Shadowing Practice: 實戰練英文｜再不停，人類要滅了？！黃仁勳痛批 AI 末世論，川普突然 CALL-IN 亂入，大罵這個陰謀只會讓美國輸！｜Nvidia、All-In Summit｜長文、註解、大字版、中英對照 - Learn English Speaking with Video
בחר מצב תרגול
כל מצב פותח חדר תרגול במסך מלא.
734 משפטים
אם המשפטים קצרים או ארוכים מדי, לחץ על ערוך כדי להתאים אותם.
1
Oh, no. This is not planned, but we know who it is.
2
Oh, no. No. Mr. President?
3
Oh.
4
Yes, sir.
5
I got to tell you something.
6
Ladies and gentlemen, GPU Jesus.
7
They love you.
8
They love you.
9
Thank you.
10
I love you back.
11
Number one podcast in the world.
12
In the world.
13
Wow.
14
We like the new jacket.
15
Well, you know.
16
You auctioned the open?
17
I just, I felt you guys needed some energy.
18
Yes.
19
I know we're talking about serious stuff here, but we need to talk about it with energy.
20
Yes.
21
Let's start with this essay from this weekend. Which one?
22
Let's start with Dario's essay.
23
Was Hemingway involved?
24
Actually, did anybody run it through Pangram?
25
I don't even know how much of it was AI helped, but that was a pretty incredible thing.
26
And then I think what a lot of people were surprised
27
by was the coalescing of the Frontier Labs around the essay itself.
28
Just, Jensen, unpack what happened, how you read it, how you interpreted it, and then we'll get into some details that were inside of it.
29
But maybe just high -level thoughts to kick it off.
30
Well, first of all, there were a lot of stuff in there.
31
Yeah.
32
And first, there's a part about safety which we have to take very seriously.
33
Safety is paramount, obviously.
34
Safety and leadership are not false.
35
They're false choices.
36
You're able to innovate quickly.
37
You're able to execute quickly.
38
And America's able to lead and to do it safely.
39
I think those are false choices, but safety is obviously important.
40
There's a matter of of internal control that I think he was speaking to.
41
Obviously, the Coxson whistleblower is very serious matter.
42
Whenever you have a whistleblower, you gotta take it very seriously.
43
I thought Coxson had great courage to put out what his concerns were.
44
And even then, there were some issues that were kind of conflated within that.
45
I think the whistleblowing is fine.
46
I think the scientific prediction about the future is less fine because it's not grounded on science, obviously.
47
And it was expressed by a scientist, but it was obviously not grounded on science.
48
And so I take issue with that.
49
But obviously the whistleblower part of it, you know, I think there's just a whole bunch of stuff.
50
Pausing, pacing, those are all the voluntary things that they could do if they feel that their company is out of control.
51
If Coxon saw something, obviously we don't know what Coxon saw,
52
but if he saw that the company was out of control and maybe it's a transition from research to engineering.
53
As you know, these labs are transitioning from research to engineering.
54
Extraordinary talent, extraordinary engineering, but obviously engineering is different than research maybe
55
that transition is clumsy you know I don't we don't know what what he saw
56
and ultimately only he knows but
57
if there was a matter of lack of control that's a
58
different topic how should the government deal with it now all of a sudden regulation
59
and rate I mean it just covers everything in one block
60
can you just help us sort of unpack we played we
61
tried to play this game actually this week on the pod
62
and it was difficult which is how do you describe like you know my mom calls me
63
and she's like Chmoth what is this whole civilizational death thing.
64
I don't know how to explain it to her.
65
So when you have very smart people like that quantize it and quantify it, I think that's probably what's perturbing to some people.
66
They're like, what does that mean, 10 % of extinction?
67
Nobody knows how to explain that to the average person, how that's even possible.
68
Well, first of all, we shouldn't because it's made up.
69
First of all, I think that we shouldn't because it's made up.
70
And these are well -educated.
71
They're called researchers.
72
Obviously, they're working in a lab.
73
And so the confluence of these words and then the prediction is alarming and troubling, and it shouldn't be done.
74
It's irresponsible.
75
Now the fact of the matter is, let's go back and look at the real facts.
76
The facts are there was a prediction that in five years time
77
radiology will be completely taken over by artificial intelligence and there'll be no radiologists in the world.
78
That has proven to be exactly the opposite.
79
We need more radiologists than ever in the world.
80
However, AI has taken over radiology completely, which is great.
81
It is automated scan reading, which is great.
82
There was a prediction that within six to twelve months, wasn't it just last year, within six to 12 months.
83
90 % of code would already be generated by AI.
84
That has turned out to be wrong.
85
Within six to nine months, that was predicted last year, 50 % of entry jobs will be wiped out.
86
That has proven to be wrong.
87
Let's see, what else has proven to be wrong?
88
I mean, all of these predictions have been wrong.
89
Right.
90
And if you're - That GPT -2 would be too unsafe to release, that Lama -3 would be too unsafe to release.
91
Oh, Juan.
92
Yeah, we've heard these.
93
Half of white collar jobs would be gone next year.
94
The jobs apocalypse.
95
We have to take accountability.
96
We have to take account for all of the stupid predictions that were made.
97
Right.
98
Somebody has to take...
99
And so we have to just keep track of all that.
100
And of course people do.
101
And remind us that those predictions are inconsistent with ultimately America winning the AI race.
102
The short form for that is some people are saying, you know, they say trust the experts and they use the analog of COVID, which again started with people that were researchers,
103
educated people that had an asymmetric awareness of the thing that the rest of us did not, saying things that ultimately turned out, we find out in facts, not to be true.
104
And so there's this war that's happening right now between the trust the experts movement and the you know Well,
105
let's just look at the actual history of these predictions and let's just think more methodically
106
What where is this coming from
107
because it's coming from inside the places that's actually making it like what do you think is the Psychological makeup
108
or what is the real incentive?
109
Maybe it's a business incentive.
110
Maybe it's a political incentive Can you just maybe guess
111
or how do you how do you think about what why they're doing this?
112
Well, first of all, I've got to tell you, these are some of the most consequential companies in history.
113
Extraordinary engineers, extraordinary researchers, really fantastic work.
114
On the one hand, I work very closely with them as companies to companies.
115
On the other hand, we have to have conversations like this in public.
116
And it's really unfortunate.
117
And I think that these companies really ought to be built the way that we used to build companies, which is in silence.
118
Right.
119
So wait, Jensen, you don't allow anybody in your organization to speak for the entire organization, especially when they're having a bad weekend or they rage quit?
120
They're not allowed to tweet on your behalf and the organization's behalf?
121
Because, well, that's what they decided when they came to work for us.
122
And we told them, this is the way you behave when you work in our company.
123
And if you like the culture of our company, which, as you know, the NVIDIA culture and the NVIDIA employee base, incredibly happy.
124
They like the fact that the company is consistent, that we're stable, that our core values are consistent with taking care of the families
125
and creating the conditions by which they can do their life's work.
126
That we do meaningful work.
127
We do it as quietly as we can and we contribute to everybody else's success, which we're very proud of.
128
And so those kind of core values people are attracted to.
129
But when you come and work in our company, there are also some things that we don't appreciate that you do.
130
Like, for example, we don't welcome political discourse inside our company.
131
Take it home.
132
You guys talk about politics outside the company.
133
Yeah.
134
We are, the company is an apolitical company.
135
You know, we're bipartisan.
136
We want America to succeed.
137
And we want whatever government is in place, we'll do everything in our power to help America succeed.
138
And so the discourse about race and religion and politics and all of that stuff,
139
we tell people, do it outside the company.
140
It's not for us.
141
In terms of maybe AI regulation, then more narrowly, Satya was here this morning and what he said is, you know, before we talk about regulation that could really stymie things,
142
why don't we just get some basics right?
143
Why don't we get measurement right?
144
Why don't we get standardization right? right um where do you
145
land on get engineering right get the engineering right right translate the research in a more predictable way so
146
that we're not fear -mongering keep it inside until we're ready
147
to expose it um what do you think the right response is you know demis had a proposal
148
which was sort of this more finra -like organization it's not clear what daria wants this transnational mutated thing
149
that has some sort of control where do you land on
150
this the sort of perspective of What do we need right now?
151
You know, regulations should solve actual problems.
152
And so the question is, what actual problems have we enjoyed?
153
And if you look at the actual problems, all of the actual problems so far have come from the labs.
154
And the reason for that, the reason for that, and just in their defense, the reason for that is because they have the most compute.
155
Right.
156
And the reason for that is because they're trying to solve the frontier problems.
157
And so in their defense.
158
And so it's sensible that the labs, the frontier labs, will be where the most danger come from.
159
It is unlikely that a high school student did something.
160
Because they just simply won't have enough compute.
161
Right.
162
And so it's unlikely that a startup will be the reason because they won't have enough compute.
163
In fact, you could look across the planet and everybody won't have enough compute with the exception of the frontier labs.
164
And so now the question is, if you look at what actually happened, and they're doing pioneering work, it's really very hard.
165
They're transitioning from research to engineering engineering,
166
I could imagine, and they're obviously building some of the most consequential technology and companies in the world.
167
They're building their company, they're building their culture, they're building the technology, they're building engineering, they're building products all at the same time.
168
And so I can understand it's a little bit hair on fire.
169
But nonetheless, the four incidents from one lab, the one giant incident from the other lab,
170
the first thing that you have to do is just root cause the problem from an engineering perspective.
171
What happened?
172
What could we have done differently?
173
And what are we going to implement and institutionalize, whether it's technology or methods or processes, and make sure that we don't let it happen again?
174
Now, I would bet you money that in every single one of those cases is within their control
175
in the future to prevent it.
176
because the alternative, if it's not in their control, and I'm sure that they are, I'm sure those four incidents won't happen again.
177
I'm sure they root -caused it and fixed it.
178
I'm sure they have now much better technology for sandboxes and runtimes and monitors and continuous monitors.
179
And so I'm certain they have much, much better technology now.
180
The alternative is also unlikely which is for them to say, look, we had these incidents.
181
After we're done analyzing it, we came to the conclusion we don't know anything that happened, and we have no idea how to control it,
182
and we're asking society for help.
183
Now, if that's the case, then we ought to, you know, a bunch of companies with engineers ought to send engineers in, and we should advise them if we can.
184
But I doubt it.
185
I think they have extraordinary people that got this handled.
186
We're not operating in a vacuum.
187
David, last night you informed me that there is a Chinese lab, the makers of GLM, who are going to put $3 billion towards a recursive self -improvement run.
188
So maybe you could tee that up for Jeff.
189
Well, that's what was announced.
190
Zipu .com, the founder, just raised $5 billion and said
191
that one of their priorities is going to be trying to get AI
192
that trains the next AI and to try and automate as much of that as possible.
193
Yeah, I think that...
194
Well, this is the new sexy phrase.
195
As you guys know, RSI is a combination of a system of ideas.
196
It starts everything with in -context stuff.
197
It starts with skills.
198
It starts with reflection.
199
It starts with reinforcement learning and synthetic data generation.
200
And these are all very sensible ideas that causes AI to get better at solving a problem over time.
201
And you could also have low rank.
202
All of that stuff doesn't include the weights.
203
You could actually improve the weights, and it's called LoRa.
204
LoRa could be improved in synthetic data generation, reinforcement learning, enhance it without training the base model itself.
205
And then over time, you could train the base model again with all of that experience.
206
And so I think it's a sensible thing
207
that you're going to use the technology to enhance productivity of all kinds of tasks, including building AI.
208
I think that's a very logical idea.
209
And I'm certain that everybody is using it in some degree.
210
It's just this phrase is now being used to weaponize the technology in some way and maybe to turn the technology.
211
It's going to spiral out of control is the impression they're trying to give, but you don't believe that's real.
212
No, no, of course not.
213
And the reason for that is because you could RSI all day long inside your company, but when you release a product, you've got to evaluate it, don't you?
214
You have to test it again, don't you?
215
You have to make sure that there's no regression, right?
216
And so the basic process of control, these labs are going to, as they move from labs to engineering, they will have much, much better control.
217
Right and when they have much better control that control comes from methods
218
and knowledge and practice and tools and technology All of those things
219
that lead to better control verification
220
and evals It's going to enable RSI to be done inside the company
221
and for good price to be released outside about open source for a second.
222
I mean this hugging face
223
We were communicating about this and I said it's going to be one of the most consequential acquisitions.
224
I don't even want to call it a transaction because I think it's more important than that.
225
Give us your first principles explanation of open source versus closed source versus open weights
226
and how the ecosystem should fit together over time.
227
The world needs both closed models and open models.
228
You want to use, I use as much closed models as I can.
229
This weekend I used four of them.
230
And they work terrifically.
231
They're frontier.
232
They're great experience.
233
They work incredibly well.
234
They're getting better all the time.
235
And the way I think about closed models is kind of like bottled water.
236
Water is free, you guys.
237
I don't know if I've told you guys, but water is free.
238
I don't want to burst everybody's bubble.
239
But water is free.
240
And this morning, I used a lot of free water taking a shower.
241
And so you use the right water in the right places.
242
And this is no different than electricity.
243
This is you know, there's no different than all kinds of commodities that we use in the world.
244
You need both Now in the case of open the reason why you need it is
245
because it could be for sovereignty reasons privacy reasons
246
Proprietary technology reasons look at the facts the facts are in the last six months
247
400 billion dollars of venture funding went into AI native companies 80 % of them use open models.
248
If not for open models, how could they build their dream?
249
Because their dream could be different.
250
Obviously, it would be different than the labs, the Frontier Labs dreams.
251
And America has so many different ways to innovate.
252
That's one of our core strengths.
253
Great ideas just coming out of the fountain.
254
And so open models enables that.
255
Open models enables every single, if we want to win the AI race, It's not about a few technology companies winning the AI race.
256
It's about every company in America.
257
Every company, every industry, every researcher, every teacher, every student, every startup.
258
Everybody wins.
259
Some of them will use closed models.
260
A lot of them will use open models.
261
There's 10 million.
262
Does it matter if the open models come from China or the U .S.?
263
Well, we're doing everything we can to make a contribution in open models.
264
However, the moment you download,
265
like, for example, probably the vast majority of the world's contribution to open source today is coming from China.
266
They just have a lot more engineers.
267
They produce everything in large scale because it's a larger country.
268
And so they produce science and math students in volume.
269
That's one of our disadvantages.
270
They're manufacturing them through amazing universities like Tsinghua University in high volume.
271
Well, they contribute open source today.
272
We download Linux, we download Kubernetes, we download all the software.
273
A lot of it has been touched by Chinese.
274
Once you download it, it's yours.
275
We fork it, we improve it, we make it ours.
276
When you download one of these Chinese models, it just happens to be made by some really great researchers in China, but it's now yours.
277
So what exactly is the race?
278
The race?
279
Yeah.
280
I think that's a really good point.
281
My point is the race is really about who exploits the technology best.
282
You know, the last industrial revolution, all of the inventors were Maxwell, Volta, Ampere.
283
None of them were American.
284
They were right the last industrial revolution came from Europe, but we exploited it We took advantage of it socially better than anybody else in the world
285
and look how it turned out for us I want to make sure that this next generation doors just like this.
286
Yeah So why are the communists getting their message out so successfully here right now?
287
You know, I think first of all, the narrative is much more practical.
288
The narrative is much more practical.
289
Nobody in China is saying that there's end of this and end of that and cataclysmic this and you know.
290
Doom or that.
291
Doom or that.
292
They're much more pragmatic about it.
293
They see AI as a technology that's going to advance their economy, advance their society.
294
And they don't have these groups who are basically saying it's going to end civilization.
295
And we're making it up.
296
The part that is frustrating is if it's true...
297
If it was true, then we ought to talk about it and go do something about it.
298
Even
299
if it's true, we ought to spend more time doing something about it than worrying a bunch of people who can't do anything about it.
300
It's our job to build it right.
301
Has there ever been a point in history where so many people have so vehemently said something that is so untrue?
302
And they're actually demonstrably untrue.
303
And it actually makes sense is untrue.
304
It's not based on science.
305
It's not based on research.
306
Everything that's based on science and research proves otherwise.
307
The fear of the frontier.
308
Humans have never been there.
309
We've never seen it.
310
Therefore, we're scared of it.
311
And therefore, it's easy to tell everyone to be scared of it.
312
It could be life experience as well, David.
313
So let me give you an example.
314
When I first graduated from school, I was an engineer and I didn't do that much typing.
315
And the reason for that is because I was the first generation before software became popular.
316
We had to go build the computers to make software pop up possible.
317
Could you imagine?
318
In this generation every single engineer who came into the world of engineering you spend all your time typing Literally, that's what you do
319
When you get a job they give you a laptop they give you a chair
320
and you start typing just you type all day long You type from the moment you wake up to the month.
321
Well, there was engineering before typing, right?
322
And so can you imagine
323
that the world has a mountain of engineering work to do where most of it is not typing anymore?
324
Sure.
325
We had busy engineers before typing.
326
I think we're going to do a lot of great engineering after typing.
327
When I say typing, I mean coding.
328
So even at NVIDIA, when software engineers talk to me, I tell them, you're just typing.
329
I've been saying that forever, but obviously for fun.
330
And I tell them, my favorite key is backspace.
331
And the reason for that is because the best software is the smallest software.
332
So I want you to use backspace as often.
333
Let's actually talk about NVIDIA.
334
Let's do a little teardown of NVIDIA.
335
So, teardown meaning just explain the pieces, because there's a lot of strategy at play.
336
Let's start at the absolute bottom.
337
Oh, no.
338
this is not planned
339
but we know who it is oh no no mr. president
340
oh yes sir I got to tell you something it I I
341
if it wasn't because of you calling I would I'm on stage with the besties I'm on stage with the besties.
342
I'm on stage with the besties.
343
I'm on stage with Sax.
344
Yeah.
345
You know, the whole group.
346
Yeah, Jason's here.
347
Chamat's here.
348
Dave and David's here.
349
Yeah, I'm sitting in front of a few thousand people.
350
And we're talking, as it turns out, we were talking about you.
351
Good job, sir.
352
Good job.
353
The fact that you saw through all of that, I mean, there's a lot of complexity and the fact
354
that a matter you saw through all of that and and I we're all just really grateful
355
tell him I said hi
356
do you want to say hi to the crowd Jason would
357
like Jason would like to put you on the even Jason
358
speaker mode how do we put how do we put on POTUS
359
and spit put them on speaker speaker yeah right into the microphone hold on sir We're getting a microphone.
360
Mr. President, you're now talking to the planet.
361
You see, the great thing about life is
362
that Jensen can develop the most complex computer chip in the world that nobody can copy for 10 years, but he can't figure out how to put me on speaker.
363
We have to remember this one.
364
So, the AI, it's almost this conspiracy, and the happiest group is China, and China is very happy.
365
And I could even say in the country, a lot of states are happy that weren't going to get anything
366
because they're being inundated by people that want to be there.
367
But now all of a sudden, you see they're building in Finland.
368
They want to build one.
369
Google wants to build a big one in Finland, which I'm not happy about because they weren't able to get permitting.
370
And I'm telling you, it's all a hoax.
371
The data centers are great and they make people wealthy
372
and they make states wealthy and it's the oil of the next 20, 25 years.
373
It's bigger than the Internet and the AI, you know, much more so.
374
And they're just playing right into the hands of a lot of people that don't want to see it happen, and that could be political people, it could also be China.
375
And we're not going to let that happen.
376
It's a hoax.
377
You're right, we're not going to let that happen, sir.
378
No, we're not going to let it happen.
379
The robots are not going to be taking over the world.
380
and that's not going to happen.
381
You know, my uncle was...
382
Hey, the top, probably maybe the best of all time, frankly, but professors at MIT for 41,
383
42 years and known as being one of the most brilliant men.
384
And he was he was there for 41 years as the top.
385
He was like at the top, top of the ladder, top of did many things.
386
Jensen knows all about it, but did many things.
387
So I have a little genetic, a little genetic strength, if you believe in the race theory
388
but i do i have genetics that explains why you know
389
so much about ai yeah well i know about ai i
390
know i also have common sense about ai uh the robots
391
will not be taken over uh the ai will not be
392
taken over the rest of the world the whole thing is a hoax now with
393
that we have to be a little bit careful we have
394
to very be you know we have to do things
395
and we have to do them prudently but
396
that doesn't mean we're going to stop an industry
397
because you know as we work on the next 10 years about how to destroy it
398
so i'm with you all the way i didn't even know
399
how you felt about it i assumed you felt the same way as me yes
400
and we if we're going to lead
401
and i have an expression it's whoever wins ai wins that's how big it is it's bigger than the internet
402
and whoever wins ai wins and we can't let this kind of stuff happen
403
and that includes very much includes data centers there are communities
404
that were dying that have data centers right now and now they're wealthy communities, really wealthy communities.
405
We're going to make sure that everybody wins in the AI race in America.
406
Every industry, every company, every state, every people.
407
Good.
408
Well, I feel strongly about it, and I have the position that can do something about it.
409
We're not going to let that stuff happen.
410
So I have no idea who's at the meeting.
411
I have no idea who the hell I'm talking to, but I'll see you. Did you hear that?
412
Did you hear that? thousands of people are clapping for you sir
413
but uh he's done an amazing job and david has done an amazing job
414
and good luck to everybody
415
and uh we're going to stay with the future the country
416
has never done better we have 20 trillion dollars of investment coming into the country
417
and that's as opposed to much less than one trillion under sleepy joe biden and that was for four years.
418
This is in one year.
419
So, you know, it's really the country has never seen anything like it, and we're going to keep it going.
420
And so thank you all very much.
421
Thank you, Mr. President.
422
I'll call you back later.
423
Thank you, Mr. President.
424
Thank you.
425
Well, that was unique.
426
I thought it was a bit.
427
Did you know that was happening?
428
I thought it was a bit.
429
You thought it was a high?
430
No, it was real.
431
I thought it was a bit at first, and I was like, put him on speakerphone.
432
Wow. And he calls you.
433
How do you think?
434
He calls you any hour at the night, right?
435
Well, we were in the Oval that time when he called you.
436
I was sleeping.
437
I was in Europe.
438
You were asleep, and he said, wake him up.
439
I felt so bad because he's like, who's coming to this dinner?
440
And we go through the list, and he's like, well, what about Jensen?
441
I said, no, sir, he's on vacation because he had to postpone this vacation for five years.
442
And he's like, get him on the phone.
443
What's vacation?
444
But why do you think he sees through the hoax?
445
It's this is the thing that is really quite an extraordinary thing.
446
It was it's polling minus 80.
447
Yeah.
448
So for anyone else that's sitting in the Oval Office, you're going to do what's popular.
449
You're representing the people.
450
This is what everyone wants.
451
They want to shut down the data centers, end AI.
452
It seems to be the popular thing in the moment.
453
But he says it's a hoax and he calls it.
454
How does he do that?
455
I got to tell you, I'm not sure.
456
And the reason for that is because a lot of people are falling for it.
457
And so the fact of the matter is it's complicated.
458
You know at first, I mean if you look at the story, if you look at the stories, it's all anchored on two things.
459
The first thing that it was anchored on was national security.
460
And recently that was all blown to bits.
461
And so that story is no longer anchored on national security.
462
Now it's anchored on safety.
463
Now if you want AI to be safe, the first thing is we need to make sure that the labs that are building it are in control.
464
That they're good tests for them.
465
if we would like to have third parties to to make sure
466
that a third party evaluate
467
if their party evaluators are available that's no different than financial control you guys know we have auditors
468
and the auditors are quite quite they don't have to be as expert as we are in our business
469
but they just have to ask the right questions
470
and I think I heard somebody say that it's good to have independent auditors
471
or evaluators but they just have to have multiple.
472
I agree with that too, just as there's multiple evaluated and auditors.
473
It makes sure that one company doesn't become you know pilled or somehow influenced for whatever reason.
474
And so you know there's a lot of different ways
475
that you could solve this and so that I think the number one thing is let's build the technology safely.
476
Let's make sure that the testing of it is safe.
477
And I recognize completely that what is being built is extraordinary.
478
But these are extraordinary companies, and we ought to hold them to extraordinary standards.
479
And they want to be.
480
I wanted to go back to open source for a second.
481
A year ago, we weren't taking it very seriously.
482
It was two years, 18 months behind.
483
The one thing that, you know, one of the, as you guys know, one of the challenges when you're on the call with President Trump is hard to say something.
484
I'm going to get in trouble for that.
485
I'm sure he's going to call me up on that. But anyhow, what I was going to tell him and all of you is that AI is creating an enormous number of jobs.
486
The thing that he wanted more than anything at the beginning of the administration and my first phone call with him, my first time I met him, is that he wants to create jobs in America,
487
he wants to re -industrialize the United States, he wants to make sure that the United States has the energy to support the next industrial revolution.
488
Without energy, there's no industrial growth.
489
And so he wants to make sure that there's energy growth, that there's job growth, that we're re -industrializing the supply chain.
490
Look at everything that we're doing right now.
491
All of it is happening right now as we speak.
492
We're creating more jobs than ever.
493
We're creating software jobs.
494
We were just talking about earlier 400 billion dollars of venture
495
financing went into the AI industry just recently yeah six months
496
well that's created a ton of jobs that's created a ton
497
of jobs it's created you know obviously enormous amount of demand for compute
498
which I'm happy about which is also
499
which is also creating a lot of demand for data centers
500
and we ought to talk about that I I think I was just, I was talking to Governor Abbott of Texas,
501
and he was, he wants to appeal to the industry to make sure
502
that we are empathetic to the small communities as we're building data centers all across America, just to be better listeners.
503
Let's actually talk about that for a second.
504
What's incredible about NVIDIA, if you break down the component parts,
505
is you've effectively had to become the bank of AI to get the ecosystem going.
506
And you've had to do it at all the levels.
507
You know, you just did this thing with Cloverleaf where you're doing Land Power Shell.
508
You did this great thing with BlackRock and Goldman and all these folks to essentially create the financing capability.
509
Walk us through your capital allocation strategy.
510
What has to happen to get a broader ecosystem of folks to be able to come in and underwrite this next phase?
511
Well, we're creating, as you guys know, this is a new industrial revolution and every aspect of it is true.
512
This new industry requires manufacturing.
513
Just as the electricity, internet,
514
and now AI, we power anything, we can find anything.
515
Now with AI we can ask and know anything.
516
Isn't that right?
517
And so that's our future.
518
We tap into the ether and we can ask it of anything we want and it could explain it to us.
519
Now, in order for that to happen, it's got to produce the intelligence.
520
And so that's a production process, which is the reason why this infrastructure has to get built.
521
But once you get the infrastructure built, the question is, what about all of the other layers across the United States?
522
This industry isn't just about the model.
523
It's not just about the chips.
524
It's mostly about the applications on top.
525
It's mostly about the infrastructure layer, or the data centers and all the infrastructure, the construction, the electricity, the power generation,
526
all of that is involved.
527
And so I look across the entire ecosystem and look for bottlenecks.
528
And if there are places where extraordinary companies are being built.
529
Constraints.
530
Constraints, extraordinary companies being built.
531
Maybe it's a supply chain that has to get scaled up so that when we're ready to deploy compute,
532
that they'll be ready for us land power shell.
533
And so this is no different than looking at the supply chain upstream.
534
You know, I probably think about the long -term supply chain more than most because our company is really large.
535
And in order for us to succeed, a whole bunch of companies has to support me.
536
You know, it's got a Corning has this, you know, Wendell Corning has to support me, Lumentum and, you know, TSMC, of course, and memory companies.
537
And so we started working with all of these companies long before the revolution
538
that the growth came so
539
that the growth could happen now i'm going now i'm doing
540
a downstream i have one cycle tends to be though
541
that the earnings over time over long stretches of time tends
542
to move up the stack right towards the application layer where
543
you can over earn for larger periods of time um i
544
mean you bought hugging face now you're sort of in the actively in the serving business i mean it seems pretty
545
natural that products like open router make a lot of sense it seems pretty
546
obvious that you know there are better versions of ways to build things like bedrock I'm sure you think about it
547
What's the natural conclusion?
548
Because it seems like the folks up here have no issue trying to move down.
549
And you have the best balance sheet, these incredible engineers, and you have the proven experience to make it right and engineer the product and get it out.
550
So how do you think about looking up and saying, I could probably do that.
551
The reason why Nvidia runs every single model in the world, we were the only, it's incredible.
552
Last year, about a year and a half ago, the only thing we ran was OpenAI.
553
Yeah, and now look at amazing models are available the metamuse is available You got grok is available grok bots incredible
554
We now run Gemini and
555
Anthropica is scaling up on our platform as well
556
Since a year and a half ago you got all these frontier AI models
557
that are now open that are available So the number of models that are growing
558
There's a whole bunch of companies that I won't mention
559
that are building frontier models as well and the number of AI labs are growing.
560
The ineffables, the reflections, the list goes on.
561
The physical intelligence, the list goes on.
562
And so all of these labs are building on NVIDIA.
563
And the reason for that is because as a company, I'd rather for us to help everybody succeed instead of taking a slice out.
564
And so we would go up as far as we need to, but as low as possible.
565
Our strategy is go up as far as we need to and as low as possible.
566
And the reason for that is because if I do that, if I solve the, if not for NVIDIA creating QDNN, all of the frameworks wouldn't exist.
567
If not for us creating Megatron core, then all of the large -scale training wouldn't have happened.
568
So we go and we invent all the technology necessary as far as we need to, and then we let a thousand flowers bloom.
569
And so that posture allows us to be, quite frankly, the only...
570
Well, look, let's be honest.
571
I agree with you.
572
The pushback would be that it really would be great to have more competition at the hyperscale layer.
573
And I think you've done a great job supporting the neoclouds.
574
There are some, and by the way, I think you introduced me to Nambias.
575
Superb.
576
Great.
577
Everything, they're amazing.
578
But we need like 50 of these guys.
579
We need 100 of them.
580
We need 1 ,000 of them.
581
And it just may take some...
582
Yeah.
583
You know, Shemar, it's just I'm surprisingly uncompetitive.
584
Really?
585
Yeah.
586
That's not my thing.
587
You know, my thing is kind of like, for example, I'd be more than happy with five hyperscalers.
588
However, the reason I noticed the early customers of all the neoclouds, all the what we call NCPs, all the early customers were the hyperscalers.
589
Exactly.
590
And the reason for that is because the hyperscalers plan once a year, but the market dynamics is so volatile right now that they're always almost wrong.
591
And so with all these regional clouds who are agile, they can move fast.
592
They know their state or they know their country.
593
They know their region.
594
They're securing land, power, and shell in a way that's hard for somebody who sits in Seattle
595
or sits in Palo Alto to be able to see the planet.
596
And so we now have basically a large -scale distributed network of companies that are building, securing land power shelf for us.
597
And now countries realize it's strategic.
598
So many countries are saying, I'm going to take my power and only give it to my own companies.
599
Well, NVIDIA is in that country as well, and we could help the NeoClouds in that country grow.
600
And so whether it's Firmus in Australia, we just did a whole bunch of stuff in Australia, brought on two more gigabytes.
601
Southeast Asia, of course, IOH and others bring on a few gigabytes.
602
And so we're building, gigawatts.
603
So we're building, you know, we're scaling up, you know.
604
It's pretty clear though.
605
I just want to get this one thing in.
606
It's pretty clear that you're going pretty high up and getting very focused on open source.
607
Obviously, you have your Nemotrons doing exceptionally well.
608
I use them often.
609
Hugging Face, Poolside and Laguna.
610
Very, very solid product that you're now acqui -hiring, hiring, whatever it is.
611
And then you have your open source stack for self -driving, also very disruptive.
612
We are the frontier model in five domains.
613
Yeah.
614
And so you don't seem to build products to get the silver medal.
615
You seem to go for the gold.
616
So are you going for the gold, and will you have the best hands -down open source model?
617
And then part B to that is, can open source catch up to Frontier Models, and are you the person to do it?
618
So the logic, Jason, is that we will build it because, one, we can, we have the skills to do it,
619
and because our customers need us to do it.
620
Right.
621
So for example, Alpamayo is the world's first thinking self -driving car.
622
And by thinking, by reasoning, you don't need as much data as, you know, you don't have to train on a few billion hours of road data.
623
Because you can reason about it.
624
Break down the problem into, oh, I've seen this before.
625
It's not exactly the same, but it's largely the same as that.
626
Okay.
627
And so Alpamayo, why is it necessary?
628
Well, there's a whole bunch of car companies.
629
Every car in the world is going to be autonomous.
630
But beyond that, every ag tech, every truck, every van, and most of them aren't big enough in scale to be able to build that whole stack.
631
So I'll build an extraordinary stack for them.
632
They do last mile adapting for their application.
633
Now everything that moves in the future could be autonomous.
634
If not for us building some of the biology models, the world wouldn't have it the ESM2 protein found language model, we created that.
635
ESM fold, open fold, alpha fold 2, all the stuff with QEquivariant, all of that stuff, technology, wouldn't have existed if we didn't build it.
636
One of my favorites, Proteina Complexa, is synthesizing next generation proteins and it's binding.
637
It's groundbreaking stuff.
638
We built that.
639
And so we'll build that because Lily needs it
640
and Mark needs it and others need it and they don't have the capability to do it or they're not yet there.
641
And so we can make a real contribution.
642
So I do everything out of need.
643
I'm not trying to disrupt.
644
I mean, we don't wake up in the morning trying to disrupt anybody.
645
We just wake up in the morning, try to help everybody.
646
Jensen, what about competitive threats that might be emerging to your core business?
647
Can you just comment?
648
We're just so nice.
649
Yes.
650
Well, I know this is...
651
Well, I actually want to just get your...
652
Let's just call it a take.
653
What's your take on TerraFab?
654
100 million square foot facility Elon's announced.
655
If anybody could do it, he can.
656
And the two of us were on a flight together to a country.
657
With a person who sometimes calls you on the phone.
658
It was a nice plane.
659
And we had like, you know, Elon likes to talk about these things.
660
And so we spent a lot of time talking about it. Is that even competitive?
661
Anybody could do it, he could do it.
662
Because you design chips, you don't fab them.
663
Could your chips be fab there or is it?
664
Well, we know a lot about process technology because we're pushing the limits of everything.
665
And because we scale at such large scale, we have incredible memory technology inside the company.
666
We're the world's best 30s company.
667
We got lots of amazing tech.
668
So your take is you've talked a lot about it?
669
So we could just, yeah, we could talk about it.
670
And you can't discourage Elon from doing it, which is one of his super power.
671
And once he decides to go do something, it's hard to stop him.
672
And so I - Can you give us your take on where China is with advanced lithography systems, native grown?
673
They're gonna get there by 2030. By 2030.
674
Yeah, and 2030 is just around the corner.
675
Yeah, also that's - Sorry, how long will all be dead at that time, so.
676
And for China, does that mean the switch is flipped and then that's all gonna go into mainland fabs almost immediately?
677
You know, the way to think about China is really good at high volume production.
678
And this is just a matter of time.
679
And, you know, I think in decades as well.
680
You know, I've been around a long time.
681
And, you know, for NVIDIA, I've got to think about what happens next decade and decade after that.
682
So two or three years, it's just a click.
683
It's nothing.
684
And so as far as they're concerned, they're already there.
685
They're already there.
686
Yeah.
687
Jensen, Elon, and Gwen.
688
We've got to run.
689
America, we've got to run.
690
Speed run.
691
We've got to.
692
Speed run.
693
Slowing down is definitely the wrong strategy.
694
I mean, it feels apparent, I think, to most of us in the industry that we're kind of in the AGI moment, and it's a definition, obviously, just as smart as any other human.
695
I think we're already there.
696
We're there, right?
697
Yeah, yeah.
698
And so then super intelligence is the next waypoint.
699
Based on what you see, based on your customer base, based on your history here.
700
And Jason, I think we're there too.
701
You think we're at super intelligent?
702
Yeah, yeah.
703
When you take a narrow segment, a narrow segment, I mean, my self -driving car, I don't want you to make me an omelet.
704
I just want you to drive the car.
705
Right.
706
That is super intelligent.
707
Super intelligent.
708
And it's better than a human.
709
Yeah, yeah.
710
Like one -tenth of the accident rate.
711
Exactly.
712
Synthesizing proteins, doing virtual screening of proteins.
713
We're already there.
714
Yeah.
715
Yeah, yeah.
716
Are you having fun being on the frontier of humanity?
717
I like it.
718
Yeah.
719
Ladies and gentlemen.
720
I like it.
721
I like it.
722
And guys, it's great there.
723
The future is great, and we want to get there.
724
Listen, a lot of us don't have to work, but I got to tell you, it's too good not to be. So fun.
725
Right?
726
And so I want to be there.
727
I want all of you guys there with me.
728
We're all going to be there.
729
We're going to be enormously successful together as a humanity.
730
And in the meantime, we got to encourage them, urge them on.
731
They're doing really, really important work, as you guys know, and I want them to succeed.
732
I also would love for us to tone down the the the drama and
733
Make and most importantly we need all of America to come with us.
734
That's how we make it Yeah, ladies and gentlemen, Jensen long
הורד אפליקציה
כל מה שאתה צריך כדי לדבר בשטף
הגיית AIציון לכל משפט
תרגול IPAשלוט בכל צליל
אוצר מיליםבנה את מאגר המילים שלך
משחק אוצר מיליםלמד תוך כדי משחק
אותו ערוץ
C11
實戰練英文 | NVIDIA 黃仁勳榮獲2025霍金獎，分享AI革命與創業心路；AI將改變所有的工作，讓你重回工作重點；想成功，一定要告訴自已這句話 | 劍橋大學演講 | 長文、註解、大字版、中英對照
Daily ConversationUniversity Students
C12
實戰練英文｜黃仁勳：你們要加油，不然我會拿走所有機會，然後我會變很有錢｜德國沒有下一個機會了｜投資德國高峰會｜長文、註解、大字版、中英對照
Daily ConversationUniversity StudentsWorking Adults
C11
實戰練英文 | SpaceX 憑什麼成為史上最大 IPO ？AI 上太空的困難點都能解決嗎？馬斯克的計劃是畫餅還是老神在在？專訪 Elon Musk 詳細說明｜完整中英對照、長文註解
IELTS PrepBusiness EnglishUniversity Students
FOR YOU
המלץ על סרטון
C11
83 משפטים
Leading AI CEOs Warn UN Security Council of Existential Risks | Sam Altman On AI Controls | N18G
Business EnglishDaily ConversationUniversity Students
C11
207 משפטים
Go’s Big Mistake
Daily ConversationUniversity StudentsWorking Adults
B25
79 משפטים
Trump abruptly storms out of NBC interview over rigged election claims | 9 News Australia
Business EnglishDaily ConversationUniversity Students
C13
933 משפטים
My Joe Rogan Appearance: What I Learned (ft. Young Jamie & Viktor Hovland)
Daily ConversationUniversity StudentsWorking Adults
B24
68 משפטים
What Happens If China Invades Taiwan? Elon Musk's AI Warning
Business EnglishDaily ConversationUniversity Students
C14
76 משפטים
The Limitless AI Lie. The Bubble Is Slowly BURSTING.
Business EnglishDaily ConversationUniversity Students
C113
55 משפטים
"Focusing is about saying no" - Steve Jobs (WWDC'97)
Daily ConversationUniversity StudentsWorking Adults
C15
237 משפטים
Khi kiến thức "bão hòa", đâu là năng lực đắt giá nhất? | đi-code ep 30
Business EnglishDaily ConversationUniversity Students
C11
358 משפטים
It’s not just you
Daily ConversationUniversity StudentsWorking Adults
![Understand AI in 14 minutes – with Anthropic's Chloe Lubinski [ARC 2026]](https://img.youtube.com/vi/aBUniZHgCnE/mqdefault.jpg)
C14
179 משפטים
Understand AI in 14 minutes – with Anthropic's Chloe Lubinski [ARC 2026]
Business EnglishDaily ConversationUniversity Students
B23
377 משפטים
Extended interview: Jamie Dimon on AI, Iran and more
Business EnglishDaily ConversationUniversity Students
אוצר מילים והערות הגייה לשיעור הזה
שיעור הדיבור הזה ברמה C1 מבוסס על הסרטון “實戰練英文｜再不停，人類要滅了？！黃仁勳痛批 AI 末世論，川普突然 CALL-IN 亂入，大罵這個陰謀只會讓美國輸！｜Nvidia、All-In Summit｜長文、註解、大字版、中英對照”. בסרטון הזה יש 734 משפטים ו-7885 מילים לתרגול Shadowing. הדיבור נמשך 45:35. הדובר מדבר בקצב טבעי של כ-173 מילים בדקה, קרוב לשיחה יומיומית. 87% מהמילים נמצאות בין 3,000 המילים הנפוצות ביותר באנגלית; כדאי ללמוד את השאר לפני שמתחילים.
אוצר מילים מרכזי בסרטון הזה
14 המילים המתקדמות ביותר בסרטון, עם ההגייה והמשמעות שלהן:
| מילה | הגייה | משמעות |
|---|---|---|
| prediction שם עצם | /pɹɪˈdɪkʃən/ | תַּחֲזִית |
| anchor שם עצם | /ˈæŋ.kəː/ | עוגן \ עֹגֶן |
| stack שם עצם | /stæk/ | ערמה |
| incentive שם עצם | /ɪnˈsɛntɪv/ | תמריץ |
| hug פועל | /hʌɡ/ | חיבק \ חִבֵּק, חָבַק |
| gigabyte שם עצם | /ˈɡɪɡəbaɪt/ | גיגבייט |
| microphone שם עצם | /ˈmaɪ.kɹəˌfəʊn/ | מִיקְרוֹפוֹן |
| doom שם עצם | /duːm/ | דִּין |
| startup שם עצם | /ˈstɑɹtˌʌp/ | סטארטאפ, חברת סטארטאפ |
| discourse שם עצם | /ˈdɪs.kɔːɹs/ | שִׂיחַ |
| autonomous שם תואר | /ɔˈtɑnəməs/ | אוֹטוֹנוֹמִי |
| educate פועל | /ˈɛd͡ʒ.əˌkeɪt/ | חִינֵּךְ |
| distribute פועל | /dɪˈstɹɪbjuːt/ | חילק |
| emerge פועל | /ɪˈmɝd͡ʒ/ | צץ, התגלה |
Phrasal verbs שתשמעו
| מילה | משמעות |
|---|---|
| go up פועל | עָלָה |
| figure out פועל | הֵבִין, מצא תשובה |
משפטים שכדאי לחזור עליהם
משפטים קצרים ושלמים מהסרטון שאפשר להשתמש בהם בשיחה יומיומית:
- Let's see, what else has proven to be wrong?
- Why don't we get measurement right?
- Why don't we get standardization right?
- You have to test it again, don't you?
- You think we're at super intelligent?
דקדוק בסרטון הזה
המבנים שהדובר משתמש בהם הכי הרבה, עם המילים המדויקות מהסרטון:
| מבנה | בסרטון |
|---|---|
| הווה מושלם (Present perfect) have/has + past participle — פעולה שקרתה בעבר ועדיין יש לה חשיבות או השפעה עכשיו | has proven · has taken · has turned |
| סביל (Passive voice) be + past participle — הדגש הוא על מה שקורה, לא על מי שעושה את הפעולה | is not planned · was expressed · be done |
| משפטי זיקה (Relative clauses) who / which + clause — מידע נוסף על אדם או דבר | difficult which is · reading, which is · people that were |
הגייה ששווה לשים לב אליה
הדובר משתמש ב-211 קיצורים וצורות מקוצרות, כמו we're, they're, don't. אמרו אותם בצורה הקצרה, כפי שאתם שומעים אותם.
- הצלילים של “th”: synthesize /ˈsɪnθəsaɪz/, synthetic /sɪnˈθɛtɪk/
- הצלילים של “sh” ו-“zh”: prediction /pɹɪˈdɪkʃən/, consequential /ˌkɑnsɪˈkwɛnʃəl/
- מילים ארוכות — שימו לב להטעמה: ecosystem /ˈikoʊˌsɪstəm/, consequential /ˌkɑnsɪˈkwɛnʃəl/, radiologist /ˌɹeɪ.diːˈɒl.ə.d͡ʒɪst/, radiology /ˌɹeɪdiˈɑlədʒi/, reinforcement /ˌɹiːɪnˈfɔːsmənt/
צלילים שדוברי עברית מתקשים בהם:
- /æ/ — פתוח יותר מה-“אֶ” העברית, לא כמו “a” ב-“cup”: anchor /ˈæŋ.kəː/, stack /stæk/, backspace /ˈbækˌspeɪs/, unpack /ʌnˈpæk/, enhance /ɪnˈhæns/
- /r/ — הלשון לא נוגעת בשום מקום, לא ה-“ר” הגרונית: frontier /ˈfɹʌntɪə/, prediction /pɹɪˈdɪkʃən/, researcher /ˈɹiˌsɝ.t͡ʃɚ/, untrue /ʌnˈtɹuː/, constraint /kənˈstɹeɪnt/
- עיצור קולי בסוף מילה — /b/, /d/, /g/, /z/, /v/ בלי להפוך אותם לאטומים: incentive /ɪnˈsɛntɪv/, fold /ˈfəʊld/, hug /hʌɡ/, synthesize /ˈsɪnθəsaɪz/, bind /ˈbaɪ̯nd/
איך לתרגל עם הסרטון הזה
- הקשיבו לכל הסרטון פעם אחת בלי לדבר, ורשמו את המילים שאתם לא מכירים.
- התחילו במהירות 0.75×, תרגלו Shadowing משפט אחר משפט, וחזרו למהירות רגילה כשזה מרגיש קל.
- הקליטו את עצמכם והשוו למקור, תוך תשומת לב למילים כמו prediction, anchor, stack.
מהי טכניקת השאדואינג?
שאדואינג היא טכניקת למידת שפה מבוססת מדע שפותחה במקור להכשרת מתורגמנים מקצועיים ופופולרית על ידי הפוליגלוט ד"ר אלכסנדר ארגואלס. השיטה פשוטה אך עוצמתית: אתה מקשיב לאודיו באנגלית ילידית וחוזר עליו בקול רם מיד — כמו צל העוקב אחרי הדובר עם עיכוב של 1–2 שניות בלבד. בניגוד להאזנה פסיבית או תרגילי דקדוק, שאדואינג מכריח את המוח והשרירים בפה לעבד ולשחזר דפוסי דיבור אמיתיים בו זמנית. מחקרים מראים שזה משפר משמעותית את דיוק ההגייה, האינטונציה, הקצב, הדיבור המחובר, הבנת השמיעה ושטף הדיבור — מה שהופך אותו לאחת השיטות היעילות ביותר להכנה לIELTS Speaking ולתקשורת באנגלית בעולם האמיתי.