Pratique du Shadowing: Elon Musk Explains How the AI Bubble Will Burst. - Apprendre l'anglais à l'oral avec la vidéo
Choisissez un mode d'entraînement
Chaque mode ouvre une salle d'entraînement en plein écran.
196 phrases
Si les phrases sont trop courtes ou trop longues, cliquez sur Edit pour les ajuster.
1
Are you worried about the Chinese catching up and getting ahead?
2
For digital intelligence, it comes down to how many chips do you have and how much power do you have?
3
China has more electricity than the United States, Europe and India combined.
4
And at some point, they probably will have a lot of compute.
5
Ergo, they will be the leaders.
6
There's a good chance that they are the leaders at some point.
7
Well, I don't know if you guys saw this one recently, but Elon Musk has just done a very spicy interview with The Economist.
8
He was fired up.
9
I would like to just admonish you and the media for the absurd characterisation of the far right.
10
Um...
11
Zero.
12
People.
13
I hear you loud and clear.
14
Zero.
15
Do you realise the media is despised?
16
It is false and misleading and nonsense.
17
Please keep this part in.
18
Yeah, he was ready to rant, that's for sure.
19
But earlier in the interview, one of the really big talking points that they covered was, of course, AI.
20
Of course it's AI, it's everywhere, it's driving my car, it's writing my emails, it's even telling me how many calories are in a glass of milk.
21
I really did look that one up.
22
But more importantly than all of that is that AI is powering a really high percentage of the stock market's returns.
23
And thanks to these seven companies, it's made the stock market look like this, with a valuation like this.
24
And that's why I found Elon's thoughts in this interview really interesting
25
because it kind of lines up exactly with what Steve Eisman has been saying recently too.
26
A guy that's pretty good at predicting stock market bubbles.
27
So in this video I want to take you through the three elements to the problem
28
that Elon discusses in this interview and what we should be looking out for as investors.
29
The first problem is arguably the biggest structural problem of this AI boom in America one that's really, really difficult to fix, and that is power.
30
But before we get into where the power is actually running out, listen to how Elon frames this whole AI race.
31
Because to be honest, the rest of this video is kind of centered around this one sentence.
32
You can think of like the constraints on AI, essentially electricity and AI chips.
33
And really, it's just those two inputs, power and chips.
34
But the scary thing is that for one of them, electricity, Elon says the US just isn't competitive in the long term and that's a problem.
35
And if you look at electricity production AI is a function of whatever the limiting factor is
36
if it's chips
37
or electricity China has far more electricity than the United States
38
in fact China has more electricity than the United States Europe
39
and India combined already my guess is China gets to four times the electricity production of the US
40
which is roughly proportionate to the population.
41
Now, in the long run, this is where AI's true bottlenecks starts to appear.
42
We know that China already outproduces the US in raw generation, but check out this Bloomberg article.
43
They note that China added the equivalent of 40% of the entire US grid's capacity in a single year
44
and has built more generation capacity in the last four years than the whole US grid combined.
45
The power gap is not closing.
46
In fact, by almost all projections, it is widening.
47
China is getting further ahead.
48
And this is a massive problem because AI eats electricity for breakfast.
49
We need stacks of it for the future of this technology.
50
And that raises the obvious question, which is why can't the US just build more power?
51
To be fair, the US certainly isn't ignoring its problems they are trying to build out the power infrastructure
52
and they recognize these power-hungry data centers are becoming a problem.
53
For example in February Trump announced the rate payer protection pledge
54
basically telling tech companies build your own power for your data centers
55
but even with this push the annoying thing is getting the new power generation needed up and online.
56
As of this year, there's roughly 2,000 gigawatts of proposed power generation sitting in US interconnection queues,
57
and the median wait to actually get switched on is roughly five years.
58
In Texas alone, ERCOT is managing a 474 gigawatt queue of large loads, and data centers make up 90% of it.
59
Some projects are facing waits of up to 12 years.
60
So the reality years, getting the new power generation online is a slow slog.
61
And this is why so many of these AI hyperscalers are turning to these things, these backup gas turbines.
62
I spoke about this in a video earlier this year, but even Elon can't wait for the grid.
63
He bought like 35 of these gas turbines for XAI's Colossus supercomputer in Memphis.
64
With these turbines, normally you'd find like one or two of them as backup generators for like a hospital.
65
But XAI had literally 35 of them humming as the data center's primary energy source.
66
So even Elon brought his own portable power plant rather than waiting for the grid.
67
That should tell you everything you need to know about how stretched the system is.
68
For digital intelligence, it comes down to how much AI compute do you have?
69
Well, how many chips do you have?
70
And how much power do you have?
71
How much electricity do you have?
72
The constraint right now, I think, is actually slightly more on power and cooling than it is on AI chips
73
because the power demands of the AI chips are very, very high.
74
We'll talk about water demands, but really the water usage of AI is negligible.
75
The electrical demand is very high.
76
And it's not just Elon either.
77
Everyone's trying to get these gas turbines now.
78
So much so, it's broken the supply chain.
79
Wait times for large turbines have stretched to five years.
80
GE Vinova's backlog is up to 116 gigawatts.
81
Siemens Energy's backlog sits around 69 gigawatts.
82
Turbine prices for plants coming online in 2030, 2031 are skyrocketing.
83
They're up 75% percent.
84
I mean, talk about a picks and shovels play.
85
Overall, really, it's GE Vinova, Siemens Energy, and Mitsubishi Heavy.
86
They are the three big dogs selling these turbines.
87
And you can see on InvestingPro, despite by normal metrics, these companies are all sitting reasonably overvalued.
88
The analysts are still seeing them as fair or undervalued because of that AI runway ahead.
89
Energy infrastructure is in hot demand.
90
When you ask Warren AI what's going on, you can see it's just two factors.
91
The data center boom has begun and thus gas turbines, which are the bridge, supply dispatchable power as renewables and the grid scales up.
92
And because of this trend, GE Venover is up 68% in the past year.
93
Siemens Energy is up 65%.
94
And to follow the energy infrastructure theme even more closely, InvestingPro also has ProPix AI, which is a model built on over 50 financial signals that surfaces those outperformers.
95
For example, in this tab, we can see a strategy called Energy Elite.
96
So if you're looking for ideas on this specific theme, being the energy infrastructure boom.
97
You can look here
98
and see how some of those businesses have compared to say market metrics like the S&P 500 energy index.
99
So it's a really fun feature.
100
I love exploring it for ideas.
101
Now I do want to mention the great thing right now is the investing.com summer sale is currently live.
102
That means you can get 55% off plus an extra 15% with my referral link in the description
103
and pinned comments so definitely check him out now is a great time
104
and thanks always to investing.com for sponsoring these videos
105
but back to it
106
so that's the power story that's elon's first point in the ai bubble debate the us is failing
107
and china is winning and that brings us right to elon's second argument
108
which is good old chip starved china and since the us has banned the export of the latest ai chips to China.
109
China is somewhat chip starved.
110
So China is chip starved but that might actually be a blessing in disguise for them.
111
So for those that don't know
112
because Nvidia makes the best AI chips in the world
113
and the US controls what Nvidia is allowed to sell to
114
China we've seen restrictions come into place banning the export of the most powerful chips to China.
115
Now the rules have actually loosened a little bit this year.
116
Trump approved limited H200 sales to China in December That started in small amounts in February, but there was a catch.
117
The US is imposing a 25% tariff on these chips destined for China So it's also worth noting Nvidia's newest chips,
118
which is Blackwell and the upcoming Rubin architecture They're still completely off limits to China
119
But here's why this is actually a problem for America
120
because the US has applied these constraints to China It has actually spurred China on to do their own thing
121
and now they are more competitive than they've ever been Now China's homegrown deep ultraviolet lithography machines,
122
the machines that print incredibly tiny circuit patterns onto the silicon wafers, they are still a fair way behind ASML, which is the gold standard,
123
particularly when it comes to performance and reliability, but SMIC, which is essentially China's answer to TSMC, they have found ways to work around some limitations.
124
So they've already demonstrated 7nm production using older DUV technology
125
and it's now starting to get there with 5nm production too
126
and right across China there's a massive effort underway to close this technology gap So
127
while the export restrictions have undoubtedly hurt China in the short
128
term they've also given the country an enormous incentive to build its own semiconductor ecosystem You know,
129
maybe China is chip starved today but in a few years from now
130
That could be a different story and that brings us to Elon's third
131
and final argument as to why the AI narrative in the US might be getting a little bit stretched
132
because despite China being chip starved right now and potentially remaining
133
that way for years to come it's pumping out AI models that are already really good and really cheap.
134
How do you think about the Chinese models and the geopolitics of it?
135
Are you worried about the Chinese catching up and getting ahead?
136
Given that the Chinese AI companies are doing as well as they are with a relatively small amount of compute, it seems that if they had a lot of compute,
137
there's a good chance that they would be the leaders.
138
And at some point, they probably will have a lot of compute.
139
So they will, ergo, they will be the leaders.
140
There's a good chance that they are the leaders at some point.
141
Listen to that.
142
Elon is confident it's only a matter of time before China is the world leader in AI.
143
And the reason for it is despite China's chip-starved AI ecosystem, they're still producing models that are very, very good.
144
I think realistically, Fable is still clearly the smartest model.
145
I think anyone realistically would say that that's still the case.
146
But the Kimi model is getting quite close to Fable.
147
So even Elon, who has every incentive to talk his own book, is saying the gap between the best US model and China's best is closing.
148
And this is exactly where what I was talking about in my Steve Eisman video comes back in
149
because Steve isn't so much worried about the technical capabilities of these Chinese models like Elon is,
150
he's worried about what happens to US LLM pricing once they are competitive.
151
AKA he's predicting a price war that will squeeze OpenAI and Anthropic.
152
And now we actually have numbers to prove this point.
153
This is from AI research firm artificial analysis.
154
From their testing, they found that DeepSeek's v4 Flash cost $0.03 per test, compared to $0.86 for Moonshot AI's Kimi K3,
155
versus $1.86 for OpenAI's ChatGPT 5.6 Sol, and $3.15 for Claude's Fable 5.
156
Now, Steve Eisman estimates OpenAI and Anthropic together account for roughly 70% of all AI-related revenue flowing into the big hyperscalers,
157
Microsoft, Amazon, Google.
158
So take the time to think about it.
159
If a cheaper, good enough Chinese model starts pulling customers away on price, that's really bad for OpenAI and Anthropics margins.
160
And if that means OpenAI
161
and Anthropics have to pull back on the hundreds of billions in future spending that they've already committed to the hyperscalers, spending that has been baked into share prices,
162
well, that could just be a catalyst for an AI correction.
163
So let's step back, let's connect the dots.
164
There's been a lot in this video, power, chips, pricing models, etc. But really, Elon's talking about three elements in one system.
165
Right now, America has the chips, but it's running out of power.
166
China has the power, but it's short on chips.
167
That's the current standoff.
168
But the problem for America is the future.
169
China's chip shortage is a solvable in years problem.
170
You know, lithography, yields, manufacturing.
171
However, America's power shortage is more of a solvable in decades problem, because he can't just build a gas turbine or a transmission line
172
or a new power plant on just a snappy short timeline.
173
So you get a power bottleneck in the US, plus China solving its chip problem, plus the models being good for a fraction of the price.
174
That is a recipe for hurt, particularly for OpenAI and Anthropic.
175
And if they can't spend the hundreds of billions they've committed to spend at the hyperscalers, then the hyperscalers suffer and the whole AI narrative starts to crack.
176
that's the story
177
and interestingly that's also why Elon is trying to build his
178
data centers in space solar is much more effective in space plus Elon's also got pre-established launch capacity through SpaceX
179
so might not be as implausible as it sounds but with
180
that said they are Elon Musk's thoughts on the AI bubble
181
that we're in I hope
182
that all made sense I hope you enjoyed it big thanks
183
again to investing.com for sponsoring remember to check out my link in the description for an additional 15% off.
184
Get in now to take advantage of their summer sale to 55% off, but that is closing in just a few days.
185
Also wanted to mention
186
that spaces are now filling up for the live Rule 1 New Money Workshop happening in Atlanta in September.
187
It's the 18th to 20th.
188
Come along for a full three-day education on the Warren Buffett strategy, including how to find great businesses,
189
how to value companies, and also how to use options strategies to squeeze more juice out of every investment you make.
190
I think you guys will really, really like learning about that.
191
I'll be there.
192
Phil will obviously be there.
193
The entire team of Rule 1 coaches will be there.
194
If that sounds like something you'd be interested in doing, please come and join us.
195
The ticket is free, so get yours today.
196
Apart from that, guys, thanks for watching, and I'll see you in the next one.
Télécharger l'application
Tout pour parler couramment
Prononciation IAChaque phrase notée
Phonétique (IPA)Maîtrisez chaque son
VocabulaireEnrichissez votre lexique
Jeu de vocabulaireApprenez en jouant
FOR YOU
Vidéo recommandée
C14
228 phrases
The AI Bubble… We Need to Talk
Business EnglishUniversity StudentsWorking Adults
C11
386 phrases
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
C13
179 phrases
Why China Will be the Big Winner of the 2020 Crisis
IELTS PrepBusiness EnglishUniversity Students
C19
171 phrases
ENGLISH SPEECH | JD VANCE: The Future of AI (English Subtitles)
IELTS PrepBusiness EnglishUniversity Students
C11
42 phrases
What 50% Unemployment Will Actually Do to the Economy
IELTS PrepBusiness EnglishUniversity Students
B24
68 phrases
What Happens If China Invades Taiwan? Elon Musk's AI Warning
Business EnglishDaily ConversationUniversity Students
C110
119 phrases
Why US Stocks May ‘Grind Higher’
Business EnglishUniversity StudentsWorking Adults
C11
394 phrases
Chip War 2.0: The Global Battle for Semiconductor Supremacy│Chris Miller (The Author of “Chip War")
IELTS PrepBusiness EnglishUniversity Students
C143
6 phrases
Ray Dalio on the AI Bubble
Business EnglishUniversity StudentsWorking Adults
C12
276 phrases
Learn English with the JOE ROGAN PODCAST — Elon Musk
Business EnglishUniversity StudentsWorking Adults
C13
176 phrases
Is Amazon Too Big?
Business EnglishDaily ConversationUniversity Students
C14
18 phrases
AI ABUNDANCE IS COMING | Elon Musk
IELTS PrepTOEFL PrepUniversity Students
Vocabulaire et conseils d’expression pour cette leçon
Cette vidéo contient 196 phrases et 2546 mots à répéter en shadowing. La partie parlée dure 13:38. Le locuteur parle vite, environ 187 mots par minute : attendez-vous à des liaisons et à des sons réduits. Seuls 81 % des mots font partie des 3 000 mots les plus courants en anglais, le vocabulaire est donc exigeant.
Vocabulaire clé de cette vidéo
15 mots de la vidéo à retenir, avec leur prononciation et leur sens :
| Mot | Prononciation | Sens |
|---|---|---|
| chip nom | /t͡ʃɪp/ | brisure, fragment |
| turbine nom | /ˈtɜː.baɪn/ | turbine |
| starve verbe | /stɑɹv/ | mourir de faim |
| infrastructure nom | /ˈɪnfɹəˌstɹʌkt͡ʃə/ | infrastructure |
| roughly adverbe | /ˈɹʌf.li/ | rudement |
| grid nom | /ɡɹɪd/ | grille |
| stretch verbe | /stɹɛt͡ʃ/ | étirer, étendre |
| invest verbe | /ɪnˈvɛst/ | investir, consacrer |
| export nom | /ˈɛks.pɔɹt/ | exportation |
| bubble nom | /ˈbʌbəl/ | bulle |
| combine verbe | /kəmˈbaɪn/ | combiner |
| essentially adverbe | /ɪˈsɛnʃəli/ | essentiellement, en essence |
| narrative nom | /ˈnæɹ.ə.tɪv/ | récit, narration |
| commit verbe | /kəˈmɪt/ | confier, commettre |
| backup nom | /ˈbækˌʌp/ | remplaçant |
Les verbes à particule que vous entendrez
| Mot | Sens |
|---|---|
| check out verbe | mater, regarder |
| get off verbe | descendre, faire descendre |
| run out verbe | épuiser, être à court |
La grammaire de cette vidéo
Les structures que le locuteur utilise le plus, avec les mots exacts de la vidéo :
| Structure | Dans la vidéo |
|---|---|
| Present perfect have/has + participe passé — une action passée qui compte encore maintenant | has just done · has built · have stretched |
| Voix passive be + participe passé — l’accent est mis sur ce qui arrive, pas sur qui le fait | was fired · is despised · is allowed |
| Propositions relatives who / which + proposition — une précision sur une personne ou une chose | question, which is · turbines, which are · ASML, which is |
Prononciation à surveiller
Le locuteur utilise 25 contractions et formes réduites, comme they've, can't, isn't. Prononcez-les sous leur forme courte, telles que vous les entendez.
- Les sons « sh » et « zh »: essentially /ɪˈsɛnʃəli/, shortage /ˈʃɔː(ɹ)tɪd͡ʒ/, restriction /ɹɪˈstɹɪkʃən/, potentially /pəˈtɛnʃ(ə)li/, transmission /tɹænsˈmɪʃən/
- Mots longs — placez bien l’accent: electricity /ɪˌlɛkˈtɹɪsɪti/, infrastructure /ˈɪnfɹəˌstɹʌkt͡ʃə/, essentially /ɪˈsɛnʃəli/, ecosystem /ˈikoʊˌsɪstəm/, architecture /ˈɑː.kɪˌtɛk.t͡ʃə/
Comment s’entraîner avec cette vidéo
- Écoutez la vidéo en entier une fois sans parler et notez les mots que vous ne connaissez pas.
- Commencez à la vitesse 0,75×, répétez phrase par phrase, puis revenez à la vitesse normale quand cela devient facile.
- Enregistrez-vous et comparez avec l’original, en faisant attention à des mots comme chip, turbine, starve.
Qu'est-ce que la technique du Shadowing ?
Le Shadowing est une technique d'apprentissage des langues fondée sur la science, développée à l'origine pour la formation des interprètes professionnels. Le principe est simple mais puissant : vous écoutez de l'anglais natif et le répétez immédiatement à voix haute — comme une ombre suivant le locuteur avec un décalage de 1 à 2 secondes. Les recherches montrent une amélioration significative de la précision de la prononciation, de l'intonation, du rythme, des liaisons, de la compréhension orale et de la fluidité.