实战练英文 | SpaceX 凭什么成为史上最大 IPO ？

实战练英文 | SpaceX 凭什么成为史上最大 IPO ？AI 上太空的困难点都能解决吗？马斯克的计划是画饼还是老神在在？专访 Elon Musk 详细说明｜

发布于 2026-06-10 · 时长 31:32

字幕摘录

时间	英文	中文
0:00	All right, well, hello, everybody, and welcome.	好了大家好欢迎
0:03	Hanging out, I got Elon and Ian Dahl with our Starlink team,	挂出,我得到了伊隆和伊恩达尔与我们的星链团队,
0:06	figured we'd check in.	以为我们会报到
0:07	这是典型的SpaceX年.
0:10	Launched a brand new vehicle, acquired XAI, now SpaceX AI,	推出全新的飞行器,收购了XAI,现在的SpaceX AI,
0:15	and now it's a terror-sized chip building project.	现在它是一个恐怖的芯片建筑项目。
0:19	And so...	所以说...
0:19	Yeah, never a dull moment.	是啊,从来没有一个无聊的时刻。
0:21	Typical year.	典型的年份。
0:22	And so let's kind of wanted to connect some of the dots	我们想把一些点连接起来

展开字幕全文（616 条）

序号	英文	中文
1	All right, well, hello, everybody, and welcome.	好了大家好欢迎
2	Hanging out, I got Elon and Ian Dahl with our Starlink team,	挂出,我得到了伊隆和伊恩达尔与我们的星链团队,
3	figured we'd check in.	以为我们会报到
4	这是典型的SpaceX年.
5	Launched a brand new vehicle, acquired XAI, now SpaceX AI,	推出全新的飞行器,收购了XAI,现在的SpaceX AI,
6	and now it's a terror-sized chip building project.	现在它是一个恐怖的芯片建筑项目。
7	And so...	所以说...
8	Yeah, never a dull moment.	是啊,从来没有一个无聊的时刻。
9	Typical year.	典型的年份。
10	And so let's kind of wanted to connect some of the dots	我们想把一些点连接起来
11	on how this all feeds into making life	如何让生活充满活力
12	multi-planetary, starting to climb up the Kardashev scale.	多行星,开始攀登卡尔达舍夫级.
13	Maybe show off some cool new AI sat stuff.	也许可以炫耀一些很酷的新AI坐的东西。
14	Let's kind of start galaxy-sized and bring people	让我们开始银河系大小并带来人
15	in with the Kardashev scale.	与卡尔达舍夫的秤。
16	What's the big picture?	什么是大局吗?
17	What is the Kardashev scale?	卡尔达舍夫的尺码是什么?
18	Like, how do you decide what progress the civilization has	比如,你如何决定文明的进步
19	made that's the most objective metric	这是最客观的尺度
20	that any alien species, say, visiting us	任何外来物种,比如来拜访我们
21	would calibrate how much progress we've	会校准我们取得了多少进展
22	made as a civilization?	成为文明?
23	And one of the most objective ways	也是最客观的方法之一
24	to do that is the amount of power	这样做是权力的大小
25	that any given civilization has been able to harness.	任何文明都能够驾驭
26	And there was a Russian physicist, actually,	有个俄国物理学家
27	by the name of Kardashev, who thought about this.	由卡尔达舍夫的名字,他想到了这一点.
28	And I think it's a good way to characterize it, which is,	我认为这是一个很好的描述方法,即,
29	you can have, you can assess how well a civilization is	你可以,你可以评估一个文明有多好
30	harnessing the power available on the planet.	利用地球上的动力
31	That's type one.	这是第一类。
32	And then type two would be, how much of the stars' power	第二型是有多少星星的力量
33	are you harnessing?	你在操纵吗?
34	And then type three would be, how much of the galaxy's power	然后第三类是银河系的功率有多大
35	are you harnessing?	你在操纵吗?
36	These are very objective and measurable numbers.	这些是非常客观和可衡量的数字。
37	So right now, we're very low on the Kardashev 1 scale.	所以,现在,我们非常低在卡尔达舍夫1级。
38	Like, if you say, like, what proportion of our planet's	比如,如果你说, 像,我们星球的比例
39	power are we harnessing?	我们掌握了力量吗?
40	It's a very, very tiny number.	这是一个非常,非常小的数字。
41	And basically, we're harnessing almost nothing	基本上,我们几乎一无所获
42	of our star's power.	我们的明星的力量。
43	So the sun is truly an immense thing.	所以太阳真的是一个巨大的东西。
44	It is difficult, with words, to characterize just	用言语来形容
45	how immense the sun is.	太阳是多么巨大。
46	But this gives you sort of a sense of scale.	但这让你有一种规模感。
47	Yeah, it's a big difficulty jump going from level one	对,从第一级跳下去困难很大
48	to level two.	到二级。
49	Very big difficulty jump, yes, and level three.	非常困难的跳跃,是的,三级。
50	And we don't even know how to do level three, really.	我们甚至不知道如何做三级,真的。
51	We'll get there.	我们会到达那里。
52	Yeah, yeah, exactly.	没错没错
53	AI will figure it out, I think.	我想AI会想办法的
54	One way to appreciate the size of the sun	欣赏太阳大小的方法之一
55	is to think about how heavy is the sun compared	想想太阳比起来有多重
56	to all the rest of the mass in the solar system.	在太阳系的其余所有质量。
57	So the sun is about 99.86% of all mass in the solar system.	因此太阳约占太阳系所有质量的99.86%.
58	And then off the remaining 0.14%, most of that	剩下的0.14%, 大部分
59	is Jupiter, one planet.	是木星,一个星球
60	So we're still lightweight.	所以我们还是轻量级的。
61	Yes.	对
62	The entire mass of Earth is in the tiny miscellaneous category.	整个地球的质量属于微小的杂类.
63	We're like Earth is a tiny dust mote compared to the sun.	我们就像地球是一个微小的尘埃相对于太阳。
64	But how much energy are we talking,	但是,我们谈论的能量,
65	like coming from the sun, especially compared	譬如从太阳而来,
66	to what we're using here on Earth?	我们用在地球上?
67	Because it feels like...	因为感觉就像...
68	Yeah.	对
69	The incident solar energy on the cross-section of the Earth	地球横截面的太阳能事件
70	is roughly a half billionth of the sun's power output.	大约是太阳能量输出的五亿分之一
71	And the vast majority of that we cannot use,	而绝大多数我们不能使用,
72	because 70% of Earth is water.	因为地球70%是水
73	Technically, our planet should be called water,	从技术上讲,我们的星球应该叫水
74	because there is 70% water.	因为有70%的水。
75	And I think an alien civilization visiting us	我觉得一个外星文明会来拜访我们
76	would be like, why are they calling it Earth when it is mostly water?	他们为什么叫它"地球"呢?
77	Were the Greenlands not green of the solar system?	格陵兰不是太阳系的绿色吗?
78	Yeah.	对
79	A bunch of the...	一帮...
80	Even where 70% water and over 30% less land,	即使有70%的水和30%以上的土地,
81	a bunch of it is Antarctica or Siberia type of thing,	一群是南极洲或西伯利亚一类的东西
82	very Northern Canada type of thing, very difficult to...	在加拿大北部,非常难...
83	Not places people typically want to live,	不是人们一般想要生活的地方
84	and you're not going to get a lot of solar power at the poles.	连电线电源都没有
85	So the actual usable area of land	因此,实际可用的土地面积
86	where you can get solar power is quite small.	在那里你可以得到太阳能相当小。
87	Anyway, in order to ascend the Kardashev scale,	总之为了提升卡达舍夫的高度
88	in order to get to any meaningful percentage	为了达到任何有意义的百分比
89	of the sun's energy harnessed, you have to go to space.	利用太阳的能量, 你必须去太空。
90	If you wanted to get to, say, a millionth	如果你想得到,说,百万
91	of the power output of the sun,	太阳的功率,
92	you would have to increase civilizational energy harnessed	要增加文明的能量
93	by much more than a million.	超过一百万
94	So we currently use much less than a trillionth	因此,我们目前使用不到万亿
95	of the power output of the sun.	太阳的功率输出
96	And a trillion is a million times a million.	而万亿是百万倍.
97	So basically, we're practically nowhere	所以基本上,我们几乎无处可去
98	on the Kardashev-2 scale, practically nowhere.	在卡尔达舍夫2级,几乎无处可见。
99	So on Kardashev scale, we're all still...	因此,在卡尔达舍夫的规模, 我们都仍然...
100	We're non-existent.	我们不存在。
101	We're non-registered.	我们没有注册。
102	We're not even...	我们甚至...
103	Yeah, we're so...	是啊,我们是如此...
104	We're not registering.	我们没有登记。
105	Not even in Microsoft.	即使在微软也没有.
106	Yeah, we're... No.	是的,我们...
107	And so to actually...	所以实际上...
108	One microsol would be an epic, epic achievement	一个微溶液将是一个史诗般,史诗般的成就
109	relative to where we are right now.	相对于我们现在所处的位置
110	Something to aspire to.	某种渴望的东西。
111	Yeah.	对
112	Yeah, that's our goal.	是的,这就是我们的目标。
113	This is, I think, both simultaneously	这,我想,两者同时
114	an incredibly adventurous goal relative to where we are,	一个令人难以置信的冒险目标相对于我们的位置,
115	and yet not particularly adventurous	却并不特别冒险
116	as a percentage of the sun's energy	占太阳能量的百分比
117	to try to achieve power harnessed being one millionth	试图实现电能的利用是第100万
118	of what the sun outputs.	以太阳为表率。
119	And so to actually start...	所以实际上开始...
120	A microsol.	一个微溶液。
121	To actually start getting there, though,	要真正开始到达那里,虽然,
122	we're not just going to throw solar arrays in space,	我们不只是要扔太阳阵列在太空,
123	try to soak up a bunch of the sun.	试图浸泡一堆太阳。
124	There has to be a need.	一定有需要的需要吗?
125	You want to go up there and do something meaningful.	你想上去做点有意义的事
126	And obviously, until this point in human history,	显然,直到人类历史上的这一刻,
127	there hasn't really been a need.	没有真正需要。
128	What has changed to make us think that maybe now's the time	是什么让我们觉得现在是时候了
129	to start trying to notch a percentage point or two?	开始尝试在一两个百分点?
130	I mean, getting to a percent of the sun's energy?	我的意思是,得到百分之一的太阳能量?
131	Maybe not a percent.	也许不是百分之一。
132	Let's go, like...	让我们去,像...
133	We'll move the decimal point back a couple of spots.	我们把小数点往后移几个点
134	So you're an extremely thick-ass civilization	所以,你是一个极厚的屁股文明
135	if you get to one percent of the sun's energy.	如果你得到1%的太阳能量。
136	And I'm like, wow.	而我喜欢,哇。
137	That civilization is going to be	那个文明将会是
138	vastly more powerful than us, to say the least.	可以说,比我们强大得多。
139	So in order to start to make some progress	为了取得一些进展
140	on the Kardashev scale,	在卡尔达舍夫号上,
141	we need to launch satellites to orbit Earth	我们需要向地球轨道发射卫星
142	and capture solar power.	并抓住太阳能。
143	And that avoids the need to build	这样可以避免建设
144	massive power plants on Earth and deal with cooling,	地球上的大型发电厂和处理冷却,
145	because cooling is actually much easier in space	因为冷却在太空中其实容易得多
146	than it is on Earth.	而不是在地球上。
147	You can just radiate to the vacuum.	你可以只是辐射到真空。
148	And so what we're proposing here	所以,我们在这里提议
149	and what we intend to do is	我们打算做的是
150	to try to climb the Kardashev scale	试图攀登卡尔达舍夫的阶梯
151	to be kind of like a respectable civilization.	就像一个可敬的文明
152	So when the aliens,	所以当外星人,
153	hopefully there are aliens out there,	希望外面还有外星人
154	and they maybe finally decide to talk to us,	他们也许最终决定跟我们谈谈
155	where we have some respectable amount	在那里我们有一些可敬的金额
156	of the sun's energy being used,	以太阳的能量发誓,
157	that's not totally pathetic,	这不是完全可悲的,
158	which is the current situation.	这就是目前的情况。
159	And so before we start sending data centers,	所以在我们开始发送数据中心之前
160	sending all this to space,	把这些都送到太空去,
161	there are some limiting factors that we got to get there	有一些限制因素,我们必须到达那里
162	that would traditionally make it so like,	传统上它会这样,
163	this is almost impossible.	这几乎是不可能的。
164	Yeah, what does it take to scale?	是啊,它需要多少规模?
165	Yeah.	对
166	So things it takes to scale are,	所以,它所需要的是,
167	you need to have a large mass to orbit capability,	需要大量质量才能进入轨道
168	which is what Starship will give us,	这就是星舰会给我们的
169	that large mass.	大质量。
170	So you ultimately need to send millions of tons	所以,你最终需要发送数百万吨
171	to orbit and beyond,	至轨道及以外,
172	and you need the power associated with that.	你需要与它相关联的力量。
173	So if you want to put out 100 gigawatts,	所以,如果你想拿出100千兆瓦,
174	or ultimately a terawatt into space from Earth,	或最终从地球进入太空的千兆瓦,
175	you will at some point need a terawatt of solar,	在某个时候你需要一兆瓦的太阳能,
176	and then you're going to need a terawatt of AI chips.	然后,你需要一兆瓦的AI芯片。
177	So the three things you need are mass to orbit,	所以,你需要的三件事是质量轨道,
178	a lot of solar power, and radiators, of course,	大量的太阳能和散热器,当然,
179	and a lot of chips.	还有很多薯条
180	All right, well, let's start taking down the list.	好了,我们开始把名单拿下来
181	So mass to orbit, that's where Starship comes in.	因此质量到轨道, 这就是星舰进来。
182	We just had first flight V3, it was awesome.	我们刚开过第一班V3航班,太棒了
183	I know you were there.	我知道你在场
184	It was crazy to see that rocket launch	看到火箭发射真是疯狂
185	in like long time coming.	很久不见了
186	What's Starship's kind of purpose of being,	星际飞船的目的是什么?
187	what is it going to be doing?	它会做什么?
188	Yeah, so Starship is going to,	是啊,所以星舰是要去,
189	it's going to revolutionize space really.	它会让空间真正发生革命化。
190	It's the first rocket design	这是第一个火箭设计
191	that is capable of full and rapid reusability.	能够完全和快速的再使用。
192	Now, reusability is the fundamental breakthrough	现在,重用是根本的突破
193	that is necessary to make life multi-planetary,	使生命变得多行星,
194	as well as to ascend the Kardashev scale.	以及提升卡达舍夫的高度
195	You simply cannot ascend the Kardashev scale	你根本无法攀登卡达舍夫的阶梯
196	unless you have a reusable spacecraft	除非你有可重复使用的航天器
197	and you cannot extend life to the moon, to Mars,	你绝不能延长月亮和火星的生命,
198	and to the rest of the solar system	以及太阳系的其余部分
199	without a reusable rocket.	没有可重复使用的火箭。
200	The cost is simply prohibitive.	代价太高了。
201	You can't make enough rockets	你无法制造足够的火箭
202	unless you can refly them.	除非你能反驳他们
203	Just like any other mode of transport,	就像其他交通工具一样
204	you can imagine that if we had to throw away airplanes	你可以想象,如果我们不得不丢掉飞机
205	every time we flew, flying would be far too expensive	每次我们飞的时候飞得太贵了
206	and basically no one would be flying airplanes.	基本上没有人会飞飞机
207	You'd be doing a whole lot more driving.	你会做很多更多的驾驶。
208	Rapid reusability.	快速复用性.
209	Every mode of transport is reusable,	每一种运输方式都可以重复使用,
210	without which it's simply not viable as a transport system.	没有它,它根本不是可行的运输系统。
211	So cars, planes, boats, forces, bicycles	所以汽车,飞机,船只,部队,自行车
212	are all obviously reusable.	很明显都是可重复使用的。
213	With rockets, it's much harder to make a rocket reusable	有了火箭火箭就更难再用
214	because Earth has a deep gravity well and a thick atmosphere.	因为地球有深重力井和厚厚的大气
215	And these make it just barely possible	这些几乎不可能
216	to achieve reusability with a rocket.	用火箭实现再使用性。
217	And there have been many prior attempts	以前也做过很多尝试
218	to create a fully reusable rocket.	以创建完全可重复使用的火箭。
219	And most of those attempts have been abandoned	这些尝试大多被放弃了
220	part way through because they didn't think	部分原因是他们没想过
221	they could succeed.	他们可以成功。
222	In order to achieve full reusability,	为了实现完全的复用,
223	everything's gotta be perfect.	一切得圆满.
224	The engines, the structure, the avionics,	发动机结构航空器
225	the choice of propellant.	选择推进剂。
226	You've got to go to extreme measures for mass optimization,	你得采取极端措施优化质量
227	which is why we have the tower catch the rocket	这就是为什么我们让塔抓住火箭
228	instead of putting on landing legs, which are heavy.	而不是把降落的腿, 这是沉重的。
229	The rocket can simply be caught by the tower.	火箭可以简单地被塔抓住.
230	And we haven't achieved full reusability yet,	我们还没有完全恢复功能
231	but we do expect to achieve that.	但我们确实期望实现这一点。
232	Hopefully later this year with Starship,	希望今年晚些时候与星舰,
233	and then you've got to achieve full reusability,	然后你必须实现完全的复用性,
234	then you've got to go a step beyond that,	那你得再走一步
235	which is make it rapidly reusable,	它能使人迅速使用,
236	such that the rocket lands, gets caught by the tower,	火箭落地,被塔所夹住
237	gets put back on the launch stand,	回到发射台上
238	and can be flown again without any refurbishment	并且可以再次飞行而不进行任何翻新
239	or laborious inspection, like an aircraft.	或像飞机一样勤劳的检查。
240	This is incredibly difficult.	这太难了
241	This is the first time that there's ever been a rocket	这是第一次有火箭
242	where that is possible.	可能的话
243	That's what makes Starship so profound.	这让星舰如此深刻.
244	And it also happens to be the largest flying object	它也是最大的飞行器
245	ever made, the heaviest flying object ever made,	有史以来最重的飞行器
246	the most powerful moving object of any kind.	最强大的移动物体任何种类。
247	Starship V3 is more than double the thrust of it,	星舰V3的推力是它的两倍
248	the Saturn V moon rocket.	土星五号月球火箭
249	By version four, we'll be pretty much three times	到了第四版,我们差不多就三次了
250	the thrust of a Saturn V moon rocket.	土星五号月球火箭的推力
251	And we expect Starship to be flying	我们期待星舰飞翔
252	more than once per hour down the road.	路边每小时不止一次
253	One of the fun facts from flight 12,	12号航班的一个有趣的事实
254	that was actually the heaviest payload SpaceX has ever flown.	这是SpaceX飞过的最重载荷
255	And that's still just a fraction of what V3 can do.	这还是V3能做的一小部分。
256	So, I mean, once we're flying massive amounts	所以,我的意思是,一旦我们飞大量
257	really rapidly, I mean, we already fly the majority	飞得很快,我意思是我们已经飞了多数
258	of payload to space with Falcon.	和猎鹰一起进入太空的有效载荷
259	Do people even really understand	人们甚至真的明白
260	what mass orbit becomes once Starship is flying?	星际飞船一飞质量轨道就变成什么?
261	It's many orders of magnitude greater	这是许多数量级更大的
262	than what is the case today.	而不是今天的情况
263	So, even with Falcon 9, Falcon Heavy,	因此,即使与猎鹰9,猎鹰重型,
264	SpaceX delivers almost 90% of all Earth mass to orbit.	SpaceX将近90%的地球质量送入轨道.
265	I think somewhere between 85 and 90% right now.	我觉得现在有85%到90%左右
266	And then most of the remaining mass, I think,	然后大部分剩下的质量,我想,
267	is launched by China, and then the rest of the world,	由中国发起,然后世界其他地方,
268	including the rest of the US, is the remaining,	包括美国其他地区,是剩下的,
269	I don't know, five to seven percent.	我不知道,5%到7%。
270	Now, with Starship, we'll be aiming to go from	现在,与星舰, 我们将瞄准从
271	somewhere around 2,500 tons a year to orbit	大约每年2500吨到轨道
272	to millions of tons per year to orbit.	至每年数百万吨进入轨道。
273	And to do so at a pretty short period of time.	在相当短的时间内完成这项工作。
274	So, we think probably we can get to a million tons	所以,我们认为也许我们可以得到一百万吨
275	to orbit per year in about three years, thereabouts.	大约在三年后每年到达轨道
276	Starship is going to take care	星舰会照顾好你的
277	of the mass to orbit limiting factor.	质量到轨道限制系数。
278	Yes.	对
279	And then power generation.	然后发电.
280	So, first, and Ian, maybe you can help.	首先伊恩也许你可以帮忙
281	People probably struggle to visualize a little bit,	人们可能很难想象
282	when you say data center in space,	当你说数据中心在空间,
283	we're not going to slap engines on a building	我们不会拍引擎在建筑物上
284	and fly it up there.	飞上去
285	These actually look pretty different.	这些看起来很不一样。
286	And so, kind of walk through how you take something	所以,那种走通过如何你拿东西
287	that's in a giant building on the ground	在地上的巨型建筑里
288	and turn it into something that's functional in space.	把它变成在太空中发挥作用的东西。
289	Yeah, I think it's pretty interesting.	是的,我觉得这很有趣。
290	A lot of people don't actually know	很多人其实不知道
291	what the inside of a data center even looks like, right?	数据中心的内在是什么样子,对吧?
292	Yeah.	对
293	It's some mythical place where the internet's	这是一些神话的地方互联网的
294	in the cloud or something.	于云何等处.
295	Yeah, some people envision wires,	是啊,有些人设想电线,
296	some people envision boxes,	有些人想象着盒子,
297	but effectively it comes down to a set number of chips	但实际上它降为一组芯片
298	and the things that we need to launch into space	以及我们需要发射到太空的东西
299	are actually quite small when we look at it.	看着它,其实很小。
300	The more challenging part is figuring out	更具有挑战性的部分正在研究
301	how do you get the power for it?	你怎么得到它的力量吗?
302	And that's where a lot of what we've worked on	这就是我们努力过的许多
303	for existing Starwing technology, the solar arrays,	现有星翼技术太阳阵列
304	are what we want to utilize that expertise	我们想利用的是这些专长
305	to be able to build a satellite	能够建造卫星
306	that can actually launch the critical components	能够真正启动关键部件
307	of the data center into space itself.	将数据中心输入空间本身。
308	We like to look at this and say,	我们想看看这个,然后说,
309	what is the actual engineering problem here?	这里到底有什么工程问题?
310	And it's really a combination of delivering power	这真的是发电的结合
311	and then taking the waste heat and energy away	然后浪费热量和能量
312	and sending it into the vacuum of space, as you mentioned.	就像你提到的把它送进太空的真空
313	Now, the AI satellite is actually much simpler	现在,AI卫星其实简单多了
314	than a Starlink satellite.	比星际迷航卫星。
315	A Starlink satellite has gigantic phase-ray antennas.	一颗星链卫星有巨大的相射天线
316	It's got parabolic antennas.	有抛物天线.
317	It's got a lot of laser links.	它有很多激光链接.
318	It's much more complicated than an AI satellite.	比AI卫星复杂得多.
319	An AI satellite is essentially a lot of solar cells,	AI卫星基本上是很多太阳能电池,
320	a radiator, and you still need some laser links,	散热器,你还需要一些激光链接,
321	but you don't have all of the super complex antennas	但你没有所有的超复杂的天线
322	that you have on a Starlink satellite.	在星际链接卫星上
323	So, I mean, given the two,	所以,我的意思是,鉴于这两个,
324	the easier one to design for is the AI satellite.	最容易设计的是AI卫星.
325	Yeah.	对
326	It's just a little bit bigger.	略显大乘.
327	It's bigger.	其脉大.
328	Just makes stuff bigger, yeah.	只是让事情变得更大,是的。
329	I was like, so we've got, this is our AI one,	我当时想,所以我们得到了, 这是我们的AI之一,
330	if you guys want to walk us through.	如果你们想带我们过去
331	Yeah, so the first thing that we're really looking at here	是啊,所以第一件事我们真的在这里看
332	is first you've got to make something compelling, right?	首先,你必须做出令人信服的事情,对不对?
333	And we thought that the right place to start	而我们以为正确的起点
334	is around the 150 kilowatt peak power level,	大约150千瓦的峰值电源水平,
335	but as we look at the workloads	但是当我们看工作量
336	with our experience with XAI,	与我们的经验与XAI,
337	we get to actually see that we can also support	我们可以看到,我们也可以支持
338	about 120 kilowatts of average compute.	大约120千瓦的平均计算。
339	There's a difference.	有分别.
340	What we're showing here is kind of a draft version	我们在这里展示的是一种草稿版本
341	SpaceX AI卫星的第一版,
342	the AI one, I guess you could call it.	人工智能,我想你可以叫它。
343	And it seems like a reasonable place to start	这似乎是一个合理的起点
344	is 150 kilowatts peak power, 120 kilowatts sustained power.	150千瓦峰值功率,120千瓦持续功率.
345	And to give you a sense of what does that actually look like	让你明白这到底是什么样子
346	in terms of the size of the radiators,	就散热器的大小而言,
347	size of the solar panels,	太阳板大小,
348	the assumptions here are 250 watts per square meter	这里的假设是每平方米250瓦
349	for the solar array	太阳阵列
350	and about 1400 watts per square meter for the radiators.	散热器每平方米约1400瓦。
351	So the radiators, these are double-sided.	所以散热器是双面的
352	Radiators are radiating both sides.	辐射器在两侧辐射
353	They're oriented knife edge to the sun	它们是朝阳的刀锋
354	and it's 1400 watts per square meter	每平方米一千四百瓦
355	is a very achievable goal.	这是一个可以实现的目标。
356	Over time, we think we can probably do above	随着时间的推移,我们觉得我们也许可以超越
357	250 watts per square meter	每平方米250瓦
358	and above 1400 watts per square meter	每平方米1400瓦以上
359	for the solar panels and radiators respectively.	用于太阳能电池板和散热器。
360	But this gives you like a,	但是这让你像...
361	yeah, this is pretty much	是啊,这是差不多
362	what the satellite's gonna look like.	卫星会是什么样子。
363	It's a lot of solar panels, radiator,	有很多太阳能电池板散热器
364	and then everything else is pretty small, like embarrassing.	然后其他的都很小,像尴尬。
365	And these are like evolutions of things	这些就像事物的进化
366	that we have actually already launched	我们实际上已经启动
367	in our Starlink constellation to date.	我们的星际迷航星座至今
368	That's really, I think the cool part to me	这真的,我觉得我很酷的部分
369	is that we're looking at solar technology	我们是在研究太阳能技术
370	that we already are going to use	已经使用
371	V3 Starlink车辆上
372	So I'm like really excited to then just take those	所以,我真的很兴奋然后只是拿这些
373	and make it bigger.	让它变得更大。
374	Yeah, probably what we want to convey here	是啊,也许我们想传达的
375	is that there's not some magic that's necessary	没有必要的魔法
376	that doesn't exist for the AI satellites.	这对AI卫星来说并不存在
377	As Ian said, a lot of this is technology	正如伊恩所说,很多都是科技
378	we've already made for the Starlink V3 satellites.	我们已经为"星际迷航V3"卫星制造了
379	So basically we don't think this is a super hard problem	所以,基本上我们不认为这是一个超级困难的问题
380	compared to things we already do.	比起我们已经做的事
381	There would also be probably something	可能还有别的
382	in the order of a terabit of connectivity,	以互联互通的顺序排列,
383	of laser link connectivity from the satellite.	卫星的激光连接
384	The 150 kilowatt peak power level	150千瓦最高功率水平
385	大致符合Nvidia GV300架会做的
386	So if you've got a GV300 with 72 GPUs,	所以,如果你有GV300 与72GPU,
387	it's peak power I think is around 140 kilowatts,	我想是140千瓦左右的峰值电源
388	but it's rarely, it's almost impossible to get it	但它很少, 它几乎不可能得到它
389	to be at that peak power.	在那个峰值电源。
390	A more reasonable operating envelope	一个更合理的操作信封
391	would be around 120 kilowatts average power.	平均功率为120千瓦左右
392	But it can peak up to 150.	但它可以达到150峰.
393	So it's basically think of it as a rack of compute in space	所以它基本上认为它是空间的计算架
394	and then you can connect these racks of compute	然后可以连接这些计算架
395	to either each other by the laser links	通过激光链接互相连接
396	or directly to the Starlink constellations.	或者直接到星际链接星座。
397	So you can close the link with the Starlink constellation	这样你就可以关闭与星链星座的连接
398	and then Starlink can then send that data to the ground	然后Starlink就可以将数据发送到地面
399	using the existing KA and KU antennas on the vehicle.	使用车上现有的KA和KU天线。
400	It also has laser links to the ground as well.	它还有激光连接地面。
401	So, and this would not be at a particularly high latency.	因此,这不是一个特别高的延迟。
402	We're talking about maybe being around six to 800 kilometers	我们谈论的可能是六到八百公里左右
403	above the earth and light travels 300 kilometers	地球之上,光线行驶300公里
404	per millisecond.	每毫秒时。
405	So that's, it's about, you know,	所以,这是,你知道,
406	three milliseconds away basically, it's not very far.	3毫秒远,基本上不远.
407	We won't worry about that too much then.	那就别太担心了
408	Sometimes people think there's gonna be some	有时人们会觉得会有
409	like high latency, I'm like, no, speed of,	像高潜伏度,我像,不,速度,
410	light moves pretty fast.	光线移动得很快
411	Light moves pretty fast, the tall one.	光线移动得很快,高个子
412	Yeah.	对
413	I think the cool thing also is the radiators themselves	我觉得很酷的是散热器本身
414	are about the same size as the existing solar rays	和现有的太阳射线大小差不多
415	for the V3 vehicle, kind of in that realm	V3型机车,在那个领域
416	where we're flying today.	我们今天要飞的地方
417	Yeah.	对
418	So I mean, they got about a 70 meter wingspan.	我的意思是,他们得到了大约70米的翼展。
419	So these are fairly large.	所以这些是相当大的。
420	We're talking about building a lot of them	我们说要造很多
421	and putting them up there.	把它们放上去
422	But you like to say like spaces in the name,	但你喜欢说像空间在名字,
423	like there's a lot of space up there.	就像上面有很多空间
424	And so even when you're talking thousands	所以即使你说了几千个
425	or even, you know, up to a million satellites,	甚至,你知道,多达一百万颗卫星,
426	you got plenty of room to move around up there.	你有很多空间移动在那里。
427	Yeah, space is really big.	是啊,空间真的很大。
428	So it's not like space is gonna get crowded.	所以空间不会变得拥挤
429	Space is enormous.	空间是巨大的。
430	Like if you zoom in close, the satellite looks big.	就像你近距离放大一样卫星看起来很大
431	But if you actually look at it relative to the Earth,	但如果你从地球来看
432	these satellites are so tiny, you can't even see them.	这些卫星太小了你甚至看不到它们
433	So they're very, very tiny compared to Earth.	所以它们与地球相比非常,非常小。
434	And I mean, we have about 10,000 starlings	我的意思是,我们有大约一万个星人
435	in orbit right now.	现在在轨道上。
436	So we've got a pretty good idea of how to operate	所以,我们有一个相当好的如何操作
437	just really large constellations and do it safely now, right?	只是真正的大星座现在安全地做,对不对?
438	We are the only operator that has any experience	我们是唯一有经验的人
439	to that scale.	达到这个比例。
440	It's a great thing that, you know,	这是一件伟大的事情,你知道,
441	we have this background so we know how tightly	我们有这个背景所以我们知道有多紧
442	we can pack the satellites and fly them safely.	我们可以把卫星打包安全飞行
443	That's a number one goal when we look at the constellation.	那是我们看星座时的第一个目标
444	We're gonna be building a lot of satellites	我们要造很多卫星
445	and we're gonna be building them here in Bastrop, right?	我们要在巴斯特罗普建造它们对吧?
446	So we've got this, which,	所以,我们得到了这个,什么,
447	so we're in the building kind of in the middle, which-	所以我们在建筑那种在中间,
448	Yeah, we're sitting in that building right now.	对,我们正坐在那栋楼里
449	This is my first time here.	这是我第一次来这里
450	The building is massive.	大楼很大
451	Like you come around the corner,	就像你来到角落,
452	you see it through the trees and you're like, oh, wow.	你从树上看到它,你就像,哦,哇。
453	But we're about to kind of put this building	但我们要把这栋楼盖起来
454	to shame, aren't we?	羞愧,不是吗?
455	Yes, we're gonna, in fact, we already have	是的,我们要去,事实上,我们已经
456	the solar manufacturing facility.	太阳能制造设施
457	It's under construction already.	现已在建.
458	And then we will be building out	然后我们将逐步建立
459	AI Stat生产大楼很快。
460	And yeah, so we expect to have the AI stat production,	是的,所以我们期望有AI Stat的制作,
461	the solar production, and all of that operating	太阳能生产, 以及所有的操作
462	at some reasonable volume by the end of next year.	到明年底,一定数量。
463	So if anybody wants to work on AI satellites,	所以如果有人想在AI卫星上工作
464	this is kind of gonna become the hub of that.	这会成为其中的中心
465	We're also, so I mean, like right behind us,	我们也,所以,我的意思是, 像在我们后面,
466	machines are humming, we're still making	机器在哼,我们还在做
467	all of our user terminals for Starlink here.	我们所有的Starlink的用户终端
468	That's not going anywhere.	那不会去任何地方。
469	In fact, we're turning on new production lines	事实上,我们正在打开新的生产线
470	for new units, right?	为新单位,对不对?
471	Yes.	对
472	In fact, these are the new Starlink terminals,	事实上,这些是新的Starlink终端
473	which we made in much higher volume	我们造化了高得多的量
474	than the current terminals.	而不是当前终端。
475	And ultimately, we think there's probably gonna be	归根结底,我们认为很可能是
476	a few hundred million Starlink terminals out there.	几亿个星际链接终端
477	And then the Starlink direct to cell constellation	然后星际链接直接到细胞星座
478	will connect directly to people's cell phones	会直接连接到人们的手机
479	and enable high bandwidth communication	并允许高频通信
480	directly from your phone to space.	从手机直接到太空
481	All right, we're two limiting factors down.	好吧,我们有两个限制因素。
482	We've got mass to orbit,	我们有质量的轨道,
483	got putting solar in a few third ones chips.	把太阳能放进第三个芯片里
484	Yes, so at least in the beginning,	是的,所以至少一开始,
485	we can obviously launch the chips	我们显然可以发射芯片
486	that are already being made.	已经做了。
487	So our current reference design	所以我们现在的参考设计
488	这是NVIDIA鲁宾芯片,
489	or it could be either GB300 or Rubin chips.	或者可能是GB300或鲁宾芯片
490	And we'll also have a reference design for TPUs	我们还有TPU的参考设计
491	and essentially you can put up,	基本上你可以忍受,
492	put any existing chips into orbit.	将现有的芯片放入轨道。
493	But the current industry seems to be,	但现在的产业似乎,
494	it seems like it's gonna, I don't know,	它似乎会,我不知道,
495	get to maybe around 100 gigawatts a year of AI compute,	一年大约100千兆瓦的AI计算,
496	but that doesn't answer the question of,	但这并不能回答问题
497	well, how do you get to a terawatt?	那么,你如何得到一个兆瓦?
498	That's why you need the TerraFab.	这就是为什么你需要TerraFab。
499	Always looking a step bigger, yeah.	总是看起来更大一步,是的。
500	Yeah, in order to get to the next order of magnitude,	是的,为了达到下一个数量顺序,
501	you need a gigantic chip factory.	你需要一个巨大的芯片工厂。
502	And to give you a sense of scale here,	为了给你一个规模感,
503	我们期望TerraFab会是
504	around 100 million square feet,	大约1亿平方英尺,
505	which is 10 times the size of the Tesla Gigafactory Texas.	这是德克萨斯州特斯拉的10倍大小
506	And what aside from just,	除了这个之外
507	I'm gonna need Starship point to point	我需要星舰指点
508	to get from one end to the other,	从一端到另一端
509	aside from just the size,	除了尺寸之外,
510	what's gonna make this unique different	是什么会让这个独特的
511	from any other chip building operation on the planet?	从其他芯片建造操作在地球上?
512	Well, I think over time,	我想随着时间的推移,
513	there's gonna be a lot of technology evolution	将有很多技术进化
514	with the TerraFab,	以大地为生,
515	but fundamentally, it's about scale.	但从根本上来说,这是关于规模的。
516	So even if there were no	所以即使没有
517	fundamental technology breakthroughs,	基本技术突破,
518	and you could simply scale	你可以简单地缩放
519	the existing chip-making technology	现有芯片制造技术
520	with a lot of difficulty	困难重重
521	to a terawatt of chip output per year.	每年的芯片产量为千兆瓦。
522	That's, if you look at it just from the logic die standpoint,	也就是说,如果你从逻辑死亡的角度来看它,
523	that's like having a billion chips per year	就像每年有十亿块薯片
524	with a kilowatt per radical.	每根基有一个千瓦
525	So it's a billion full radical equivalent chips	所以这是十亿个完全激进的筹码
526	each doing a kilowatt,	各作千瓦,
527	and then you're gonna need a lot of memory to go with that.	然后,你会需要很多的记忆与去。
528	A lot of people today even think	今天很多人甚至认为
529	orbital data centers were like a decade away.	轨道数据中心就像一个十年后。
530	Yeah, I think we wanna try to give people	是的,我想我们想给人们
531	a sense of the timeframe.	一种时间范围的感觉。
532	At least the timeframe we're aiming for.	至少我们的目标时间范围。
533	I mean, people should take this	我的意思是,人们应该采取这个
534	with a grain of salt to some degree	一定程度上还有一粒盐
535	because this is just our best guess.	因为这是我们最好的猜测
536	So this is not a promise of what we'll do.	所以,这不是一个承诺我们将做什么。
537	This is what we are going to try to do	这就是我们要做的
538	and think we probably can do,	并且认为我们也许可以做到这一点,
539	which is to get to roughly an annualized rate	也就是达到大致的年化率
540	of a gigawatt per year by the end of next year	到明年年底每年1千兆瓦
541	in terms of space AI compute,	在空间方面 AI计算,
542	and then aspirationally scale that	然后满怀希望地衡量
543	by an order of magnitude per year.	每年按数量顺序计算。
544	So in two and a half years hitting an annualized rate	所以在两年半内, 达到年化率
545	of 10 gigawatts a year to space,	每年10千兆瓦的空间,
546	and three and a half years, maybe 100 gigawatts.	三年半也许100千兆瓦
547	And then depending upon what progress there is	然后看进展如何
548	in chip making in the rest of the world	在世界其他地方制造芯片
549	and with the terafab going beyond that	并随着terafab 超越了这一点
550	to scale to a terawatt per year,	以年均千兆瓦为单位;
551	which is a thousand gigawatts,	这是一千千兆瓦,
552	which is that's twice the current electricity consumption	就是电量的两倍
553	of the United States.	美国常驻联合国代表
554	I think there will be an appetite for that, but we'll see.	我想会有胃口的但是我们会知道的
555	It's a lot of satellites.	卫星数量众多.
556	I don't know what it's gonna think about,	我不知道它会怎么想,
557	but maybe do a lot of simulations or something.	但也许做很多模拟什么的。
558	So after we've, you know,	所以,我们之后,你知道,
559	worked through all the limiting factors,	克服了所有的限制因素,
560	we've kind of topped out what we can do on earth.	我们掌握了我们的能力
561	What is the next step to, again,	下一步是怎样?
562	try and actually notch maybe some percentage points	尝试并实际注意一些百分比
563	towards becoming Kardashev level two?	成为卡达舍夫二级?
564	Why stop there?	为什么停下来?
565	Yeah.	对
566	Why think small?	为什么觉得小?
567	Because a terawatt actually is very small.	因为一兆瓦实际上非常小。
568	I don't want you to think small.	我不想让你觉得小。
569	Let's not think small.	让我们不要想小。
570	So there is, in order to get to another three orders	所以有,为了得到另外三个订单
571	of magnitude to thousand X from a terawatt per year,	从每年千兆瓦到千X,
572	the only way that we can really see that you can achieve	只有这样我们才能真正看到你能实现
573	that is on the moon with a mass driver,	那是在月球上,有一位大型的司机,
574	essentially where you do local production	主要是在当地生产的地方
575	of photovoltaics and radiators on the moon.	月球上的光伏和散热器
576	Maybe you bring the chips from Earth	也许你从地球带来了筹码
577	or you could conceivably make the chips on the moon.	或者你可以想象在月球上制造芯片
578	But you need most of the mass to be made on the moon	但你需要大部分质量在月球上制造
579	so you don't have to transport it to the moon from Earth.	所以你不用从地球把它运到月球
580	And then because the moon has no atmosphere	因为月亮没有大气层
581	and only one sixth Earth's gravity,	只有六分之一的地球引力
582	you can accelerate the AI satellites into deep space	可以加速AI卫星进入深空
583	without a rocket.	没有火箭。
584	So you can basically shoot them into space	这样你就可以把他们射入太空了
585	using an electromagnetic gun, like a rail gun type.	使用电磁枪,就像铁轨枪型.
586	I mean, it's basically a linear electric motor	我的意思是,它基本上是一个线性电动机
587	is the way to think about it.	这是思考的方法。
588	As I think we can show people.	我想我们可以给大家看
589	I mean, if that doesn't get you excited for the future,	我的意思是,如果这不让你兴奋的未来,
590	I don't really know what will.	我真的不知道会怎样
591	I'm fired up to see a mass driver on the moon.	我被炒鱿鱼了看月球上的大众司机
592	That would be very cool.	这将是非常酷。
593	Yeah, sci-fi future.	是啊,科幻的未来。
594	It would also mean that if we're bringing	这也意味着,如果我们带来了
595	that amount of mass to the moon,	即月亮的质量,
596	it would mean that anyone who wants to go to the moon	意思是任何想登月的人
597	will be able to go to the moon.	能够去月球。
598	And I think that'd be pretty cool.	我觉得这很酷
599	Yeah, I'm gonna be jumping first in line to get up there.	是啊,我要去跳第一条排到那里。
600	Yeah, I mean, you never want to go to the moon	是啊,我的意思是,你永远不想去月球
601	at least once, I think.	至少有一次,我想。
602	Just once.	仅此一次.
603	Yeah, you can move there if you want.	是啊,你可以在那里移动,如果你想。
604	Go live on the moon.	去住月华.
605	We'll see.	等着瞧吧
606	Thanks, guys, for chatting with me for a little bit.	谢谢,伙计们, 和我聊了一会儿。
607	All right.	好吧,我们走。
608	Excited to see a whole new kind of satellite,	很高兴看到全新的卫星
609	a whole bunch more Starship launches,	一大堆更多的星舰发射,
610	more chips, more solar, more everything.	更多的芯片,更多的太阳能,更多的一切。
611	It's a big future,	这是一个巨大的未来,
612	but I'm excited to see everybody at this company	但我很高兴看到这家公司的所有人
613	go out and build.	出去建设。
614	All right, sounds good.	好吧,听起来不错。
615	It's exciting.	令人兴奋.
616	Great. Thanks, guys.	不错谢谢

该视频共有字幕 618 条。解锁更多字幕为会员功能，请移动到价格

下载安装