好的模型的输出，这种生成输出的转换过程，可能就是你说的‘秘制配方’——以无人驾驶为例，但同样适用于人形机器人等任何场景。推理环节的关键在于输出质量和速度。

Right? Yep. That output from the model that's already been trained months or years ago or whatever, that conversion to produce the output is what you're saying might be actually the secret sauce specifically for, you know, driverless cars in this example. But it could be for anything, whether it's a humanoid robot. That inference step, it comes down to the quality of the output and the speed of the output.

如果为此专门设计推理芯片，那确实可能成为核心竞争力。你刚才想表达的是这个意思吗？

And if you got a specialized inference chip for that, that really might become the secret sauce. Is that kinda where you're going with your comment?

正是如此。想象一辆自动驾驶汽车——它绝不能依赖与某个中央数据中心的通信来获取转弯指令。

Very much so. If you think about you've got a car driving itself. Yeah. You cannot have that car communicate with some central data center somewhere for instructions on the next turn. Yeah.

必须实现毫秒级响应，明白吗？所以计算必须本地化。人形机器人同理，任何在现实世界执行AI推理的终端都是如此。

It has to be, you know, millisecond instant. Right? So you need that compute on the vehicle. Same thing with humanoid robots. Same thing with really anything else that's going to be running AI inference that's doing something in the real world.

可以保持与数据中心的连接，但绝不能依赖它来操控车辆或机器人。对吧？所以归根结底是能效问题。想想马斯克最近谈到的与微软竞争的话题。

You might have a connection to the data center, fine, but not to operate that vehicle or that robot. Yeah. Right? So it all comes down to energy efficiency. Think about too the topic that Elon has been talking about lately, this competition to Microsoft.

他提出的这个所谓'宏硬'概念，与微软完全相反。不是使用预先开发的软件，而是让AI实时运行软件。是的，对吧？要实现这一点，又需要非常强大的推理芯片来支撑这种操作。

This idea of what he's calling macro hard, the exact opposite of Microsoft. So instead of having pre developed software, you basically have AI run the software on the fly. Yeah. Right? And in order to do that again, you're going to need very powerful inference chips to be able to do that kind of thing.

当然，要在价格上与微软竞争，产品必须非常便宜才行。

And certainly, you're going to compete against Microsoft on price, it's going to be have to be very cheap.

对，对。这很有道理。整个发展速度太令人着迷了。

Yeah. Yeah. That makes a lot of sense. Yeah. This is fascinating, this whole rate.

我好奇的是，市场热议的这件事最让人震惊的部分——埃隆是如何说服黄仁勋提供这么多芯片，让xAI如此快速扩张的？我在想是否存在某种交易，比如'我要关闭Dojo项目，不会在这个领域与你竞争，你将独占鳌头'。

And I wonder how much of this the thing that the market was talking about that just kinda blew everybody's mind is how did Elon convince Jensen to give him so many chips and scale up x AI so quickly? I wonder if there was somewhat of a, hey. I'm gonna shut down the dojo. I'm not gonna compete with you in this space. You're just gonna be the winner.

'我会退出这个赛道。但话说回来，我明天就需要你提供这么多芯片。'不知道你是否听说过相关传闻？我记得Dojo项目关闭是在八月份，而xAI最后版本训练完成是在那之前。

I'm gonna step away from it. But by the way, I need this many chips from you, like, tomorrow. I wonder have you heard any types of rumors, you know, dating back? And I know that this dojo shutdown was in August was, like, when this happened and when he trained the last version of the x AI. It was prior to that.

但我在想，这到底是他决定不参与这个领域的竞争，还是他确实意识到自己实力不济。

But I wonder if there was some type of I'm not gonna play in this space or if he was just truly he knew he was shellac.

这是个好问题，普雷斯顿。我猜这是逐步演变的过程。他们可能意识到无法跟上英伟达的步伐，团队成员也开始陆续离开。

Yeah. It's a good question, Preston. My guess is that this evolved over time. I think they were probably seeing that they couldn't really keep pace with where NVIDIA was. Members of the team started leaving.

所以我认为事情变得复杂了。也许埃隆看到了另一条路径，再次通过推理端。如果你打造出这款惊人的推理芯片，就能将其扩展成训练超级计算机。是的，我不确定他是否真的对黄仁勋说过‘你这样做我们就会那样做’之类的话。

So I think it just became difficult. And I think maybe Elon saw perhaps another pathway to, again, through the inference side. And if you build this amazing inference chip, then you can scale that up Mhmm. To become a training supercomputer as well. So, yeah, I I don't know exactly whether he would have actually said to Jensen, you know, we will do this if you do that.

我不确定是否有必要。我想黄仁勋知道埃隆的能力。谁不想让埃隆·马斯克成为自己最大的客户之一呢？天啊。

I don't know if that was necessary. I think Jensen knows what Elon was capable of. And wouldn't you love to have Elon Musk as you're one of your largest customers? Oh my god.

没错。顺便说，我们刚读了《思考机器》这本书，讲的就是节目里的黄仁勋，太精彩了。

Yeah. By the way, we just read this book. It's called The Thinking Machine. It's about Jensen Huang on the show. It was amazing.

这家伙简直是运营者中的运营者。说到这个领域的竞争优势或护城河，很大程度上取决于生产线的执行力。埃隆经常这么说。看看特斯拉，我觉得人们低估了这点，因为很多人不理解复杂终端产品生产线的卓越运营。而这正是埃隆真正闪耀的地方——生产线执行。

I mean, that guy is an operator of operators. Speaking of which, when you look at, like, a competitive advantage or, like, a competitive moat in this space, so much of it just comes down to the execution of production line. You hear Elon say this all the time. And so when you look across Tesla, I would imagine, I think it's probably discounted because I think a lot of people don't have an appreciation for just operational excellence on a production line, especially one that's producing a complex end item. And that's where Elon really shines is just in the execution of these production lines.

我甚至无法想象他现在手头有多少这样的生产线，简直难以想象。这是你们会涉及的话题吗？或者你们认为这确实是埃隆相对于领域内其他竞争者的护城河？

And I can't even imagine how many he has, like, in his stack at this point. It's unimaginable. Is this something that you ever cover or that you get into or something that you think is in fact a competitive moat of Elon's versus some of the competition in the space?

确实如此。或许我们可以稍微拓宽下这个问题。我详细讨论过这个观点：埃隆几乎是以一己之力带领我们从旧体系过渡到新体系。

Yeah. Very much so. And maybe let's maybe broaden out the question a little bit here. I have talked about at length about this notion that Elon kind of single handedly is taking us from one system, this old system that we have, to a new system.

嗯。

Mhmm.

这其中有许多不同的层次，对吧？从他们的生产线来看，特斯拉的一项创新就是制造生产机器的机器。他们能够重新设计整条生产线。

And there's many different levels. Right? So it starts at kind of if you think about their production line, one of Tesla's innovations is building the machine that builds the machine. Right? They're able to sort of redesign the production line.

他们制造这些机器是为了更高效地生产车辆，其制造系统的效率之高，是其他公司难以企及的。

They build the machines so that they can make these vehicles more efficiently, and they've got this efficiency in their manufacturing system that really no other company can really achieve.

这种递归式的表述简直太疯狂了。是的，太疯狂了。在商业史上我们从未见过类似的事情。

The recursiveness of that is of that statement is just so insane. Yeah. So insane. Yeah. We've never seen anything like that in the history of business.

没错。特斯拉最惊人的地方在于，如果市场上没有现成的设备供应商，他们就自己造。然后其他企业竞相模仿。我们正从手工制造时代——手工制造也包括机器人——

That's right. The amazing thing about Tesla is if there's if there isn't a company out there that provides the equipment, they just build it themselves. Right? And then people try to copy them. So we're going from this world of manual manufacturing and manual can include robots.

转向这种超级自动化的制造系统。SpaceX在得克萨斯州巴斯特罗普的星链终端工厂就是例证，年产量我记得是1000万台还是更多。整个流程高度自动化，原材料从工厂一端进入，星链终端就从另一端产出。

It's sort of this hyper automated manufacturing system. And we've seen that with SpaceX when the Starlink terminal in Bastrop, Texas. I forget what the I think it's 10,000,000 terminals a year or something, or maybe it's more. But it's basically highly automated. The raw materials go into one side of the factory, and the Starlink terminal comes out the other side.

未来人形机器人也将采用类似制造模式。全球可能会有数十家工厂开始量产这类产品。广义上说，比如SpaceX正推动我们从单行星经济迈向多行星经济——虽然现在还处于非常非常早期的阶段，对吧？

We're going to see that same kind of manufacturing for the humanoid robots. And we're going to see probably dozens of factories around the world start to build these things at some point. But just broadly, SpaceX, for example, is going to take us from a single planet economy to eventually a multi planet economy. We're in early, early, early stages of that. Right?

是的。特斯拉能源正在引领我们从化石燃料世界转向可再生能源，这非常重要。嗯。对吧？涉及的领域实在太多了。

Yeah. The Tesla energy bringing us from this fossil fuel world to renewables, and that's very important. Mhmm. Right? There's so many different things.

想想The Boring Company，如今我们拥有的是二维城市交通网络，即地面道路。但最终，它将发展为三维立体交通。

If you think about The Boring Company, today we have a two dimensional urban transportation network, roads on land. Eventually, it's going to be three d.

你知道吗，Cerne？我上周就在纳什维尔，有人告诉我他们已经开始在纳什维尔为The Boring Company修建通往机场的隧道了。说实话我之前完全没听说过这事，可能是我自己没太关注这方面，但感觉像是人们看了演示后惊呼‘哇，这太有意思了’就立刻行动了。

And you know what, Cerne? I so I was literally in Nashville last week, and I had a person telling me that they are have already started work on The Boring Company on a tunnel in Nashville out to the airport. And this is something, you know, I had really haven't heard, and maybe that's my own lack of focus kind of in the area, but it almost seems like people saw a demo. They were like, oh, wow. That's really interesting.

但他们没意识到这些项目已经在多个城市实际推进了，这一切正在发生。

But they don't realize that they are actually moving out on projects in cities now, like it's already happening.

没错。拉斯维加斯有个大型项目在进行中，纳什维尔也是，还有其他几个地区也在筹备。目前还处于早期阶段，因为隧道掘进机的速度必须提升——现在简直慢得像蜗牛爬。嗯。

Yep. There's a big project in Vegas that's ongoing, that Nashville, and I think there's some other areas as well that are in the works. We're still in the early phases of this because the tunnel boring machine really needs to become faster. These things move at literally a snail pace. Mhmm.

实际上比蜗牛还慢。所以需要进一步创新，首先要解决人类不能待在隧道里的问题，机器工作时人在隧道里非常危险。理想状态是这台机器能持续掘进，并同步完成隧道建设。

Actually, slower than a snail pace. So they need to make some further innovation to make sure that, first of all, you can take humans out of the tunnel, very dangerous for humans to be in the tunnel when these things are working. But you want this machine that can continuously bore the tunnel and build out the tunnel behind it as it's going through.

嗯。

Mhmm.

The Boring Company在这方面取得了重大进展，这非常令人期待。试想，如果能把大量车辆交通转入地下，城市将会变得多么美丽宁静。对吧？这一天即将到来。

And The Boring Company has really made some great strides in that, so that would be very interesting. Just think about it. If we can put a lot of the vehicle traffic underground, just imagine how beautiful and quiet cities would be if you could do that. Right? That's coming.

我不清楚具体的时间框架。可能是在十年、十五年后，我们才能真正开始看到这方面的加速发展。嗯。从宏观角度来看，所有这些技术正推动我们进入一个能源成本将下降并趋近于零的时代。运输成本也将下降并趋近于零。

I don't know what the time frame on that is. It might be, you know, ten, fifteen years from now in terms of when we really start to see an acceleration in that. Mhmm. From a big picture standpoint, all these technologies is driving us to a situation where the cost of energy is going to come down and trend towards zero. The cost of transportation is going to come down and trend towards zero.

人工智能将使智力成本下降并趋近于零。对吧？而人形机器人——虽然我们还没讨论到——它将推动劳动力成本下降。嗯。长期来看会趋近于零。这些因素的结合将把我们从受能源、智力和劳动力严重限制的世界，带入一个这些资源都极度充裕的世界。

AI is going to cause the cost of intelligence to come down and trend towards zero. Right? And then the humanoid robots, we haven't talked about yet, but humanoid robots, it's going to drive down the cost of labor Mhmm. Trend towards zero over time. The combination of all that takes us from a world that is very limited and constrained by energy, intelligence, labor, right, to a world then that has an abundance of all those things.

如果我们只考虑能源——因为能源对一切都至关重要。是所有事物的基础。想象一下如果拥有无限能源，你能做到什么。是啊。

If we think about energy, just focus on energy because energy is so critical to everything. Everything. Yeah. Imagine if you had unlimited energy, what you could do. Yeah.

你可以为全球所有数据中心供电。可以毫无顾虑地为比特币网络提供能源，再也不用担心能耗问题。谢天谢地，那些反对声音已经基本消失了。对吧？但更重要的是你能实现其他真正重要的事。

You could power up all the data centers in the world. You could power up the Bitcoin network without any worry about using too much energy. Those arguments have kinda gone out the window, thankfully. Right? But you could do other things that are really important.

你可以随心所欲地进行海水淡化。世界上将不再有任何地方面临水资源短缺。如果你愿意，甚至可以在沙漠中培育森林。嗯。凭借无限能源。食物供应也会更加充足。

You could desalinate as much water as you wanted to. You would have no water shortage or scarcity anywhere in the world anymore. You could grow forests in the deserts if you wanted to Mhmm. With unlimited energy. You can the food would also be more abundant.

所以我们正处在从旧体系向新体系转变的关键时刻。而埃隆通过他旗下的多家公司，正在众多领域推动前沿发展。

So we are on the cusp of, again, moving from one system to another. And Elon, through his various companies, is pushing the forefront in many, many areas.

是啊。没错？我百分之百赞同你说的这些。但持怀疑态度的听众会说：是的，但这将彻底取代人类——在这个体系中人类如何创造价值并获得报酬？因为他们可能无法与经过优化的人形机器人竞争。

Yeah. Right? So I'm with you on all of that 100%. But the naysayer that would be listening to this is saying, yeah. But you're gonna completely displace humans as to, like, how do they add value and get paid in this system because they're not gonna be able to compete with an optimist robot humanoid robot potentially.

对吧？这就引出了全民基本收入的讨论，进而指向比特币之类的东西。但我觉得，人们听到这些时更大的担忧是，他们会在内心自问：我要失业了，然后我该怎么办？

Right? Which then comes up the UBI conversation, which then points to Bitcoin and, like, all these other things. But Yeah. I think the bigger concern for people when they hear all this is just I think they're looking internally and they're saying, I'm gonna lose my job. And, like, then what in the world am I gonna do?

因为我在这些人形机器人面前毫无竞争优势。这种对话该怎么进行？老实说，先生，我最常听到这种反应是当我兴致勃勃地和妻子讨论这些时，她总会回到这个观点，而我给不出满意答案。我真的不知道该如何回应。

Because I don't have a competitive advantage against these humanoid robots. How do you have that conversation? And honestly, sir, where I hear this the most is when I I get all excited about this stuff and I'm telling my wife, and my wife, like, always comes back to this point, and I don't have a good answer for her. I don't really know what to say.

就像从前大家都务农时，有人坐在屋里说：'我见过拖拉机，它会抢走我们的工作，我们该怎么办？'嗯。

Well, sitting in a house some time ago when everybody used to work in farms. Mhmm. And you come along and you say, I've seen a tractor, right, and it's gonna take our jobs. What are we going to do? Mhmm.

当时你也给不出答案。嗯。你无法告诉那个担心失去农场工作的人，拖拉机将以人类和耕牛永远达不到的效率接管耕地。关于人类未来去向和新工作岗位从何而来，答案并不明确。是的。

You wouldn't have had the answer. Mhmm. You wouldn't have been able to tell that person that was worried about losing their farm job and tractors taking it over and plowing the fields way more efficiently than humans ever could and oxen ever could. The answer's not clear in terms of what humans are going to do and where the new jobs will come from. Yeah.

对吧？所以这是其一。纵观人类历史，每次重大技术进步最终创造的工作岗位总是多于被取代的。想想电话发明时那些女性接线员，后来我们发明了自动交换机。

Right? So that's one thing. Throughout human history, every time there's been a massive technological advancement, we've always created more jobs than we displace. Right? Think about all the women that were switchboard operators with the telephone, and then we invent a automatic switchboard.

七十五万人很快失业，但他们在其他领域找到了工作。首先要认识到的是：前景是模糊不清的。我们无法预测新岗位会出现在哪里。互联网诞生时，谁能预料到你我会从事现在这样的工作？

Three quarters of a million people were put out of work pretty quickly. They found jobs in other areas. So that's the first thing to recognise is that it's murky, it's unclear when we look forward. We can't predict where the new jobs were. Who would have said that when the internet was built, that you and I would be doing something like this?

是啊，没错。

Yeah, right.

对吧？谁会想到有人会在YouTube上花时间做节目、讨论各种话题呢？是啊，没人能预料到。确实没有。

Right? That there'd be people that spend time on YouTube and do shows and talk about things. Yeah. No one would have guessed it. No.

所以在AI领域，我认为同样会出现一些出人意料的进展。没错。但即便如此，让我们往更悲观的方向设想：假设AI在各方面都比人类强，无论是体力劳动还是不需要体力的工作，比如像我这样的财务顾问工作。

So in the world of AI, I think there'll be some surprises along that front as well. Yeah. But even so, let's go down that darker road where we say, okay, AI is more capable than humans on every level. Whether it's labor or even jobs that don't require labor. A job like mine for example, as a financial advisor.

假设AI比我聪明高效得多，能完成我所有的工作，让我失业。从某个角度看，如果事情能更便宜高效地完成，世界其实会变得更好。问题就变成了我该如何生存？进而演变成如何将这些技术红利分配给全人类？听埃隆谈论这个话题时，他说我们正在迈向的不是全民基本收入，而是全民高收入的世界。

Let's say the AI is just so much smarter and more efficient than I am and can do everything that I can do and puts me out of business. Well, on one level, the world is actually better off if we can do things more cheaply and more efficiently. It just becomes a question of how do I survive? And then that becomes a question of then how do we distribute the benefits of these technologies to humanity at large? When I hear Elon talk about it, he says, we're actually headed for a world of not universal basic income but universal high income.

不过他并未详细说明这个转变过程。从供给端我能理解——如果我们能降低能源、劳动力、运输等所有成本，就应该能以极大丰富度生产任何所需物品。一切都将变得充裕。是的。

Now, he hasn't really shared much detail on sort of how he sees that transitioning. I understand it from a certainly from a supply perspective. We should be able to produce anything and everything we want at massive abundance if we can drive down the cost of all those things we just talked about energy, labour, transportation. Everything should become abundant. Yeah.

举个很好的例子：回顾电话通讯市场，过去长途电话要按分钟计费。而现在便宜到根本不需要计量，只需支付固定费用。

And the one example that's a really good one is if you look back at, like, telephone market, the communications market, we used to pay by the minute for long distance calls. Yeah. Now it got it gets to the point where it's too cheap to meter. You just pay a flat fee. Yeah.

无论你从纽约还是加州打电话都无所谓了，谁在乎呢？每月只需付一笔固定费用。今天我们花大价钱购买的食物，很可能也会变成这样。

Doesn't matter anymore whether you're calling from New York, California. Who cares? It's just one flat fee per month. Yeah. And that may well be the case too with things that we pay a lot for today like food.

如果食物变得充裕廉价，你只需每月向杂货店支付固定费用，走进去随便拿取所需。

If food becomes abundant and cheap, you just pay a monthly flat fee to the grocery store and you walk in and take whatever you want.

意思是，举个例子。这些想法如此陌生，但我认为你的观点相当突出。当你回顾历史，观察那些看似技术将取代大量人力的时刻，而芒格和巴菲特在多年的股东大会上也曾谈及这一点。你知道吗？他们讨论过这个观点。

Mean, ideas example. These ideas are so foreign, but I think your point is pretty salient. It's when you go back in time and you look at these other moments where it looked like technology was gonna displace a whole lot of people, and I've heard Munger and Buffett cover this in some of their shareholder meetings through years. You know? They've talked about this idea.

而他们不断回归的一个核心是：我们甚至难以略微理解哪些机会或价值主张会自然浮现，供人类利用。若否认这种情况反复发生，某种程度上是徒劳的。对此持反对意见的人可能会说，我们从未创造过比人类更聪明的智能，因此这次不同——谁知道呢？也许这是个合理的论点，但我是个乐观主义者。

And that's the one thing that they also keep coming to is it's really hard to kind of even remotely understand what opportunities or value props kinda pop out naturally for humans to then take advantage of. And to act like it hasn't happened time and time again, you know, is somewhat of a fool's errand. I think the contrarian to to that would say, well, we've never built intelligence that is smarter than humans, and so this therefore, this is different, which who knows? Maybe it's a valid argument, but I'm an optimist. I know.

有时可能对自己不利，但我确实是个乐观主义者，我认为

To my own detriment maybe sometimes, but I'm an optimist and I I think

你必须如此，普雷斯顿，因为不得不这样。看看现状吧。这有点像潘多拉魔盒。盒子已经打开，无法再合上了。

you have to be, Preston, because You do. Let's look at this. Yeah. This is kind of like Pandora's box. The box has been opened and it's not going to be closed.

不。如果从传统对抗角度思考，比如以中美竞争为例。嗯哼，明白吗？如果我们不发展这些技术——无论是比特币，还是马斯克通过特斯拉、SpaceX等公司在做的——中国人就会抢占先机。

No. So if you just think about it from this traditional adversarial situation where let's say it's us against China, for example. Mhmm. Okay? If we don't pursue these technologies, whether it's Bitcoin, whether it's what Elon is doing across those companies with Tesla, SpaceX, the Chinese will.

嗯哼。那么你希望中国拥有比我们更先进的人工智能吗？希望他们的火箭技术超越我们吗？希望他们在数字安全和数字货币领域比我们更胜一筹吗？答案显然是否定的。

Mhmm. So do you want Chinese to have a better AI than us? Do you want the Chinese to have better rockets than us? Do you want the Chinese to have a better form of digital security and digital money than us? The answer is no.

我们必须确保与所有竞争者保持同步。因此我们绝不能对这些技术视而不见。不，绝对不行。无论是否情愿，这些技术都带有某种战争与博弈的色彩。

We have to make sure we're competitive with everybody else. And so we can't really afford to stick our head in the sand on any of this stuff. No. No. This is you know, whether we like it or not, there's an element of sort of warfare and battle to all these technologies.

埃隆曾谈到，如果我们失去优势，比如在无人机领域，那么我们就会变得非常容易受到无人机攻击。是的，对吧？能源生产也是如此。如果我们放任中国在能源生产增长方面遥遥领先而我们停滞不前，那么很快我们在任何领域都将无法与中国竞争。

And Elon has talked about that if we lose superiority, for example, in drones, then we become very vulnerable to drone attacks. Yeah. Right? And the same thing with energy production. If we let the Chinese run away with growing the energy production and we don't, then suddenly we're not going to be competitive against the Chinese really in anything.

让我们稍作休息，听听今天赞助商的消息。

Let's take a quick break and hear from today's sponsors.

你是否曾对比特币挖矿感兴趣？作为一名矿工，过去几个月我一直在使用Simple Mining，体验简直是无缝衔接。我选择自己想要的矿池挖矿，比特币直接发送到我的钱包。总部位于爱荷华州锡达福尔斯的Simple Mining提供高端白手套服务，面向从个人爱好者到大型矿工的所有人。他们已运营三年半，目前管理着超过1万台比特币矿机。

Have you ever been interested in mining Bitcoin? As a miner myself, I've been using Simple Mining for the past few months, and the experience has been nothing short of seamless. I mine with the pool of my choice, and the Bitcoin is sent directly to my wallet. Simple Mining based in Cedar Falls, Iowa offers a premium white glove service designed for everyone from individual enthusiasts to large scale miners. They've been in business for three and a half years and currently operate more than 10,000 Bitcoin miners.

由于爱荷华州丰富的风能资源，他们的电力供应超过65%来自可再生能源。他们不仅通过顶级托管和现场维修服务简化挖矿流程，还通过将你的运营作为商业运作来帮助你获得财务收益。这种方法能带来显著的税收优势，并提高你的投资盈利能力。你是否曾担心维护矿机的复杂性？他们已为你解决后顾之忧。

Based in Iowa, their electricity is over 65% renewable, thanks to the abundance of wind energy. Not only do they simplify mining with their top notch hosting and on-site repair services, but they also help you benefit financially by running your operations as a business. This approach offers significant tax advantages and enhances the profitability of your investment. Do you ever worry about the complexities of maintaining your mining equipment? They've got you covered.

前12个月内所有维修费用全包。若遇到停机情况，他们将给予补偿。如果你的矿机暂时不盈利，可以无罚金暂停运行。当你准备升级或调整设备时，他们的专属市场提供便捷的二手设备转售渠道。加入我和众多满意矿工的行列，简化你的比特币挖矿之旅。

For the first twelve months, all repairs are included at no extra cost. If you experience any downtime, they'll credit you for it. And if your miners aren't profitable at the moment, simply pause them with no penalties. When you're ready to upgrade or adjust your setup, their exclusive marketplace provides a seamless way to resell your equipment. Join me and many satisfied miners who have simplified their Bitcoin mining journey.

立即访问simplemining.i0/preston开始体验。再次重申：simplemining.i0/preston即可开始。SimpleMining，让一切变得简单。

Visit simplemining.i0/preston to get started today. That's simplemining.i0/preston to get started today. With SimpleMining, they make it simple.

作为小企业主，你没有提前下班的奢侈。你的生意需要你24/7全天候关注。因此当你招聘时，你需要一个和你一样努力的合作伙伴。这个招聘伙伴就是LinkedIn职位。当你下班时，LinkedIn开始上班。

As a small business owner, you do not have the luxury of clocking out early. Your business is on your mind twenty four seven. So when you're hiring, you need a partner that works just as hard as you do. That hiring partner is LinkedIn jobs. When you clock out, LinkedIn clocks in.

领英让免费发布职位变得简单，您可以分享到人脉网络，并在同一平台管理优质候选人。领英甚至能协助撰写职位描述，通过深度候选人洞察快速触达目标人群。您可选择免费发布职位或付费推广以获取三倍优质申请者。使用领英的企业中，72%的中小企业表示其能帮助找到高质量人才，让您对获得最佳申请人充满信心。了解为何我们及超过250万家中小企业选择领英进行招聘。

LinkedIn makes it easy to post your job for free, share it with your network and get qualified candidates that you can manage all in one place. LinkedIn can even help you write job descriptions and quickly get your job in front of the right people with deep candidate insights. You can post your job for free or promote it to get three times more qualified applicants. And with LinkedIn, you can feel confident that you're getting the best applicants as 72% of small and medium sized businesses using LinkedIn say it helps them find high quality candidates. Find out why our business and more than 2,500,000 other small businesses use LinkedIn for hiring today.

免费发布职位请访问linkedin.com/studybill。即linkedin.com/studybill免费发布职位。条款与条件适用。想象一下：午夜时分，您躺在床上浏览新发现的网站，将心仪商品加入购物车。准备结账时，突然想起钱包在客厅，而您根本不想下床去拿。

Post your job for free at linkedin.com/studybill. That's linkedin.com/studybill to post your job for free. Terms and conditions apply. Picture this, it's midnight, you're lying in bed scrolling through this new website you found and hitting the add to cart button on that item you've been looking for. Once you're ready to check out, you remember that your wallet is in your living room and you don't want to get out of bed to go get it.

正当您要放弃购物车时，那个紫色购物按钮出现了——它已保存所有支付和配送信息，让您足不出被窝就能快速完成交易。这就是Shopify。包括我在内的众多商家选择它是有原因的：从结账到创建独立店铺，Shopify让一切变得更简单。

Just as you're getting ready to abandon your cart, that's when you see it, that purple shop button. That shop button has all of your payment and shipping info saved, saving you time while in the comfort of your own bed. That's Shopify. There's a reason so many businesses, including mine, sell with it. Because Shopify makes everything easier, from checkout to creating your own storefront.

Shopify是全球数百万企业的商业后盾，支撑着美国10%的电商交易。从美泰、Gymshark等知名品牌到我这样刚起步的商家都在使用。Shopify为您提供全球转化率最高的结账系统。让Shopify助您实现商业构想，日后您定会感谢我。立即注册享受每月1美元试用，开始销售之旅：shopify.com/wsb。

Shopify is the commerce platform behind millions of businesses all around the world and 10% of all e commerce in The US. From household names like Mattel and Gymshark to brands like mine that are still getting started. And Shopify gives you access to the best converting checkout on the planet. Turn your big business idea into reality with Shopify on your side and thank me later. Sign up for your $1 per month trial and start selling today at shopify dot com slash wsb.

网址是shopify.com/wsb。好了，回到节目。

That's shopify.com/wsb. Alright. Back to the show.

让我们宏观审视他构建的技术体系复杂性。我想从商业角度分析：纵观他正在推进的所有项目——特斯拉汽车、自动驾驶、电动卡车、能源业务、Optimus人形机器人（Dojo项目刚被搁置）——未来三到五年内，您认为哪些技术会真正崛起？我先为观众列举这些项目，帮助他们全面理解他的布局。

So let's zoom out and kinda look at the intricacy of, like, what he's building. I wanna kinda look at it from a business standpoint. Which one of these technologies do you kinda see really starting to take off in the coming, call it, three to five years when you just look across the whole array of things that he's I mean, he's working on? Let me just name some of this for the audience so that they can really kinda wrap their head around the big picture of what he's doing. You got Tesla auto.

包括特斯拉汽车、自动驾驶系统、电动卡车业务、特斯拉能源、Optimus人形机器人（Dojo项目已暂停）。

You got Tesla autonomy. You got Tesla robo trucking and the cars. You have Tesla Energy. You have the Optimus humanoid robot. Dojo just got shelved.

你们有SpaceX。有星舰。有星链，所有互联网都从天上来。还有XAI。还有Neuralink。

You have SpaceX. You have Starship. You have Starlink, all the Internet coming from the sky. You got XAI. You got Neuralink.

还有X平台，他正在收集所有数据。他正在涉足X支付。还有无聊公司。还有超级高铁。这个清单可以一直列下去。

You got X where he's collecting all the data. He's stepping into x payments. You got the boring company. You got the Hyperloop. The list just goes on and on and on.

而我认为你的专长，Searn，在于你能把这些不同的技术拼凑在一起，能看到系统级思维带来的系统之系统价值正在显现。这就是我认为华尔街所忽视的地方。我猜你同意吧。他们只是孤立地看待每个事物，然后说，哦，好吧，我不知道。

And where I would say your expertise, Searn, is you're able to kinda piece all of these different technologies together, and you can see the system of systems value that's coming and emerging out of that systems level thinking. And that's where I think Wall Street is missing it. I'm I'm assuming you agree. They're just looking at each thing individually and saying, oh, yeah. Well, I don't know.

但当你把它们像拼图一样组合起来时，你说的是这些碎片组合后产生的超越个体价值的涌现价值。所以真正的问题是：在这些拼图碎片中，你认为未来三到五年内哪一个会真正爆发成为下一个S曲线？

But when you piece it together as a puzzle, you're saying there's this emergent value above and beyond the individual value that comes out of the hole. So really long question. Which one of those puzzle pieces are the next s curve kind of thing that you think is gonna really pop in the coming three to five years?

简短回答的话，我认为是自动驾驶出租车。但实际上，他们在你刚才提到的所有领域都会取得惊人进展。不过让我先谈谈自动驾驶出租车的现状。目前特斯拉制造并销售车辆给个人用户，利润微薄。

The short answer, I think, is robotaxi. But in reality, they're going to make incredible progress on everything you just mentioned. So but let me just talk about the robotaxi situation Yeah. How it stands today. So today, Tesla makes and sells vehicles and sells them to people and makes a little bit of profit.

假设特斯拉以4万美元售价卖出一辆车，净利润可能是4000美元。其中可能包含政府补贴和税收优惠等对特斯拉有利的部分。利润率大概10%左右，如果运气好的话。

Let's say Tesla makes a car that they sell for $40,000, they might make a $4,000 net profit on that car. And some of that might include government subsidies and tax credits and that kind of stuff that benefit Tesla. It's like a 10% margin. Yeah. If they're lucky.

明白吗？更进一步说，假设为了简化模型，消费者平均每五年换购一辆新车。那么特斯拉从每个人身上每五年才能赚到4000美元。现在转向自动驾驶出租车模式的话...

Okay? And further, you might buy a new car, let's just say, you know, to make this easy every five years. So Tesla's making $4,000 once every five years if they sell cars per person. Yeah. Move to robo taxi situation.

明白吗？他们打算用于此的车辆电池容量大约是Model y的一半。所以制造成本会很低。对吧？假设这辆车在五年内能产生20万美元的利润。

Okay? The cars that they're going to use for that have a battery capacity of about half of the Model y. So the cars will be cheap to make. Okay? And let's say the vehicle, over a five year period produces $200,000 in profit.

不是每年，而是整个五年周期。是的。懂吗？所以你看这是巨大的增长。对吧？

Not per year, but just over the five year period. Yeah. Okay? So you're looking at a massive increase. Right?

从每五年4000美元到20万美元。

$4,000 once every five years to 200,000.

嗯。

Mhmm.

对吧？你正从向挑剔且口味多变的消费者卖车，转向制造赛博舱...是的，一种非常实用基础的工具，

Right? You're moving from selling cars to consumers, and consumers are picky and taste change, to making cybercaps Yeah. A very utilitarian basic machine that

有个优势是它们看起来都一样。对吧？是的。所以不用考虑那些。对。

can give you a There's advantage that they all look the same. Right? Yeah. So not to think about any of that. Yeah.

对吧？因此如果你考虑到电池含量减半的事实，实际上相当于五年周期内从4000美元到40万美元。这是卖车利润的100倍。特斯拉即将通过Robotaxi实现的就是这种突破。嗯。

Right? And so if you adjust for the fact that the battery content is half, then actually it's like equivalent to like $400,000 over a five year period compared to 4,000. It's a 100 x the profitability of selling a car. That's the kind of unlock that Tesla's on the cusp of with Robotaxi. Mhmm.

就市场机遇而言，这不仅关乎颠覆优步和共享出行公司，而是颠覆整个交通体系。包括个人拥有的车辆。有些家庭拥有两到三辆车。初期，人们会处理掉第三辆车，因为可以乘坐机器人出租车，这很合理。

And in terms of the market opportunity, it's not just about disrupting Uber and rideshare companies. It's about disrupting all transportation. Your personally owned vehicle. There are people that own, you know, families that own two or three vehicles. Initially, you'll get rid of the third vehicle because you can take robotaxi and that makes sense.

随着机器人出租车成本下降，人们会放弃第二辆车。当成本持续降低时，他们将舍弃第一辆车。一旦机器人出租车变得普及且极其便宜，人们基本上会在所有场景中使用它。我一直强调经济规律终将胜出——如果能每年节省5000美元交通开支，人们必然会选择改变。

As the cost of robotaxi comes down, people are gonna ditch the second vehicle. As the cost of robotaxi keeps coming down, they're gonna ditch the first vehicle. They're just going to use robotaxi basically in every situation, once they become ubiquitous and super cheap. And I've always said that economics wins here. If you can save, you know, $5,000 a year on your transportation costs, you are going to do it.

人们将发现可以节省开支。他们会乐于不再拥有汽车、支付保险，也不必处理停车烦恼和各类付款事项。'必须拥有私家车'的观念很快就会成为过时的想法。

People will discover that they can save money. People will happily not own a car, pay insurance, and deal with all the headaches of parking the car and, you know, paying for all the payments. This idea that we have to own our vehicles actually is kind of it's gonna be an antiquated idea very quickly.

先生，我想用具体数字帮助大家理解。如果数据有误请纠正：当前使用优步的价格约为每英里2-3美元。若购买4万美元的私家车，使用五年后转售，按美国人年均行驶里程计算，每英里成本约65-80美分。

Sir, and I wanna stick some numbers on this for people so that they can think in their own terms. So when you go out and correct me if any of these numbers are off. If you go out and you get an Uber today, you're paying anywhere from, call it, two to three dollars per mile in order to get that Uber. If you buy your own car, and let's say you buy a $40,000 car, you hold it for five years, you sell it, you get whatever, you know, it's worth at that point after five years, and you drive it an average American amount over those five years. This would cost you about 65¢ to 80¢ per mile.

优步价格在2-3美元区间，而自有车辆五年后转售的成本约65-80美分。埃隆认为能将机器人出租车成本降至25-30美分/英里，这相当于将个人拥车成本直接减半。

So Uber's Uber's in that 2 to $3 range. If you own your own car and then you sell it after five years, it's about 65¢ to 80¢. Elon's thinking he can get these robo taxis down to about 25 to 30¢ per mile, which, I mean, you're very generically, it's half the cost of going out and owning your own vehicle and dealing with all of that, personally.

还有一点，普雷斯顿，这个平台还能叠加其他服务。当自动驾驶车辆网络普及后，它们可以配送食物、快递，成为各类服务的载体。同时也可作为广告平台使用。

And the other thing too, Preston, that happens is that this becomes platform to layer other services on. So if you've got this ubiquitous network of vehicles driving around Yeah. They can deliver food, they can deliver packages, you can layer on all kinds of other services on top of this platform. Yeah. You can also use it as an advertising platform.

所以我曾说：最终用户可能获得免费乘车服务，前提是接受广告投放。毕竟车辆运营成本极低，收入来源可以从乘客转向广告商。

So I've actually said that eventually, you may be able to get free rides on this network if you're willing to subject yourself to advertising. Wow. Right? Because the costs of the car is so cheap to operate. Instead of getting paid by the customer, you can just get paid by the advertisers.

餐厅可能会付费给特斯拉在车内打广告。对吧？或者餐厅甚至可能付钱让特斯拉推荐说，‘嘿，我们建议你去XYZ披萨店。’然后就直接带你去。就像在无人驾驶出租车里能体验到的服务种类，其实是无限的。

Restaurants might pay Tesla to advertise in the car. Right? Or restaurants may even pay Tesla to say, hey, we'd recommend going to XYZ pizza place. And it just takes them there. Like, the kind of experiences you can have in a robotaxi, you're actually kind of unlimited.

Sern，你能想象他已经拥有xAI并且对你喜欢什么了如指掌吗？是的。甚至知道你想看到什么类型的广告？他将会掌握所有这些信息。

Sern, can you imagine the fact that he has x AI and already has a very good understanding of what it is you like? Yeah. What it is you wanna be advertised on? Like, he's gonna know all of that.

而且你还能和车对话，Preston。你可以和无人驾驶出租车进行交流。

And you'll be able to talk to the cars too, Preston. You'll have conversation with the robotaxi.

这些东西会获取你所有的地理位置信息，比如你通常在哪里吃饭。是的。天啊。

These things are gonna get all your geo, like where you typically eat. It's Yeah. Holy crap.

假设你旅行到一个城市。你飞到芝加哥然后说，‘带我去城里最好的披萨店。’接着Grok会和你聊天，问‘你喜欢这种吗？还是那种？’你们可以对话，然后车就直接带你过去。

Let's say you travel to a city. You fly to Chicago and you say, take me to the best pizza place in the city. And then Grok is talking to you and saying, well, do you like this? Do like that? You can have a conversation, and then the car then just takes you there.

我看过它能充当导游的样子，会说‘顺便一提，这栋建筑建于1938年，曾是全球第二高楼’，或者根据你的兴趣介绍各种信息。它变成了一个提供各类服务的平台。

I've seen the way, it can act as a tour guide and say, oh, by the way, this building was built in 1938, it was once the second tallest building in the world or, you know, whatever you're interested in, it becomes a platform for all these different kinds of services.

对吧？我看到一些人在车里使用Grok AI的视频，至少网上看到的反馈——不知道是不是营销推送——但网评好得出奇。人们说‘感觉就像和真人对话一样’。是啊，懂吧？人们会对它说‘嘿’。

Right? Saw some of the videos of people using the Grok AI in the car, and the feedback, at least that I saw online, I don't know if I was being fed this for marketing purposes or what, but the feedback that I was seeing online was extremely favorable. People were like, it felt like I was having a conversation with a real person as I was Yeah. You know? People were telling it, hey.

带我去一个你认为我会喜欢、但在自己镇上从未去过的地方，他们真的带我去了某处。然后他们走进去，反应就像是——我在网上看到的一些反馈实际上非常正面。我觉得我们正处在这整个趋势的风口浪尖上。

Take me somewhere that you think I would like that I've never been before in my own town, and it literally took them somewhere. And then they went in, and they were like, this was so, like, some of the feedback that I've seen online was actually really favorable. And I think we're just on the cusp of, like, where this is all going.

目前Grok还没有直接连接到汽车的操控系统。所以你可以和Grok对话，就像和笔记本电脑、台式机或手机聊天一样，但它并不能控制车辆。我认为最终两者会建立连接，届时你将能与Grok进行有趣的对话，比如：'嘿，刚才为什么别那辆车？你没看见吗？'或者'能不能开慢点？'

Like, this is not Right now, Grok is not connected directly to the operation of the car. So you can talk to Grok, and it's like talking to, you know, just laptop or a computer or your phone, but it's not controlling the car. Eventually, I think there'll be a link between the two, and you'll be able to have an interesting conversation with Grok about, hey, why did you cut that other car off back there? Didn't you see it? Or, you know, can you drive a little slower?

我对你现在的驾驶方式感到不安。或者类似的情况。你将能与车辆进行这样的对话。对吧？

I'm not comfortable with the way you're driving right now. Or or whatever it is. You'll be able to have this conversation with the vehicle. Yeah. Right?

除了查询餐厅信息之类的基本功能外。我认为当实现车辆对话功能时，那将会是个很酷的时刻。

In addition to just being able to query it with information about restaurants and different things. That's going to be a neat moment, I think, when that happens.

好的。所以汽车是其中一部分。我认为卡车方面可能进展稍慢些，帮我理一下时间线——关于卡车的情况。嗯，请继续。

Okay. So the cars is one piece of it. The other part that I think is maybe a little bit further behind, help me out understanding kind of the timeline, but the trucks. So, yeah, go ahead.

特斯拉正在内华达州建设年产能5万辆半挂卡车的工厂。这个工厂很可能在今年底或明年初完工，之后开始提升产量。假设到2027年，他们每年能生产5万辆半挂卡车。

Yeah. Tesla is building a factory in Nevada that has a capacity of 50,000 semi trucks a year. That factory will likely be finished maybe the end of this year, early next year. They'll start ramping up production. So let's say by 2027, they're producing 50,000 semis a year.

要知道美国有相当多的卡车。要用电动卡车替换整个车队需要很长时间。与此同时，它们也会像无人出租车一样——终将成为自动驾驶卡车。正如这将彻底改变运输成本，它也将革命性地降低货运成本。

You know, there's quite a few trucks in The United States. It's gonna take a while to replace the fleet with electric trucks. At the same time, they will also be like the robotaxi. They will be autonomous trucks at some point. And just like how that's going to revolutionise the cost of transportation, it's going to revolutionise the cost of trucking.

我们确实能在卡车运输成本上实现巨大节省。而卡车运输成本会导致我们购买的所有商品价格上涨，对吧？如果卡车运输费用高昂，在全国范围内运输货物成本高企，就会让我们购买的所有东西都变得更加昂贵。因此，这将成为贯穿整个经济的强大通货紧缩力量，对所有消费者都有利。

We can really get massive savings on the cost of trucking. And of course, trucking costs cause inflation in everything we buy. Right? If trucking is expensive and moving goods around the country is expensive, makes everything we buy that much more expensive. So this is going to be a massive deflationary force throughout the economy that's gonna be good for all consumers.

所以每个美国人都将受益于特斯拉降低卡车运输成本的能力。当前卡车运输行业在运营成本方面存在严重问题。大多数卡车运输公司利润微薄。燃料成本、司机成本——据我计算，司机成本大约每英里1美元。

So every American will benefit from Tesla being able to reduce the cost of trucking. The trucking industry right now has a real problem in the cost to operate. Most trucking companies are not very profitable. The cost of fuel, the cost of the driver. Cost of the driver, I think, works out something like a dollar a mile.

没错。我就是这个意思。大约每英里70到90美分。

Yeah. That's what I'm saying. It's about 70¢ to 90¢ per mile.

对，加上福利可能1美元左右。自动驾驶卡车能消除这部分成本。当然柴油成本也非常高昂。

Yeah. Including benefits, maybe a buck. Yeah. And so robot trucking gets rid of that. And then, of course, the diesel cost also is huge.

如果是电动卡车，燃料成本就能大幅降低。而且维护成本也是——传统卡车维护成本很高因为发动机会磨损。电动卡车由于运动部件更少，维护需求会低得多。

If they're electric trucks, you have massive savings on the fuel costs. Yeah. And then, of course, the maintenance. The maintenance on these trucks is pretty high because these engines wear out. Electric trucks won't have as much maintenance as fewer moving parts.

需要维护的部件就是更少。

There's just less stuff to maintain.

是啊。只要更换电池和新轮胎就能继续行驶。对。明白了。嗯。

Yeah. They'll just swap the battery and put new tires on it and keep rolling. Yep. Okay. Yeah.

所以我现在看到的干线运输每英里总成本大约是1.6到2美元，而特斯拉自动驾驶卡车预计会降到60到80美分。也就是说成本能降低50%到65%。虽然不像个人用车那样有10倍的惊人改进，但50%的降幅也很可观。不过据我理解，这其实是更稳定的收入来源，因为合作对象都是大型供应商，比如亚马逊或沃尔玛这些巨头。

So the all in cost per mile that I'm seeing for the line haul is, like, anywhere from a buck 60 to $2 a mile, and the Tesla Robo trucking is expected to come down to about 60¢ to 80¢. So we'll say about a 50 to 65% reduction in cost. Not as much as the individual car for people getting around, which, you know, was mind blowing, like, 10 x improvement. This is about 50%. But I guess my understanding is that this is a much more reliable stream of income because you're dealing with big vendors and, like, once they come into this, call it Amazon or Walmart or whoever.

只要这些大企业进入这个领域并稳定下来，发现成本大幅降低后，整个业务就会顺畅运转。所以

Like, once they come into this space and they stay there and they see that it's a lot cheaper, it's like it's humming along. So

没错。要知道卡车运输成本哪怕降低10%都意义重大，足以引发巨大变革。50%的降幅简直是颠覆性的。

Yeah. I mean, a 10% reduction in trucking cost would be huge. I mean, would drive massive change. Yeah. The 50% is absolutely enormous.

是的。卡车运输市场的不同之处在于全是商业运营商。不像我们私家车可以低成本运营。嗯。

Yeah. The thing that's different about the trucking market is you've got it's all commercial operators. Right? It's not like we personally own our own vehicles and we can operate them pretty cheaply. Mhmm.

对吧？现在用优步和Lyft可比开自家车贵多了。但卡车市场不存在这种情况，整个行业都是商业化运营，目前平均每英里收入大概在2到2.3美元。

Right? And Uber and Lyft right now are way more expensive to use than just driving our own personal vehicles. You don't really have that in the trucking market. It's all commercially operated. The industry's earning at an average, say, dollars to $2.02 30 a mile.

只要你能把成本控制在行业平均水平之下，就能抢占市场份额。随着成本持续下降，那些传统运营商就会失去竞争力，最终被淘汰出局。整个行业会迅速转向电动卡车和自动驾驶卡车。

And as long as you can come under that, you're gonna gather market share. And as you keep sort of driving that down a little bit, all those players are going to be They're not going to be economic anymore. You're going to drive them out of business. The whole industry is going to shift towards electric trucking and autonomous trucking pretty fast.

崔，我们先插播一段赞助商内容。商业的未来会怎样？问十个专家能得到十一种答案。牛市熊市争论不休，真希望有人能发明个预言水晶球啊。

Trey Let's take a quick break and hear from today's sponsors. What does the future hold for business? Ask nine experts and you'll get 10 answers. Bull market, bear market, it goes on and on. Can someone invent a crystal ball?

迄今为止，已有超过42,000家企业通过Oracle的NetSuite为其业务未来保驾护航。作为排名第一的AI云ERP系统，它将会计、财务管理、库存和人力资源整合到一个流畅的平台上。通过这一体化的业务管理套件，您将获得单一数据源，从而拥有做出快速决策所需的可见性和控制力。借助实时洞察和预测功能，您能以可操作的数据预见未来。若我需要此类产品，这正是我会选择的解决方案。无论您的公司盈利数百万还是数亿美元，NetSuite都能助您应对即时挑战并把握最大机遇。

Until then, over 42,000 businesses have future proofed their business with NetSuite by Oracle, the number one AI cloud ERP bringing accounting, financial management, inventory, HR into one fluid platform. With one unified business management suite, there's one source of truth giving you the visibility and control you need to make quick decisions. With real time insights and forecasting, you're peering into the future with actionable data. And if I had needed this product, it is exactly what I would use. Whether your company is earning millions or even hundreds of millions, NetSuite helps you respond to immediate challenges and seize your biggest opportunities.

说到机遇，请立即在netsuite.com/study下载《CFO人工智能与机器学习指南》。该指南在netsuite.com/study可免费获取。Netsuite.com/study。作为创始人，您正快速前进——无论是实现产品市场契合度、筹备下一轮融资，还是敲定首笔大额企业交易。但随着AI加速初创企业的开发和交付速度，安全要求比以往任何时候都更高。

Speaking of opportunity, download the CFO's guide to AI and machine learning at netsuite.com/study. The guide is free to you at netsuite.com/study. Netsuite.com/study. As a founder, you're moving fast, whether that's towards product market fit, your next round, or your first big enterprise deal. But with AI accelerating how quickly startups build and ship, security expectations are higher than ever.

正确处理安全和合规问题能推动增长，但若拖延太久则会阻碍发展。Vanta专为快速发展的团队打造深度集成和自动化工作流，助您快速通过审计，并在模型和客户演变过程中通过持续监控确保安全。我欣赏Vanta为企业节省时间和金钱的承诺——近期IDC白皮书显示，Vanta客户每年可获得535,000美元的收益，平台成本仅需三个月即可收回。立即访问vanta.com/billionaires，通过Vanta初创企业计划节省1,000美元，加入10,000余家正在使用Vanta实现规模扩张的雄心勃勃的企业。限时优惠请访问vanta.com/billionaires。

Getting security and compliance right can unlock growth or stall it if you wait too long. With deep integrations and automated workflows built for fast moving teams, Vanta gets you audit ready fast and keeps you secure with continuous monitoring as your models and customers evolve. I love Vanta's commitment to saving companies time and money as a recent IDC white paper found that Vanta customers achieve $535,000 per year in benefits and the platform pays for itself in just three months. Go to vanta.com/billionaires to save $1,000 today through the Vanta for Startups program and join over 10,000 ambitious companies already scaling with Vanta. That's vanta.com/billionaires to save $1,000 for a limited time.

想象这个场景：午夜时分，您躺在床上浏览新发现的网站，将心仪商品加入购物车。准备结账时，突然想起钱包落在客厅，而您不愿起身去取。正当要放弃购物车时，那个紫色购物按钮出现了——它已保存所有支付和配送信息，让您安享床榻之便。这就是Shopify。

Picture this, it's midnight, you're lying in bed scrolling through this new website you found and hitting the add to cart button on that item you've been looking for. Once you're ready to check out, you remember that your wallet is in your living room and you don't want to get out of bed to go get it. Just as you're getting ready to abandon your cart, that's when you see it, that purple shop button. That shop button has all of your payment and shipping info saved, saving you time while in the comfort of your own bed. That's Shopify.

包括我的企业在内，众多商家选择Shopify有其原因。因为它让一切变得更简单——从结账到创建专属店面。Shopify是全球数百万企业的商业后盾，支撑着美国10%的电子商务交易。从美泰、Gymshark等家喻户晓的品牌，到像我这样刚起步的企业。Shopify还为您提供全球转化率最高的结账系统。

There's a reason so many businesses, including mine, sell with it. Because Shopify makes everything easier, from checkout to creating your own storefront. Shopify is the commerce platform behind millions of businesses all around the world and 10% of all e commerce in The US. From household names like Mattel and Gymshark to brands like mine that are still getting started. And Shopify gives you access to the best converting checkout on the planet.

让Shopify助您将宏伟商业构想变为现实——您稍后会感谢我的。立即注册享受每月1美元试用，开始销售之旅：shopify.com/wsb。重复一遍：shopify.com/wsb。好了，回到节目。

Turn your big business idea into reality with Shopify on your side and thank me later. Sign up for your $1 per month trial and start selling today at shopify.com/wsb. That's shopify.com/wsb. Alright. Back to the show.

关于机器人出租车领域，竞争对手是谁？现阶段谁能对他构成实质威胁？

On the robotaxi piece, who's the competitors? Who could give him a run for his money at this point?

嗯，这是个好问题，因为中国可能有一些竞争对手，而且我认为中国人可能会对特斯拉控制这个市场有些犹豫。因此，我预计中国会确保他们有一些本土的人才，并且有一些公司在那里展现出相当令人印象深刻的成果。但客观地说，特斯拉遥遥领先于这些公司。是的，有很多公司都想参与其中。

Well, that's a good question because there might be some competitors in China, and I think the Chinese might be a little hesitant to have Tesla control that market. So I would expect the Chinese to kind of ensure that they've got some homegrown talent there and there's some companies that are showing some pretty impressive results there. But Tesla, objectively, is way ahead of both of those companies. Yeah. You've got a lot of companies that want to be part of this.

以Waymo为例，在美国，Waymo的问题在于他们有一个非常昂贵的解决方案，一辆配备了大量传感器和各种附加设备的汽车。特斯拉基本上只使用摄像头和一台电脑。而Waymo需要数千美元的其他设备，你必须对现有车辆进行改装，然后加入车队。特斯拉自己制造车辆，它们从生产线下线时就已经准备好投入使用了。

If you look at Waymo, for example, in The US, the problem with Waymo is that they have a very expensive solution, a car with a lot of sensors and a lot of bells and whistles. Tesla is basically just using cameras and a computer. And Waymo has thousands and thousands of dollars of other equipment that you have to take existing cars, modify them, and add them to the fleet. Tesla makes their own vehicles, and they roll off the assembly line ready for service.

它们自己开到目的地

They drive themselves to where they

它们可以自主驶离。是的，他们现在在工厂里就是这么做的。它们自己开到停车场，等待卡车来装载。所以Waymo在成本上真的无法与特斯拉竞争，而且我不认为他们将来能做到。

They drive themselves off autonomously. Yeah. They're doing that today in the factory. They drive themselves to the lots to be picked up by the trucks. So Waymo really can't compete with Tesla on cost, and I don't know that they'll ever be able to.

是的。因为，再说一次，他们没有自己的制造工厂。

Yeah. Because, again, they don't have their own manufacturing plants.

是的。你提到了备受争议的激光雷达。汽车上是否需要激光雷达？而埃隆一直坚定地认为，嘿，如果后面有一个人，他们使用的是图像、眼睛，你知道的，可见光频谱感知，那么为什么汽车还需要其他东西呢？尤其是如果汽车上基本上布满了‘眼睛’，在可见光频谱下观察的话。

Yeah. You're getting into the highly debated LIDAR. Do we need LIDAR on the car or not? And Elon has always been of the firm opinion that you just hey. If there's a human back there and they're using image, eye, you know, visible light spectrum sensing, then why in the world do I need anything other than that with the car, especially if I have basically eyeballs all over the car viewing in the visible light spectrum?

是的。普雷斯顿，埃隆还谈到的另一个风险是，如果两个传感器之间出现分歧怎么办？是的，你如何解决这个问题

Yeah. The other risk too, Preston, that Elon has talked about is that what happens if you have the disagreement between the two sensors? Yeah. How do you resolve that

实时？我不太接受这个论点。我觉得这更像是他在故作姿态，说明安全性足够，并试图暗示这样反而可能降低安全性。但如果你用足够多的数据训练它——毕竟AI总是信息越多表现越好，对吧？

real time? I don't know if I buy that argument. I think that's more him posturing for why it's safe enough and, like, kind of raising it as if it could potentially make it less safe. But I think if you trained it on enough data I mean, AI is always better when you give it more information. Right?

所以我不认为这会造成那种对立。

So I don't think that it's gonna have that rivalry.

如果属实的话，我见过一些能自圆其说的案例。但重申下，我对这个领域的技术原理并不专业。当然，参数越多、输入AI模型的数据越丰富，效果理应更好。不过确实要注意避免混淆，比如一个传感器显示是墙，另一个却显示是塑料袋。

If it's true I've seen some examples where that makes sense to me. But, again, I'm not an expert in the mechanics of this area. Certainly, the more parameters, the more inputs you give into the AI model, the better off it should be. Yeah. But certainly, you wanna be careful about confusing it and saying one sensor is saying that's a wall, another sensor is saying that's a plastic bag.

这要怎么解决？嗯...如果处理不当，可能立即就会出问题。确实。

And how do you resolve that? Mhmm. You resolve it in the wrong way, suddenly you could have a problem. Yeah.

这个论点很有道理——想象一下塑料袋盖在激光雷达传感器上的场景。没错，这确实是个关键点。

Well, that's a good argument because the yeah. The plastic bag imagine a plastic bag, like, being over top of the LiDAR sensor. And yeah. No. That's a great point.

还有这种情况。我的意思是，这时车辆只能减速。就像挡风玻璃被遮挡时，你会尽量安全快速地减速靠边。自动驾驶车辆在摄像头或传感器受损时也必须采取同样措施。

That too. I mean, that case, the car would just have to slow down. I mean, just like if if something came on the windscreen of your vehicle, you just try to slow down as safely and as quickly as you can, pull off to the side of the road. The same thing would have to happen with an autonomous vehicle if its cameras or sensors were degraded in some way. Yeah.

知道吗？比如下雨时很多人说这些车辆无法运行，其实不然——它们能在一定程度内运行，只在认为不安全时停驶。而多数人类司机在危险天气里反而会继续驾驶。

You know? I think too, like for example, during rain, a lot of the people say, well, these vehicles are not going to be able to operate when it's raining. And it's not true they can, but to a point. And they stop operating when they think it's unsafe. Most humans actually continue to drive actually even in those unsafe conditions.

因此，即便车辆在某些时段不运行，这项技术也将在多个层面上使出行变得更加安全。

So it's actually going to make travel a lot more safer on all kinds of different levels even if the vehicles don't operate during periods of time.

那么我们现在进展如何？他们已经在奥斯汀进行试运行了。还要多久我们才能看到这些车辆真正上路？特斯拉电动皮卡真的让我很惊讶，突然间你就发现它们无处不在。

So where are we at? They're already doing test runs in Austin. How much longer until we're gonna start seeing these things, like, show up? I was really surprised with the Tesla truck. It was just like you started seeing them everywhere, like, out of nowhere.

自动驾驶出租车也会这样吗？它们会突然出现，几个月内就遍布各地？还是说会按城市逐步缓慢推广？

Is that kinda what we're gonna see with the robotaxis where they just kinda, like, show up and, like, they seem to be kind of everywhere within months, or is this gonna be a slow rollout city by city?

是的。目前奥斯汀和硅谷地区都在进行试点运营，这两个区域是我们看到的主要测试地。埃隆显然计划尽快在多个地点推出自动驾驶出租车服务。

Yeah. They're operating right now in a pilot service in Austin. They're operating in pilot service in the Silicon Valley area. Those are the two areas we're seeing testing in a bunch of other places. Elon certainly has designs on rolling out robotaxi in multiple locations as quickly as possible.

但现实是，我认为他们必须确保安全性达到最高标准。最不该做的就是仓促行事。即便比大众预期晚六个月或一年也完全可以接受，对吧？

But the reality is I think they need to make sure that it's as safe as it can possibly be. The last thing they wanna do is rush this. And if they are six months, a year later than everybody's expectations So be that's fine. Yeah. Right?

这方面必须慎之又慎。与此同时，他们正在改进基础的全自动驾驶模型，我们都迫切期待着第14版本的发布，

You really have to be careful with this. At the same time, they're improving the underlying full self driving model. We're all anxiously awaiting for version 14,

希望很快就能看到。这个版本相比之前会有哪些预期改进呢？

and hopefully, we'll see that soon. Anything expected with that version versus

嗯，埃隆提到过这将使模型考量的参数数量增加10倍，因此系统性能应该会大幅提升。正如你所说，给AI提供的信息越多，结果就应该越好。埃隆还曾评论说这辆车似乎具有感知能力。所以我们都在急切等待亲身体验，看看它到底是什么感觉。

Well, Elon's talked about how it's going to 10 x the number of parameters that the model considers, and so it should make the system that much better. Again, to your point, the more information that you give AI, the better the outcome should be. And Elon has made sort of comments that the car seems sentient. So we're all anxiously awaiting Hold on. Experience this for ourselves and see what it's like.

因为这本质上就是把AI嵌套进一个载体里，也就是汽车。是啊，他提到过哪些具体例子说明这东西...

Because you basically have AI nested itself into a body, which is a car Yeah. He's like, what examples does he have of this thing is

目前我们缺乏具体案例，只有他的评论。而且要知道，他比我们早几个月看到这些。所以我们迫切想了解FSD新版本的真实表现。

We were short on on examples at this point. It's just comments from him. What? And, you know, he's seeing things months before we get to see it. So we're anxiously awaiting what this new version of FSD is really like.

好吧好吧，这到底是什么意思啊？

Well well, yeah, what in the world does that even mean?

没错。显然目标是让汽车驾驶比人类更出色、更安全。我认为在很多方面它已经做到了。是的。

Yeah. I mean, the goal, obviously, is for the car to drive better and safer than a human. And I think in many respects, it already does. Yeah.

具体数据是怎样的？这方面你听到什么消息...

What are those stats? What are you hearing from that regard as far as

这个嘛，情况比较复杂。很多是人们发布在X平台上的个人经历，描述车辆在特定条件下的表现。我自己就有过一次体验。

Well, that's mixed. I mean, a lot of it is the personal stories that people post on X. Right? And how the car behaves in certain conditions. You know, I had one experience.

举个例子，我当时在田纳西州的高速公路上，车速大概保持在75英里每小时，也就是限速标准。突然间，我驾驶的Cybertruck感觉加速了。还没等我反应过来，另一辆车以约140英里的时速从我旁边呼啸而过。那辆车原本紧贴在我右侧车道后方，超车时突然切到了左侧车道。

For example, I was on a highway in Tennessee and I was doing probably 75 miles an hour, whatever the speed limit was. And all of a sudden, felt my I drive a Cybertruck. I felt my Cybertruck accelerate. And before I could even realise what was going on, another car passed me, probably going a 140 miles an hour. But it was coming up right behind me in the right lane and then suddenly turned into the left as it passed me.

但卡车系统预判那辆车可能会追尾，于是精准计算出应该加速来避免碰撞。等那辆车超过去后，卡车又自动降回到原定车速。这种人类根本无法察觉和反应的预判行为...

But the truck thought that that car might actually rear end me. So it correctly calculated that it should accelerate so that No that impact would be way. Right? And then as soon as that car passed me, the truck slowed back down again to the speed. So that kind of behaviour that humans, we just can't see, we can't react to it.

直到那辆车超过去我才注意到它。类似的情况还有——比如系统发现人行道上有孩子走向马路，而人类司机因为厢式货车遮挡根本看不见。它能识别到有孩子可能突然从盲区窜出，于是提前减速。

I didn't see that car until it passed me. Right? Wow. So there's stories like that and there's other things where it sees kids that walk onto the street, the human driver never saw because they were blocked by a van or something. It saw the kid approaching the van from the sidewalk, recognized that there's a child behind there that might pop out, therefore the car slowed down.

这类案例很多。不过Preston，确实也存在车辆在某些场景下表现怪异的情况。当然，目前还不是尽善尽美的。

There's things like that. But also, Preston, there's also examples of the car kinda behaving weird in certain situations. Mhmm. Of course. So, you know, not everything is perfect yet.

我记得有数据显示这车每英里事故率远低于人类驾驶？这个数据准确吗？

I think I read somewhere that the car is demonstrating per mile way lower crash rates than a human driver. Is that accurate? Is that true?

这个说法是成立的。虽然我们拿不到原始数据，但根据特斯拉公开的信息，我们认为属实。是的。

I think that's fair to say. Again, we don't have our hands on the actual underlying data. Based on what Tesla shares with us, we believe that to be true. Yes. Yeah.

没错。而且根据特斯拉披露的数据，优势幅度其实相当显著。

Yeah. And in quite a significant way too, by the way, based on the data that Tesla shares.

是的，很有意思。好的，所以这是两个例子。你还有其他接近市场或即将呈现S型增长曲线的案例吗？

Yeah. Interesting. Okay. So those were two examples. Do you have any others that you think are close to market or something that's going to be an s curve in the Yeah.

估值方面？嗯，能源存储业务。特斯拉生产这些巨型储能包，它们的尺寸基本上就是一个能装在平板拖车上的箱子大小。

Valuation? Well, the energy storage business. So Tesla makes these mega packs. They are the size there's a box, basically the size that fits on a flatbed trailer. Okay.

设计得这么大是因为电池本身就又大又重。你希望它们尽可能大，但又不至于大得难以运输。这些巨型储能系统能让我们储存能量并在需要时使用。比如风电和太阳能只在特定时段发电，这时储能系统就完美解决了问题。

And they're designed to be that big because batteries are big and heavy. You know, you want them to be as big as possible but not so big that you have problems moving the system. These mega pack systems basically allow us to store energy and use it whenever needed. So it's perfect, for example, if you have wind and solar that only generate certain times of the day, you can store the energy and use it later. That's great.

埃隆说过，如果我们部署足够的电池储能，实际上可以在不新增发电设施的情况下，让美国发电容量翻倍。因为我们有很多季节性/时段性闲置产能。但有了储能系统，就能让这些设施持续满负荷运转并储存能量。这是在现有能源基础设施上快速提升产能的捷径——当然我们仍需新建，但最立竿见影的方案就是给所有设施加装储能系统。

Elon has said that if we applied enough battery storage, we could actually double the generating capacity of The United States without adding any more generation. Because we have a lot of sort of capacity in places that's excess, that's only used in certain times of the year, certain times of the day, for example. But with battery storage systems, you could run those at capacity and store that energy all the time. So that's one quick way to accelerate the energy generation infrastructure that we have already in place without building any more. Now, of course, we have to build more, but an easy win would just be to add battery storage to everything.

所以我认为特斯拉能源业务应该会加速增长。虽然已经在增长，但我认为未来增速还会更快。

And so I think that Tesla Energy their energy business should see accelerated growth. It already has, but I think even further from here.

你觉得这是不是埃隆绝口不提比特币的主要原因？某种程度上说，这是比特币挖矿的竞品吧？

Do you think that this is one of the main reasons why Elon just, like, doesn't talk about Bitcoin? Is this is a competitor to in a way, this is a competitor to Bitcoin mining?

确实。每次思考比特币和特斯拉的关系时我都有些困惑。要知道特斯拉在2021年2月买过比特币，当时比特币市值接近万亿美元。嗯...

Yeah. It puzzles me a little bit when I think about Bitcoin and Tesla. I mean, certainly, Tesla bought Bitcoin, I think it was in February 2021. You know, that time Bitcoin's market value was like a trillion dollars or approaching a trillion. Mhmm.

他们出售了一些。埃隆对此的一条评论是，等比特币网络的能源来源超过50%是可再生能源时再告诉我。这就是他的不满所在，对吧？自那以后，他对此就没什么讨论了。

They sold some. And one of Elon's comments about it was, well, tell me when the Bitcoin network has crossed 50% renewable energy sources. That was his kind of his beef. Right? And since then, there's not really been much discussion from him about it.

是的，对吧？我一直在鼓动特斯拉应该开始购买更多比特币，不一定非要成为一家比特币储备公司，虽然那样也很有趣，但可能有点过头了。但即便只是增持更多比特币，因为我的立场是，随着我们接近AGI和ASI的世界——嗯——那将在许多方面是一个非常不确定的时期。

Yeah. Right? Now, I've been agitating a little bit that Tesla should start buying more Bitcoin, not necessarily become a Bitcoin treasury company, although that could be interesting too, but that may be taking a little bit too far. But even just acquiring more Bitcoin because my stance on this is that as we approach this world of AGI and ASI Mhmm. That that is going to be a very uncertain time in many ways.

我希望任何公司进入那个时期时都尽可能拥有更多资本。嗯。更不用说，随着我们接近那个人形机器人劳动力或仅仅是AI普遍带来颠覆的世界，法定货币的价值可能会进一步受到威胁。他们将不得不——

And I would want any company to go into that with as much capital as possible. Mhmm. And not to mention that as we approach that world where we get disruption from humanoid labor or even just AI in general, the value of fiat money is going to potentially be compromised that much more. They're gonna have

——通过印钞来抵消所有正在发生的技术性通缩。是的。

to offset all of the technological deflation that's happening with Yeah. Printing.

没错，百分之百。所以唯一能安全存储资本的地方就是比特币。是的，那将不受影响。

Yeah. 100%. So the only place within the store capital is going to be in Bitcoin. Yeah. That's gonna be safe from that.

是的。现在特斯拉应该尽可能多地重新投资于他们的业务。这才是限制他们的因素。他们没有回购股票，而是在重新投资业务。

Yeah. Now Tesla should reinvest in their business as much as possible. That's what's limiting them now. They're not buying back stock. They're reinvesting in their business.

目前，他们拥有约370亿美元的现金储备，我想大概是这个数字，还有10亿美元的比特币。所以我希望看到他们逐步增加比特币持仓。像英伟达这样积极回购股票的公司，我认为犯了大错。我们已经看到，投入AI基础设施的资金需求巨大且每个季度都在增长，预估数字越来越高。英伟达现在放弃他们的资本，我认为实际上是他们的重大失误。

Now, they have a cash hoard of about 37,000,000,000, I think it is, you know, and a billion in Bitcoin. So I'd like to see them over time build up their Bitcoin position. Certainly companies like Nvidia that are aggressively buying back stock, think are making a big mistake. And I think we're seeing already that the amount of money that needs to go into AI infrastructure is so large and growing every quarter. We get the estimates are higher and higher and higher that for NVIDIA to give away their capital right now, I think is actually a big mistake on their part.

是的。不，我认为他们会没事的。我不认为这会让公司陷入风险，但他们可能会错过未来的机会。

Yeah. No. I think they'll be fine. I don't think it's putting the company at risk, but they may be missing out on future opportunities.

是的。显然我同意这一点。那是个铁杆比特币支持者。对吧。是的。

Yeah. I would agree with that, obviously. That's a hardcore Bitcoiner. Right. Yeah.

我想和你讨论的最后一个话题是Optimus机器人。我觉得这太令人震撼了。从工程角度看，光是看那些设计视频——甚至不限于特斯拉的，还包括其他从事这一领域的团队——比如他们如何设计人形机器人的脚踝，或是手上那些关节所需的惊人工程量。我记得埃隆最近说过，设计人形机器人的手极其困难，疯狂地难，要协调所有关节，你根本想象不到其中需要多少工程投入。

One last topic that I wanna get into this with you on is the Optimus robot. I find this mind blowing. I find from an engineering standpoint, just watching videos on here's how we designed and not even from Tesla, but just like other people that are operating in this space. Here's how we designed the ankle for our humanoid robot in the crazy amount of engineering into just some of the joints in the hands. I think Elon recently said that designing a humanoid robot's hand is crazy difficult, insanely difficult to get all the joints and, like, you just don't understand how much engineering goes into something like that.

和我们聊聊你对这个项目发展阶段的看法，比如时间线。距离它真正上市还有多远？目标市场会是什么？以及它的用途和功能。

Talk to us about where you see this start off with, like, the timeline. How far out are we from this thing kinda really going to market? What would that market be? And then its use and utility.

首先补充你刚才说的，我问过埃隆造这个是不是比造星际飞船更难？他说不，那是个更困难的问题。所以这样对比来看...对吧？

First of all, just to add on to what you said, I think Elon, when I asked about was it harder than making Starship? And he said, no. That's a harder problem. So to put it in context Yeah. Okay?

没错。所以这可能是埃隆现在研究的第二难的项目。或许人工智能也算一个。我不太确定。

Yeah. So it's the second hardest thing that Elon's working on, I suppose, right now. Yeah. Maybe AI's up there too. I'm not sure.

人形机器人这事非常有趣。在我看来，这是人类历史上首次开发出最终能在所有层面超越人类劳动力的东西。我们将创造一个在任何方面都比人类做得更好的系统。它能随时保持工匠水准，永远不会出现马虎的工作。

The humanoid robot thing is really interesting. In my view, for the first time in human history, we're developing something that can outcompete human labor eventually at all levels. We're going to be developing a system that can do absolutely everything better than a human can. It can be a craftsman at all times. There'll be no sloppy work.

它将每一次都把事情做到完美。不仅如此，它每天大概能工作二十二到二十三小时。是的，单凭这一点，就能与人类竞争。即使运行速度只有人类的三分之一，一个机器人也能在二十四小时内胜过人类。

It'll be doing things perfection every single time. Not only that, it can work probably twenty two to twenty three hours a day. Yeah. So right there, you can compete against humans. Even if it was operating at one third the speed, one robot could outcompete a human over a twenty four hour period.

没错。因为人类每天大概只工作八小时左右。但随着时间的推移，它不仅会达到人类的能力水平，甚至可能超越人类。所以时间线很有意思，因为他们现有的机器人实际上已经很有用了。

Yeah. Right. Because humans only work, let's say, eight hours a day or something like that. But over time, not only will it be as capable of a human, it'll probably become more capable. So the timeline's interesting because the robots that they have today actually are useful.

它们已经展现出能完成有用工作的能力。哪怕只是拿起一个箱子从A点搬到B点，这也是有价值的工作。很多人在工厂和仓库整天做的就是这些，分拣包裹之类的。现在的机器人已经能做到这些了，也许速度不如人类快，但它们不需要那么快。

They are already showing that they can do useful work. Even just picking up a box or something and moving it from point a to point b is useful work. There's a lot of people that do that all day in factories and warehouses and stuff, sorting packages, all that kind of stuff. The robots today can already do that. Now, maybe not as fast as a human, but again, they don't have to be as quick.

人类有各种各样的问题，对吧？我们要休息，会疲劳，会犯错。有人旷工，有人在工作中制造麻烦，甚至醉酒或吸毒后上班。

Humans have all kinds of problems, right? We take breaks, we get tired, we make mistakes. Some people don't show up for work. Some people cause problems at work. They show up drunk or drugged.

这些人形机器人不会有这些问题。所以时间线关键在于：埃隆想什么时候把机器人推向市场？它需要达到多高的生产力和能力水平？今年早些时候我们预计年底前他们会生产五千到一万台机器人，部署在特斯拉自家工厂和供应商处。但由于重新设计手部等原因，这个计划已经推迟了。

All those issues that you won't have with humanoid robots. So the timeline is kind of one of those things is at what point does Elon want to bring a robot to market? How productive or how capable does it need to be? Now, we were expecting earlier this year that by the end of this year, they would have made five to 10,000 robots, and that they'll be deployed within Tesla's own factories and maybe their suppliers. They've stepped back from that because they've been redesigning the hand and so on.

所以时间线稍有延后。我认为最终我们会得到能力更强的机器人，或许2026年才能真正开始量产。埃隆常说通常需要三个版本才能推出商用级机器人，这有点像iPhone的发展历程。

So the timeline has been pushed back a little bit. I think ultimately we're going to have a far more capable robot. And so perhaps we start seeing some production in earnest in 2026. Elon has always said it typically takes three versions before you get a commercial ready robot. To me, this is a bit like iPhone.

对吧？第一代iPhone问世后，下一代就大幅改进，到第三代越来越好。我认为这些机器人一旦开始量产就会有相当不错的需求。有趣的是，随着它们能力提升，其价值也会越来越高。

Right? You come out with the first iPhone, and the next one's that much better. And then iPhone three, and it just keeps getting better and better and better. So I think there'll be actually pretty good demand for these robots as soon as they can really begin making them. And then an interesting thing happens is as they become more capable, they become more valuable.

因此，他们每台机器人实际能赚取的金额最初将会上升。就收入而言，每台机器人每年可能高达15万美元。我一直认为他们应该采用机器人即服务的模式。因为如果你出售机器人，随着机器人能力提升，可能会错失很多潜在收益。当然，我想也可以对软件更新收费。

So the amount of money that they can actually earn per robot will go up initially. Could be as much as $150,000 a year per bot in terms of revenue. And I've always said that they should do a robot as a service. Because if you sell a robot, you might be giving up a lot of upside as that bot becomes more capable. Now, I suppose I could always charge for the updates.

对吧？如果他们出售一台每小时创造5美元价值的机器人，没问题，你按这个功能支付相应价格。但如果想要每小时创造10美元价值的版本，就得为软件升级付费。

Right? If they sell a bot that does, you know, $5 an hour work, fine. You pay a certain price for that. It does $5 an hour work. But if you want the $10 an hour work, you're gonna have to pay for the software update.

在我看来，更简单的方式是直接提供机器人租赁服务，你只需按小时支付机器人的功能使用费——就像现在企业按小时雇佣工人那样。普雷斯顿，我想提一个特别有趣的现象：那些拥有大量员工的企业，虽然员工是公司资产，但他们不再出现在公司资产负债表上。作为劳动力，员工的价值并未体现在财务报表中。

To me, an easier method is just to provide robots as a service, and you just, you know, pay for the capability of the bot on a per hour basis. Kind of like companies do today on for workers on a per hour basis. Let me just touch on, Preston, something that's really interesting that's happening here. If you think about companies that have all these workers, yes, they're assets of the company, but they're not on the company's balance sheet anymore. There's no value of the employees as a workforce that shows up on the company's balance sheet.

也许足球俱乐部除外，对吧？他们为球员支付转会费，还能通过交易球员获利。但普通企业的员工不是这样运作的。而当你拥有机器人时，首次实现了劳动力的资本化。想象一下：如果一个机器人年利润2.5万美元，按40倍市盈率计算...

With the exception perhaps of a soccer or football club, right, where you pay for these players and you get to trade them and you get the money for the players. That doesn't work that way for employees with companies. But what happens is when you have a robot, you get to capitalise that labour for the first time. Right? So think about what happens if you have a robot that makes 25,000 a year in profit, and you use a PE multiple of 40.

那就是百万美元的价值。特斯拉每生产百万台机器人，潜在市值增长就可能达到万亿美元——无论是特斯拉本身还是使用这些机器人的企业。

That's a million dollars in value. For every million robots that Tesla makes, that's a trillion dollars in new market cap potentially. For whoever owns it. Yeah. For Tesla and then also for the companies that are using it.

他们，呃，他们

They, you know, they

哦，你是说如果他们采用租赁模式？对，如果出租的话确实如此。

Oh, you're saying if they would, yeah, if they would rent it out. Yeah.

没错。所以，我们正在见证人类历史上首次可能出现的劳动力资本化。我曾说过，Optimus（特斯拉人形机器人）终将成为特斯拉市值的主要组成部分，或许在未来某个时间点（比如2040年）占到其市值的80%左右。

Yeah. Right. So what we're seeing for the first time in human history is potentially the capitalization of labor. And I have said that Optimus will become the vast majority, maybe 80% of Tesla's market value at some point in the future, let's say by 2040,

甚至可能更早。据我理解，埃隆完全赞同这个观点。他发表过相关声明。是的。

maybe even earlier. My it's my understanding. Elon is in complete agreement with that. He has statements. Yeah.

对吧？

Right?

是的。不是要自夸，但我一年前就说过这个观点，而埃隆最近也提到了。当然，埃隆的远见远超于我，他并非受我启发。关键在于：全球劳动力市场约占世界GDP的50%。

Yeah. Now not to toot my own horn here, but I was saying this, you know, a year ago, and Elon said it recently. Now, again, Elon's way ahead of me. He's not listening to me. But this is a very important point because market for labor in the world is about 50% of world GDP.

如果我们能将其中大部分转化为机器人劳动力，这部分价值将被资本化，相当于凭空创造市值。这就是为什么我认为特斯拉面临巨大机遇——如果他们能成功量产数百万台机器人，每百万台就意味着万亿美元市值。顺便说，埃隆提过要生产数十亿台机器人。想想他通过SpaceX开展的登月、火星计划、小行星采矿等事业，我把太空看作新大陆。届时派去的不会是人类，而是机器人军团。

If we're going to turn a good portion of that into robotic labor, it will be capitalized, and we're going to create market value out of thin air. This is why I think that Tesla has such an incredible opportunity because if they can successfully make millions of robots, or every million that they make is a trillion dollars in market value. Elon, by the way, has talked about making billions of robot. And when you think about too what he's doing with SpaceX and going to the moon and going to Mars and potentially mining asteroids and all the other things, I think of space as kind of like a new continent. And so you're not gonna send people, you're gonna send the bots.

没错。地外应用对机器人的需求简直不可估量。

Yeah. So the off planet uses for robots are just enormous.

完全同意。我想用具体数字帮助大家理解：今年他们的机器人生产目标是5,000台。

Yeah. Right? Totally. I wanna tell people just some numbers on this so that they can kind of wrap their head around it. So this year, they were shooting for 5,000 of these robots to be built.

看起来他们在六月份的生产线上遇到了些麻烦。当时正在组装一千台设备，我猜他们后来回去重新设计了。所以现在传出的数字可能是到今年年底能生产两千到五千台这样的设备。也许吧。到2026年，他们的目标是年产两万到五万台，而到了2027年，则要超过十万台。

It looks like they hit a snag in the production line back in June. They were assembling a thousand units, and I guess they went back for a redesign on that. And so the numbers that are being tossed around now are maybe between two to 5,000 of these things made by the end of this year. Perhaps. Next year, by the 2026, they're shooting for 20 to 50,000 units produced, and then by 2027, over a 100,000 units.

是的。到2030年，他们预计每年能生产一百万台。你大概知道吗，就是随口问问，他们现在每年生产多少辆汽车，好让大家有个概念...

Yeah. By 2030, they're looking at a million units a year being produced. Do you know how many off the top of your head, how many cars they produce per year right now just so people can kind of

对。目前年产量略低于两百万辆。哇。顺便说一下，一个Optimus机器人的重量大约是特斯拉Model 3的3%。所以假设他们每生产一辆车，就能制造十个机器人。

Yeah. It's a little bit less than 2,000,000. Wow. And by the way, a robot, an Optimus robot weighs about 3% of a Tesla Model three. So let's say, for example, that for every car they made, they can make 10 robots.

对吧？虽然重量只有3%，但假设占用了他们10%的产能。所以每生产一辆车就能造十个机器人。如果他们生产一百万个机器人，就相当于生产了十万辆汽车。

Right? So even though it's 3% of the weight, let's say it's 10% of their production capacity. So Yeah. For every car and robots. So if they make a million robots, that's like making a 100,000 cars.

对特斯拉来说，生产十万辆车并不困难。这么说吧，一百万听起来很多，但其实只相当于多生产了十万辆车。或者如果他们每辆车能造二十个机器人，那就相当于生产了五万辆车。

That's not difficult for Tesla to make a 100,000 vehicles. Just to put in context, like a million of something sounds like a lot. Yeah. But that's equivalent to making another 100,000 cars. Or if they can make 20 robots for every car, that's like making 50,000 cars.

他们第一年就生产了那么多电动皮卡，而那还是个非常复杂的产品。他们设计Optimus时要让它能高效快速地生产，甚至可能由Optimus自己来制造机器人，让机器人来造机器人，对吧？如果能做到这点会非常有趣。但他们在早期面临的挑战之一，是如何为这些产品扩大供应链。毕竟这是从零开始打造的新产品。

They made that many Cybertrucks in the first year, and that's a very complex product to make. They're designing Optimus to be made very efficiently, very quickly, and maybe even by Optimus itself, the robots building the bots, right? And that'll be very interesting if they can do that. But the one thing that they're going to have a challenge with, at least in the early years, is scaling up the supply chain for these things. You're building a new product from scratch.

这可能是个挑战。所以在最初几年，我认为我们应该保持较低的期望。但十年后，他们每年可能会生产高达一亿台这样的设备。

It could be a challenge. So in the early years, I think we should keep our expectation pretty low. But ten years from now, they're gonna be pumping out potentially a 100,000,000 of these things a year.

那么用例是一开始就直接进入工厂吗？这是最初的主要用户吗？我猜它不会进入家庭。

And is the use case going to the factories right out of the get go? Is that the primary user from the beginning? I'm assuming it's not going into homes.

嗯，有些其他机器人公司首先专注于家庭市场，我认为这对他们来说是明智的，因为他们可能无法在工厂领域与特斯拉竞争。是的。但最终，我们会在任何需要人类劳动力、存在短缺的地方看到它们。目前制造业就存在劳动力短缺。

Well, there are some other robot companies that are focused on the home first, and I think that's smart for them because they probably can't compete with Tesla on the factory side. Yeah. But eventually, we're going to see them everywhere. Everywhere where human labor is needed, where there's a shortage. There's a shortage in manufacturing right now.

此外，有很多人类从事肮脏、危险或极其枯燥的工作。人们其实并不想要这些工作。看看沃尔玛的员工流动率，每年高达40%。

Also, we have a lot of humans that do jobs that are dirty, dangerous, really boring. Humans don't really want these jobs. Yeah. Look at the turnover that Walmart has. It's like 40% a year.

对吧？货架补货、卸货卡车、搬运物品——这些工作对机器人来说再合适不过了。

Yeah. Right? Stocking shelves and unloading trucks and moving stuff around. It's perfect work for robots. Yeah.

明白吗？所以你不需要解雇任何人，这些岗位会自然减员。我认为大公司会迅速采用这些机器人。但最终，很多人也期待家用机器人，比如作为老人看护或儿童陪伴。

Right? And so you don't have to fire anybody. It's just the natural attrition of these jobs. So I think these big companies will adopt these robots very quickly. But ultimately, there are a lot of people excited to have robots in the homes as caregivers for elderly, as cures for kids.

比如我的孩子们就特别喜欢机器人，他们巴不得家里有个能聊天、帮忙收拾玩具的机器人。所以我认为各个领域都会有巨大需求。但经济回报最显著的是三班倒的工厂。

Like, my kids love robots. They would love to have one in the house that they could talk to and get the robot to pick up their toys and that kind of stuff. Right? So I think there'll be a big demand from all different corners. But the economic win is factories that run three shifts.

没错。因为这些机器可以每天工作22、23个小时。天啊。所以工厂可能愿意每年支付几十万美元购买机器人，但初期很少有消费者愿意承担这个价格。

Yeah. Because you can use these things, you know, twenty two, twenty three hours a day. Good lord. And so a factory may be willing to pay a couple $100,000 a year for a robot. There are not many consumers that would be willing to pay that much initially.

没错。对吧？随着生产规模扩大，这些东西的成本会大幅下降，最终变成另一种消费品。

Yeah. Right? Over time, as they scale up production, the cost of these things will come way down, and it'll become another consumer product.

确实。先生，我非常享受这次对话，我们以后一定要再约，或许可以每季度一次。天啊，您对这些话题的广博见识令人惊叹，非常感谢您抽空来分享这些见解。

Yeah. Sir, I've thoroughly enjoyed this con we have to do this again in the future, maybe quarterly or something. But wow. You're just a breadth of knowledge knowledge on all of these different topics. I can't thank you enough for making time to come on and talk about this.

如果想让大家能关注您或了解更多信息，您有什么推荐渠道吗？

If you wanna give people a hand off to where they can follow you or learn more, what do you got?

对，在X平台上找我就行。我的账号是@CERNBasher，用我的本名注册的——顺便说一句，Basher确实是我的真名，我很喜欢这个名字。

Yeah. Just find me on X. My at CERN Basher, my first name and last name, at CERN Basher. That is my real name, by the way. I love it.

有些人以为我是粒子对撞机之类的，毕竟叫Basher嘛，但其实不是，这就是我的本名。

Some people think that I'm a particle accelerator, you know, Basher, but no. That's my real name.

这个名字在科技圈简直完美，先生。我想也是。好的，我们会附上您X账号和公司的链接，这些建议都非常棒。

It's perfect for being in the tech space, sir. I suppose it is. Alright. We'll have links to your ex account and also your company. They're brilliant advice.

再次感谢您参加节目，先生，这次谈话真的非常愉快。

And thank you, sir, for coming on the show. This was a really pleasant conversation.

这是我的荣幸，普雷斯顿。非常感谢你。

It was my pleasure, Preston. Thank you very much.

感谢收听TIP节目。请务必在您喜欢的播客应用上关注Infinite Tech，不错过我们的任何一期节目。访问investorspodcast.com可获取节目笔记和相关课程。本节目仅供娱乐，决策前请咨询专业人士。

Thank you for listening to TIP. Make sure to follow Infinite Tech on your favorite podcast app and never miss out on our episodes. To access our show notes and courses, go to the investorspodcast.com. This show is for entertainment purposes only. Before making any decisions, consult a professional.

本节目版权归The Investor's Podcast Network所有。未经书面许可不得进行联合播出或转播。

This show is copyrighted by The Investor's Podcast Network. Written permissions must be granted before syndication or rebroadcasting.