注：本文内容基于马斯克在 G20 峰会期间与美国总统科技政策顾问、白宫科技政策办公室主任迈克尔・克拉茨欧斯（Michael Kratsios）的对话实录整理，核心覆盖监管环境、AI 经济效应、人形机器人落地与全球电力缺口四大议题。在 "AI 泡沫" 争论与 "AI 实物化"（AI Infra）抢装潮并行的当下，这场对话给出了一个少见的完整框架：技术革命的速度，最终由监管、资本与电力的供给速度共同决定。

引言

G20 峰会期间，美国白宫科技政策办公室主任迈克尔・克拉茨欧斯邀请马斯克参与对话。克拉茨欧斯在开场时明确表示，经济增长是政府关注的核心驱动力，而技术是其中的重要组成部分。马斯克则把问题归结为一道简单的算术题：增长更快的一方，最终会压倒增长较慢的一方。

在对话中，马斯克提出，软件正逼近他所说的 "Stockfish 时刻"：按他的判断，未来 12-18 个月内，AI 可能在软件开发及其他数字化任务上全面领先人类。他还预测，十年内全球人形机器人数量将超过十亿台，单台生产力约为人类的五倍；与此同时，AI 芯片扩张速度与电力供给增速之间的错位，可能在 2027 年形成至少 15 吉瓦的电力缺口。

这场 G20 对话真正值得企业管理者关注的，不只是几个惊人的数字，而是背后的资源约束：监管决定创新能否启动，资本决定技术能否成长，电力和供应链决定技术能否落地。以下内容根据本次访谈内容整理而成，文末附本次访谈中英文对照完整实录。

一、"默认合法" 还是 "默认非法"

被问及 "什么样的国家能让创新真正落地" 时，马斯克的回答坦率而直接：新事物应当 "默认合法"，而不是 "默认非法"。他以欧盟为例，认为高强度监管会显著拖慢新技术的落地速度。

在马斯克看来，监管更像减速带，而不是闸门：它未必能阻止技术发展，却可能显著改变技术落地的速度与地点。

更具启发性的是他的 "树苗与大树" 框架：大企业通常能够接触国家高层，初创公司却很难获得同等的政策通道。多数国家容易把资源过多地投入 "大树"，却没有给 "树苗" 足够的成长空间。

对企业而言，这意味着选址、注册和组织架构搭建等环节，正在成为影响创新效率的关键变量。判断一个国家是否值得布局，不只是看它能给大企业多少补贴，还要看它是否为无名的初创公司保留了足够开放的政策通道。

二、算力竞争的物理底层：15GW 电力缺口

访谈中颇具分量的内容，是马斯克围绕算力落地约束这一话题展开的回应。在谈及这一话题时，他直接点明核心矛盾："实际上，确实存在一场电力危机。"

他的推演链条只有三步，每一步都有数字：

• 马斯克援引的分析师共识认为，到 2027 年，AI 芯片可能面临至少 15 吉瓦的电力缺口；
• AI 芯片产能年增速约为 40%-50%，全球多数区域电力供应增速约为 10%-20%，两者之间存在明显错位；
• 马斯克称，谷歌、Anthropic 等公司已经在向 SpaceX 租用算力，而 SpaceX 之所以能够较早部署相关算力，关键在于自行建设了发电设施。

底层迁移由此发生：算力竞争从 "谁拿到 GPU" 变成 "谁能搞定电"，能力圈正向 "能源动员" 延伸。GPU 出口禁令把全球算力切成两个平行市场，电力、监管、芯片可及性共同决定位置。各国新的筹码，由此浮现 —— 谁能把电建好、开放给 AI 公司，谁就在新版图里占位。

三、"Stockfish 时刻"：数字能力的 12-18 个月窗口

谈到 AI 的能力边界，马斯克做了一个未来大概率会被反复引用的类比 ——Stockfish。Stockfish 是手机上就能击败世界冠军卡尔森的国际象棋引擎。

马斯克的预测：到明年某个时候，AI 软件将达到这一级别 —— 人类写软件将完全无法与 AI 竞争。配套量级：AI 让全球 GDP 增 20%-30%，约每年 20-30 万亿美元；到明年年底，AI 将胜任所有 "数字化" 的工作。

他把世界区分为 "数字世界" 和 "原子世界"。AI 可以在较短时间内复制和扩散数字能力，但物理世界需要建设工厂、组织供应链、部署设备，并完成大量现实世界的物流和生产活动。

推论因此变得反直觉：纯数字能力的窗口正在快速缩短，而 "原子侧" 的时间差，恰恰可能成为数字原生公司暂时无法复制的护城河。软件正从稀缺的竞争壁垒，逐渐变成更易获得的基础能力；价值将更多向 "软件 + 原子" 的结合处迁移。

四、十年百亿台机器人：乘法公式与递归爆发

如果说 AI 对数字经济的 20%-30% 增量已经足够震撼，马斯克对机器人的估算则是另一个数量级：经济总量可增长 10 倍甚至更多。

可拆解的公式是人形机器人价值≈AI 软件 ×AI 芯片 × 机电灵巧度（尤其手部）—— 三项均指数级提升，乘法意味着任何短板都被放大，任何突破都被放大。

引爆机制是递归：机器人制造机器人，"一开始非常缓慢，随后爆炸性增长"。

马斯克预测，十年内全球人形机器人数量将超过十亿台，单台生产力约为人类的五倍，总生产力将超过全人类之和 —— 这是保守估计。

启示藏在曲线形状里：递归制造前期是漫长、看似停滞的爬坡，正是过去十年机器人行业被资本冷落的阶段；判断一条赛道，不是看它现在跑多快，而是看它是否在递归曲线的左侧。而供应链高度全球化、搬运原子的成本，既是拖延，也是掌握原子能力者的护城河。

十年前华盛顿谈的是 "自动化工厂"，十年后话题已是通用机器人。范式迁移，往往就在 "再等十年" 的错觉里完成换挡。

中英文对照版全文

Kratsios： First, our first speaker is Elon Musk as the CEO of Tesla and SpaceX. Elon has shown that conquering challenges of physics and engineering can be easier than cutting through red tape. And I think his dedication to doing both whoever continues to inspire people around the world. So thank you, Elon, for joining us.

克拉茨欧斯： 首先，我们的第一位嘉宾是埃隆・马斯克 —— 特斯拉和 SpaceX 的首席执行官。埃隆已经向世人证明，攻克物理和工程上的难题，有时甚至比突破繁文缛节还容易。而他在两方面都全力以赴的精神，持续激励着世界各地的人们。所以，感谢埃隆的出席。

Kratsios： We were just wrapping up our introductory session where. We made clear that under this presidency, we believe that economic growth is one of the most important drivers for all of us, and we believe that technology is a big piece of that.

克拉茨欧斯： 我们刚刚结束了介绍环节。在这一环节中，我们已经明确了一点：在这届政府任内，我们相信经济增长是对所有人最重要的驱动力之一，而技术是其中非常重要的一部分。

Kratsios： So you've been on the front lines of this for quite a while. And I think one of the key questions, I think, for the group and something that we want to talk to you about was, as you have driven innovation across multiple countries around the world. You know, what, in your opinion, separates countries where innovators can successfully turn breakthroughs into deployed technologies from those where progress actually stalls? Because I think there's a lot of lessons Learned about what people, governments should be doing and what maybe they shouldn't be.

克拉茨欧斯： 你在这个领域已经深耕很久了。我想问你的一个关键问题，也是在座各位都关心的：当你在全球多个国家推动创新时，你认为，那些能让创新者成功把突破转化为实际落地技术的国家，和那些进展停滞不前的国家，区别究竟在哪里？因为我觉得，关于政府应该做什么、也许不应该做什么，这里面有大量经验教训值得总结。

Musk： I think a lot of people talked about this and I think some of it, frankly, is pretty straightforward, is that you have to have an environment that's relatively free of regulation, meaning that new things must be default legal as opposed to default illegal. So in the EU, for example, we find that there's the regulation level is extraordinarily high and things are generally default illegal. And this inhibits progress with, of new technologies. It's, it slows it down. Doesn't, it doesn't ultimately stop it, but it slows it down quite considerably.

马斯克： 很多人聊过这个话题，坦率地说，其中一部分答案相当直白 —— 你需要一个相对不受监管束缚的环境，也就是说，新事物应当 "默认合法"，而不是 "默认非法"。比如在欧盟，我们发现那里的监管水平极高，新事物普遍处于 "默认非法" 的状态。这会严重拖慢新技术的发展步伐 —— 虽然不至于彻底阻止进步，但会大幅延缓进程。

Musk： There, of course, you need to have a, you need to have venture capitalists and an environment that is supportive of new companies. You can think of new companies like little, like they're like small saplings in a forest. So what most. Countries tend to do is they tend to provide too much support to the large existing trees in the forest and not enough to the small saplings. But the large trees don't need this support. It's the small saplings that do the startups. And so the system should be generally biased towards supporting. The small trees as opposed to the large ones, but that is rarely the case because the, the large, the large companies have access to usually they have access to the leadership of the countries and the small the small startups do not. So you really need to foster. The growth of young companies and take active steps in that regard. And like I said, make things default legal, not default illegal.

马斯克： 当然，你还需要风险投资，需要一个对新公司友好的环境。你可以把新公司想象成森林里的小树苗。而大多数国家倾向于做的，是给森林里那些已经长成的大树过多的支持，却没能给小树苗足够的养分。但大树其实并不需要这种支持 —— 真正需要的是那些初创公司，也就是那些小树苗。所以，整个体系应当总体上偏向扶持小树，而不是大树。但现实往往并非如此，因为大公司通常能够直接对话国家高层，而初创小企业却做不到。所以你确实需要去扶持年轻公司的成长，并且要主动采取措施。就像我说的，要让新事物 "默认合法"，而不是 "默认非法"。

Kratsios： One question that a lot of the countries here face is a question around adoption. I think there's a general recognition that adopting emerging technologies can be very beneficial to economic growth. No matter what sort of shape, size of any country is, you know, how do you think about adoption? What country, what should countries be doing to encourage adoption? Does it go back to the same sort of regulatory structures? Or how do you think about adoption as a key driver for growth?

克拉茨欧斯： 在座很多国家都面临一个关于 "采用" 的问题。大家普遍认识到，采用新兴技术对经济增长非常有益，无论这个国家是什么形态、规模大小如何。那么你怎么看待 "采用" 这件事？各国应该怎么做来鼓励对新技术的采用？这是否又回到了监管结构的问题上？或者说，你如何看待 "采用" 作为推动增长的关键动力？

Musk： Yeah, I think you wanna have a. Lean forward, try new technologies, approach to new technologies otherwise as opposed to be somewhat stuck in the past. Naturally, new technologies need, need to be, need some encouragement and I think should be embraced. We're gonna see and are seeing, in fact, significant productivity gains from. Artificial intelligence, and we'll see very dramatic gains in productivity from robotics. You know, things like with Tesla self driving car is gonna be a tremendous boon or is a tremendous boon already to users, and I think humanoid robotics will be true. Just an incredible change, I think.

马斯克： 是的，我认为你应该采取一种 "向前倾" 的姿态 —— 积极尝试新技术，以进取的态度对待新技术，而不是多少有些固守在过去的思维里。新技术天然需要一些鼓励，而且我认为它应当被拥抱。事实上，我们已经看到、并将继续看到人工智能带来的显著生产力提升，而机器人技术将带来更为惊人的生产力飞跃。比如特斯拉的自动驾驶汽车，对用户来说已经是一项巨大的福祉，我认为人形机器人也将带来难以置信的改变。

Musk： Just to give you some sense of scale here, I think the, I think AI will probably increase the global economy by 20 to 30%. That's my rough estimate, meaning on the order of 20 to 30 trillion dollars per year. And. And AI will be able to do anything digital, anything that does not require shaping of atoms by hand, probably by the end of next year.

马斯克： 为了让大家有个量级上的概念：我认为人工智能很可能让全球经济总量增长 20% 到 30%—— 这是我的粗略估计，也就是每年大约 20 到 30 万亿美元。而且，大概到明年年底，AI 将能胜任所有 "数字化" 的工作，也就是所有不需要用手去塑造原子的事情。

Musk： So as I'm sure people know, AI is already incredibly good at software and it's getting to the point where. And I won't just be good at software, it'll be what I call Stockfish level good. So Stockfish is a chess program that can beat the world's best chess players very easily. In fact, at this point, you could run Stockfish on your phone. And beat Magnus Carlsen at chess. So at some point next year, is my prediction, software will be so good, AI software will be so good that it will be Stockfish level good, meaning that it is impossible for a human to compete in writing software with AI. But like AI will just crush all humans as software. And I think it will be extremely good, possibly Stockfish level, but certainly extremely good at all forms of reasoning and anything digital. Literally in 12 to 18 months, I what one of this.

马斯克： 我相信大家都知道，AI 在写软件方面已经出色得令人难以置信，而且它正朝着更高的水平迈进。它将不只是 "擅长" 写软件，而是会达到我所说的 "Stockfish 级别" 的强。Stockfish 是一个国际象棋程序，可以轻松击败世界上最顶尖的棋手。事实上，现在你在手机上运行 Stockfish，就能在象棋上打败马格努斯・卡尔森。所以我预测，到明年某个时候，软件 ——AI 软件 —— 会好到 "Stockfish 级别"，意思是人类在写软件这件事上将完全无法与 AI 竞争，AI 会像碾压人类棋手一样碾压所有人类程序员。我认为它会极其出色，也许能达到 Stockfish 级别，而且至少在各类推理和一切数字化的事情上都会极其出色。真的，就在未来 12 到 18 个月内。

Musk： So actually, I'll let me just one more things that my apologies. The so 20-30% increase of total global economy. So this is quite a lot of prosperity we're talking about here just from digital AI, but from robotics, from humanoid. Basically think of like a general purpose robot AI, I think we'll see many.

马斯克： 所以，抱歉，让我再补充一点。我刚才说的是全球经济总量 20% 到 30% 的增长 —— 仅仅来自数字化的 AI，这已经是相当大的繁荣了。而如果算上机器人、算上人形机器人 —— 基本上你可以把它理解成通用型机器人 AI—— 我认为我们将看到更多。

Musk： Multiples of the global economy, meaning like you can increase the economy by a factor of 10 or more. These are mind boggling numbers.

马斯克： 是全球经济的倍数增长，也就是说，你可以让经济总量增长 10 倍甚至更多。这些数字真的令人难以置信。

Kratsios： Yeah, I was just about to say that. I mean, one of the stats we gave to the group here was that in the four years since sort of ChatGPT was launched, think we, there's, you know, over a billion people around the world already using AI. There's been very quick uptake around the country on, around the world on that.

克拉茨欧斯： 是的，我正想说这个。我们给在座各位分享过一个数据：自从 ChatGPT 发布以来的四年里，全球已经有超过十亿人在使用 AI。全球各地对 AI 的接受速度非常之快。

Kratsios： I guess the question to you is, I think a lot of people think about sort of the next chapter being in this sort of applied or sort of physical AI, where do you see sort of robotics trending? How quickly is it gonna be implemented? I think I remember in the first Trump administration, we were talking about sort of automated factories. And now we're a year later, 10 years later. So how, how, you know, how quickly do you think these changes are happening in the sort of physical AI world?

克拉茨欧斯： 我想问你的问题是：很多人认为下一个篇章是 "应用型 AI" 或 "物理 AI"，那你觉得机器人技术会朝哪个方向发展？它落地会有多快？我记得在特朗普第一个任期内，我们就在谈论自动化工厂了。现在十年过去了。那么，你认为在物理 AI 的世界里，这些变化发生的速度会有多快？

Musk： So anything physical always takes longer than anything which is digital. You know, it's, when you solve something digitally, it's just software that you can easily copy across other computers. When it's physical, you've got to build up an entire massive supply chain. You've got to SH, you've got to move a lot of atoms, all of, and Su, the supply chains are very much global at this point. So. That's why it takes longer.

马斯克： 任何物理层面的东西，永远比数字层面的东西花的时间更长。你知道，数字层面的解决方案只是软件，你可以轻松地把它复制到其他电脑上。但物理层面的东西，你得建立一整套庞大的供应链，得搬运大量的原子，而且如今的供应链已经高度全球化。所以，这就是为什么它会花更长时间。

Musk： Nonetheless, when you think of humanoid robotics way the way think about, I think, I think the right framework is consider that the usefulness of a humanoid robot, a general purpose robot is going to be roughly the digital, eh, the AI software, how, how good is the AI software times how good is the AI chip in the robot times how good is the electromechanical dexterity, especially of the hands.

马斯克： 不过，要思考人形机器人，我认为正确的分析框架是这样的：一个人形机器人 —— 也就是通用型机器人 —— 的实用价值，大约等于三项因素的乘积：AI 软件有多强，机器人里搭载的 AI 芯片有多强，以及机电灵巧度有多高 —— 尤其是手部的灵巧度。

Musk： Now all three of those things are improving exponentially, and the usefulness the robot is those three things multiplied by each other. Then when you make the robots, the robots will, the will will the robots will start manufacturing the robots. So you get a recursive effect. So it starts off very slowly, but then it grows at an explosive rate.

马斯克： 而这三项因素都在呈指数级提升，机器人的实用价值就是三者相乘的结果。当人类制造出机器人之后，机器人又会开始制造机器人，于是产生了一种递归效应。所以这个进程一开始会非常缓慢，但随后会以爆炸性的速度增长。

Musk： So if you say like 10 years from now, I would say there are well over a billion humanoid robots. And the productivity per robot will be probably five times that of a human. Meaning that the productivity in 10 years of humanoid robots, and I think this is a conservative estimate, by the way. This is, this one I'd be willing to. To put serious money betting on that there will be at least a billion robots in 10 years, and that those robots will be at least five times the output of a human, meaning the billion humanoid robots will be more productive than all humans combined.

马斯克： 所以如果说到十年后，我的判断是：人形机器人的数量将远超十亿台，而每台机器人的生产力大约是人类的五倍。也就是说，十年后人形机器人的总生产力 —— 顺便说一句，我认为这是一个保守的估计 —— 这个预测我愿意押上重注：十年内至少会有十亿台机器人，每台机器人的产出至少是人类的五倍。这意味着，这十亿台人形机器人的生产力，将超过全人类生产力的总和。

Kratsios： Shifting gears just a little bit to. To a question that's kind of facing the US today, data centers have been a big political issue over the last six to eight months here in the United States, and I think there's a general understanding that in order. Order to drive and power the AI revolution that's coming, we need to have the electricity and the data center compute capacity to do the training and the inference of all this AI.

克拉茨欧斯： 让我们稍微换个话题，聊一个美国当前正面临的问题。在过去六到八个月里，数据中心在美国一直是个重大的政治议题。我认为大家有一个共识：为了驱动并支撑即将到来的 AI 革命，我们需要足够的电力和数据中心的算力，来完成所有这些 AI 的训练和推理。

Kratsios： You know, how do you think about this particular issue? Where are we on the curve of kind of how much we built versus how much we need? And as government leaders here think about how to prepare their economies and build the right power and data infrastructure, how should they be thinking about where sort of compute and power needs are gonna be in the next few years?

克拉茨欧斯： 那么你怎么看待这个具体的问题？我们已经建了多少、还需要多少，我们现在处在这条曲线的什么位置？作为在座的政府领导人，在考虑如何让本国经济做好准备、建设合适的电力和数据基础设施时，他们应该如何看待未来几年算力和电力需求的走向？

Musk： Well, there actually is a quite a crisis of power. So the, this is a. In fact, this is something if you if you just sort of follow the AI topic on the X platform, which by the way is where almost all of the AI discourse takes place. You, you, you, I think you get a very good sense for where things are headed. So that's how I get my news, and it's incredibly good. Everyone who's anyone in AI posts on X. So that's why I recommend like just go on the AI topic on X and you'll understand all these things and get it a a day by day account of things.

马斯克： 实际上确实存在一场电力危机。事实上，如果你在 X 平台上关注 AI 话题 —— 顺便说一句，那里几乎是所有 AI 讨论发生的地方 —— 你就能对事情的走向有非常清晰的感知。我自己就是这么获取资讯的，而且信息质量高得惊人。AI 圈里所有有分量的人都在 X 上发帖。所以我建议大家直接去 X 上看 AI 话题，你就能明白这一切，并且获得逐日的实时动态。

Musk： The, the, and what the consensus is at this point is that there will be a significant power shortfall. I. Next year, so not like distant future. There's expected to be, I believe the consensus estimate among analysts that follows the AI space very closely is that there will be at least a 15 gigawatt shortfall of power in 2027. For AI chips, so this is perhaps an obvious thing that one would expect to occur because the rate at which AI chips is being produced is, has been rising incredibly rapidly. They're sort of rising on the order of 40 to 50 percent a year, but the power available outside of China. Has been rising at like 10 to 20% here. So obviously the faster rising thing will eventually overwhelm the slower rising thing.

马斯克： 而目前业界的共识是：明年就会出现严重的电力短缺 —— 这不是遥远的未来。我相信，那些密切关注 AI 领域的分析师们的共识是：到 2027 年，AI 芯片的电力缺口将至少有 15 吉瓦。这其实是可以预见的结果，因为 AI 芯片的产能增速快得惊人 —— 每年大约增长 40% 到 50%；而中国以外的电力供应，每年只增长 10% 到 20% 左右。很明显，增长更快的那一方，最终会压倒增长较慢的那一方。

Musk： And so there, there, in fact, even, I would say at this point, there are challenges with power even before next year, which is why Google and. Anthropic and many other companies are actually leasing compute from SpaceX because we've been able to turn on AI better than anyone else so far, but this is by constructing our own power plants is the only way we were able to do it. So now China does have a tremendous amount of electricity. But due to GPU export bans, one cannot, you know, establish data centers with the latest chips in China. So, so US, but so really the consideration is what sort of electricity growth is there outside of China, and that is currently a significant shortfall.

马斯克： 而事实上，我甚至想说，在明年到来之前就已经出现电力方面的挑战了。这也是为什么谷歌、Anthropic 等许多公司现在都在向 SpaceX 租用算力 —— 因为到目前为止，我们比任何人都更早、更好地让 AI 运转了起来，而我们能做到这一点，唯一的方式就是自建发电厂。中国确实拥有巨量的电力，但由于 GPU 出口禁令，你无法在中国部署使用最新芯片的数据中心。所以真正要考量的是：中国以外的电力增长情况如何 —— 而目前那里确实存在显著的缺口。

Musk： Relative to AI chip production. So there is, this creates an opportunity, I think, for countries around the world just to save, to just sort of, if they're interested in AI data. The centers to pro, to construct a lot of power and offer that to AI companies. And in exchange, of course, that these AI data centers would be taxed and have to pay, you know, reasonable fees and stuff. But it does create an opportunity for a lot of countries.

马斯克： 相对于 AI 芯片的生产而言。所以我认为，这为世界各国创造了一个机会：如果它们对 AI 数据中心感兴趣，就可以大量建设电力设施，并把电力提供给 AI 公司。当然作为交换，这些 AI 数据中心会被征税，需要支付合理的费用等等。但它的确为很多国家创造了机会。

Kratsios： Absolutely. Well, thank you so much for your time. It means a lot that you could join us here and really appreciate it. Thank you so much.

克拉茨欧斯： 完全同意。好的，非常感谢你抽出宝贵时间。你能出席对我们意义重大，我们真的非常感激。非常感谢。

Musk： You're most welcome. Thank you.

马斯克： 你太客气了。谢谢。

（完）

注：以上访谈全文来自网络，仅供交流学习。

结语

这场 G20 对话为企业管理者提供了一个清晰的行动框架：12-18 个月的数字能力窗口正在快速关闭，纯软件岗位将首当其冲面临 AI 替代；而 "软件 + 原子" 的结合处，恰恰是下一轮竞争的护城河所在。电力作为算力竞争的物理底层，正在成为各国和企业争夺的新筹码；人形机器人的递归爆发曲线，则预示着未来十年生产力格局的根本性重塑。

对空天产业而言，这一框架同样适用：航空航天制造作为典型的 "原子密集型" 产业，其供应链、工艺和工程能力构成了难以被纯数字能力快速复制的壁垒；而 AI 在设计、制造、运维等环节的渗透，正在加速这一产业的效率提升。空天界将持续关注 AI 与空天产业融合的最新动态，为行业提供深度洞察与产业连接。