擎天柱：有史以来最伟大的产品？

女士们先生们，欢迎埃隆·马斯克。好的。你在帕洛阿尔托，而不是华盛顿特区。

Ladies and gentlemen, please welcome Elon Musk. All right. Good. All right. Where are you? Alto. You're in Palo Alto and not Washington DC.

我在帕洛阿尔托的特斯拉全球工程总部。

I'm at Tesla Global Engineering Headquarters in Palo Alto.

是的，所以不再去华盛顿特区了。你回到工作中了，很专注。

Yeah. So, no more Washington DC. You're back at work. You're focused. Yeah.

呃，是的。我自五月以来就没去过华盛顿。

Uh, yeah. I haven't been to DC since May.

好吧。那真是一次疯狂的支线任务。有什么从华盛顿特区经历中学到的教训吗？

Okay. Uh, that was a hell of a side quest. That was a good. Any lessons from your time in Washington DC?

呃，政府基本上无法修复。仅代表我个人。我支持大卫的崇高努力，有才华的人进入政府是好事，但归根结底，看看我们高得离谱的国债，利息支付已经超过了国防部的预算，所以如果人工智能和机器人不能解决我们的国债，我们就完蛋了。

Uh, the government is basically unfixable. Elon only. I support David's noble efforts and it's good to have talented people in the administration, but at the end of the day, if you look at our national debt, which is insanely high, the interest payments exceed the US defense department budget, and Nikki Bryzy, so if AI and robots don't solve our national debt, we're toast.

这是个很好的过渡。我认为 Optimus 将成为人类历史上最伟大的产品。进展如何？你多少精力花在 Optimus 上？时间线是怎样的？我想你们在第三版，也许是第四版。告诉我们一切。

Which is a great segue. Optimus is, I think, going to be the greatest product in the history of humanity. What's the progress like and how much of your cycles are going specifically to Optimus? What's the timeline? I think you're on version three, maybe four. Tell us everything.

呃，好吧，全部讲完要花很长时间。

Uh well, yeah, everything would take a long time.

我们有时间。

We've got time.

嗯，我们正在最终确定 Optimus 第三版的设计，那将是一个非常卓越的机器人。它将拥有人类的手工灵巧性，即非常复杂的手。人工智能大脑能够导航和理解现实，并且它将大规模生产。这是缺失的三样东西。如果你看任何其他机器人公司，他们都缺少这三样。这是三个真正困难的事情。实际上，在这一点上，我花在 Optimus 上的脑力可能比任何其他单一事情都多。这是为了解决现实世界的人工智能，Optimus 的所有机电问题，供应链和生产挑战，因为人形机器人不存在现成的供应链。所以我们必须从头开始重建，这需要大量的垂直整合。Optimus 中的任何执行器都无法从现有供应链获得。所以，但我认为可以准确地说，如果成功，Optimus 将是有史以来最大的产品。

Um, we're finalizing the design of Optimus version 3, and that really is going to be a very remarkable robot. It will have essentially the manual dexterity of a human, meaning a very complex hand. The AI mind can navigate and comprehend reality, and it will be made in very high volume. Those are the three things that are missing. If you see any other robotics company, they're missing those three things. Those are the three really hard things. And I actually spend, at this point, it might be more of my mental cycles than anything else, any other single thing, on Optimus. That's solving for real-world AI, all of the electromechanical issues of Optimus, the supply chain and production challenges, because there is no supply chain that exists for humanoid robots. So we have to recreate it from scratch, which requires a lot of vertical integration. None of the actuators in Optimus are available from an existing supply chain. So, but I think it is accurate to say that if successful, Optimus will be the biggest product ever.

大规模生产后的成本，2030 年，每台机器人 4 万美元。你认为第一批会卖多少钱？我们什么时候能买一个在农场干活？

And the cost of it at scale, 2030, $40,000 a robot. What do you think the first wave of them will cost? And when will we be able to buy one to work on the ranch?

我认为一旦年产量达到 100 万台，边际生产成本大概在 2 万美元左右。这取决于你在机器人 AI 芯片上花了多少钱。而且你需要在执行器上实现很多效率提升。每只手臂有 26 个执行器，比如 26 个电机、变速箱和功率电子器件。所以，但 AI 芯片会很贵，可能占物料清单的 5000 或 6000 美元，甚至更多。但我认为在每年 100 万台的规模下，生产成本大概在 2 万美元左右，也许 2.5 万，类似这样。价格将取决于需求。

I think that the marginal cost of production once you hit a million units per year is probably around the $20,000 range. It sort of depends on how much you spend on the AI chip in the robot. And you need to achieve a lot of efficiencies in the actuators. There are 26 actuators per arm, like 26 electric motors, gearboxes and power electronics. So, but the AI chip will be pretty expensive, like that might be $5,000 or $6,000 of the bill of materials, maybe more. But I think at volume, at a million units a year, the production cost is probably on the order of $20,000, maybe $25,000, something like that. And price will be a function of demand.

埃隆，你能给大家解释一下为什么手部如此重要，为什么执行器设计如此独特，为什么这么困难，为什么没人做，以及为什么你必须从那里开始才能正确构建机器人的其余部分。

Elon, can you maybe explain to everybody why the hand is so important to get right and why the actuator design is so unique and why it's so difficult, why nobody makes it, and why you have to start there almost to build the rest of the robot properly.

嗯，事实证明人类的手进化成了极其精密的机器。你的手实际上是一个了不起的东西。仔细看看你的手，想想你能用手做多少事情，非常多。

Well, it turns out the human hands are incredibly evolved to be this incredibly sophisticated machine. Your hand is actually a remarkable thing. Look closely at your hands and think of all the things you can do with your hands, which is a lot.

我能想到很多事。

I can think of many things.

是的，我刚才在想一些事。

Yeah, I was just thinking about something.

你知道，你的手是非常多功能的工具。是的，你可以击掌。

You know, your hands are very versatile instrument. Yeah, you can give him a high five.

非常多功能。你可以挥棒球棒，可以穿针引线，可以弹钢琴或小提琴，可以拆卸或组装汽车。手是非常多功能的工具。而且手的大部分肌肉实际上在前臂。所以你的手有点像木偶。肌肉来自前臂，拉动肌腱，肌腱也因进化而非常出色。所以你有一个肌腱网。人类的手有大约 27 或 28 个自由度，取决于你怎么数。太神奇了。所以为了创造一个通用的人形机器人，你必须解决手的问题。

Very versatile. You can swing a baseball bat, you can thread a needle, you can play the piano or violin, you can disassemble or assemble a car. The hands are incredibly versatile instruments. And most of the muscles of the hand are actually in the forearm. So your hand is kind of like a puppet. The muscles come from the forearm and pull the tendons, which are also incredibly good due to evolution. So you have this web of tendons. The human hand has something like 27 or 28 degrees of freedom, depending on how you count. It's amazing. So in order to create a robot that can be a generalized humanoid, you must solve the hand problem.

是的。我们有，它有手，需要手。所以这是不是像你最初建造特斯拉时那样，供应链不存在，现在你必须出去找人合作，建立所有这些垂直整合，获得支持。是不是真的无处可寻，你必须从头建立这一切？

Yeah. We had, it's got hands, needs hands. And so is it like when you were first building Tesla where the supply chain doesn't exist and now you have to go out and find folks to work with and build all this vertical integration, get support. Is it literally like it's just nowhere to be found and you're going to have to build all of this stuff up?

是的，我们实际上花多少钱都买不到执行器。它们根本不存在。尽管有成千上万种不同尺寸和形状的电机。

Yes, we could not actually buy the actuators for any amount of money. They simply didn't exist. Even though there are, I don't know, 10, 20,000 electric motors out there of various sizes and shapes.

我们不得不从零开始设计每一个电动机、变速箱和控制电子设备，基本上是从物理第一性原理出发。

We've had to design every electric motor, gearbox and controlling electronics from scratch, basically from physics first principles.

好消息是你在过去几十年里积累了大量工厂经验。那么，与 Cybertruck、Model Y、Model X、Gigafactory 相比，这有多难？

The good news is you've got a lot of experience with factories over the last couple of decades. So how challenging is this versus Cybertruck, Model Y, Model X, Gigafactory?

是的，那个被称为法贝热彩蛋的 Model X。它比那些都难。难得多，显著地。星舰？不，星舰更难。所以介于 Model X 和星舰之间。

Yeah, the Faberge Egg known as the Model X. It's harder than any of those things. Much harder, significantly. Starship? No, Starship's harder. So somewhere between a Model X and a Starship.

硬件和软件哪个更难？

What's harder, the hardware or the software?

目前，我们在硬件的最终设计上挣扎。就像我说的，主要是手。不是说机器人其他部分不重要，但手，包括前臂，占了整个机器人工程难度的大部分。

Right now, we're struggling with the final design of the hardware. Like I said, it's really primarily the hand. Not to dismiss the rest of the robot, but the hands, inclusive of the forearm, are a majority of the engineering difficulty of the entire robot.

假设你克服了硬件挑战。基于 LLM 的所有进展，你能免费获得多少？消费者能跟这个机器人互动、跟它说话、让它做事，它能理解吗？

Let's assume you get past the hardware challenges. How much do you get for free based on all the progress with LLMs? Will consumers be able to interact with this robot, talk to it, ask it to do things, and it'll understand?

哦，是的，没问题。

Oh yeah, no problem.

我注意到你在网上花了很多时间在 AI 上。

You're spending a lot of time with AI, I noticed online.

没那么久。也许我对 Bunning Grom Imagine 有点过头了，但是……

Not that long. Maybe I went a little over the top with Bunning Grom Imagine, but...

但说真的，那些角色和这些机器人似乎是……我想你可以得到 Annie 的具身。为什么是人形，Elon？你可以做出比人类更好或更简单的东西来完成特定任务。你如何决定把它做得像人类一样？

But in all seriousness, those characters and these robots seem to be... you could get the embodiment of Annie, I suppose. Why the human form factor, Elon? You could make something better or simpler than a human for specific tasks. How do you decide to make it just like a human?

嗯，如果你想做人类能做的所有事情，结果是你需要一个类人机器人。如果你只想做一部分，那就容易得多。但事实证明，人类进化成这种形状和能力是有充分理由的。有五根手指和拇指是有价值的，甚至小指也很有用。脚趾更多是为了激进行走，但手指……另外，人类为我们设计了世界。所以如果你能制造一个类人机器人，它将立即与我们为世界建造的东西向后兼容。

Well, if you wanted to do all the things that a human can do, it turns out you need a humanoid robot. If you want to do just a subset, that's much easier. But it turns out humans evolved to this shape and capabilities for good reasons. There is value to having five fingers and the thumb, and even the pinky is quite useful. Toes are more for aggressive walking, but the fingers... Also, humans have designed the world for us. So if you can make a humanoid robot, it'll be immediately backwards compatible with what we've built the world for.

正是如此。

Precisely.

机器人还有另一部分：LLM、手的驱动，还有运行它的硅片。有 Dojo，你在 X 上发了 AI5 和 AI6。你似乎对硅片层的方向非常兴奋。能告诉我们吗？正在建造什么？是补充还是潜在的长期竞争对手？

There's another part of the robot: the LLMs, the actuation in the hands, but also the silicon that runs it. There was Dojo, and you posted on X AI5 and AI6. You seemed incredibly excited about the direction of the silicon layer. Can you tell us about that? What is being built? Is it a complement or a potential long-term competitor?

在特斯拉，我们有两个不同的芯片项目：一个用于训练侧的 Dojo，另一个是我们称为 AI4 的推理芯片，目前在所有车辆中出货。我们正在最终确定 AI5 的设计，它将是从 AI4 的巨大飞跃。根据某些指标，AI5 的改进将是 AI4 的 40 倍。这是因为我们在 AI 软件和硬件上以非常精细的粒度紧密合作，所以我们确切知道限制因素在哪里。AI 硬件和软件团队正在共同设计芯片。

At Tesla, we had two different chip programs: one for Dojo on the training side, and what we call AI4, our inference chip, currently shipping in all vehicles. We're finalizing the design of AI5, which will be an immense jump from AI4. By some metrics, the improvement in AI5 will be 40 times better than AI4. This is because we work closely at a very fine grade level on the AI software and hardware, so we know exactly where the limiting factors are. The AI hardware and software teams are co-designing the chip.

所以硅片有 40 倍的改进。对观众来说，这是否意味着 FSD 质量和特斯拉驾驶员安全以及机器人质量的量级提升？当你投入生产时，这一切体现在哪里？

So a 40x improvement in the silicon. For the audience, does that mean an order of magnitude increase in the quality of FSD and safety for Tesla drivers, and the quality of the robot? Where does it all manifest when you bring it into production?

准确地说，40 倍是与 AI4 最差的限制相比，即运行 softmax 操作。目前我们必须在模拟模式下用大约 40 步运行 softmax，而在 AI5 中只需几步本地完成。AI5 还将轻松动态处理混合精度模型。就名义原始算力而言，算力提升 8 倍，内存提升约 9 倍，内存带宽提升约 5 倍。但由于我们解决了 AI4 的核心限制，你将 8 倍的算力改进乘以另外 5 倍的改进，这些改进来自对 AI4 中当前次优的非常细粒度硅片级别的优化。这就是 40 倍改进的来源。

To be precise, the 40x is compared to the worst limitation on AI4, which is running the softmax operation. We currently have to run softmax in around 40 steps in emulation mode, whereas that'll be done in a few steps natively in AI5. AI5 will also easily handle mixed precision models dynamically. In terms of nominal raw compute, it's eight times more compute, about nine times more memory, roughly five times more memory bandwidth. But because we're addressing core limitations in AI4, you multiply that 8x compute improvement by another 5x improvement from optimization at a very fine grain silicon level of things currently suboptimal in AI4. That's where you get the 40x improvement.

我相信目前车内的 AI4 芯片将实现至少是人类两到三倍、甚至可能 10 倍的自驾安全性。相关软件将在未来几个月内发布。版本 14 将是特斯拉软件自版本 12 以来最大的升级。我们将参数数量提高一个数量级。使用了大量强化学习。你可以把 AI 看作压缩现实的一种方式，其中一些压缩步骤损失太大，我们已经解决了。所以通过这些更新，到今年年底，你的车会感觉像是有知觉的。

I am confident that the current AI4 chips in the cars will achieve self-driving safety that is at least two to three times that of human, and maybe even 10x. The software for that will be released over the next few months. Version 14 will be the biggest upgrade in Tesla software since version 12. We are increasing the parameter count by an order of magnitude. There's a lot of reinforcement learning that's been used. You can think of AI as a way of compressing reality, and some of those compression steps were too lossy, and we addressed that. So over these updates, your car is going to feel like it is sentient by the end of the year.

是的，说实话，已经感觉这样了。我在行业新闻中看到你花了大约 170 亿美元购买频谱，以使你的卫星和 Starlink 网络直接连接手机。一两年后会是什么样子？我们会放弃 Verizon 账户，直接扩展我们的 Starlink 账户吗？

Yeah, it feels that way already, to be honest. I saw in the trades that you spent about $17 billion on some spectrum to enable your satellites and Starlink network to connect directly with phones. What will that look like in a year or two? Are we going to drop our Verizon account and just expand our Starlink account?

谢谢。

Thank you.

我们有点希望，因为 Verizon 有点烂。

We're kind of hoping, because Verizon kind of sucks.

你们中有多少人想要一部星链手机？谁想要星链手机？这在技术上可行吗？我知道你们看不见，但所有人都举手了。

How many of you want a Starlink phone? Who wants a Starlink phone? Is it technically possible? I know you can't see it, but it's everyone.

是的，好的，酷。所以这是一个长期的事情。它将允许 SpaceX 直接从卫星向手机提供高带宽连接。但手机需要一些硬件改动。由于当前手机不支持这些频率，芯片组必须修改以添加这些频率，这大概需要两年时间。所以能够使用所获频谱的手机大约在两年后开始发货。我们还需要建造能通过这些频率通信的卫星。因此，我们并行地建造卫星，并与手机制造商合作，将这些频率添加到手机中。然后卫星和手机将很好地握手，实现高带宽连接。但最终效果是，你应该能在任何地方用手机看视频。

Yeah, all right, cool. So this is a kind of a long-term thing. It will allow SpaceX to deliver high bandwidth connectivity directly from the satellites to the phones. But there are hardware changes that need to happen in the phone. Since these frequencies are not supported in current phones, the chipset has to be modified to add these frequencies, and that probably is a 2-year time frame. So the phones that are able to use the spectrum that was acquired probably start shipping in around 2 years. And we also need to build the satellites that are going to communicate on those frequencies. So, in parallel, we're building the satellites and working with the handset makers to add these frequencies to the phones. Then the satellites and the phones will handshake very well to achieve high bandwidth connectivity. But the net effect is that you should be able to watch videos anywhere on your phone.

哇。那太疯狂了。这些频率在室内建筑里能用吗，就像你现在的手机那样？

Wow. And it's going to be crazy. And do these frequencies work indoors inside buildings, like your phone currently does?

如果你在厚金属屋顶的建筑物里，那不行。但普通住宅，可以。

If you're in a building with a thick metal roof, then no. But normal homes, yes.

埃隆，你的愿景是不是这样：我们不再需要 AT&T 账户，然后在英国或印度时漫游，而是可以直接与星链达成一个协议，最终在全球范围内使用？不是今天，但某个时候。这是最终目标吗？我们不再需要区域运营商，而是有一个全球运营商，那就是你。

Elon, is your vision for this that instead of having an AT&T account and then roaming when you're in the UK or India, we could have one direct deal with Starlink that works all over the world eventually? Not today, but at some point. Is that the end goal? That we don't need a regional carrier, we have a global carrier and that would be you.

那会是选项之一。明确一下，我们不会让其他运营商倒闭。他们还会存在，因为他们拥有大量频谱。但是的，你应该能够拥有一个星链账户，就像你有 AT&T、T-Mobile 或 Verizon 一样，它与家里的星链天线配合使用，提供免费 Wi-Fi，也能在手机上使用。这将是一个全面的解决方案，既满足家庭高带宽需求，也满足手机直连高带宽需求。

That would be one of the options. To be clear, we're not going to put the other carriers out of business. They're still going to be around because they own a lot of spectrum. But yes, you should be able to have a Starlink account, like you have an AT&T or T-Mobile or Verizon, that works with your Starlink antenna at home, free Wi-Fi as well as on your phone. It would be a comprehensive solution for high bandwidth at home and for high bandwidth direct to cell.

你会买下一些运营商来获得更多频谱吗？也许你可以买下 Verizon。

Could you buy some carriers to have more spectrum? Maybe you could buy Verizon.

并非不可能。我想这可能会发生。

Not out of the question. I suppose that may happen.

我们来谈谈星舰。你刚刚进行了一次非常出色的发射。它距离可预测并准备投入商业使用还有多远？

Let's talk about Starship. You just had a really phenomenal launch. How close is it to being predictable and ready to go in a commercial setting?

我认为我们明年会回收飞船。我们还有一次星链版本二的发射；版本二设计中只剩下一枚助推器和一艘飞船。之后就是版本三，这是一个巨大的升级，因为它配备了猛禽 3 发动机。火箭上几乎所有东西在版本三中都变了。所以版本三可能会有一些初期磨合问题，因为这是一个彻底的重新设计。但它能够完全可重复使用地将超过 100 吨送入轨道。我认为，除非我们遇到一些非常重大的挫折，SpaceX 将在明年展示完全可重复使用能力，捕获助推器和飞船，并将超过 100 吨送入有用轨道。

I think we'll recover the ship next year. We've got one more launch of the Starlink version two stack; there's only one booster and ship left in the version two design. Then thereafter it's version three, which is a gigantic upgrade because it's got Raptor 3. Pretty much everything changes on the rocket with version 3. So version 3 might have some initial teething pains because it's such a radical redesign. But it's capable of over 100 tons to orbit fully reusable. And I think unless we have some very major setbacks, SpaceX will demonstrate full reusability next year, catching both the booster and the ship, and being able to deliver over 100 tons to a useful orbit.

目前世界上最好的火箭在运载吨位方面能做到多少？

What does the best rocket in the world do now in terms of tonnage to space?

就商业火箭而言，有猎鹰重型，在侧助推器可重复使用的情况下，大约能运 40 吨。

In terms of commercial rockets, there's Falcon Heavy, which with side booster reuse will do about 40 tons.

所以这是五倍大？

So this is five times bigger?

嗯，有效载荷是两倍半大，但星舰将是完全可重复使用的。

Well, two and a half times bigger in payload, but Starship would be full reusability.

明白了。所以一切都返回。

Got it. So everything comes back.

埃隆，在失败发射发生爆炸后，就是最近那次大爆炸，有很多声明说会有环境、FAA 等各种审查。但再次回到发射台的速度快得惊人。你们怎么这么快就回来了？不仅是技术和工作方面，还有监管审批方面？

Elon, after the explosion that happened with the failed launch, the more recent one with the big boom, there was a lot of proclamations that there's going to be environmental and FAA and all these other sorts of reviews. The recovery back to the launchpad again was incredibly fast. How did you get back so fast? Not just technically and work-wise, but regulatory clearance-wise?

嗯，确实有很多问题和审查。我们都通过了。这要归功于 SpaceX 团队。他们工作非常努力，完成了下一次飞行和助推器的测试，并将其送上发射台并成功飞行。非常为 SpaceX 团队感到骄傲，他们在恢复方面做得非常出色。

Well, there were a lot of questions and reviews. We got through them all. Credit to the SpaceX team. They worked incredibly hard and got the next trip and booster tested and on the pad and flown. Huge credit to the SpaceX team. Very proud of them for doing such a great job recovering.

制造完全可重复使用的轨道火箭是有史以来最热门的工程问题之一，当然也是最困难的工程项目之一。至少能排进前三。这是 SpaceX 从一开始就有的目标，从 2002 年开始。23 年后的今天，我们走到了这里。这是一段漫长的旅程，团队非常有才华，是有史以来最优秀的火箭工程师团队。我认为明年我们终于能够实现完全可重复使用。

Creating a fully reusable orbital rocket is one of the hottest engineering problems ever, certainly a candidate for most difficult engineering project ever. It's on the podium at least. That's been the goal of SpaceX from the beginning, from 2002. Here we are 23 years later. It's a long journey with a super talented group, by far the most talented group of rocket engineers ever assembled. And we're finally next year, I think we'll be able to achieve full reusability.

埃隆，从现在到完全可重复使用，你关注的主要技术障碍是什么？有没有一些让你绞尽脑汁的关键难题，还是更多是处理一系列经验教训并将其整合到下一次发射中？

Elon, what are the big technical blockers that you're focused on between now and full reusability? Are there some showstoppers where you're obsessing over trying to figure out, or is it more about getting through a laundry list of learnings and integrating them into the next launch?

对于飞船的完全可重复使用，热防护罩还有很多工作要做。从来没有人制造过完全可重复使用的轨道热防护罩。航天飞机的热防护罩每次飞行后都需要九个月的维修。所以从来没有人制造过完全可重复使用的轨道热防护罩。

For full reusability of the ship, there's still a lot of work that remains on the heat shield. No one's ever made a fully reusable orbital heat shield. The shuttle heat shield had to go through nine months of repair after every flight. So no one has ever made a fully reusable orbital heat shield.

那是一个材料科学问题还是工程问题，还是两者都有？

And is that a material science problem or an engineering problem or both?

这是一个材料科学工程问题。我们确实在研究基础物理。

It's a material science engineering problem. We're really looking at the fundamental physics here.

再次从物理第一性原理出发：想办法制造一种能承受高温、非常轻、不将热量传递到主结构、且保持完整性的东西。然后上升时如果遇到雨，瓦片不会溶解。有很多不同的问题。你需要在不进行繁琐检查的情况下知道这些瓦片是否正常工作。所以真的需要数万块瓦片全部正常工作，不需要像航天飞机那样逐一翻新或检查。

Again, physics first principles: trying to figure out how to make something that can withstand the heat, is very light, doesn't transmit heat to the primary structure, and whose integrity is maintained. Then as you ascend, if you hit rain, the tiles don't dissolve. There are a lot of different issues. You really need to know these tiles are working without laborious inspection. So it really needs to be that tens of thousands of tiles all work and don't need to be refurbished or checked one by one, as was the case with the shuttle.

是的。

Yeah.

我们能换个话题吗？你谈到了特斯拉，然后是 SpaceX。现在我想问你一些关于 Grok 和 XAI 的问题。你想给我们一个更新吗？我想你谈到了下一个模型的状态，你说了一些令人难以置信的事情。我仍然认为人们没有真正理解：将会有一次新的训练运行，你预计不是从普通网络和 Common Crawl 开始，而是使用大量合成数据。请告诉我们 Grok 的进化情况、这项创新以及为什么它如此重要。

Can we maybe switch now? You talked about Tesla, then SpaceX. Now I'd like to ask you some questions about Grok and XAI. You want to give us an update? I think you talked about where the next model is and you said something incredible. I still don't think people really understand: there's going to be a next training run where you expect not to start from common web and Common Crawl, but an enormous amount of synthetic data. Just tell us how the evolution of Grok is going, this innovation, and why it's so important.

是的。我们使用大量推理算力和推理来审视所有源数据，这实际上是人类知识的语料库。然后我们思考每条信息，补充缺失的内容，纠正错误，并从训练数据中移除虚假信息。例如，以维基百科为例，但这适用于书籍、PDF、网站、各种形式的信息。Grok 使用大量推理算力来查看一个维基百科页面，并判断：哪些是真实的、部分真实的、虚假的或缺失的？然后重写页面，移除虚假信息，纠正半真半假的内容，并补充缺失的上下文。

Yeah. So we're using a lot of inference compute and reasoning to look at all the source data, which is really the corpus of human knowledge. Then we think about each piece of information, add what's missing, correct mistakes, and remove falsehoods from the training data. For example, take Wikipedia, but this applies to books, PDFs, websites, every form of information. Grok uses heavy amounts of inference compute to look at a Wikipedia page and say: what is true, partially true, false, or missing? Then rewrite the page to remove falsehoods, correct half-truths, and add missing context.

顺便说一句，埃隆，你能把这个发布出来吗？我们能创建一个类似 Groipedia 的东西吗？我的意思是，那会很棒。维基百科太有偏见，而且是一场持续不断的战争。如果某件事被纠正，五分钟后就会有一群人试图恢复原状。它已经变得极度党派化，到处都是活动家。所以如果你把维基百科修复成真相的来源，把它发布出来让全世界看到就太好了。

Well, Elon, by the way, could you just publish that? Could we create like a Groipedia? I mean, that would be great. Wikipedia is so biased and it's a constant war. If something gets corrected, five minutes later there'll be an army of people trying to revert it. It's become hyperpartisan and there are activists all over it. So if you fix Wikipedia as a source of truth, it'd be great to publish that just so the world has it.

是的，尤其是我们的个人简介页面，简直是一场灾难。

Yeah, especially for our bio pages, which are a disaster.

好的，我会和团队谈谈，比如 Groedia 之类的。这就是 Groedia 版本。会很有趣的。是的，就把它放出来几分钟。

All right, I'll talk to the team about that, like Groedia or whatever. This here's the Groedia version. It'd be interesting. Yeah, and just have it out there for a few minutes.

关于训练 Grok 5，你正在扩大孟菲斯 Colossus 超级集群的规模。你能给我们一个更新吗？另外，我们在缩放定律方面处于什么位置？如果扩大集群规模，你会得到更强大的 AI 模型吗？是否存在收益递减点？需要多少算力？如果你投入两倍的算力，模型会好 10%还是 100%？是对数线性吗？硬件扩展还有多少潜力？

In terms of training Grok 5, you're scaling up your supercluster in Colossus in Memphis. Can you give us an update on that? And also, where are we in the scaling laws? If you scale a bigger cluster, do you get a more powerful AI model? Is there a point of diminishing returns? How much more compute? If you throw twice as much compute at it, do you get a 10% better model or 100% better model? Is it log-linear? How much more juice is there left in scaling hardware?

我认为算力与智能之间存在一个自然对数函数关系。打个比方，10 倍的算力会使智能翻倍。这可能是一个粗略的经验法则。但这仍然意味着从 100 IQ 到 200 IQ，仍然是一个很大的进步。我认为我们会看到智能持续扩展，直到利用太阳的大部分能量进行计算，最终是银河系的大部分能量，也就是 K2、K3 级别的算力。一旦你把 AI 视为智能整体升级的一部分，而不是一个终点，人类智能也随着人口增长和信息存储而扩展。现在，由于人口下降和低增长率，人类智能正在趋于平稳，实际上还会下降。我的猜测是，最快明年我们就会有 AI 在任何事情上比任何单个人类都聪明。

I think there's a natural logarithmic function associated with the amount of compute. For argument's sake, 10x more compute will double the intelligence. That might be a rough rule of thumb. But that still means you go from 100 IQ to 200 IQ, still a pretty big deal. I think we'll see intelligence continue to scale all the way up to where most of the power of the sun is harnessed for compute, and then ultimately most of the power of the galaxy, sort of K2, K3 scale compute. Once you think of AI not as a destination but as part of the overall escalation of intelligence, human intelligence has also scaled as population increased and we stored more information. Now, due to population declines and low growth rate, human intelligence is plateauing and will actually decline. My guess is we might have AI smarter than any single human at anything as soon as next year.

哇。

Wow.

是的，然后大概五年内，比如 2030 年，AI 会比所有人类加起来还聪明。

Yeah, and then probably within five years, say 2030, AI is smarter than the sum of all humans.

你认为人类因为 AI 的进化而衰落吗？你认为地球生态系统正在发生我们并不真正理解的进化吗？

Do you think humans are on the decline because AI is evolving? Do you think there's an evolution of the ecosystem on Earth that we don't really understand?

也许我们潜意识里知道它要来了。我希望出生率能回升。我显然是提高出生率的坚定支持者。

Maybe we implicitly know it's coming. I hope birth rates turn around. I'm a big proponent of increased birth rate, obviously.

那你为此做了什么吗？

Well, you doing anything about it?

是的，我在努力树立一个好榜样。我们在这次会议上进行了一次意想不到的深入讨论：自杀式同理心、西方、下降的出生率。我注意到你对此非常活跃，还有开放边境。开放边境就像让入侵者进来。这三者可能是同一回事吗？

Yeah, I'm trying to set a good example. We had a big conversation at this conference we didn't expect: suicidal empathy, the West, declining birth rate. I noticed you've been pretty active about it and open borders. Open borders is like let the invaders in. Could all three be the same thing?

西方似乎有很多自杀式症状。最明显的是出生率低于更替水平。显然，如果这种情况无限期持续下去，西方将无法繁衍足够的人口来替代自己。但还有其他事情：边境完全开放，以至于西方文化和社会结构开始瓦解。你在欧洲尤其能看到这一点，英国、法国或德国的本土文化正开始可能被那些被引入且不融入的文化所取代。还有犯罪。现在社交媒体上有一个案例，一个名叫伊琳娜的年轻女子在地铁上被毫无意义地杀害。这本身已经够可怕了，但精英媒体却拒绝报道，好像它不存在一样。

It seems like there are a number of symptoms of the West being suicidal. The most obvious one is the birth rate is not at replacement level. Obviously if that continues indefinitely, the West will literally not reproduce enough to replace itself. But there are other things too: the fact that borders were totally opened to the point where Western culture and social fabric start to come apart. You see this especially in Europe, where indigenous cultures of the UK, France, or Germany are starting to potentially be taken over by cultures of people who are brought in and aren't assimilating. You have crime. We have this case on social media right now, this young woman, Irina, who was killed in a senseless way on a subway. That's horrific enough in itself, but then the elite media just refused to cover it, like it didn't exist.

所以，你看到犯罪问题没有被解决甚至没有被承认，完全没有承认。这几乎就像我们在试图否认螺旋式下降的现实。所以你有所有这些数据点，似乎表明西方是自杀式的，或者似乎不想保卫自己或繁衍自己。听着，我想这个房间里的每个人都认为生活很棒，对吧？我是说，它相当美好，值得活下去。当 Alex Karp 今天早些时候在这里为西方辩护时，那是会议上最响亮的掌声之一。所以，我想我们可能真的不明白发生了什么。你的看法是什么，Elon？你对西方自杀的看法是什么？

So you have this issue of crime that's not being addressed or even acknowledged, and no acknowledgement of this. It's almost like we're trying to deny the reality of the spiral. So you have all these data points that seem to suggest that the West is suicidal or doesn't seem to want to defend itself or propagate itself. Look, I think everyone in this room thinks that life is awesome, right? It's pretty great and worth living. When Alex Karp was here earlier today defending the West, that got some of the loudest applause at the conference. So I guess we probably don't really understand what's going on. What's your take, Elon? What's your take on the suicide of the West?

我非常担心。我非常担心。我认为西方的行为与自杀无异。但至少在美国，普遍有一种乐观感，但你上次和一个住在欧洲的欧洲人谈话，对方是乐观的，是什么时候？

I'm very worried about it. I'm very worried about it. I think the actions of the West are indistinguishable from suicide. But at least in America, there's generally a sense of optimism, but when's the last time you talked to someone from Europe who lives in Europe who's optimistic?

有一阵子了。几十年，哪怕一个。很罕见。

Not for a while. Decades, even one. It's rare.

所以我认为，除非人们对未来有乐观和使命感，否则自杀可能就是会发生的事。生孩子是对未来的乐观行为。所以如果你不乐观，那么……所以我认为我们可能需要给人们一种对未来的乐观和兴奋感，以及相信未来会比过去更好，这样他们就会更有兴趣生孩子。

So I think unless people have a sense of optimism and purpose about the future, suicide might be just what happens. Having a child is an act of optimism about the future. So if you're not optimistic, then... So I think we need to maybe give people a sense of optimism and excitement about the future and a belief that the future will be better than the past, and they'll be more interested in having kids.

宗教在过去是否扮演了角色，Elon，来安抚人们，让他们有那种感觉，当他们……

Did religion play a role in the past, Elon, to kind of placate and make folks feel that way when they...

是的，我想是的。自然厌恶真空，如果你拿走宗教，那么我认为你会得到一些东西来代替它，实际上比之前更糟。它基本上是破坏性的。你得到觉醒思想病毒填补了宗教曾经拥有的空洞。取代宗教，你得到这些反乌托邦的事实上的宗教，它们非常自我毁灭。所以我认为也许某种宗教复兴，或者至少我们需要的，是一些连贯的哲学，让人们能够兴奋起来。对我来说，这是一种好奇心的哲学。我对宇宙的本质感到好奇，我想走出去，我希望人类在那里探索星空。也许遇到外星文明。也许在某些情况下，我们看到一个早已消亡的外星文明的废墟，但它们曾经非常强大，持续了 1000 万年。就像你在《星际迷航》或非反乌托邦科幻书、电影或节目中看到的那种东西。所以我有一个好奇心的哲学：我只想知道发生了什么。为了知道发生了什么，我们必须扩大意识的范围和规模。我们必须扩展意识。我们必须成长。我们必须成长人类，我们必须延伸人类，以便理解和领悟宇宙，甚至是我们应该问关于那个答案——宇宙——的什么问题。道格拉斯·亚当斯的书《银河系漫游指南》实际上是一本伪装成幽默的深刻哲学书。他试图表达的观点是，问题才是真正困难的部分。答案就是宇宙。答案就是你周围看到的一切。但有一个我们不知道要问的问题。

Yeah, I think so. Nature abhors a vacuum, and if you take away religion, then I think you get something in its place which is actually worse than what was there before. It's destructive basically. You get the woke mind virus filling the hole that religion used to have. Taking the place of religion, you get these dystopian de facto religions that are very self-destructive. So I think perhaps some sort of revival of religion, or at least what we need, is some coherent philosophy that people can get excited about. For me, it's a philosophy of curiosity. I'm curious about the nature of the universe and I want to go out there and I want humanity to be out there exploring the stars. Maybe meeting alien civilizations. Maybe in some cases we see the ruins of a long-dead alien civilization but they were very strong for 10 million years. The kind of stuff that you see in Star Trek in a non-dystopian sci-fi book or movie or show. And so I have a philosophy of curiosity: I just want to know what's going on. And in order to know what's going on, we must have an increase in the scope and scale of consciousness. We must expand consciousness. We must grow. We must grow humanity and we must extend humanity in order to comprehend and understand the universe, or even what question we should ask about the answer that is the universe. Douglas Adams' book The Hitchhiker's Guide to the Galaxy is actually a deep book on philosophy disguised as humor. The point he was trying to make was that the questions are the really hard part. The answer is the universe. The answer is everything you see around you. But one of the questions we don't know to ask.

现在有些问题我想我们确实知道。我想知道：物理学的标准模型关于宇宙起源是否正确？我们真的是 138 亿岁吗？宇宙如何终结？它是以热寂还是其他方式结束？一个黑洞。我们可能是。Elon，你能谈谈整个模拟问题吗？我们是一个模拟吗？也许。你认为我们首先在哪里找到答案？在 AI 中还是在星空中？因为你显然在追求两者。

Now some of the questions I guess we do know. I'd like to know: is the standard model of physics correct about the origins of the universe? Are we actually 13.8 billion years old? How does the universe end? Does it end in a heat death or in some other way? A black hole. We might be. Elon, can you talk about the whole simulation question? Are we a simulation? Maybe. Where do you think we find the answer first? In AI or in the stars? Because you're pursuing both obviously.

我希望更多人能支持好奇心的哲学。因为我认为它非常令人兴奋。而且本质上是乐观的。因为关于宇宙的本质有一种奇妙的惊奇感。当你揭开宇宙中的某个秘密时，那太棒了。一个全新的理解世界就打开了。我们过去甚至不知道所有大陆在哪里。地图上会写着‘这里有龙’，我们只知道当他们朝那个方向航行时，他们没有回来。我有点觉得月球基地，或者这次真正去月球，会是朝着正确方向迈出的一大步。你仍然有月球计划。那是什么状态？还在议程上吗？

I hope more people can get behind a philosophy of curiosity. Because I think it's very exciting. And inherently optimistic. Because there's this amazing sense of wonder about the nature of the universe. And when you uncover some secret in the universe, that's amazing. And a whole world of understanding is opened up. We used to not even know where all the continents were. The map would say 'here be dragons' and all we knew is that when they sailed in that direction, they didn't come back. I kind of feel like the moon base, or just going to the moon for real this time, would be a big step in the right direction. You still have the moon planned. What's the status of that? Is that still on the agenda?

是的，我认为拥有……我认为我们想尝试作为文明达到新的高度。所以我认为去月球没问题，但我们应该去月球是为了建立一个月球基地，比如一个月球研究基地。月球有些部分可能比地球的部分更古老。如果我们有一个月球上的科学基地，我们可能会更了解宇宙的本质。那会很酷。然后我们显然想超越月球去火星，并在火星上建立一个自给自足的城市。我确实认为人类命运有一个分叉路口，如果我们能在火星上建立一个自给自足的城市，关键测试是如果来自地球的补给船因任何原因停止到来，火星是继续繁荣还是消亡？在火星能够独立繁荣和成长的那一刻，意识的可能寿命会大大增加，因为我们不再依赖地球上一切顺利。地球上总是存在自我毁灭的可能性，比如第三次世界大战、超级病毒或像毁灭恐龙那样的流星。我们从化石记录中知道，有很多次大规模灭绝事件。

Yeah, I think having... I think we want to try to reach new heights as a civilization. So I think it's fine to go to the moon, but we should go to the moon in order to establish a lunar base, like a lunar research base. There are parts of the moon that are perhaps older than parts of Earth. And we might understand more about the nature of the universe if we had a science base on the moon. That would be very cool. And then we obviously want to go beyond the moon to Mars and build a self-sustaining city on Mars. I do think that there is a fork in the road of human destiny where if we can establish a self-sustaining city on Mars with the key test being if the resupply ships from Earth stop coming for any reason, does Mars continue to prosper or does it die out? At the point at which Mars is able to prosper and grow on its own, the probable lifespan of consciousness is dramatically greater because we are no longer dependent on everything going right on Earth. There's always some possibility of self-annihilation on Earth with World War III or a supervirus or a meteor like the one that destroyed the dinosaurs. We know from the fossil record that there've been many mass extinction events.

所以我想问的问题是：文明是否会继续上升，使我们能在文明弧线下降之前让火星实现自给自足？因为让生命成为多行星物种的机会之窗，在地球 45 亿年历史中首次出现。

So the question I was wondering about is: will civilization continue to ascend such that we can make Mars self-sustaining before the civilizational arc descends? Because the window of opportunity to make life multiplanetary exists now for the first time in the 4.5 billion year history of Earth.

是的。

Yeah.

埃隆，假设我们到了那里，你也在那里。你会是资深政治家，拥有火星的道德权威。你会如何管理火星？

Elon, let's assume that we get there and you're there. You'd be the elder statesman, you'd have the moral authority of Mars. How do you run Mars?

但我想再次强调，火星的治理形式或早期谁在那里并不那么重要。真正重要的是火星能够自给自足，我们真正成为一个多行星物种，实现行星冗余。这样，如果发生什么事——显然我们应尽一切可能确保地球上的生活美好——但总存在地球毁灭事件的风险，比如自我毁灭或自然灾害。一旦我们成为多行星物种，意识的可能寿命会急剧增加，关键考验是：如果补给船停止到来，火星能否生存？所以首批火星任务并不那么重要。重要的是你能否将足够的吨位运到火星，使火星能够独立繁荣。这意味着它必须拥有文明的所有要素。不仅仅是需要在火星上建芯片厂或造船厂，而是需要具备建造一切的能力。

But I want to emphasize again that the form of governance on Mars or who's there in the early days is less important. What really matters is that Mars is self-sustaining, that we are truly a multiplanet species, and that we've achieved planetary redundancy. So that if something happens—and obviously we should do everything possible to make sure life on Earth is great—but there's always some risk of an annihilation event on Earth, like self-annihilation or some natural disaster. The probable lifespan of consciousness increases dramatically as soon as we are a multiplanet species, with the key test being: can Mars survive if the resupply ships stop coming? So getting the first missions to Mars is not that important. What matters is whether you can get sufficient tonnage to Mars such that Mars can prosper on its own. That means it has to have all the ingredients of civilization. It's not just that you need to build a chip factory or a ship fab on Mars, but you need the ability to build everything.

你对时间尺度有概念吗？假设星舰从 2026 年开始投入使用。然后会有一系列测试。显然会有很多早期测试。我们只有特定的发射窗口。所以有很多时间限制。在你看来，这是 50 年的事？还是 150 年的事？是我们这一代还是我们孩子那一代的事？如果一切顺利，你认为那个时间点在哪里？

Do you have a sense of the time scale? Let's assume Starship is at a state starting in 2026. Then there's going to be a bunch of testing. Obviously, there's going to be a bunch of early testing. We only have certain launch windows. So there's a bunch of time constraints. Is this a 50-year thing in your mind? Is it a 150-year thing? Is it something for our generation or our children's generation? Where do you see that point if it's optimally possible?

如果一切顺利，我认为可以在 30 年内实现。

If things go our way, I think it can be done in 30 years.

哇。

Wow.

前提是每次火星转移窗口（每两年一次）的运载吨位呈指数增长。每两年行星排列一次，你可以前往火星。我认为大约需要 15 个，也许少至 10 个，但 10 到 15 个火星转移窗口。如果每次火星转移窗口的吨位都呈指数增长，那么大约 25 年内就有可能让火星实现自给自足。

So provided there's an exponential increase in the tonnage to Mars with each successive Mars transfer window, which is every two years. So every two years the planets align and you can transfer to Mars. I think in roughly 15, but maybe as few as 10, but 10 to 15ish Mars transfer windows. If you're seeing exponential increases in the tonnage to Mars with each Mars transfer window, then it should be possible to make Mars self-sustaining in about 25 years.

太棒了。难以置信。好了，女士们先生们，埃隆·马斯克，我们回城后再见。我们想念你。

Amazing. That's incredible. All right, ladies and gentlemen, Elon Musk, we'll see you when we're back in town. We miss you.

下次我们当面见。

We'll see you in person next time.

谢谢你，兄弟。好了。

Thank you, brother. All right.