Prática de Shadowing: OpenAI: A Bubble Bigger Than Dotcom - Aprenda a falar inglês com vídeo
Criando lição...
1
We have no current plans to make revenue.
2
We have no idea how we may one day generate revenue.
3
We have made a soft promise to investors that once we've built this sort of generally intelligent system,
4
basically, we will ask it to figure out a way to generate an investment return for you.
5
Sam Altman, his plan for paying back OpenAI's investors was to ask AI.
6
They've now got this AI bubble so big that even the US Treasury is now quietly sounding the alarm.
7
Back in June, OpenAI confidentially filed to become a public company.
8
This came a week after Anthropix's same announcement.
9
And while the company claimed they were keeping their options open on the timeline, other sources reported Sam Altman was pushing for as early as September.
10
That was until an independent reporter, Ed Zittron, got his hands on OpenAI's financials, Which is one of those things where we knew it was bad,
11
we just didn't know how bad.
12
Last year, OpenAI burned around $38 billion.
13
Now, much of these costs came from turning the company from a non-profit into a for-profit.
14
But as Ed Citrin so concisely put it, despite these changes, they've remained profitless.
15
They turned from a non-profit to a for-profit, though they've remained profitless.
16
A few days after these numbers leaked, reports came out that OpenAI was delaying going public,
17
exciting SpaceX's volatile stock, which if you haven't seen looks more like a crypto pump and dump than a legitimate company stock.
18
But it's more likely that the reason was simply the cat was out of the bag.
19
OpenAI was wanting to go public at a 1 trillion valuation
20
despite spending 21 billion dollars to make just 13 billion in revenue.
21
Earlier this year, it was reported that Sarah Fryer, OpenAI's own chief financial officer, warned Sam Altman against going public this year.
22
Her reasons were quite simple.
23
Sora being a failed project, executives leaving the company, at the time the Elon Musk trial,
24
and of course the $1.4 trillion commitment that they couldn't pay.
25
Seems reasonable.
26
But as a result of her pushback, it was reported that Sam Altman excluded her from joining investor meetings to discuss spending with OpenAI investors,
27
which is literally what she's in charge of.
28
So the woman with the most information warned privately while Sam Altman continue to publicly praise AI.
29
This plays out across the entire industry, but maybe the CFO is wrong.
30
Maybe the demand is there and the losses are temporary.
31
So let's check, who is actually paying for this?
32
OpenAI currently has the largest market share at 53.9%, with Gemini at 27.9%, Claude 9.2%, and Grok at 2.4%.
33
Despite this, only around 5% of ChatGPT users actually pay for a subscription
34
and notably more users are switching to Claude or Gemini.
35
So I just switched from ChatGBT to Claude as I'm sure many of you have.
36
I can't lie Claude is a game changer man like that ChatGBT stuff that's in the past.
37
As someone who literally went viral for teaching people how to
38
study with ChatGBT I fear I might be switching over to Claude because she is that girl.
39
So even after spending six billion dollars on sales
40
and marketing they couldn't convince their users to upgrade to the 20 paid version it's almost as
41
if people don't find it that valuable which is even more concerning
42
when you consider
43
that 20 a month isn't anywhere near what these ai companies need to charge to be profitable
44
so if users aren't willing to pay who is the obvious answer is businesses, but they're not seeing the value either in may this year
45
ibm survey of ceos revealed only 25 of ai initiatives have delivered expected return on investment
46
and only 16 have scaled enterprise wide
47
and where gains do exist they're narrow the gain concentrates in code generation customer support assistance
48
and drafting most other roles show no measurable payback yet
49
which makes sense chat gpt claude gemini these are all large language models
50
so language-based tasks like coding is where we see some utility
51
but these companies are pushing these models to be something they're not they've overhyped them
52
and as a result users corporations investors they're not getting the value they were promised
53
and at the same time companies like meta are announcing they
54
will sell their ai data center compute to other companies despite saying this just last year even with the capacity
55
that we are bringing online in 2025 we are having a hard time meeting the demand
56
that teams have for compute resources across the company they went
57
from not having enough compute for their own internal use at meta to now needing to sell that compute
58
because they have too much this comes after xai did the
59
same thing they're now leasing their data center compute to Anthropic and Google.
60
So let me get this straight.
61
Users don't want to pay $20 a month.
62
Enterprise companies are not seeing a return.
63
Meta and XAI, who are supposedly leaders in AI, and yet they don't even have a use case for all these data centers they've been building,
64
so they're leasing them to their competitors.
65
And the cherry on top is that costs have exploded.
66
OpenAI, Anthropic, Microsoft, they've all moved away from a flat rate subscription model to a usage-based billing for their enterprise customers.
67
With AI now being billed by usage, one company accidentally spent $500 million in a single month using Claude.
68
They told their developers to code as much as possible with AI, so they did.
69
We went from use AI for everything to let's measure our usage.
70
But the result is companies are now avoiding American AI models and opting instead for Chinese models like GLM,
71
DeepSeek, and Kimi, which are either free or very cheap.
72
Coinbase announced they cut their AI spend 50% by switching to GLM and Kimi.
73
Cursor admitted they built their Frontier coding model on top of Kimi, and even Microsoft is testing DeepSeek and Copilot as a cheaper option.
74
Chinese models grew from about 1% of token usage on OpenRouter in late 2024 to over 60% by early 2026,
75
surpassing US models for the first time.
76
But it gets worse.
77
It's not just users who are moving away from ChatGPT, it's OpenAI's own partners.
78
You may remember Apple's Siri was supposed to be powered by ChatGPT, but fast forward to June this year, it's now powered by Google's Gemini.
79
And last year, OpenAI acquired the AI hardware company IO for $6.4 billion.
80
They wanted physical devices running chat GPT like from the movie Her, devices that always listen and continuously respond.
81
Key detail, IO's founder is Johnny Ive, the man who designed the iPhone.
82
And what do we learn this month?
83
Apple is suing OpenAI.
84
Alleging OpenAI asked job candidates from Apple to share details about secret projects
85
and to to bring device components and prototypes to their interviews.
86
But it doesn't stop there.
87
Microsoft is no longer OpenAI's exclusive cloud provider and will no longer pay a revenue share to OpenAI.
88
Every few months, Microsoft and OpenAI have to release some joint statement about their strained relationship, which is exactly what companies do when everything is fine, right?
89
Elon Musk has sued them, now Apple is suing them, and if that's not enough,
90
then there is the extremely high turnover with key leaders leaving on a monthly basis.
91
Turns out OpenAI is a dumpster fire of a company, but zoom out and they represent the entire AI industry.
92
And this isn't just me saying this.
93
The people who regulate the U.S economy are now quietly saying the same thing.
94
A leaked draft report from the U.S.
95
Treasury Department outlines the risk that AI firms pose to the economy.
96
The analysts found that AI firms are more deeply entrenched in the U.S economy than their dot-com predecessors.
97
A downturn in the AI market would send shockwaves throughout the entire economic ecosystem.
98
Stock markets, private credit markets, companies financing data center buildouts, cloud providers, chip manufacturers, and utilities would all feel the effects.
99
What I find most concerning about this leak is that the report was put together by Treasury analysts for Secretary Scott Besant,
100
Fed Chair Kevin Walsh, and various other federal financial regulators, the very people who are supposed to be regulating the U.S economy.
101
So while Scott Besant's analysts were working on this report, here's what he said in New York.
102
And we are seeing that the hyperscalers are going to spend at least $750 billion this year.
103
Obviously, a lot of that's import, so it won't go directly into GDP.
104
But we are seeing the here and now, whether it's in the construction jobs, in the capital markets, the things that are happening.
105
And I think that we all just have to have an open mind in terms of where this can take us.
106
The Treasury Secretary seemingly went into this interview knowing his own
107
analyst had warned the AI bubble is now bigger than the dot-com bubble.
108
And yet he sat there and praised hyperscalers spending more than $750 billion on data centers alone this year.
109
And then had the audacity to say we all just need to have an open mind about where this can take us.
110
The people with the raw numbers, the data, the truth about how big this bubble really is are privately discussing how bad it is
111
while publicly lying to our faces.
112
Telling us to have an open mind when it's our retirements on the line.
113
It's our tax dollars partially funding these AI data centers, it's our electricity bills increasing to power them.
114
When this AI bubble bursts all the wealthy people sitting in that room will be just fine.
115
In fact, they'll buy up more houses when everything crashes.
116
But it's us, the everyday people, who will have fewer jobs, higher grocery prices, a wrecked retirement, and less stability overall.
117
In the report, the analysts argue that AI firms pose significant risk to the entire system if financial conditions change,
118
productivity goals are missed, or various choke points stymie growth.
119
We have already checked all of these boxes.
120
First condition, missed productivity goals.
121
We've covered that one.
122
Only a quarter of enterprise AI initiatives deliver the expected return.
123
A perfect example of this, Pizza Hut franchisees are suing after they claim they were forced to use an AI delivery management platform for increased productivity,
124
but instead resulted in delivery delays and unhappy customers, costing more than $100 million in losses.
125
Second condition we've already met, choke points, stymie, and growth.
126
Oracle borrowed $43 billion this year to build data centers.
127
But there's a problem with this.
128
Oracle is using these billions to build data centers to then lease back to other AI companies.
129
This unlocks $638 billion in future revenue, and half of that is supposed to come from OpenAI alone,
130
who we know is also burning through billions.
131
And let's not forget, this is the same industry where two of the five leading AI companies, Meta and xAI, are already leasing their data centers.
132
Breaking news on the Bloomberg terminal, Oracle and OpenAI are apparently ending plans to expand that Texas data center site.
133
This is Stargate.
134
Remember, it was just yesterday where Brody Ford here at Bloomberg reported
135
that Oracle was planning to cut thousands of jobs as it
136
was dealing with the cash crunch from a massive AI data center expansion effort.
137
And now a new report out saying that Oracle
138
and OpenAI have scrapped plans to expand that flagship AI data center in Texas.
139
And that's before we factor in the growing backlash against AI data centers.
140
$64 billion worth of data center projects have been blocked or delayed amid local opposition.
141
Choke points everywhere you look.
142
Which brings us to the third condition, which we've already met.
143
Financial conditions changing.
144
The money is drying up.
145
Back in March, Reuters reported that OpenAI was offering private equity firms preferred equity stakes with a guaranteed minimum return of 17.5%.
146
OpenAI isn't just running out of money, they're running out of financing options.
147
And let's consider if the current investors were truly confident in this company,
148
they would put more of their money in instead of stepping aside to allow private equity to extract 17.5%.
149
So how can Sam Altman guarantee such a high return?
150
Well, despite what he said at the beginning of the video, it's quite clear that the plan was always to go public.
151
But why would we, the public, want to own stock in a company that has no profits?
152
No money?
153
They're broke.
154
But OpenAI wanting public funds really shouldn't come as a surprise.
155
Last year, Sarah Fryer, the CFO, made comments about a federal backstop, Meaning the government would pay for OpenAI's debts if OpenAI couldn't.
156
She later walked those comments back, but fast forward to this month, we learned that OpenAI has offered the US government a 5% stake in the company.
157
This is yet another example of billionaires privatizing the earnings and socializing the losses.
158
If the government takes an ownership stake in OpenAI, the company officially becomes too big to fail.
159
And who prevents it from failing?
160
The taxpayers.
161
And if you think the government taking an ownership stake in a company is unlikely,
162
well, since January 2025, the U.S government has announced investments worth $26.7 billion across 30 deals involving direct ownership of companies.
163
The U.S government is now apparently a private equity firm.
164
Sam Altman planned to ask ChatGPT how investors would get their money back.
165
Maybe someone should ask it, who's left holding the bag?
166
Reportedly that the administration and potentially some sort of sovereign wealth fund would take a 5% stake in open AI.
167
I mean, is that a situation that you would like to see happen or do you think that's more disruptive?
168
No, I don't think there's any need to.
169
They talk about, I think Sam Orman said, well, we can share the benefits of AI and the profits of AI.
170
What profits?
171
What returns?
172
That 5% will have to get congressional approval.
173
The sovereign wealth fund is still an idea at this point.
174
But fundamentally, large language models are not the future.
175
The only reason big tech is investing in this is that they've run out of hyper-growth ideas.
176
They don't have a next iPhone.
177
They don't have a new Google search.
178
So they've put over a trillion dollars with trillions more to come into a kind of a dead-end industry
179
because when that ends, they'll have to admit that they don't have anything else.
✨ Vídeo recomendado
B21
216 frases
The Leaked Audit That Exposes OpenAI's Real Financial Crisis
Business EnglishUniversity StudentsWorking Adults
B21
57 frases
William Blair’s Louie DiPalma: We estimate SpaceX will spend about $50 billion on AI capex this year
Business EnglishDaily ConversationUniversity Students
B22
20 frases
Is the AI Boom Just Another Bubble?
Business EnglishDaily ConversationUniversity Students
B21
112 frases
‘We don’t have months’: Bernie Sanders sounds alarm on AI's 'extinction' threat
Business EnglishDaily ConversationUniversity Students
B21
81 frases
How A Single Shell Company Drained $1.6 Billion From The Pentagon
Business EnglishDaily ConversationUniversity Students
B21
40 frases
I MADE $15B WITH CHAT GPT
Business EnglishDaily ConversationUniversity Students
B22
178 frases
Against All Odds: Never Give Up When Life Gets Hard | Elon Musk Motivation
Business EnglishDaily ConversationUniversity Students
B21
85 frases
Significant slowdown in AI payoff could tip economy into a recession, says Apollo Global's Slok
Business EnglishDaily ConversationUniversity Students
B22
190 frases
He’s always wrong
Business EnglishDaily ConversationUniversity Students
B21
261 frases
GMA First Look Full Broadcast - Sept. 9, 2026
Business EnglishDaily ConversationUniversity Students
B23
269 frases
The AI Industry is a Complete Mess
Business EnglishDaily ConversationUniversity Students
B21
190 frases
This 25-Year-Old Built a 2-Year Moat Nobody Can Bet Against | Corgi, Nico Laqua & Emily Yuan
Business EnglishUniversity StudentsWorking Adults
Por que esse vídeo é ótimo para praticar fala em inglês?
O vídeo sobre a OpenAI é uma excelente fonte de material para quem quer melhorar a pronúncia em inglês e a fluência. A linguagem é conversacional, com expressões coloquiais e variações de tom que refletem a naturalidade da fala cotidiana. Além disso, ele mistura tópicos técnicos (como "market share" e "large language models") com comentários pessoais, o que ajuda a treinar a compreensão de diferentes registros. Se você pratica shadowspeaks (imitar a fala em tempo real), essa combinação de formalidade e informalidade é perfeita para desenvolver flexibilidade na fala.
Desmembrando frases naturais
Vamos analisar 3 frases do vídeo que mostram como o inglês é falado no dia a dia:
- "I can't lie, Claude is a game changer, man." A expressão "I can't lie" (não posso mentir) é muito usada para dar ênfase a uma opinião sincera. "Game changer" (mudança de jogo) é uma metáfora coloquial para algo que transforma a situação. A pausa com "man" no final deixa a frase mais informal e próxima da fala real.
- "It's almost as if people don't find it that valuable." "It's almost as if" (é quase como se) é uma estrutura usada para sugerir uma ideia sem afirmá-la diretamente. "That valuable" (tão valioso) usa "that" em vez de "so" para enfatizar a baixa percepção de valor, um detalhe que torna a fala mais natural.
- "Where gains do exist, they're narrow." A inversão "do exist" (existem de fato) é usada para dar ênfase a uma exceção. "Narrow" (limitado) é uma palavra simples mas eficaz para descrever ganhos restritos, mostrando que o inglês não precisa ser complexo para ser claro.
Rotina de prática com shadow speak
Para usar esse vídeo no seu treino de shadow speech, siga esse passo a passo:
- Escute e repita frases curtas: Selecione trechos de 5-10 segundos (como o comentário sobre o Claude) e reproduza, pausando e repetindo imediatamente. Foque na pronúncia das vogais (como "game" e "change") e no ritmo das pausas.
- Grave a sua voz: Após repetir, grave a sua versão e compare com o original. Perceba diferenças no tom (por exemplo, a ênfase em "literally" no vídeo) e ajuste sua fala para se aproximar do natural.
- Trabalhe com expressões coloquiais: Frases como "that's in the past" (isso já passou) ou "she is that girl" (ela é a pessoa certa) são chaves para falar como um nativo. Pratique essas expressões até que elas saiam naturalmente, sem hesitação.
Essa rotina ajuda a melhorar não só a pronúncia, mas também a fluência e a compreensão de contextos reais. Lembre-se: o shadowspeak é sobre imitar não só as palavras, mas também a emoção e a entonação do falante!
Gramática neste vídeo
As estruturas que o falante mais usa, com as palavras exatas do vídeo:
| Estrutura | No vídeo |
|---|---|
| Present perfect have/has + particípio passado — uma ação passada que ainda importa agora | have made · they've remained · haven't seen |
| Voz passiva be + particípio passado — o foco está no que acontece, não em quem faz | was reported · were promised · being billed |
| Orações relativas who / which + oração — informação extra sobre uma pessoa ou coisa | financials, Which is · someone who literally · valuable which is |
O que é a Técnica de Shadowing?
Shadowing é uma técnica de aprendizado de idiomas com base científica, originalmente desenvolvida para o treinamento de intérpretes profissionais. O método é simples, mas poderoso: você ouve áudio em inglês nativo e repete imediatamente em voz alta — como uma sombra seguindo o falante com 1-2 segundos de atraso. Pesquisas mostram melhora significativa na precisão da pronúncia, entonação, ritmo, sons conectados, compreensão auditiva e fluência na fala.