Hang tight while we fetch the video data and transcripts. This only takes a moment.
Connecting to YouTube player…
Fetching transcript data…
We’ll display the transcript, summary, and all view options as soon as everything loads.
Next steps
Loading transcript tools…
Пузырь ИИ вот-вот лопнет? Почему под угрозой не только профессии, но и вся экономика | Falcon Finance | YouTubeToText
YouTube Transcript: Пузырь ИИ вот-вот лопнет? Почему под угрозой не только профессии, но и вся экономика
Skip watching entire videos - get the full transcript, search for keywords, and copy with one click.
Share:
Video Transcript
Video Summary
Summary
Core Theme
Artificial intelligence is rapidly transforming the global economy, creating immense value and new opportunities while simultaneously posing significant challenges related to resource consumption, job displacement, and economic inequality.
Mind Map
Click to expand
Click to explore the full interactive mind map • Zoom, pan, and navigate
Economists are used to arguing about oil,
gold and gas. But today the focus is on
another resource: intelligence.
Just not human, but artificial.
While everyone is arguing whether it will replace people, we
decided to calculate what it will cost
and what it could turn into. Each
GPT chat query requires energy several times
greater than a typical Google search.
The electricity required to train a large
model costs millions of dollars, and
some countries already use
more energy for data centers than for entire cities. Behind the
beautiful promises of free AI
lies a huge industry.
Tech giants are spending hundreds of
billions of dollars on servers and chips,
building airport-sized data centers, and
buying up processors for the price of a sports car each
. At the same time, new
professions are emerging with salaries in the hundreds of thousands of
dollars for the ability to properly
communicate with a neural network, while in other places,
people are losing their jobs en masse because
the algorithm makes them faster and cheaper. And here's what's
surprising. Microsoft has compiled a
list of dozens of professions. who is in
danger and who is protected. The result
was paradoxical. It is not
the workers who are under threat, but the intellectuals: journalists,
translators, analysts, even
programmers. Those who
work with their hands are safe: nurses, mechanics,
cooks, cleaners. It turns out that
artificial intelligence is changing not
physical labor, but mental labor, and
much faster than we thought. By
some estimates, 80% of workers
today already feel the influence of Jesus. And all of
this, of course, sounds like abstract
numbers: millions, billions, percentages. But
how can we understand what they mean for
us personally? After all, the story of artificial
intelligence is not just about chips and
data centers. This is about how
the rules of the economy change imperceptibly. Some
professions disappear, and others appear
with astronomical salaries. What
was considered safe yesterday is
under attack today. Today we
will try to uncover the real economy of
artificial intelligence. From salaries in
the hundreds of thousands of dollars to countries where
data centers consume a fifth of all
energy. from the race for chips to
corrections that can turn the market upside down
overnight. And most importantly, we’ll figure out
how to notice these changes
before they directly affect you. But
before we begin, a friendly word
of advice. To gain a deeper understanding of
these processes, I recommend subscribing to
our Falcon Finance Telegram channel. There
we explain how the new
economy works, what trends are influencing the
labor market, where investments are flowing, and
why there's always a strategy behind technology presentations
. It's all
about finance, the global economy,
cryptocurrencies, and how economic
news and events influence our
investment decisions. Understanding these
mechanisms means being ready for
change. The link will be in the description. The
QR code will be here on the screen.
Be sure to check it out. So, over the
last couple of years, artificial
intelligence has ceased to be a prediction. It is
already built into the economy, and
our reality tomorrow depends on how we learn to work with it today.
Corporations realized this before anyone else. They
invest billions and are already dividing the world into
winners and losers. At first
glance, it seems like this game is somewhere
far away, but in fact, it is this race that
determines the economy of the entire world. The best
example is Nvidia. Until recently, it was a
manufacturer of chips for video games.
Today its market capitalization is
$4.3 trillion, more than Apple. And
all because Nvidia turned out to be the
only one who knows how to make
special processors. Graphics
chips ideal for training
neural networks. The whole world suddenly realized that without
these chips there would be no GPT chat or
artificial intelligence industry. But it's
important to understand that Nvidia's rise is
only part of the picture. Behind it stands a whole
parade of corporations that invest
hundreds of billions of dollars and divide the
future of the economy among themselves. Amazon, Google,
Meta and Microsoft will spend $
251 billion on capital expenditures in 2024
. 62% more than the
previous year. And Microsoft is already planning
to spend 80 billion by the end of 2025. Where is
all this money going? The answer is simple:
data centers are hangars the size of airports,
filled with thousands of servers. Each server
costs as much as a good house, a quarter of a million
dollars. And the chips for them are so
scarce that the price on the black market
has jumped from $10,000 to $40,000
apiece. It's a real race for
infrastructure. And in this race there is a
clear division. The first echelon is the
technology giants. They're building
empires out of data centers and buying up GPUs,
those same Nvidia graphics chips, like
gold during a crisis. Amazon
Web Services controls 30% of the global
cloud computing market. Microsoft
Azure - 21%, Google Cloud - 12%. Together
this is 2/3 of the entire market. But let's see
how much it costs to rent this intelligence.
Open AI, for example,
charges $1.25 per million
input tokens and $10 per million
output tokens for access to GPT5. What does it mean? A token is
roughly a word or part of a word. 1,000
tokens is a page of text. At first
glance, it seems like pennies per request. But if your
business processes a million requests a
day, the bill at the end of the month could
reach $100,000. It's not cheaper in the clouds
. Amazon's own
server rental for EI costs around $60 per
hour, which is half a million per year for one
powerful computer. And
hundreds or thousands of such servers are needed. Microsoft and Google have
roughly the same numbers. In other words,
intelligence is now sold like
electricity or land rent, and its
cost has already become a separate line item for the
company's expenses. It is measured, as
you understand, not in kopecks per request, but
in tens and hundreds of thousands of dollars per
month. But even this is not the limit.
Laboratories are already testing biological
computers based on living
brain neurons. CarrkallapS grew
rat brain cells in the lab, and they
learned to play virtual ping-pong in
just 5 minutes, faster than
computer algorithms, which
take an hour and a half. And Startup Finals
Park creates microbrains that work like
biocomputers. One hour of renting these
brains costs $500. But
they require a million times less energy than
servers. Imagine that in
a few years you will be able to rent not
computing power, but literally
brains in banks. Jokes about how it's a shame you
can't give a person brains
are becoming less funny. And
while corporations fight for
infrastructure, people are facing a
new reality. Bank of America launched
launched
Eric, a virtual assistant, in its mobile app, and he has already
conducted over a billion conversations with
clients. As a result, the cost
of processing one request fell and
the efficiency of operators increased. According to the
National Bureau of Economic
Research, using EI in
call centers increases the productivity of
agents by an average of 14%, especially among
newcomers. In retail, artificial
intelligence generates product descriptions.
eBay increased conversion by 3%
using only automatically generated
texts. It seems like a small thing, but 3% of a
billion-dollar turnover is tens of
millions of dollars in additional
profit. And in jurisprudence, JP Morgan
has introduced the Coin system, which
analyzes contracts. It saves
360,000 hours of work per year for traditional lawyers
, removing half of the routine tasks and
freeing these specialists for more
complex work. At the same time,
completely new professions are emerging on the market.
Just a couple of years ago, no one
had heard of them, but today the
world's largest companies are competing for these specialists.
For example, an industrial engineer. His job
seems simple. He chooses the right
words to communicate with the neural network, but in
reality it is subtle engineering. How to tune
a model to produce accurate and
useful results. In Silicon Valley, they
are willing to pay up to
$300,000 a year for this. For comparison, this is
several times higher than the average
developer salary. Would you like to work as an
industrial engineer, or does it seem like just
another trendy profession for a
couple of years? What do you think? Be
sure to share your thoughts in
the comments. It's interesting to see
how ready people are for change, or if
you're already in the process. Another rare role is AI
Safety Engineer. Its job is to ensure
that the neural network does not provide toxic
advice, encourage users to take
risks, or violate laws. In essence, this is a
new form of ethical oversight of
algorithms. The average salary here is about
about $250,000.
$250,000.
Open AI, Antropic, and Google are competing for such specialists. Here's
another example. Synthetic
dataset engineer. Imagine that to train a
model, you need huge amounts of data.
But real data is often restricted by
copyright, confidentiality,
or simply does not exist in the required
volume. Then such specialists create
artificial data sets that are as
close to reality as possible. Salary from
180 to 220,000 dollars per year. And this is
a profession that
simply did not exist just a few years ago. Do you see a
pattern? All these roles have
one thing in common. Routine
intellectual labor is becoming cheaper, but
the ability to control AI is becoming more valuable than
gold. But there is also a downside.
Small and medium-sized businesses find themselves
trapped. They don't have the money to build
their own data centers, so they
rent capacity from giants. And this is where what experts
experts
call vendor login comes into play. Dependent on
one technology provider.
Imagine a startup that
runs entirely on Amazon services. The entire
customer base, payment system, and internal
processes are all stored and managed in their
cloud. Switching to another platform is
technically possible, but it takes months of
work, millions of dollars in
data transfer costs, and the risk of business disruption during the
migration. According to Garner, about 70% of
small businesses in the US are critically dependent
on a single cloud provider. This
means that if Amazon, Microsoft, or
Google suddenly raise prices or
change their rules, businesses simply won't be able to
switch from their services. This is how a
new form of economic dependence is created.
Previously, businesses were dependent on
suppliers of raw materials or energy.
They now depend on access to computing
power, a resource without which
modern businesses can no longer survive.
It turns out to be a paradox: everyone invests,
but only a few know how to
make money from it. At the same time, we
cannot stop, because competitors are not sleeping.
While companies grapple with how to
profit from artificial intelligence,
electricity bills are rising
exponentially. And it's
not just corporations that foot these bills;
we all do. Imagine you
are writing a simple request to the GPT chat. Make
me a vacation plan for Italy.
You will receive a ready answer in seconds. Everything seems
easy and free. But at that moment,
somewhere in the data center, dozens of
graphics processors turn on. They consume
tens of times more energy than a
regular Google search.
Water is needed to cool servers. It is estimated that one liter is needed for
every several dozen requests. For
one person it is almost unnoticeable. But
when there are billions of such requests per day, the
carbon footprint is comparable to the emissions of
entire countries. And this is just the beginning, because
training models is even more expensive.
For example, training GPT3 required
about 1.3 GWh of electricity. That's equivalent to the
annual consumption of 120 American
homes. The carbon footprint was 550 tons of
carbon dioxide. It's as if more than
100 cars drove non-stop
all year round. And then the
daily work of the models comes into play. This is where
artificial intelligence begins to change
entire countries. The situation in Ireland is
critical. Data centers
consumed 5% of all electricity in 2015, and by
2023 it will be 21%. This is more than all the residential
buildings in the country. In fact, Ireland's data centers alone
draw more energy from the grid
than a city with a population of over a million. In the United States,
the story is no less telling. Let's take the
state of Iowa. Microsoft has invested
more than $10 billion in
data center construction here. Local authorities were
delighted and gave the company 100 million in
tax breaks, counting on thousands of
jobs. What was the result? There are
fewer than 300 permanent jobs. Yes,
around 1,600 people were employed during construction, but these were
temporary contracts.
Microsoft received more than $100 million in tax breaks
. If you divide this amount by
the number of permanent employees, it comes to
tens of millions for each position.
A figure that local residents and
journalists have called highly controversial. But
the consumption of resources is even
more impressive. In July 2022 alone, when
Microsoft was training GPT4, servers
used 43.5 million liters of water per
month. For comparison, this is the monthly
consumption of a city with a population of 30,000
people. In the Netherlands, data centers
have also become the largest consumers of energy.
In some regions, they consume
more electricity than all residential
buildings combined. And this is no longer a
story of individual corporations. For countries,
artificial intelligence is becoming a
macroeconomic factor. Ireland
has effectively become an energy
colony of data centers. The Ava spend
billions of dollars but create almost no
jobs. In the Netherlands, authorities
are forced to halt investments
to avoid overloading the grid. In other
words, artificial intelligence already
impacts countries' GDP, energy, and employment
as much as oil or gas.
Only this new raw materials economy
is built not around wells, but around
data centers. But what is happening to
the planet as a whole? If individual countries
are already feeling the strain, what is the
global scale of the problem? Today, the
world's data centers consume 1.5% of the world's
electricity, more than all of
Argentina. The International Energy
Agency predicts that by 2030
the figure could exceed 3%.
Global emissions from the industry
are estimated at 300 million tons of carbon
dioxide per year. It's as if all the
cars in the UK suddenly
doubled their mileage. And even companies
that promise to operate on green
energy can't always operate without the
conventional grid. The sun doesn't always shine,
the wind doesn't always blow, but servers run
24 hours a day, 7 days a week. It's
not easy with water either. Modern
data centers use evaporative
cooling. In arid regions this
becomes a critical factor.
In Arizona alone, the combined consumption of the centers
is estimated at 200 million liters per day. This is
equal to the daily water consumption
of a metropolis. But there is a second cost, a
human one, which reduces not only the
corporation's electricity costs, but
also its people costs. Open AI estimates that
up to 80% of American workers are
potentially affected by
AI in the future, and in the
ACP, about a quarter of jobs
are in areas at high risk
of automation. This means that over the
next decade, millions of
professionals could lose their traditional
job responsibilities. What happens to
employees who are laid off?
Most often, no golden cushion is
provided. Sometimes companies
offer retraining, but there are
almost no mass compensations. And while some
professions are disappearing, others are just
emerging with salaries of hundreds of thousands of
dollars, but accessible only to a narrow circle of
specialists. For most workers,
this means a sharp rise in inequality and
the need to acquire new
skills themselves to remain in the market. And here is the
contradiction. On the one hand,
artificial intelligence must solve
humanity's problems, but on the other hand, it itself
creates new ones: energy,
environmental, and social. And this is where
the question arises: if corporations
are cutting corners on people and jobs are
disappearing, how can the average person protect
their capital in such a turbulent
economy? For many,
cryptocurrency is one of these methods.
In fact, this is an asset that does not
depend on the decisions of central banks. But there is a
nuance. Most people enter the market
blindly and lose money. To
avoid this, my
team and I created our crypto community a few years ago.
There, we share analytics, analyze
working strategies, setups, real-world
cases, and show how global events
affect cryptocurrency prices. We
already have more than 12,000 participants. We've
built an entire ecosystem with
training for both beginners and
seasoned marketers. The most important thing is that all of this is
absolutely free, without any
attempts to sell you anything. The link
will be in the description. The QR code will be here on
the screen, so be sure to
join in. But if you look at the
bigger picture, the problems don't end with
jobs and money. Artificial
intelligence is also having an impact on the environment, and this is
no longer a threat to individual professions, but to
entire countries. When it comes to the growth of
artificial intelligence, we most often
think about computing, money, and
speed. But there is another boundary
that can stop the IBOM earlier
than crises or competitors. This is
ecology. It all comes down to simple things:
electricity, water and air.
Data centers are growing faster than
power grids can modernize. And
some countries have already started saying,
"Enough!" The first serious restrictions
appeared in Europe. In 2019,
authorities in Amsterdam froze
data center construction for a year.
The reason is simple: the network couldn't cope.
Meta planned a gigantic facility, but
the project was never realized. And when
the moratorium was lifted, the rules tightened.
Now construction is allowed only in
special zones and only under the condition of
high energy efficiency. For example, the
energy efficiency coefficient, an indicator of how much
energy is spent on the servers themselves and on
everything else: cooling, lighting,
infrastructure. In the Netherlands,
the bar was set at no more than 1.2.
This means that for every kilowatt of
server operation, a maximum of 0.2
0.2
kW can be spent on everything else. Failure to meet
the conditions will result in a refusal to obtain permission. In
Ireland, the situation is even more dire. The
national network operator, Airgri,
has effectively suspended the connection of
new data centers around Dublin. The network
simply can't handle the load. Want to
build? Prove that you will have your
own renewable energy power plant
. USA are following the
same path. In Virginia, where
most American data centers are concentrated, a
law was passed: all new facilities with a
capacity of over 50 MW must
run entirely on green energy by
2026. It sounds noble, but in
practice this means billions of dollars in
additional investment in solar and
wind farms. California has gone even
further. Companies are required to publish
reports on their water and
electricity consumption. And for now, this is just
transparency, but the next step is obvious:
limits and fines. And Europe is preparing an even
more radical tool: an assessment of the
environmental footprint of each major AI
model. How much energy was spent on
training, how much CO2 was emitted, what is the
carbon footprint of each query. If
the figures are too high,
the model can simply be banned from
use in the European Union. And this
fundamentally changes the economy. Previously, everything
was decided by computing power and access
to data. Now a third factor has appeared:
environmental efficiency. Microsoft
is experimenting with seawater cooling
. Google with geothermal energy.
Meta promises to make it by 2030
Data centers are carbon negative, and
Microsoft is water positive. But any of
these initiatives means additional
costs. And this raises the question:
will environmental restrictions lead to
innovation or, on the contrary, hinder
development. Startups are already emerging
that are creating effective EI.
They use model quantization,
data compression, and specialized chips. They are trying to
achieve the same quality with lower
energy consumption. But for most
small companies, such technologies are
too expensive. It turns out that
the same giants with
long-term money and infrastructure are winning again.
The environment, which was supposed to be a
brake on EI, may turn out to be its
new filter. And this filter works
simply. Only the richest survive.
But all these rules and new standards
affect not only corporations.
Ultimately, the bill for green EI is ours.
When Microsoft or Google are forced to
build their own power plants and
implement complex cooling systems, these
costs are transferred to the cost of
services. Chatbot subscriptions become
more expensive. Businesses that use
the cloud are raising prices for their services.
Even Utility bills rise faster
where data centers overload
the power grid. It's a paradox.
Environmental restrictions are supposed to
protect people and the planet, but in
reality they only highlight
inequality. Large corporations will most likely
survive. They have the money, but
in the end, ordinary people will have to pay.
The history of technology always repeats itself.
We saw it with railroads in the 19th
century, with the internet in the 1990s, and now we
see it with artificial intelligence.
First, euphoria, then collapse, and only then a
golden age. Economist Carlotta Perez
described this cycle very accurately. Every
technological revolution goes through the same
stages. A breakthrough, a manic
phase of investment, the collapse of the bubble, and only
then mature implementation. By the way, if this
logic seems familiar to you, it's no
coincidence. In our separate video about
financial bubbles, we analyzed in detail
how this mechanism works, from the tulip
mania to the dot-com era. And there you can see that
every time euphoria, no matter how new it may
seem, always passes The
same scenario. We'll definitely leave a link
somewhere here, guys, in the corner, on
the screen, in the description under this video.
Be sure to watch it, because
today artificial intelligence is following the
same trajectory. Look at the numbers.
In 2019, venture capital investments in
artificial intelligence amounted to about
$40 billion. In 2025, it's already over $100
billion in just six months. Open AI
is valued at $300 billion, Antropic at $60
billion, although their actual revenue
is measured in hundreds of millions. It's a
classic bubble. Remember Wework, the
coworking network that was called
the offices of the future. Its value
reached $47 billion at its peak, but just a
few months later, the valuation collapsed by
about three-quarters to $10-12 billion. For
investors, this became a symbol of how
quickly even the loudest
promises deflated. Or take the very recent case of
Humane AI. Former Apple top managers
raised $230 million. investments.
But the product turned out to be a failure:
slow, overheating, and with a
$24-a-month subscription. By August,
there were more returns than sales, and in
February 2025, the company sold
the rest of the business to HP for $116 million—
five times less than the investment.
Stories like these are alarming,
because only 5% of AI companies today
actually earn more than 10% profit.
The rest are either experimenting or
simply burning through cash. This can't go
on forever. What will be
the trigger? Completely different options are possible
. For example, strict
regulation, when the government sharply
limits AI development. Or, say, an
energy crisis, if it becomes clear
that the planet simply can't
withstand the race for computing
power, or simply disappointment from
investors when promises don't match the
actual returns. And, most likely, a collapse is
almost inevitable. But this is not the end, but
rather a transition. After the dot-com boom,
Google, Amazon, and eBay survived. After the correction, the
leaders will also survive. And those with
real products and a stable
economy. NVIDIA and Microsoft will most likely
remain, and hundreds of overvalued
startups will disappear in a matter of months.
It is after this that the golden age will begin,
when artificial intelligence will cease
to be hype and become infrastructure,
like electricity or the internet. According to
international economists,
mature artificial intelligence is capable of adding
adding
almost 1% per year to productivity growth for
20 years. This is trillions of dollars for the
global economy. But all this will come
later. Now we are living in a mania phase, and the
only question is who will survive
the coming winter. And we are used to
thinking that technology arrives
gradually. First as a toy, then
as a tool, and only then does it become
a necessity. But with artificial
intelligence, everything is different. It is already integrated into
the economy and is changing its pace. Like,
you know, an accelerator in a reactor. For the first time
in decades, we live in an era when
not only work and money, but the very
understanding of value is changing right before our
eyes. Knowledge that previously accumulated over
generations can now be Rent for a
subscription. And this isn't just a new
business model; it's a new way
to organize the world. And in such a reality, no
one can be sure of
tomorrow. What seems safe today
may disappear tomorrow, and what
seems like a temporary fad
may become the norm in a year. Therefore, the main
question here isn't whether a machine will replace you
. The main question is: what place
will humans occupy in a system where intelligence
has ceased to be our unique
advantage? Preparing for such a
future is impossible with a textbook, but there are
steps that really help.
For starters, I think it's worth at
least a general understanding of how
language models work, what
agents are, and trying to apply them to
your tasks. Let these be the
simplest things: automating routine tasks,
speeding up text processing, or simplifying
data organization. Look at the
professions emerging around
artificial intelligence. Some places need
people who help companies
use neural networks correctly. Others need
those who check the quality of
algorithms or train employees
to use new tools. You don't
necessarily have to change your career drastically, but
you can figure out that This will give you a better idea of
where to take the first step.
Start, I don't know, with a
little research. Read the stories of
those already working in these fields.
See which path resonates with you
most. It's equally important to understand your
strengths. Just
ask yourself honestly: what do you do
best? And how can you develop it so that it's
hard to do without you. For
some, it's the ability to find common ground with
people, for others, the ability to
organize a process and bring it to a
result, while others have taste, a sense of
style, or a keen eye. And these things
are born from experience and character, which is
why they're so valuable. And finally, it's
important not only to learn the tools, but
also to change your attitude toward change,
because the world is changing too quickly
to expect any ready-made recipes
or, for example, to hope for
stability. Therefore, remember that those
who remain curious and flexible
adapt more easily and quickly find
their place in the new reality. Friends,
where do you think artificial intelligence will lead us
? Will it be a
tool that expands
everyone's capabilities? Or, for example,
will it become a lever of power in the hands of a select few?
And in general, are you ready to accept this
rate of change? Be sure to share
your thoughts in the comments. It will be
interesting to compare points of view, so to speak
. If you found this analysis helpful, be
sure to support the video with a like. This
really helps us conduct new and
interesting research on economics and
technology. As always, all useful links are
in the description below this video. And
Click on any text or timestamp to jump to that moment in the video
Share:
Most transcripts ready in under 5 seconds
One-Click Copy125+ LanguagesSearch ContentJump to Timestamps
Paste YouTube URL
Enter any YouTube video link to get the full transcript
Transcript Extraction Form
Most transcripts ready in under 5 seconds
Get Our Chrome Extension
Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.
