推荐微信扫码登录以获得更好的体验!另:🎉建议更新到最新版,以体验新功能!🎉 版本更新记录→

【双语音+文稿】埃隆·马斯克最新访谈:马斯克准确预测了未来三

共 1 集

剧集目录

集标题时长
P1【双语音+文稿】埃隆·马斯克最新访谈:马斯克准确预测了未来三1:01:46

P1 · 【双语音+文稿】埃隆·马斯克最新访谈:马斯克准确预测了未来三年(罗恩·巴伦完整采访) p01 英文版 (P1)

字幕摘录

时间英文中文
0:00When I was preparing for this conversation, one of the things I just visualized was E.T.当我准备这段对话时, 我刚刚想象到的一件事是E. T.
0:07So E.T.所以说E. T.
0:09going across the moon, you know, with a boy on a bicycle and a contraption to communicate穿过月球,你知道, 和一个男孩骑着自行车和机械通信
0:15很明显是Starlink的前身
0:20So I guess that didn't work.所以我猜这行不通。
0:28So Elon uses the first principles about everything should be done the right way,所以伊隆运用关于一切的第一原理 应该正确行事
0:36not the fastest way, not the quickest way, not the cheapest way, the best way.不是最快的方式,不是最快的方式, 不是最便宜的方式,最好的方式。
0:41And everything looks at all over again.而一切众生复观.
0:43So if someone makes an investment on his behalf所以如果有人替他投资
0:46for one of his businesses, it's always viewed in the terms of, is this the best we can do?对于他的一个企业, 它总是被看 在条件,这是我们所能做的最好的?
展开字幕全文(979 条)
序号英文中文
1When I was preparing for this conversation, one of the things I just visualized was E.T.当我准备这段对话时, 我刚刚想象到的一件事是E. T.
2So E.T.所以说E. T.
3going across the moon, you know, with a boy on a bicycle and a contraption to communicate穿过月球,你知道, 和一个男孩骑着自行车和机械通信
4很明显是Starlink的前身
5So I guess that didn't work.所以我猜这行不通。
6So Elon uses the first principles about everything should be done the right way,所以伊隆运用关于一切的第一原理 应该正确行事
7not the fastest way, not the quickest way, not the cheapest way, the best way.不是最快的方式,不是最快的方式, 不是最便宜的方式,最好的方式。
8And everything looks at all over again.而一切众生复观.
9So if someone makes an investment on his behalf所以如果有人替他投资
10for one of his businesses, it's always viewed in the terms of, is this the best we can do?对于他的一个企业, 它总是被看 在条件,这是我们所能做的最好的?
11Best we can do, the fastest we can do, the most effective.力所能及,最快,最有效.
12So one thing I would say is that所以我要说的是
13he's a builder.他是一个建筑师。
14He's probably the only builder I'm aware of in the United States who builds stuff.他可能是我在美国唯一认识的建筑师 他建造东西。
15But builds stuff that's not just regular stuff, like 10 million square foot buildings, 5 million,但建造的东西不只是普通的东西, 像1000万平方英尺的建筑,500万,
1620 million.两千万
17Now he's got 100 million square foot building for TerraFab that he's thinking about现在他有1亿平方英尺的建筑 为TerraFab,他在想什么
18or going to build.或将要建设。
19And so the question, and no one else has done this.于是问之,亦无他人所为.
20No one.没有人。
21And anything还有一切
22like it.喜欢它。
23And so the question is, of course, and when you're doing such a thing, then you get所以问题当然是 当你做这种事时 你就会得到
24scaling benefits.扩大效益。
25You get to buy things and produce things much more effectively than you你买东西,生产东西 比你有效得多
26could if you were a smaller entity.如果你是一个较小的实体。
27So the question is, how do you manage to do this?所以问题是,你怎么做到的?
28And why didn't anyone else ever try to do anything like this?为什么没人想做这种事?
29We're trying to get我们试图得到
30silicon to come back to the United States again.硅再次回到美国
31We need it.我们需要它。
32We need power.我们需要力量
33We need silicon.我们需要硅。
34No one's doing this except for you.除了你 没人会这么做
35How come?为什么?
36Well, that's actually, that's not quite accurate其实,这不太准确
37in that TSMC and Samsung are building fabs in the U.S.在那个TSMC和三星正在美国建造法布
38So TSMC is building a large collection of因此,TSMC正在建造一个大型的集合
39fabs in Arizona.亚利桑那的法布斯
40Some of them are, I believe, operational.我相信,其中一些行动正在展开。
41And Samsung has some operational三星已经开工了
42fabs in Texas and is building, or has built, I should say, a fab for Tesla that will be德克萨斯州的法布斯正在建造 或者已经为特斯拉建造了一个法布
43dedicated to the AIS 5 and AIS 6 chip designs that Tesla's done.致力于特斯拉完成的AIS 5和AIS 6芯片设计.
44So, but relative to what's所以,但相对什么
45being done in Asia, the amount of chip fabrication in the U.S. is very small.在美国制造的芯片数量很小。
46And even if you take即使你接受了
47the best case assumptions of all the chip fabs that have been announced, that will be built in所有已公布的芯片的最好案例假设,将建在
48America, that's still a tiny fraction, perhaps in the order of 10%, optimistically 10% of what's美国,那还是很小的一部分, 也许在10%左右, 乐观的10%是什么
49in Asia.在亚洲。
50So as I think most people know, Taiwan does an incredible amount of heavy lifting on chip所以,我想大多数人都知道,台湾在芯片上 做了惊人的举重
51production.生产。
52TSMC is an amazing company, truly incredible.TSMC是一个惊人的公司,真的不可思议。
53The sheer number of, a variety of数量众多,种类繁多
54chips and the magnitude and the volume is incredible, what that TSMC does.芯片 尺寸和体积都令人难以置信 TSMC所做的一切
55Same thing for也一样
56China.中国.
57Great job, too, on the memory side.伟大的工作,在记忆方面。
58But it's overwhelmingly built in Asia.但是它绝大部分建在亚洲。
59My concern really is that there is always some geopolitical risk in that region.我真正关切的是,该地区总有一些地缘政治风险。
60And I think我觉得
61that that risk may come to fruition at some point, in which case, if we don't have all the风险可能会在某个时候产生, 在这种情况下,如果我们没有所有
62chips in America, then we may not have any chips at all, or very few, the 10% number type of thing.在美国,芯片,那么我们可能没有任何芯片 任何芯片,或非常少, 10%的数字类型的东西。
63And then there's the, maybe looking ahead, you know, past any sort of medium term geopolitical然后是,也许展望未来,你知道, 通过任何一种中期地缘政治
64bumps in the road, we're simply, even if you exclude geopolitical crisis issues, the rate at我们只是,即使你排除了地缘政治危机问题,
65which we're making chips is simply too slow.我们做的芯片太慢了
66So if you want to get to, say, a terawatt a year of chip所以,如果你想得到,比如,一个千兆瓦一年的芯片
67production, which is our goal, that's the reason it's called the TeraFab.生产,这是我们的目标, 这就是为什么它被称为TeraFab。
68当我说TeraFab,
69there's three possible meanings to TeraFab.泰拉法布有三个可能的含义.
70One is terror, T-E-R-R-O-R, which I think sometimes一个是恐怖,T -E -R
71people are like, why would you want a TeraFab?人们喜欢, 你为什么想要一个寺院?
72还有T -E -R
73but it's actually T-E-R-A, as in terawatt.但它实际上是T -E -R -A, 像兆瓦。
74So, but arguably all three apply.所以,但可以说这三个都适用。
75So,这么说
76it's terrifyingly large.这是可怕的大。
77So, the rest of the world is making, we'll make this year所以,世界其他地方正在做, 我们将在今年
78around 20 gigawatts of AI chips, maybe 25, and then next year, depending on大约20千兆瓦的AI芯片, 也许25, 然后明年,取决于
79whose projections you believe, it's maybe around 35 to 45 gigawatts of AI chips next year.明年大概是35到45千兆瓦的AI芯片
80But if you want to do a terawatt, you need 30 times what current global production will be但是,如果你想做一个千兆瓦, 你需要30倍于当前全球产量
81roughly next year.大约明年。
82For just us.就为了我们
83We're building that for us.我们正在为我们建造它。
84Well, yes, we're building...嗯,是的,我们在建...
85I mean,我是说...
86what the annual production is in one plant, which you think is going to do年产量在一个工厂里,你认为会做什么
87as big a production as the GDP in the world, as a GDP for the 20th largest country, one plant.产量和世界GDP一样大 作为第20大国家的GDP,一个工厂
88Yeah, it might even make the top 10.是啊,它甚至可能 达到前十。
89So, obviously, a plan like this, I think,因此,很明显,这样的计划,我想,
90if it fails, it'll be crazy, and if it succeeds, it'll be audacious.如果失败了,那会很疯狂,如果成功的话,就会大胆.
91So, it sort of depends on the outcome.所以,这在某种程度上取决于结果。
92I think it will succeed, at least to some degree.我认为它将取得成功,至少在某种程度上。
93Hope so.希望是这样
94It's going to cost a trillion dollars.将得万亿.
95But we won't need that money immediately.但我们不需要钱
96We'll be building phases.我们将建造阶段。
97And we are looking at ways to而我们正在寻找方法
98change the manufacturing process for chips, to improve the efficiency of chip production改变芯片制造流程,提高芯片生产效率
99as we've done in many other arenas.就像我们在很多其它竞技场里所做的那样
100Just one example.举个例子
101Right now, the amount of time that a现在,时间的长度
102wafer in a chip app spends simply being transported and waiting is more than half the time.芯片应用中的面包 仅仅被运送 等待的时间超过一半
103So, more所以,还有
104than half of what a chip is doing in a chip app is nothing.超过芯片在芯片应用中所做的一半是无用的.
105It's either being moved around or要么被移动 要么
106waiting for a machine.等待一个机器。
107So, I think there's basically a lot of potential for improvement所以,我认为基本上有很多改进的潜力
108in chip fabrication.在芯片制造。
109Why don't these people do it themselves?为什么这些人不自己动手?
110If they have plans to just sit with如果他们有计划坐在一起
111chips half the time, just not moving, what's to stop them from improving their own systems?芯片一半的时间,只是不移动, 是什么阻止他们改进自己的系统?
112Well, no, actually, if you visit a fab, you'll see that the wafers are moving a lot.不,实际上,如果你去拜访一个家庭, 你会看到,饼饼 移动了很多。
113They're他们是
114and waiting.等待着
115But the amount of time that atoms are being changed on the wafer in a way that is但原子在华夫饼上被改变的时间
116useful is less than half the time.有用时间不到一半。
117And I'd say sometimes it's maybe 20% of the time that我会说,有时,它可能是20%的时间
118something useful is happening to the atoms on that wafer.磁盘上的原子发生了一些有用的事。
119The reason that they want to他们想要的原因
120try something adventurous with chip apps is that these things cost a tremendous amount of capital尝试一些冒险的芯片应用 是这些东西花费巨大的资本
121and if they have a recipe that works, they simply don't want to risk something new.如果他们有一个有效的食谱, 他们只是不想冒新的风险。
122So, generally the preference in the chip industry is to repeat what they did last time with minor芯片行业的偏好是 重复他们上次用小的
123changes and not take any risks.而不是冒任何风险。
124So, we're building a research chip app in Texas, just in Austin,我们在德克萨斯州建造一个研究芯片应用程序 就在奥斯汀
125at the Gigafactory Texas.在德克萨斯州
126That research fab is going to try many different approaches to chip这个研究法布会尝试很多不同的芯片方法
127application.应用。
128And many of those approaches won't work, but some of them I'm sure will work.其中许多方法不会奏效, 但我相信其中一些方法会奏效。
129And then we will translate those improvements to the Tariff Fab itself, the production Tariff Fab,然后我们将把这些改进 翻译为关税法布本身, 生产关税法布,
130which is in Grimes County in Texas, near Texas A&M.它位于德克萨斯州格里梅斯县,靠近德克萨斯A&M.
131And that's where we'll scale up那就是我们扩大规模的地方
132production of our AI chips.生产我们的AI芯片。
133I think one of the things that's going to be interesting about the我认为其中一件事情 是要去有趣的
134research fab in Austin is that under one roof we will have the ability to make在奥斯丁的研究中,我们在一个屋檐下 将有能力制造
135logic, memory, packaging, and do the lithography so that we can design new chips逻辑,内存,包装,并做拼图,这样我们就可以设计新的芯片
136and under one roof make the whole chip tested with a very fast iteration cycle.在一个屋檐下 整个芯片测试 使用非常快的迭代周期。
137Something that would normally take a year, I think we may be able to reduce to a month.通常需要一年的时间,我想我们可以缩短到一个月。
138So, for any given technology endeavor, the pace of progress is determined by因此,对于任何特定的技术努力来说,进展速度是由以下因素决定的:
139how many iterations can you do and how much progress can you make between those iterations.你可以做多少次迭代, 以及你能够在这些迭代之间取得多少进展。
140So, if you can improve the speed of iteration and how many iterations can you do per year or所以,如果你能提高重复的速度 以及每年能做多少重复,或者
141per unit time and the progress between each iteration, in any given technology you will每个单元时间和每个迭代之间的进度,任何特定技术,你都会
142accelerate your progress at a rate far greater than anyone else.以比任何其他人快得多的速度加快进度。
143If you're able to如果你能够
144iterate an order of magnitude faster, you as a first approximation will advance your technology以更快的速度排列一个数量级, 你作为第一个近似值将推进你的技术
145order of magnitude faster.数量顺序更快。
146So, most people, there's a tremendous amount of inertia因此,大多数人, 有大量的惯性
147in business because if someone makes an effort to spend money and it's not successful,因为如果有人努力花钱 却不成功
148then that's a black mark on them.那他们身上就有一个黑印
149And if it is successful and they're trying to get the top job如果成功,他们想要获得最高职位
150of the business, then it just means that they have to work harder.他们必须努力工作
151So, if they're successful,所以,如果他们成功的话,
152they work harder and if it fails, they're finished for their opportunity for the top job.他们更努力地工作,如果失败了,他们就完了,以获得顶级工作的机会.
153Elon has established a culture in his business, Business Is, where it's always about伊隆在生意上建立了一种文化 是,它总是关于
154come to me with ideas and then if you think it's successful, and I believe that's the case,向我提出一些想法,然后如果你认为它是成功的, 我相信就是这样,
155then we'll throw away what we've done before and start all over again.然后我们放弃我们以前做过的一切 重新开始
156It's always about它总是关于
157doing the best thing.做最伟大的事情。
158One of the things that's interesting to me about your culture is that you你的文化让我很感兴趣
159introduced me to Key Luck, the person who's in charge, Intel, who's now going to have an介绍我认识Key Luck, 负责的人,Intel, 他现在要有一个
160对TerraFab感兴趣
161And so, I spent some time with him and he comes to visit me and I said,于是,我和他呆了一段时间, 他来看我,我说,
162so what's it like to work with Elon?和伊隆合作是什么感觉?
163He says, well, other people come to visit my facility他说,嗯,其他人来参观我的设施
164in Oregon and they come and they walk around for an hour and a half and they leave.在俄勒冈,他们来了, 他们走来走去一个半小时,然后就离开了。
165And Elon comes伊隆来了
166and then he sits there, he was there for four and a half hours and he interviews one engineer然后他坐在那里, 他在那里4个半小时 和一位工程师面谈
167after another, after another, after another, questioning everything.复次复次复次复次复次问一切法.
168That's all concept.这就是所有的概念。
169So,这么说
170everything that he has to do is speed.他必须做的都是速度
171And then when I went to Texas, to Boca Chica, to visit然后当我去德克萨斯 博卡奇卡 拜访
172their car factory that they were building there, while it was under construction,他们在那里建造的汽车工厂 正在建设中
173they introduced me to the head of that facility, the manager, a giant man.他们把我介绍给那个设施的负责人 经理 一个巨人
174And of course, everyone当然,各位
175seems like a giant to me, but he was a really giant guy.在我看来,他是个巨人 但他是一个真正的巨人。
176And he had been previously the head他以前是头
177of production at Ford.在福特生产。
178And so, I said, well, so what's it like, difference between Ford and Tesla?我说,福特和特斯拉有什么区别?
179He said, well, when I was at Ford and I had a change, I had to go through six different levels他说,当我在福特,我有一个变化, 我不得不经历六个不同的关卡
180to get a change approved.以便获得批准。
181When I have to have a change here, then all I do is walk upstairs当我要换衣服的时候,我只好上楼
182and there's Elon.还有伊隆
183And so, when a plan to something under construction, Elon's not some distant guy,所以,当一个计划 在建设中的东西, 伊隆不是一个遥远的家伙,
184he's there.他在那边
185He even has an Airstream.他甚至有一个气流。
186He lives in an Airstream.他住在一个气流。
187Do you know any other executives你认识其他高管吗?
188of any big companies, potentially the largest company in the world, going to live in an Airstream任何大公司,潜在的 世界上最大的公司, 会生活在气流
189to make sure that all the questions that are taking place in making a facility are going to确保所有在设施建设中出现的问题
190answered fast?快速回答吗?
191So, it's amazing culture, amazing person, and amazing people who's recruited.所以,这是惊人的文化,惊人的人, 和惊人的人谁被招募。
192So,这么说
193this whole people who work there, they're just brilliant people.他们都是聪明人
194It's harder to get a job there在那里找工作比较难
195than get into Harvard.而不是进入哈佛大学。
196Of course, I don't know why people would go to Harvard anyway anymore.当然,我不知道为什么人们会去哈佛了。
197Well, by the way, Airstreams are actually pretty nice.顺便说一下 气流其实不错
198So, I wouldn't say I'm like,所以,我不会说我喜欢,
199no, I'm not that bad, honestly.不,我没那么坏,说实话。
200I don't kind of like my little Airstream.我不太喜欢我的小气流
201But it's actually parked right next to the power plant.但实际上它停在发电厂旁边.
202So, we built this power plant that's所以,我们建造了这个发电厂,那就是
203producing over a gigawatt.生产超过千兆瓦。
204And probably it'll go to one and a half to two gigawatts of internal可能要到1到2千兆瓦的内部
205power plant production in about a fifth of the time that such a thing would normally be built.电厂的产量在一般建造时间的五分之一左右。
206So, built in about a year, as opposed to normally about five years to that.所以,建于大约一年, 而通常大约五年。
207And it's not as though somebody has to come to me upstairs.也不像有人要上楼来找我
208There is no upstairs.没有楼上
209So,这么说
210we have another floor in the, I don't know, Memphis plus this two thing.孟菲斯还有两层楼
211I walk the entire site我走遍整个网站
212sometimes two, sometimes three, sometimes four or five times a day.有时2次,有时3次,有时每天4到5次.
213So, the entire site,所以,整个网站,
214the power plant and all the buildings.电厂和所有建筑
215And I also insist on doing at least我也坚持至少做
216some of the work myself manually.一些工作我自己手动。
217So, I've pulled power cables the size of an anaconda.所以,我已经拉电线 大小像一个Aaconda。
218They're incredibly heavy and long.他们非常沉重和长。
219And I've done the fiber optic cabling.我做了光纤电缆。
220I've done electrical switches.我做了电开关。
221I've worked on the power plant management software.我研究过电厂管理软件
222So, I try to所以我试着
223do at least a little bit of everything so that I'm not numb to the pain of the people who are至少做一点一切 这样我就不会麻木于 那些人的痛苦
224actually doing it.实际做到这一点。
225And I have some understanding of the difficulty and what can be done to improve我对困难有一定理解,可以做些什么来改进
226the situation.情况
227So, it's not as though I'm sitting in an office and somebody has to come to me.所以,这不是像 我坐在办公室 和有人来找我。
228I'mи琌
229sitting in an office.坐在办公室里
230So, let's talk a little bit about data centers, which are now in the news.因此,让我们来谈谈一些数据中心, 这些中心现在在新闻里。
231Yeah, popular topic.是啊,流行话题。
232How to make friends and influence people.如何交朋友,影响人.
233The pitchfork level for data用于数据的草叉级别
234centers is pretty high.中心相当高。
235Yeah, I'd say.是的,我会说。
236But if you want to have AI, you've got to have data centers.但如果你想有AI, 你必须有数据中心。
237And so, let's talk about them first on this planet.因此,让我们先在地球上谈论它们。
238So, Elon believed that the roadblock所以,伊隆认为路障
239that would be in the way of data center construction would be having enough chips,这将阻碍数据中心的建设 将有足够的芯片,
240enough compute actually.算够了
241And so, what he did is that a lot of people, we have Brock.所以,他做的是,很多人, 我们有Brock。
242因此,他站在了OpenAI,Anthropic的前列.
243Those guys were his guys.那些家伙是他的手下。
244And then,礛
245of course, he doesn't like OpenAI anymore.当然 他不再喜欢OpenAI了
246Yeah, well, I did start OpenAI as an open是啊,我确实开始OpenAI作为一个开放
247source nonprofit, and now it is a closed for maximum profit entity.来源非营利,现在它是一个关闭的最大利润实体。
248So, it's literally the所以,这是字面上的
249cul-de-sac of what I started.我开始的事
250Right.对
251And then, Anthropic was founded by seven people that left然后,Anthropic 由七个人创建 离开
252OpenAI because they didn't trust Sam Altman for good reason.OpenAI 因为他们不相信山姆·阿尔特曼是有原因的.
253They said this publicly.他们公开说。
254That's not a secret.这不是秘密
255所以,实际上,Anthropic是OpenAI的副手。
256So,这么说
257yeah.对
258I guess I should have just started OpenAI as a regular C corp instead of as a我想我应该刚开始 OpenAI 作为正规的C公司而不是一个
259nonprofit.非营利组织。
260But, you know, I thought this would be something that would be a philanthropic thing但是,你知道,我认为这是 东西会是一个慈善的事情
261for the world.为了世界
262I didn't anticipate it would one day turn into a for-profit that is a closed我没想到有一天会变成 一个盈利的,一个关闭的
263source.来源.
264And when you left because you were disappointed in the direction, they offered当你因为对方向失望而离开时, 他们提出
265you an ownership interest, and you said, no, I don't want it.而你说,不,我不想要它。
266And so, you left.所以 你离开了
267Yeah.对
268I mean,我是说...
269well, I was confused as to how you could have a commercial interest in a nonprofit.我当时很困惑 你怎么会对一个非营利组织有商业利益
270That didn't seem to make sense to me.这对我来说似乎不合理。
271You know, I provided almost all the capital in the beginning你知道,我提供了几乎所有的首都 在一开始
272for the company.为公司。
273It was my idea, and I named it, and recruited the key people,这是我的主意,我给它取了名字, 并招募关键的人,
274and taught them everything I know.教他们我所知道的
275But apart from that, nothing.但除此之外,无所事事.
276So, to go back to data centers.所以,回到数据中心。
277So, you thought that the idea about putting all this AI in所以,你认为的想法 将所有这些AI
278operation was that we need data centers, and people want to build them, and it takes them操作是我们需要数据中心 人们想要建立它们 需要它们
279ordinarily about two and a half years to build one, and it costs maybe $60 billion to build it,通常大约两年半的时间来建造一个, 建造它可能需要600亿美元,
280$50 or $60 billion to build a gig.500或600亿美元 建造一个演出。
281And then the people who build them for hyperscalers and rent然后建造它们的人 超规模的和出租
282them, in order for them to get the capital, they need to borrow it, or get it in some way他们为了拿到资本, 需要借钱,或用某种方式
283that they have to have long-term leases, five years at least, to get them.他们必须拥有长期租约,至少5年,才能得到这些租约。
284So, the rents people所以,他们租人
285pay in five-year leases are maybe $12 billion a year for a $60 billion investment.5年的租金可能为每年120亿美元,投资600亿美元。
286What we do我们做什么?
287instead is that we build them, and we build them because we have the use for them, because we're而是我们建造它们,我们建造它们, 因为我们有它们的用途,因为我们是
288integrated for our cars, for our robots, for our rockets, for everything.我们的汽车,机器人,火箭,一切。
289And because we need them,因为我们需要他们,
290then we can take risk of building them without knowing that we have contracts for other people.这样我们就可以冒着风险去建造它们 而不知道我们和别人有合同
291And what happens, though, is we have them built, and they're not.事情是这样的,我们造了它们,它们却不是。
292So, it takes us somewhere,所以,它带我们到某个地方,
293it was taking us four months, and now it's taking us eight or nine months to build.我们花了四个月的时间 现在我们花了八个或九个月的时间来建设
294So,这么说
295it takes them two and a half years, it takes us eight months, nine months.花了他们两年半的时间 我们花了八个月九个月
296It costs them $60花了60块
297billion, it costs us $35 billion.耗资三百五十亿
298And the demand for compute is so great that people who build对计算的需求是如此之大 以至于人们建设
299centers are dark because they don't have the power to turn them on.中心是黑暗的,因为它们没有 启动它们的能力。
300And we'll rent them to compute我们租他们来计算
301until we need it for ourselves, for GROC, and we'll rent it to them for somewhere around $50直到我们自己需要它,为GROC, 我们会租给他们约50美元左右
302billion a year.一年十亿
303So, we're getting $50 billion a year for $35 billion dollar investments.所以,我们每年得到500亿美元 用于350亿美元的投资。
304So,这么说
305revenues for SpaceX are about to explode, going up.SpaceX的收入即将爆炸, 上升。
306Yes, they are exploding.是的,他们正在爆炸。
307So, I think, yeah,所以,我认为,是的,
308Q2 to Q3 revenue increase was, I think, over 80% this year.我认为,从Q2到Q3的收入增长今年超过80%。
309I'm not sure what the exact numbers我不知道确切的数字
310will be, Q3 to Q4, but probably something like 70% Q3 to Q4.将会是,Q3到Q4,但大概是70%的Q3到Q4.
311And I would say maybe it's something我会说,也许这是什么
312similar, Q4 to Q1.类似,Q4 到Q1。
313It is extraordinarily difficult to bring on compute at scale.计算规模非常困难。
314And I think there's a lot of speculators out there who think you can simply order some GPUs,我认为有很多投机者认为 你可以简单地订购一些GPU,
315hand a purchase order to some company, to build a data center, and just rake in cash.给某个公司发个订单 建立一个数据中心 然后用现金
316And the还有
317there's a meme that I think is quite funny, which I think is called the Jose meme,有一个meme 我觉得很有趣, 我认为这叫做Joseme,
318where there's, you may have seen it before, there's one guy digging the hole in the road,在那里,你可能见过它, 有一个家伙挖 在路上的洞,
319and there's like eight people watching him dig the hole.有八个人看着他挖洞
320So, now, meaning like there's one person所以,现在,意思是像有一个人
321doing actual work, and then eight people watching the guy do the work.做实际工作,然后八个人 看家伙做的工作。
322Now, unless you increase the现在,除非你增加
323amount of work done does not increase.完成的工作量没有增加。
324And so adding more people who simply, you know, issue所以,增加更多的人 谁简单的,你知道,发行
325purchase orders to Jose does not result in more things occurring, only if there are more Jose's,给Jose的订单不会导致更多的事情发生 只有Jose的,
326more work.还要多工作
327So, I aspire to be Jose.所以,我渴望成为何塞。
328SpaceX想成为Jose.
329We do actual work.我们做实际工作。
330This is a really这是真的
331big deal.说来话长
332And with the massive expansion and construction in the United States, frankly,而随着美国的大规模扩张和建设,坦率地说,
333it's not as though there's been this massive influx of people who do actual work at a physical并不是说有大批人涌入 做实际工作的人
334level.级别。
335But there's been a massive increase in the number of people who issue purchase orders.但发布定购单的人数却大增.
336So, that means that the people doing actual work are spread incredibly thin.所以,这意味着做实际工作的人 分布得非常稀薄。
337And so you have a large所以你有一个大
338number of partially completed data center projects throughout the country, because at this point,全国部分完成的数据中心项目的数目,因为目前,
339the peanut butter is spread so thin you can barely taste it.花生酱散发得很薄 你几乎无法品尝
340区别在于在SpaceX
341we actually are used to doing incredibly challenging physical work in terms of building我们实际上习惯于在建筑方面 做极具挑战性的体力工作
342giant rockets and vast launch complexes.巨大的火箭和巨大的发射复合体。
343And Tesla's also very good at large scale而特斯拉也非常擅长大规模
344construction.建筑业。
345And probably the two best companies at physical hardware on Earth outside of China也许是中国以外地球 物理硬件最好的两家公司
346是SpaceX 和 Tesla 。 (笑声)
347I don't even know who would be number three.我甚至不知道谁是第三位
348And, you know, I think even而且,你知道,我觉得甚至
349Tesla might be the best in the world, even including China, where China is incredibly good.特斯拉或许是世界上最好的,甚至包括中国在内,中国是极好的。
350It's worth noting that, you know, the Tesla Model Y is the best selling EV in China.值得注意的是,你知道,Tesla Model Y是中国最畅销的EV.
351Well, you taught them a lot in China.你在中国教过他们很多
352This is like winning a land war in Asia, classically difficult.这就像赢得一场亚洲的土地战争, 古典上的困难。
353And yet we kind of are,然而,我们有点,
354actually, in that Model Y is the best selling car in China.实际上,在Y型车是中国最好的销售车型。
355Think about that for a second.考虑一下
356And let me tell you, the Chinese car companies are super good.让我告诉你,中国汽车公司是超级优秀的。
357Yeah, the person from Ford went there.对,福特的人去了那里
358He said he was blown away by the quality of his cars in China.他说自己在中国的汽车质量被吹走.
359China's awesome.中国棒极了
360I love China, by the way.我爱中国,顺便说一句。
361It's inspiring to go there.去那里很鼓舞人心
362Well, most people who build in China, they have to have Chinese ownership在中国建设的大多数人 必须有中国人的所有权
363of a significant portion of the operation in China.在中国的很大一部分业务。
364And they were so anxious.他们很焦虑
36549% is what49%是怎样的
366you're allowed to own in China normally.通常允许你在中国拥有
367And Tesla, they wanted us to be there so badly特斯拉,他们非常希望我们在那里
368that they let us go 100% ownership for us.他们让我们拥有100%的所有权
369Yes.对
370That took quite a few years of meeting with花了好几年时间和
371Chinese officials.中国官员.
372And I actually, you know, some of the people in the government actually,实际上,我,你知道, 一些人在政府实际上,
373you know, they actually literally yelled at me, like spit flying type of thing, that we had to你知道,他们实际上 字面上吼我, 像吐口水飞的东西, 我们必须
374get sort of, you know, married to a Chinese partner.和中国合伙人结婚
375And I'm like, well, we're just a babyи琌
376company.连队。
377And I don't think babies should get married, you know, because I would look funny我觉得孩子不应该结婚 因为我看起来很搞笑
378going down the aisle with a baby.带着婴儿走下过道
379And they're like, you're not a baby.他们喜欢,你不是一个婴儿。
380And I'm like, no,而我喜欢,没有,
381we're a baby.我们是婴儿。
382And, you know, babies should just, you know, not get married.而且,你知道,孩子应该 只是,你知道,不要结婚。
383It's too young.太年轻了
384So这么说
385eventually they said, okay, fine, we'll change the law.最终他们说,好吧,好吧,我们会改变法律。
386And then you can own the factory.然后,你可以拥有工厂。
387And还有
388mostly it was like, we just need to be able to move fast, you know, and make decisions.大部分时候,就像,我们只是需要 能够快速行动,你知道的,并作出决定。
389And if如果
390you've got to always ask someone else for permission to do the decisions, then how are你总得征求别人同意 做决定,然后怎么样?
391you going to, you know, it's difficult to innovate if you don't actually have your hands on the你要知道,如果你真的没有 手放在
392steering wheel.方向盘。
393Well, these days, you don't need your hands on the steering wheel because of self现在,你不需要手放在方向盘上 因为自己
394driving.开车
395But that metaphor is not going to make any sense in the future.但是,这个比喻在未来没有任何意义。
396Yeah.对
397So we're talking这么说吧
398about revenues going up a lot.关于收入上升很多。
399For SpaceX, it was, I think, 15 or 16 billion last year.对SpaceX来说,我想去年是150亿或160亿。
400And this year还有今年
401it's going to be ending the year 26 at $100 billion or so of annualized revenues.26年的年收入将达到1000亿美元左右
402And then礛
403next year, we think it's going to be between 500 billion and a trillion.明年,我们觉得它将会在5000亿到1万亿之间。
404So we've gone from 15我们从15岁开始
405billion to somewhere around 100 billion annualized to 500 billion to a trillion.每年有1000亿到5000亿到万亿
406And that's just这就是
407getting started.开始
408And what's really cool about selling, renting, renting our data centers卖,租,租我们的数据中心
409to competitors, that basically they're paying us 50 billion for something cost us 35.对于竞争者来说, 他们基本上要给我们500亿 因为我们付出了35。
410And we're我们是
411going to have collections of cash coming in for renting data centers that we still own.将收取现金 来租我们仍拥有的数据中心。
412So we got所以,我们得到了
413for free, right?免费的,对吧?
414Well, I mean, I think just to comment on a few other things.嗯,我的意思是,我想只是 评论一些其他的事情。
415So we are tracking所以我们在追踪
416to have an annual rate of revenue that is in excess of $100 billion in December.12月年收入超过1000亿美元。
417Obviously,显然
418December is pretty close.十二月已经很近了
419So I have good intent to that.所以,我有一个很好的意向。
420So unless there's some sort of, you know,所以,除非有某种,你知道的,
421natural disaster or something, we will meet that target.自然灾害什么的,我们将达到这一目标。
422And then based on the plans that we have然后根据我们的计划
423next year, it seems quite likely that our annual recurring revenue in December of next year will明年12月我们每年的经常性收入
424exceed 500 billion.超过5000亿
425Now, this is not a promise.现在,这不是一个承诺。
426This is what I think will happen.这正是我想会发生的事。
427This is my best这是我最好的
428guess.猜对了
429This is what I want to distinguish between this is like a promise to do something versus这就是我要区分的 这就像一个承诺 做一些事情与
430what I think will occur.我认为会发生什么。
431What I think will occur is an ARR in December of next year of 500 billion.我认为,明年12月将发生5000亿的ARR。
432And I simply said, a trillion is not out of the question.我只是说,万亿并非不可能。
433Meaning it was previously out of the意思是,它以前是 出自
434question and it is no longer out of the question.这已经不是问题了。
435It's maybe 2% or something likelihood.也许有2%的概率
436But it's不过
437interesting that it's not 0%.有趣的是,它不是0%。
438That's the interesting part.这是有趣的部分。
439And the probability is increasing机会在增加
440with time.与时间。
441So what my biological neural net predicts for digital super intelligence in terms因此,我的生物神经网络预测 数字超级智能的方面
442of its effect on the global economy is about a 20 to 30% increase.它对全球经济的影响大约增加了20%至30%。
443Without taking robotics into不带机器人进入
444account, I think around a 20 to 30% increase, which would be roughly a $25 trillion increase大约要增加25万亿美元。
445in the global economy.在全球经济中。
446And this will happen very fast.这会很快发生的。
447So that's a $130 trillion global economy因此,这是一个130万亿美元的全球经济
448and you think that digital AI is going to increase that by 25, 30%?你认为数字AI会增加25%,30%?
44920 to 30%.20%到30%。
450Yes.对
451That's my guess.这就是我的猜测。
452What's your guess for how much would increase the economy?你猜会增加多少经济?
453I think probably 20 to 30%.我想大概是20%到30%
454I mean, it could be higher than that.我的意思是,它可能比这更高。
455But it probably doubles the productivity of white但它可能两倍于白人的生产力
456color work.颜色工作。
457So that's just digital.所以这只是数字。
458That's not physical.这不是物理的。
459That's digital.这是数字化的。
460数字AI.
461Physical is something we're going to be dominant in, I think.体质是我们要主导的东西,我想。
462Yes.对
463Well, the way to think about那么,如何思考的方式 考虑
464physical AI in the form of robotics is that any given economy is the number of people in it机器人形式的物理AI是任何特定经济都是其中的人数
465the average productivity per person.人均生产率。
466So if you have a humanoid robot that is extremely productive如果你有一个人造机器人 极富生产力
467and can work 24-7, so it's working 168 hours a week, it doesn't take vacation or bathroom breaks每周工作168小时,不休息或洗手间
468or need to eat food or anything like that.或需要吃食物或类似的东西。
469And the speed at which you can operate is faster than运行速度比
470human speed.人类速度。
471You can expect roughly a productivity per week that is on average five times better than每周的生产率平均比
472a person.一个人。
473Roughly.大约吧
474Because a person is working 40 hours a week and doesn't need to take breaks,因为一个人每周工作40小时 不需要休息
475obviously.很明显
476And doesn't move as fast as a robot.而且移动速度也不如机器人快.
477So I say a minimum productivity expectation for所以我说,最低生产力期望
478a humanoid robot would be five times that of a human.人造机器人是人类的五倍
479Then of the, you know, take the population然后,你知道, 人口
480of the United States, say how many people are either in school or retired?美国有多少人在学校或退休?
481Well, that's about嗯,这是关于
482half of the people, about half the people are either in school or retired.一半的人,大约一半的人不是在学校,就是退休了。
483And of those who还有那些
484are working age, not everyone works or is engaged in highly productive employment.工作年龄,不是每个人都工作或从事高生产力的就业。
485So then you're这么说你是
486left with perhaps around, I don't know, 120 million people in America that have a net可能还有一亿二千万美国网民
487positive output of goods and services.货物和服务的正产出。
488So then you say, well, how many robots then would be那你就说 有多少机器人
489needed to match that productivity?需要与生产率相匹配吗?
490Well, I guess you'd need, you know, about 20 to 30 million robots我想你需要两千到三千万个机器人
491to match the goods and services output of the United States.与美国的货物和服务产出相匹配。
492And we would build those, that's expensive to build a robot than to build a car.我们会建造这些, 造机器人比造汽车贵。
493Yes, initially, I'd say about the same.是的,一开始,我也会这么说
494But then over time, it would be much less.但随着时间推移,情况会少得多。
495So 30,000 cost3万块
496for a robot on the way to 20.给20号公路上的机器人
497Yeah.对
498And we would rent those or we would sell them or what would we我们会租这些 或者我们卖它们,或者我们是什么
499do?是吗?
500Do we have an idea?我们有主意吗?
501I think probably rent them.我想也许租的
502To stay on the data centers, then if people are留在数据中心 如果有人
503upset about data centers in their communities, is there any way that since these are so profitable因为他们社区里的数据中心很烦恼,
504for us, why wouldn't we just say, okay, here, your utility bills in your community will stay flat对我们来说,我们为什么不只说,好吧, 在这里,你的公用设施账单 在你的社区将保持平稳
505and we'll subsidize it.我们会给它补贴。
506And we will use that subsidy and we will also give you 50 million我们会用那份补贴,我们也会给你五千万
507dollars for your schools and hospitals.给学校和医院的钱
508Do we have any programs like that where we have gone我们有这样的节目吗?
509communities and say, yes, we are going to help you in your communities and we'll give you more并说,是的,我们会帮助你 在你的社区,我们将给你更多的
510jobs.任务。
511We're really a job creating company.我们真的是一个创造就业的公司。
512No one creates jobs the way we do.没有人像我们这样创造工作机会。
513Is there anyone,有没有人
514anyway, I don't think we've done, I don't think our industry has done a good job不管怎么说,我觉得我们没有做, 我认为我们的行业没有做得很好
515about explaining how this is a benefit of people, not harming them.解释这如何是人民的利益,而不是伤害他们。
516Yeah, well, actually locally,是啊,其实当地,
517in the Memphis area, I think we're actually quite popular.在孟菲斯地区,我觉得我们其实相当受欢迎。
518It's actually, the popularity is there,其实,人们的欢迎是在那里,
519the lack of popularity is where there aren't data centers and they think the data center will have缺乏受欢迎程度是没有数据中心的地方 他们认为数据中心将会有
520a bad effect.一个坏的影响。
521But in Memphis, you know, Memphis is booming because of space sites.但是在孟菲斯,你知道,孟菲斯因为空间站点而兴旺.
522The restaurants餐厅
523are full, the hotels are full.旅馆都满了
524We're paying at times twice the prevailing wage that people我们有时要付两倍的薪水
525were earning before.以前是赚钱的
526Why don't people write about this?为什么人们不写这个?
527Why aren't there stories about this?为什么没有这个故事?
528Why is this not?为什么不呢?
529There are some stories that affect.有一些故事影响。
530We've also helped out with local schools我们还帮了当地的学校
531and the tax revenue that the county and the city are earning is tremendous.郡和城市的税收是巨大的。
532So I think this is maybe this popularity thing is a temporary concern because there actually is所以我觉得这也许这个受欢迎的东西是一个暂时的担心,因为实际上有
533tremendous prosperity that's brought by the data centers.数据中心带来的巨大繁荣
534And if you talk to the people in town,如果你和镇上的人说话
535they all feel it because their businesses are booming.他们都感受到了 因为他们的生意蓬勃发展
536So then go to data centers in space,然后去太空的数据中心
537the obvious solution.很明显的解决方案。
538So people don't have to worry about them in their communities anymore这样人们就不用在社区里担心他们了
539and if they want, if they don't want to.如果他们愿意,如果他们不愿意。
540And for us, it's less expensive to put them in space.对我们来说,把它们放进太空比较便宜.
541And every time we put up new generations, they become more and more powerful than they had been每一次我们安置新生代, 他们就变得比他们更强大
542before.之前,我刚说
543Isn't that, and everyone says, well, it's 10 years away.不是么,大家都说, 10年了。
544And he says, no, we're starting,他说,不,我们开始,
545Google has already tested with our set with a satellite that we've run for them.Google已经用我们运行的卫星测试了我们的设备。
546It's been已经是
547effective.有效。
548And so we're going to be starting to put them up in 27 or 28.所以我们要开始把它们放在27或28。
549How long before impact?撞击要多久?
550Yeah.对
551是的,目前有人测试了Nvidia GPU。
552There's a satellite with有一个卫星与
553an Nvidia GPU up there that we launched and there's a satellite we recently launched with a Google我们发射的NvidiaGPU最近用谷歌发射了一颗卫星
554TPU that, and these are operating into the best of my knowledge that they're working fine in space.TEPU,这些正在运行 在我的最了解 他们的工作在太空。
555We expect to start launching the first Starmind computers, which we've co-designed with Nvidia,我们期望开始推出第一台星明电脑, 我们已经与Nvidia共同设计,
556probably, I'm just guessing here, but in the third quarter of next year.也许,我只是猜测这里, 但是在明年的第三季度。
557And we'll start off slow,然后我们慢慢开始
558but then it will ramp to significant volume probably in 28 and then very large volume in 29.但随后它可能会在28年,然后在29年, 大幅攀升。
559And really in order to get to the terawatt per year scale, or even I'd say maybe the 200 gigawatt而且真的为了达到每年的千兆瓦, 或者甚至我会说,也许200千兆瓦
560per year scale, you really need to go to space because they're just, there isn't enough power每年的尺度,你真的需要去太空 因为他们只是,没有足够的能量
561on the ground to realistically get to those levels per year.每年实际达到这些水平。
562Just as a frame of reference,作为参考
563the power output, the average power consumption in the United States is just over美国平均电力消耗刚刚结束
564500 gigawatts.五百千亿瓦.
565So if you look, the peak power capability in the United States is roughly所以,如果你看看, 美国的峰值电源 大约是
5661.2 terawatts, but average power consumption is just over 500 gigawatts.1.2千兆瓦,但平均耗电量刚刚超过500千兆瓦.
567So you can really see that这样你才能看出来
568if you're doing say 200 gigawatts a year of AI, that would be like doing a 40% increase in the如果你说每年200千兆瓦的人工智能, 那就等于增加40%
569solar panels and you put them in space.太阳能电池板 你把它们放进太空
570Because it's essentially, it's always sunny in space因为它本质上是, 它总是阳光在太空中
571and the solar panels can always face the sun directly.太阳板总是可以直接面对太阳
572They're not off angle to the sun.他们不是离角到太阳。
573And you don't have atmospheric attenuation of solar power.你没有在大气中减少太阳能。
574So you get 30% more solar power in所以你可以得到30%的太阳能
575space.空间。
576It's sunny all the time.阳光普照
577There's no seasonality.无季节性.
578You can get away with a very tiny你可以带着一个很小的
579battery pack because you don't need to last through the night.电池包 因为你不需要熬夜
580The cost of intelligence情报费用
581in space is dramatically less than the ground once you have a low cost per ton to orbit.一旦你每吨轨道成本低, 空间中的成本就大大低于地面。
582So Starship, because it will be the first fully reusable rocket, will lower the cost of access因此,星舰,因为它将是第一颗完全可重复使用的火箭, 将降低访问成本
583to space by a factor of 100 or maybe more.以100倍或更多乘以空间。
584I think actually in the limit, maybe within我认为实际上在极限,也许在
585say probably two or three years, the cost of a ton to orbit will be less than the cost of说大概两三年,一吨到轨道的成本将低于
586air freight from California to Australia.从加利福尼亚到澳大利亚的空运。
587So NASA said that in 2010, the cost to get to orbit美国航天局说,在2010年,进入轨道的成本
588was $18,500 a kilogram.一公斤18,500块
589That was because you use a rocket one time.那是因为你曾经用过火箭
590And then when we came out with然后当我们出来的时候
591a rocket Falcon 9, you could use over and over again, it became, I think it was $2,500.猎鹰9号火箭 你可以反复使用 它变成了,我想是2500美元。
592And then礛
593went to $1,500.去了1500块。
594And now it's going to go to $100 or $1,500.而今将至一百或一千五百.
595I think we'd get to under $100и稱и莱赣Τ100遏
596a kilogram to orbit.a 进入轨道。
597So it's important to just appreciate how profound full reusability is.所以重要的是要明白 完全的可重复性有多深
598So with Falcon 9, we have, I guess, probably around 80% reusability.因此,与猎鹰9, 我想,我们大概有 大约80%的复用性。
599So the booster comes back and所以助推器又回来了
600you'll probably see in the videos of the Falcon 9 booster landing.在猎鹰9型助推器降落的视频中 你可能会看到
601And then the fairing or nose cone然后是圆锥或鼻锥
602that contains the satellites, those fairing halves land by parachute in the ocean and we包含卫星的,那些美容 一半的降落伞 在海洋中,我们
603refly those.换掉那些
604But we do not refly the upper stage.但是我们不反驳上台。
605So we're unable to recover the upper stage of因此,我们无法恢复上层
606Falcon 9.猎鹰9号
607And the cost of the upper stage is roughly the cost of a medium-sized jet.而上层的造价大约是中型喷气机的造价.
608So that means the minimum cost per flight for Falcon 9 is a medium-sized jet.因此,这意味着猎鹰9的每次飞行的最低成本是一架中型喷气机.
609Now that's still现在还是这样
610much better than anything else.比其他任何东西都好
611But it's safe to say that there wouldn't be a lot of air travel if但可以肯定的是,如果
612every time you flew somewhere, you had to throw away a medium-sized jet.每次你飞到某个地方 都要丢掉一架中型飞机
613You're like, I think I'll你就像,我想我会
614drive.开车
615So now with Starship, the entire system is designed for reusability.所以现在与星舰一起,整个系统都是为可重复性设计的.
616And not just reusability,而且不只是重复性,
617but extremely rapid reusability.但是非常快速的复用性。
618The reason to catch the booster and the ship抓住助推器和飞船的理由
619with the tower arms, we call it the chopsticks.用塔的手臂,我们称之为筷子。
620It's kind of like Mr. Miyagi and the Party Kid有点像宫城先生和党内小子
621catching the fly, but just make it bigger.抓住苍蝇,但只是让它更大。
622And the reason for that is that the rocket takes原因就是火箭会
623off from the launch pad.从发射台关闭。
624So the rocket is placed on the launch pad by the tower and then is caught所以火箭被放置在发射台旁边 然后被抓住
625by those same tower arms and placed back on the launch pad when it returns from space.当发射台从太空返回时,
626So you don't just have full reusability, you have extremely rapid reusability.所以,你不只是有完全的复用性, 你有极快的复用性。
627You really couldn't你真的没办法
628get faster reusability than landing back where you took off.得到更快的复用性 比你降落的地方。
629And this in principle would allow这原则上是允许
630us to refly the rocket, the giant booster, the super heavy booster, once an hour.我们每个小时一次 反射火箭 巨型助推器 超级重型助推器
63110,000 launches10 000次发射
632a year.一年时间。
633Yes, probably more in the end, but yes, more than a launch an hour.是的,也许更多在最后, 但 是的,不止一个小时的发射。
634So when I was young,所以当我年轻的时候,
635people would line up to watch every now and very rarely a rocket.人们现在会排队观看,很少看到火箭。
636And we could do one an hour.我们可以做一个小时。
637Doesn't even make the newspapers anymore.连报纸都做不了了
638Yes.对
639Well, this will be quite, I mean,嗯,这将是相当的,我的意思是,
640you really want to watch this one.你真的想看这个
641This one's cool.这一个很酷。
642I mean, it's the largest我的意思是,这是最大的
643a flying object ever made.飞行器曾经制造。
644And it's the highest thrust object ever made by humans of any kind它是人类制造的 最高推力物体
645whatsoever.随便
646So yeah, this thing weighs over 5,000 tons on Liftoff and it's going straight up,这东西在升降机上重量超过5000吨 直升
647you know, fast.你知道,快点。
648And then that giant thing comes back, you know, it comes back from space,然后那个巨大的东西又从太空回来
649you know, like five minutes after Liftoff, that booster is coming down one way or another.你知道,像5分钟后,升降机, 该助推器正在下降 一种或另一种方式。
650That's going to, you know, hopefully in a good way, but it's coming down one way or another.这将是,你知道, 希望在一个好的方式, 但它正在下降的方式。
651So这么说
652excitement is guaranteed.兴奋是有保障的。
653Success is not always guaranteed, but excitement certainly is.成功并不总是得到保证,但激动是肯定的。
654It's very cool when you see it come back and lands in the same place where it took off.当你看到它回来 降落在起飞的同一地点时 感觉很酷
655Yeah.对
656The chopsticks.筷子
657Yeah, it looks crazy.是啊,它看起来疯了。
658You know, when I first proposed this to the team,当我第一次向团队提出这个建议时
659they looked at me like I had lost my mind, but I'm like, look, I think this can be done, you know.他们看着我 就像我失去了我的头脑, 但我喜欢,你看, 我想这是可以做到的,你知道的。
660There's nothing that's preventing it from happening from a physics standpoint.从物理学的角度来说,没有什么能阻止它发生。
661And the only thing I can think of that would be slightly better than catching it with the tower我唯一能想到的 就是比用塔抓住它好一点
662arms would be literally landing on the launch pad itself.武器会真正降落在发射台上
663But then, you know, if you get something但是,如果你得到的东西
664wrong, then you damage the launch pad and that would take a lot of work to fix.错误的,然后你破坏 发射台, 这将需要很多工作 修复。
665So if we just所以,如果我们只是
666catch it with the arms, that's a little better because we don't smash the launch pad necessarily用手臂抓住它,那会好一点,因为我们不打碎发射台
667if something goes wrong.如果出事的话
668But that's the difference of minutes between catching with the arms and但是,这是 分钟的区别 抓住手臂和
669simply landing in the actual launch pad it took off from.只是降落在它从实际发射台上。
670Now, much harder than the booster is现在,比助推器更难
671catching the ship.抓住飞船。
672There's a lot of things that have to be done that have never been done before.诸有众生得无上菩提.
673No one has ever made a truly reusable orbital heat shield.没有人做过真正可再使用的轨道热屏蔽。
674So the shuttle heat shield required所以需要穿梭热屏蔽
675nine months of repair between flights.飞行间隔9个月
676So even though the shuttle orbiter portion did come back,所以即使太空梭轨道飞行器部分真的回来了
677you know, they'd lose the main tank, which is the primary airframe or the primary sort of你知道,他们会失去主油箱, 这是主机体或主型
678primary structure really to which everything attacked on every flight.每一次飞行都会受到攻击
679But they get the但他们得到
680refurbishment of the heat shield before they can fly again.在它们能再飞之前整修热屏蔽
681So no one has ever made a fully所以从来没有人做一个完整的
682reusable orbital heat shield.可重复使用的轨道热屏蔽。
683It's an incredibly difficult thing to do.很难做到
684And it's something I've我也有这种感觉
685been working on for a decade, basically.基本上已经工作了十年了
686It's so hard.难曰.
687One of the hardest technical problems最大的技术问题之一
688I think humans have ever tried to solve.我认为人类曾经试图解决。
689But we've got an incredibly talented team that has worked但是我们有一个非常有才华的团队 曾经工作过
690incredibly hard for a very long time.非常艰难的很长一段时间。
691And we've been able to iterate, go back to iteration.并且我们一直能够 提振,回到迭代。
692We've我们已经
693had many, we've been able to do many iterations.有过很多,我们做了很多迭代。
694Whereas the shuttle couldn't really iterate.而太空梭却无法真正承受
695They always have people on board the shuttle.他们总是有人在太空梭上。
696So they couldn't take any risks because somebody所以他们不能冒险 因为有人
697might get killed.可能会被杀的
698Whereas with Starship, we were taking many risks in order to explore the和星舰一起 我们冒了很多风险来探索
699physics of what's needed to achieve full reusability of the heat shield and the entire热屏蔽和整个热屏蔽的 完全重用所需的物理
700vehicle.车辆。
701So we've actually achieved sort of precise soft landings of the ship several times因此,我们实际上已经实现了 某种精确的软着陆 飞船几次
702{\fn华文仿宋\fs16\1cHD1D1D1}到海里去
703If there had been a tower there, the ship would have been caught.如果那里有一座塔,那艘船就会被抓住.
704So that's maybe an interesting point to make.这或许是一个有趣的点。
705That if the ship thought it was landing with如果飞船认为它与
706a simulated tower and had a tower been there, it would have been caught.一个模拟的塔 如果有一个塔在那里, 它会被抓住。
707Now we want to be现在我们想成为
708very sure that everything's going to work when we bring it back to the actual tower so that we don't非常肯定,一切都会工作 当我们把它带回真正的塔 所以,我们不
709smash the tower or rain debris on people's heads, which would definitely affect our popularity.粉碎塔或雨屑 在人们的头上, 这肯定会影响我们的受欢迎。
710So we want to be highly confident in the reliability of the ship.因此,我们希望对舰只的可靠性具有高度信心.
711And I think we're我认为我们
712finally there where we think it is worth, with ship 42, which hopefully is prophetic because终于到了我们认为值得的地方, 第42号船, 希望它是预言的,因为
713you know 42 is kind of a special number, that we can try to catch it with the tower.你知道42是一个特殊的号码, 我们可以尝试抓住它与塔。
714So that's这么说吧
715what we're going to do in about a week from now.一周后我们会做什么
716We're all rooting for it.皆悉拥护.
717So Starship星舰
718is the great unlock here.这是伟大的解锁在这里。
719Just as an aside, I was watching the young person who was heading the顺便说一句,我正看着那个领导着
720Starship operation and he was speaking to a group of people in the audience wearing sport jackets星舰行动 他和一群穿着运动外套的观众说话
721and suits and white shirts.还有西装和白衬衫
722And he came on and he was wearing engineer boots and blue jeans他穿着工程师靴子和蓝色牛仔裤
723and a plaid shirt and unshaven and really long hair.和一件衬衫, 和不剃发和真正的长发。
724And he starts talking and he said,然后他开始说话 他说
725F this and F that and F this.F这个和F那个和F这个。
726And then I'm looking at him and saying, oh my God, that's me when I was然后,我看着他 说,哦,我的上帝,这是我 当我是
72740.40. 萨尔瓦多
728Really good guy.真是个好人
729Drives motorcycles.开摩托车
730So this is the great unlock.这就是伟大的解锁。
731But all the things that Elon但是,所有的东西 伊隆
732does, they might not make sense to you at the very beginning when you hear about them.你听到他们的消息时,他们可能就不理解。
733And then you还有你
734say, oh wow, I see.说,哦,哇,我明白了。
735He probably was thinking about that all along.他可能一直在想这个
736And so when you think about所以当你想到
737reusable rockets, when you talk about so much cost to get to space, the only ones who can go可重复使用的火箭, 当你谈论这么多的费用 到达太空, 唯一可以去的火箭
738to space, that was $18,000, the only ones who go to space are governments to protect their countries前往太空的只有政府 来保护自己的国家
739or it could be a television company who spends $2 billion on a satellite so it brought us to one或者可能是一家电视公司 在卫星上花费了20亿美元 所以它给我们带来了一个
740place.地点。
741No one else.无别者.
742And so at bringing down the cost so dramatically, then all of a sudden you如此惊人地降低成本,突然间你
743启用Starlink。
744And Starlink, you have 10,000 satellites now on the way to 20,000.还有Starlink 你现在有1万颗卫星 准备发射到2万颗
745And that's这是
746what I thought.和我想的一样
747And now Elon says, no, it's not going to be 20,000.现在伊隆说,不,它不会是2万。
748It's going to be 100,000.将谓十万.
749Why is it going to be 100,000?何故为十万.
750Because we thought in the past that the revenues he'd be generating因为我们过去以为他能创造收入
751from Starlink were going to be from consumers and government and enterprise.从Starlink 将来自消费者 以及政府和企业。
752And now, though, he现在,虽然,他
753says that agentic is going to be multiples of what we thought about before.说特工会是我们之前所想的数倍
754And every time you...每当你...
755Yeah, well, my understanding is that actually agentic internet traffic already exceeds是的,我的理解是,实际上 代理互联网流量已经超过
756human traffic, if I read that correctly, from Cloudflare.人类交通,如果我读的正确,从云雀。
757我相信Cloudflare预言
758that in five years, agentic AI traffic will be 1,000 times more than human traffic.五年后,人工智能的代理流量将比人流量多1000倍。
759And if anything, I think that is underestimating the amount of agentic AI traffic.如果有的话,我认为这低估了 人工智能的流量。
7601,000 times.诵一千遍.
761And how many years?还有几年?
762Five?五个?
763Yes.对
764So we've been thinking about Starlink as being所以,我们一直认为星际链接是
765a business that could do a trillion dollars in revenues and make seven or eight hundred一个能赚到一万亿美元 赚到七八百块的生意
766billion dollars in EBITDA in 10 years and be worth 14 or 15 trillion.10年后的EBITDA 价值达14或15万亿美元
767All of SpaceX is now worth现在所有的SpaceX都值了
768less than two trillion.不足两万亿.
769So that was one opportunity.因此,这是一个机会。
770But now,但现在
771I think we're probably really light on how much SpaceX, how much Starlink can be worth.我认为我们可能真的轻 多少SpaceX, 多少Starlink是值得的。
772But because you're adding so many more satellites, now we've got space centers, data centers.但是因为你增加了这么多的卫星, 现在我们有空间中心,数据中心。
773Yeah, I mean, Starlink revenue is growing very rapidly by any sort of normal business standards.是啊,我的意思是,Starlink的收入 快速增长 通过任何种类的正常业务标准。
774It's like you're roughly doubling revenues per year at large scale.这就像你大约翻倍 收入每年大规模。
775And I remember when that我记得,当
776used to be a big deal.从前是大事.
777It wasn't that long ago that doubling revenues with high gross margin不久前,收入翻一番,毛利润很高
778and positive cash flow every year would normally be an impressive thing.而每年的正现金流通常都令人印象深刻。
779Although I have to say,虽然我不得不说,
780with AI, it does exceed that.与AI,它确实超过这一点。
781So with AI, you have order of magnitude per year growth potential.所以在人工智能方面,你每年的增长潜力是按数量排列的。
782So and I mean, if you think about like things I've said, because a bunch of things I've said所以,我的意思是,如果你想 喜欢我说过的事情, 因为我说了很多事情
783sound extremely crazy.听起来非常疯狂。
784But I think we'll get there.但我想我们会到的
785You know, sometimes I'm a little optimistic on有时候我有点乐观
786the time frame, but I always deliver in the end.时间框架,但我总是在最后交付。
787So, you know, there's and also I'm not always所以,你知道,有,而且我并不总是
788late.迟了
789I'm trying to say that.我试图说。
790Because the press always reports on when I'm late, but they never因为媒体总是报道我迟到的时候,但他们从来没有
791report on when I'm early.早点报告
792And actually I'm early quite often.其实我经常早到
793So but the press, you know,所以,但媒体,你知道,
794is a sort of they never write positive articles.这是一种他们从来没有写正面的文章。
795They just ignore the times when I'm early and他们只是忽略了我早起的时候
796only print the times when I'm late.我迟到的时候才写
797So it seems like I'm always late, but that's actually not看来我总是迟到 但其实不是
798the case.这个案子
799My predictions, I aim for my predictions to be at the 50th percentile, which means that我的预测,我的目标 我的预测是 以50个百分点,这意味着
800they are basically half right with respect to the predicted time frame.就预测的时间范围而言,它们基本上一半是正确的。
801And that's about how这就是如何
802things have turned out.事情变了
803So, you know, we have 11,000 satellites up there right now, primarily所以,你知道,我们有11000颗卫星 现在在那里,主要
804the version two satellite.第二版卫星
805We've just started launching the version three satellite.我们刚刚发射三号卫星
806Version版本
807three satellite has over 10 times the capability of a version two satellite and we'll be launching三颗卫星的能力是第二版卫星的十倍以上,我们将发射
80810 times as many of them.其数十倍.
809So the amount of bandwidth that Starlink will produce in, I don't因此,星际链接产生的带宽量,我不
810know, two or three years will be a hundred times more bandwidth than it currently offers.两年或三年的带宽比现在高一百倍
811And还有
812yeah, so I think you said like what's the revenue per gigabit per second type of thing?是的,所以我想你说,像什么 收入每千兆塔每秒的事情?
813It's going to be high.将至高处.
814I mean, and then you say how many gigabits are you transferring?我的意思是, 然后你说多少千兆塔 你转移?
815It's这是
816by AI robots that far exceeds human demand by like literally a thousand.由AI机器人 远超人类需求 几乎千人。
817And just the Starlink communications revenue will be something staggeringly high.星际链接的通讯收入会很高
818But if you look at the AI stuff, if we do say a hundred gigawatts a year to space of AI compute,但是如果你看看AI的东西, 如果我们真的说100千兆瓦 每年的空间 AI计算,
819so what's the value per gigawatt of that compute?那计算值是多少?
820Well, even if it went down to ten dollars,就算跌到十块钱
821that would be a trillion dollars a year.一年一万亿美元。
822Extra.额外。
823And cumulative.和累积.
824So every year would be another所以,每年会是另一个
825trillion dollars because the satellites don't vaporize.因为卫星不会蒸发
826So you'd be stacking a trillion dollars所以你会堆积一万亿美元
827a year of revenue per year.每年收入一年。
828And every five years?五年一次吗?
829If it went down to ten dollars, if it was say at如果跌到10美元,如果它说
830fifty, you'd have five trillion.五十,你会有五万亿。
831So how are telcos ever going to meet demand for the service如何满足对这个服务的需求
832with the infrastructure that they have?他们拥有的基础设施?
833And if they have spectrum, for example,如果他们有光谱,例如,
834which is a public good, why would the government allow that to continue to be the case这是公益事业 政府为何会允许这种情况继续发生
835when there's a need for a lot more capacity than they can possibly provide?当需要 更多的能力 他们可能提供?
836Yeah, well, should we pass that question?是啊,我们应该通过这个问题吗?
837I think SpaceX will be able to get very far with the我想SpaceX能飞得很远
838spectrum that it's acquired.它的光谱是获得的。
839But, you know, at a certain point, if we're able to make much better但是,你知道,在某个时候, 如果我们能做得更好
840use of the spectrum than say a legacy telco is, we could just acquire that legacy telco and use使用频谱 而不是说一个遗留的telco是, 我们可以仅仅获得 遗留的telco 和使用
841spectrum, you know, ten times more effectively.光谱,你知道,十倍有效。
842And we have we've been thinking that for the我们一直在想
843data center.数据中心。
844So we're thinking about 10,000 to 100,000 for Starlink ultimately.因此,我们想 大约一万到十万 为星际链接最终。
845And every time,每次都这样
846every five years, we replace the satellites that exist, replacing them with ten times the capacity每五年一次 我们用十倍的能力来取代现有的卫星
847of what we're taking down, de-orbiting.正在拆除的轨道。
848So those are thousands of huge amounts of capacity we're这些是数千个巨大的容量,我们
849going to be providing.将会提供。
850Yes.对
851We still see opportunity for order of magnitude improvements in the我们仍然看到机会 在规模上改善
852satellite's capability.卫星的能力。
853So I think Starlink version four will have ten times the capability of version所以,我认为Starlink第四版将拥有十倍的版本能力
854three, especially with the long distance laser links.三,特别是长距离激光连线.
855So our current laser links are因此,我们目前的激光链接
856around 100 gigabits a second.大约100千兆巴一秒。
857Our next generation laser links will be ten terabits a second,我们下一代的激光链路 将每秒10塔比特,
858a hundred times better.一百倍的改进。
859And then we'll put ten of them on.后加持十门.
860So, you know,所以,你知道吗,
861so we get a little short of time, but there's a couple things more.所以,我们得到一点点 时间,但还有几件事。
862So when people are worried所以当人们担心的时候
863about data centers and about AI, people in China aren't worried about, you know, about having too关于数据中心和AI, 在中国人们并不担心, 你知道,有太
864much data centers and too much power.很多数据中心和太多的动力
865So in China, they have about three times the amount of power所以在中国,他们拥有大约三倍的能量
866that we do.我们这样做。
867And they're adding as much power a year, about 300 gigabits, compared to our 550他们每年增加大约300千兆比特 和我们的550比
868that we have in total.我们总共有。
869They're adding 300 a year.他们每年要加300元
870I'm not sure if that's a steady state power number,我不确定这是否是一个稳定状态的电源数字,
871but there are certainly increases.但肯定有增长。
872I don't think it's quite that high.我不认为它相当高。
873It was a little lower than这比
874that.那个
875But it is a massive number.但它是一个巨大的数字。
876So China at this point, I think, is roughly three times the因此,我认为,中国目前大约是
877output of the United States, and it's trending to get to at least four times the total power美国的输出, 并且它的趋势 达到至少四倍的总功率
878output of the United States.美国的产出。
879And industrial capacity to first approximation is proportionate to工业能力与
880your electricity output.你的电力输出。
881So, and then chips, they're behind.所以,然后芯片,他们落后。
882For now.现在。
883And they're moving他们正在移动
884forward fast.快速前进。
885And one executive at SpaceX a number of years ago told me that几年前SpaceX的一位主管告诉我
886she thought that the only thing standing between my grandchildren speaking Mandarin她认为我孙子之间唯一会说普通话的东西
887还有SpaceX
888Do you believe that?你相信吗?
889Well, I think there's a merit to learning Mandarin,我觉得学普通话是有优点的
890either way.不管怎么说
891It's a very interesting language.语言非常有趣.
892But yeah, I think in terms of competing with China,但是,我想在与中国竞争方面
893from an electricity generation standpoint, and just overall how much digital intelligence从发电的角度来说, 总体来说,有多少数字智能
894have you harnessed for economic activity, I think the only way to be competitive with利用你来从事经济活动 我认为唯一的办法就是与
895China is to take the intelligence to space.中国要把情报带到太空
896So there, we can sort of bring on在那里,我们可以 有点带
897a few hundred gigawatts a year, and like potentially, you know, a terawatt a year.每年几百千兆瓦, 和潜在的,你知道的,每年一兆瓦。
898So if we're able to do a terawatt a year of intelligence to space, then所以,如果我们能做一个千兆瓦 一年的智能到太空,那么
899in that scenario, unless China can do a terawatt a year to space, then we will have在这种情况下,除非中国每年对太空做一个千兆瓦,否则我们将
900an economic output in excess of China.超过中国的经济产出。
901So that's a very big number, of course.当然,这是一个非常大的数字。
902That's saying这就是说
903we do twice as much digital super intelligence per year as there is currently the average power我们每年做的数字超级智能两倍于现在的平均功率
904usage of the United States.美国的使用。
905So the average power usage of the United States is 500 gigawatts.因此,美国的平均功率使用量为500千兆瓦.
906A terawatt is a thousand gigawatts.千兆瓦就是千兆瓦
907So with a terawatt per year, that would be因此,每年有一个千兆瓦,这将是
908doing twice the current average power usage of the United States to space with digital将美国目前平均电能的两倍用于数字空间
909super intelligence every year.每年的超级情报
910These are like staggering numbers, basically.这些就像惊人的数字,基本上。
911So two more questions.故又二问.
912Why do we need so much power?为什么我们需要这么多的力量?
913Why do we need so much chip?为什么我们需要这么多芯片?
914Why?为什么?
915What's何谓
916power in chips?芯片里的动力?
917The demand will be, this is going to sound somewhat circular, but需求将会是, 这听起来有点循环,但
918the demand for intelligence, for digital super intelligence will come from digital super对智能的需求,对数字超级智能的需求 将来自数字超级
919intelligence.情报
920Which is for medical purposes, which is for, I mean, what does that mean?这是为了医疗目的 我是说 这是什么意思
921It means that super intelligence will saturate all human needs by far.这意味着超级智能会饱和所有人类的需求,直到现在.
922There will be universal将实现普遍化
923high income, universal extremely high income, universal medical care that is better than the高收入、普遍极端高收入、普遍医疗照顾比
924best medical care available on earth today.今天地球上最好的医疗护理
925Where you could wish for a castle or whatever you want,在那里你可以希望 一个城堡或你想要的,
926basically.基本上说
927And the digital super intelligence can make it for you as a side project.而数字超级智能可以让你成为一个侧面项目.
928So I guess if you zoom out a lot, you sort of think, well, okay, as civilizations progress,所以,我想如果你放大了很多, 你会觉得,嗯,好吧,随着文明的进步,
929you can think of the progression of civilization as how much energy has been harnessed by that你可以把文明的发展看作是 利用了多少能量
930civilization.人类文明
931And so currently, just to show you how much room for growth there is in our因此,目前, 只是为了向你们展示 我们的成长空间有多大
932of power harnessed by our civilization, by a factor of a million, we would still be using我们文明所利用的力量, 以百万的系数,我们仍会使用
933much than a millionth of the sun's energy.太阳能量的百万分之一
934That's a lot of room for growth.令增长广大.
935And you'd obviously, you know, completely saturate anything anyone could ever wish for on earth.你显然会,你知道, 完全饱和 任何任何人可能希望在地球上。
936And that point, it's really the digital super intelligence serving the digital super intelligence.而这一点,其实是数字超级智能服务于数字超级智能.
937And our economy will seem quaint, quaint and like, and our money system will seem like,我们的经济看起来很奇特,很奇特, 和我们的货币系统似乎,
938like, you know, exchanging seashells for currency.像,你知道,以海贝换取货币。
939That's what our monetary system will seem like我们的货币体系就会是这样
940to super intelligence.到超级智能。
941You know, technically, that's the thing that used to happen.你知道,在技术上,这是过去发生的事情。
942Well,那个
943I mean, we used to money used to be like, shiny metal that people would give to each other,我的意思是,我们过去的钱 像,闪亮的金属 人们会给对方,
944like, I'll give you this shiny silver piece of metal.喜欢,我会给你 这闪亮的银块金属。
945And then, and then you have like,然后,然后,然后你喜欢,
946shiny gold pieces of metal.闪闪发亮的金属片
947And, and you have to carry these shiny bits of metal from one place而且,你必须携带 这些闪亮的金属片 从一个地方
948to another and in order to do transactions, which is really onerous, frankly, and easy to rob.为了进行交易 交易非常繁重 坦率的说 容易抢劫
949You know, if you've got gold coins or silver coins, but that used to be how people你知道,如果你有金币或银币, 但过去人们是这样
950people used to do transactions, which is really onerous, frankly, and easy to rob.人们曾经做交易, 这真的很繁重,坦率, 很容易抢劫。
951would pay for things.会付出代价的
952Must have been incredibly difficult to buy things买东西一定很难
953long distance.长距离
954Because you've got to transport all the silver and gold因为你得把所有的金银都运走
955over the ocean, and the ship sinks,在海洋之上,有船舶沉没,
956and you don't even know what's going on.你甚至不知道发生了什么
957It's not like you can pull them on your cell phone你又不是用手机拉他们
958or something like that.或者类似的东西。
959So obviously, it'd be fun for me to talk to you forever.很明显,我永远跟你说话会很有趣
960But I'm out of time, and so I have just one last question.但我没有时间了 所以我还有最后一个问题
961So when you're 83 like I am, how do you所以,当你83 像我一样,你如何
962want people to think about you?想让别人想到你?
963So you're working at these crazy hours you're making.所以你是在这些疯狂的时间里工作的
964Nobody would work the way you work to just accumulate money,没有人会像你那样工作 积累钱财
965I think.我觉得
966And so you think that whatever you have所以你认为无论你有什么
967is going to be highly taxed at the end,最终会被高度征税,
968but you have a lot of money ultimately going但你有很多钱 最终去
969to foundation, your foundation.到基金会,你的基金会。
970What is, how do you want people to think about you?你希望人们怎么看待你?
971How will you want people to think about you when you're 83?八十三岁时,你希望人们如何看待你?
972Well, I guess I would like, there's perception and reality.嗯,我想,我想, 有感觉和现实。
973I would like the reality, whatever the perception is,我希望现实,不管是什么感觉,
974but I'd like the reality to be that I made但我希望现实是 我创造了
975a significant contribution in the furtherance of civilization.是对推进文明的重大贡献。
976That, you know, I think, yeah.这,你知道,我想,是的。
977Thank you so much for being here today.谢谢你今天能来
978Well, thank you very much.非常感谢
979Thank you.谢谢
该视频共有字幕 979 条。解锁更多字幕为会员功能,请移动到 价格