Home
Login
Register
Learning Content
Loading...
Corporate AI Is A Delusion. $600 Billion Just VANISHED. - Video học tiếng Anh
Practice Listening
Practice Listening
/
Video
/
The Infographics Show
/
Corporate AI Is A Delusion. $600 Billion Just VANISHED.
Corporate AI Is A Delusion. $600 Billion Just VANISHED.
Select learning mode:
View subtitles
Pick word
Rewrite word
Highlight:
3000 Oxford Words
4000 IELTS Words
5000 Oxford Words
3000 Common Words
1000 TOEIC Words
5000 TOEFL Words
Subtitles (270)
0:00
Big tech just burned through hundreds of billions building the infrastructure for AI.
0:04
And the return so far? A fraction of the cost. Nowhere close to break-even. It might be one
0:10
of the most expensive bets in modern history. Beneath the polished interface is something
0:15
most people never think about… a physical system with real-world limits. Data centers
0:20
burning through enormous amounts of electricity. Power grids pushed closer to their capacity just
0:25
to keep everything running. And it relies on a hidden army of exploited human labor.
0:30
You don’t get to stay outside of it. You’re already interacting with it every day.
0:35
The world has already committed to a system it may not be able to stop.
0:38
This is the trillion dollar AI lie.
0:42
Chapter 1: The Receipt Nobody Wants to Read
0:45
In 2024, analysts at Sequoia Capital posed a simple question. If AI companies keep spending the
0:51
way they are right now, how much money would the industry need to make every year to justify it?
0:56
Their answer was about $600 billion a year.
1:00
That’s not what AI companies are earning today.
1:03
That’s what they’d need to earn for all of this to make financial sense.
1:06
Right now, generative AI is already making real money. Billions in annual
1:11
revenue across companies like OpenAI and Anthropic. Analysts
1:14
predict Ai-driven profits may rise into the trillions by the mid-2030s.
1:19
But a lot of the reported revenue isn’t even profit at all.
1:22
Anthropic recently scaled its revenue to hundreds of millions per month in 2025. It sounds
1:28
incredible until you realize it was also expected to lose billions over the course of the year. In
1:33
mid-2025, OpenAI secured $10 billion dollars of funding. But since its operating costs were so
1:39
steep, it was asking for another $8.3 billion just months later. And Elon Musk’s xAI? It was
1:46
reportedly burning through more than a billion every single month… just to keep the lights on.
1:51
It’s a serious problem.
1:53
One that points to something people don’t like to talk about.
1:56
In traditional software, once you build a product adding more users is cheap.
2:00
That’s why companies like Microsoft and Adobe can become insanely profitable.
2:04
The cost of serving the next customer is basically nothing.
2:08
AI? It doesn’t work like that.
2:10
Every time you ask a question, it costs money. Real money. The answer comes from a delicate
2:16
interplay of computational power, electricity, and infrastructure. Which means scaling doesn’t
2:21
automatically make things more efficient. In some cases, it does the opposite.
2:25
The more people use the system, the more expensive it becomes to run.
2:29
The largest companies and investment groups in the world have already committed,
2:33
with annual spending pushing past $400 billion. AI growth speculation has become a core engine
2:38
of the S&P 500. It becomes even sketchier when you realize that the broader market,
2:44
built up of retirement accounts, index funds, and otherwise “safe” investments, are all now
2:49
deeply exposed to the whims of a technology that hasn’t even proven it can pay for itself at scale.
2:55
If the economics of generative AI’s future are so uncertain, why does every corporate
3:00
CEO in the tech world seem so hell-bent on staking everything on its future?
3:04
According to a recent survey, roughly 90% of CEOs say AI will fundamentally change their
3:10
companies by 2028. It’s a huge number. But when you look at the actual financial data,
3:15
the reality is brutal. Only 25% of AI initiatives are actually delivering their expected ROI.
3:22
9 out of 10 executives believe the technology is
3:25
essential. But only about 1 in 4 can explain how it actually makes money.
3:30
What once felt like a strategy now seems like corporate peer pressure.
3:34
And you see it playing out in real time. Each company
3:36
tries to outdo the next rushing out AI initiatives, talking about “AI-first”
3:41
strategies, and buying up massive quantities of GPUs because not buying them looks worse.
3:46
If this amounts to the corporate version of a gold rush, nobody wants
3:50
to be the one who showed up without a shovel. That’s where we’re at right now. Infrastructure
3:54
is being built up at full speed. The spending is already committee and seed money raised.
4:00
But the returns are barely keeping pace.
4:02
Chapter 2: Success That Makes You Poorer
4:05
Over the past few years, a quiet industry has emerged to support AI systems. Things
4:10
like data labeling, content moderation, and output verification have all grown as
4:15
the core tasks that make models usable in the real world. They’re basically
4:19
the difference between a raw system that produces chaotic, inappropriate outputs
4:23
and one that can answer questions without embarrassing the company that built it.
4:27
All that work happens outside the model itself.
4:30
In offices and call-center-style environments in places like Kenya,
4:33
the Philippines, and India, thousands of workers spend their days correcting
4:37
the mistakes that AI systems still make constantly. In some documented cases,
4:42
they’re paid just a few dollars an hour to filter out violent or explicit content.
4:46
According to one report, content moderators in second and third-world countries navigate
4:51
a combination of “psychological trauma, poverty wages, and the suppression of
4:56
union organising conditions” considered intolerable under Western labor laws. It
5:01
sounds less like a futuristic breakthrough and more like something out of the 19th century.
5:05
That's because in some ways it is.
5:07
The marketing language around AI suggests autonomy. Machines that can think, learn,
5:12
and act, replacing human effort altogether.
5:15
The reality is closer to something older and more familiar. It’s a layered system,
5:19
where the most visible tip of the iceberg is clean and efficient and the least
5:24
visible part - the system’s inner core - is messy, labor-intensive, and easy to ignore.
5:29
That doesn’t mean the technology is “fake.” But at second glance,
5:33
the word “artificial” is doing an awful lot of heavy lifting.
5:36
A large portion of the work that makes these systems run is still very human.
5:40
That human system doesn’t disappear as the system scales. If anything,
5:44
it just becomes more important.
5:46
When the system relies on both massive compute infrastructure and ongoing human input,
5:51
then the cost structure starts to look very different than the one most people imagine.
5:55
Rather than a clean, self-improving machine, you get something closer to
5:59
a hybrid: part automated, part manual support… both constantly maintained.
6:04
At the beginning of the AI boom, researchers found out that a huge percentage of companies
6:08
calling themselves “AI Startups” weren’t actually using any meaningful AI in their core products.
6:13
In some cases, it was as high as 40%. That didn’t necessarily
6:17
mean fraud. But it did hint at something even more important.
6:21
From the very beginning, “AI” became a signal. A way to attract funding, justify higher valuations,
6:27
and position themselves in a market where everyone was suddenly supposed to have an AI story.
6:32
By 2024, roughly 78% of all companies reported using AI in some form. Just a year later,
6:40
61% of all global venture capital was flowing into AI-related businesses.
6:44
You might expect that after all this adoption, all this investment,
6:48
we’d be seeing clear, consistent returns. Instead, the gains are highly concentrated,
6:53
with about 75% of the financial benefits reaped by just 20% of the companies.
6:58
As it turns out, the majority of companies talking about AI aren’t
7:02
actually making that much money from it. They’re experimenting, deploying features,
7:06
and integrating tools, but not transforming their business like everyone thought.
7:11
In the early days, companies claimed AI without really using it.
7:14
Today, companies are using it, but don’t know how to make money off it.
7:19
And that’s a much bigger problem.
7:20
Because now the stakes are higher, the expectations are enormous,
7:24
and the gap between what’s actually happening and what was promised is getting harder to ignore.
7:29
Chapter 3: The Energy Wall
7:31
The deeper constraint on all of this isn’t financial, but physical.
7:36
For years, a simple comparison was repeated that a single AI query can use around ten times the
7:41
electricity than a standard web search. That number comes from early estimates,
7:45
and like most simple comparisons, it’s been argued over, updated,
7:49
and stretched depending on who is making the case.
7:52
But the part that hasn’t really changed is the assertion that AI workloads are simply heavier.
7:57
By 2022, data centers were already consuming roughly 460 terawatt-hours
8:02
of electricity globally every year. That’s about the same as an entire
8:07
country like Germany or Japan, or 2% of total global demand.
8:11
By this year, that number could land somewhere between 620 and
8:15
1,000 terawatt-hours, depending on how aggressively AI keeps growing.
8:19
When you look into a data center, how that energy is used is somewhat surprising.
8:24
40% of the electricity is used for actual computing. The “AI doing its thing” part
8:30
Another 40% is used just to cool the machines down.
8:34
The remaining 20% goes to everything else: moving the data around; keeping the systems stable;
8:39
basically making sure the whole operation doesn’t collapse under its own complexity.
8:43
So nearly half the energy we’re pouring into AI…isn’t
8:46
making it any smarter. It’s just keeping it alive.
8:49
Now zoom out one level.
8:51
A single large “hyperscale” data center,” the kind companies are building for AI, can consume
8:56
100 megawatts of power or more. 1 megawatt can typically power over 150 U.S. homes. 100? Over
9:04
a year, that’s akin to the electricity needed to charge well over 200,000 electric vehicles…
9:10
For just one building.
9:11
And there are over 8,000 data centers worldwide, with a third of them sitting
9:16
in the United States alone. Worryingly, studies are now showing that the vast majority of these
9:21
are located in climates considered way too hot for efficient operation.
9:25
It’s time to stop thinking about the “cloud” as something abstract. We’re talking about
9:29
a growing network of massive, power-hungry facilities, clustered in specific regions,
9:35
pulling from the same grids that supply homes, schools, and businesses.
9:39
The demand isn’t close to being evenly distributed, either. In areas with
9:43
concentrated pressure, city officials are forced to make trade-offs in real time:
9:48
Do you expand the grid?
9:49
Do you raise power prices?
9:51
Do you slow down development and limit job growth?
9:54
None of those are easy answers. And none of them deal with the issue of water, either.
9:59
Remember the 40% of energy used simply to cool these mammoth data centers? That often depends
10:04
on huge volumes of water moving through the system, constantly cycling to carry the heat away.
10:10
A single large data center can use millions of gallons of water per day,
10:14
about the same as a town of 30,000 to 50,000 people. Over a year,
10:18
even a mid-sized facility can burn through around 100 million gallons just to stay cool.
10:24
And most of that water doesn’t come back.
10:27
In many systems, 70 to 80% is lost to evaporation, effectively disappearing into the air. Meanwhile,
10:33
roughly 75 to 90% of data centers rely on water-based cooling, often pulling from the
10:39
same rivers and municipal supplies that serve local communities. In at least one Oregon town,
10:44
a single company’s data centers consumed over 25% of the city’s water supply.
10:49
The same system that promises infinite scale is drawing from very finite supplies.
10:54
Power grids that expand overnight, and water systems are already under pressure. The more
10:59
the system grows, the more it pulls. And right now, it’s not obvious where the ceiling is.
11:04
And while all of that strain is building in the physical world,
11:07
something else is happening inside companies. Because while AI is
11:11
pulling more from the outside… it’s also quietly pulling something from the inside.
11:16
Their data. Chapter 4: Security Self-Sabotage
11:20
According to recent data, about 34.8% of employee inputs into
11:25
AI now contain sensitive information. That’s up from just over 10% in 2023.
11:31
A third of everything being pasted into your AI chatbot of choice are legal documents,
11:36
customer data, medical records, source code, and contracts.
11:39
What’s worse, 83% of the companies doing this
11:42
have absolutely zero technical controls in place to stop it.
11:46
They can send out policy emails until the cows come home. But the next day they’ll
11:50
turn around and tell those same employees to be faster, more productive and efficient.
11:54
The employees now have to decide: deadline…or policy? They pick the deadline, every time.
12:00
The scale of this security self-own is starting to look like a slow-motion disaster.
12:04
Over 225,000 Chat GPT credentials have already been found for sale on dark web marketplaces,
12:11
often harvested from compromised machines or reused passwords. Major companies like Apple,
12:16
JP Morgan, and Goldman Sachs have either restricted or banned tools
12:20
like ChatGPT internally after realizing what was happening.
12:24
Samsung learned the hard way.
12:26
In early 2023, 3 employees read an email lifting a previous Chat GPT ban and thought,
12:32
we’re home free. The first went and uploaded proprietary source code to debug this problem
12:38
they were having. Another copied notes from an internal company meeting into their chat feed,
12:42
while a third used Chat GPT to “identify defective equipment in a semiconductor line.”
12:48
Pretty soon, the ban was back. An internal survey quickly showed that
12:52
65% of Samsung employees thought AI tools were a security risk.
12:57
The most unsettling part of all this is what most people misunderstand about the
13:01
core disconnect between user privacy and the way AI systems are designed to work.
13:06
AI models are trained on massive datasets. On consumer plans like
13:10
OpenAI’s Free and Plus version, consumers allow the company to
13:14
use their conversations to train future models by default. Sure, you can opt out
13:19
by navigating a maze of settings. But most users aren’t even aware that toggle exists.
13:24
“Until you do, every document, every contract snippet, every
13:28
client detail you type becomes potential training material,” warns one report.
13:33
Once trade secrets are used in training or processing an AI model,
13:36
the damage can be effectively permanent.
13:39
You can’t just delete it. It’s like trying to unmix paint.
13:43
That’s what makes this far more sinister than a traditional data breach. Those are rare,
13:48
visible, and fixable, for the most part.
13:50
Compromising user security is normal for AI companies. HIPAA noncompliance
13:55
is built into the workflow. That’s why some security researchers are already
13:59
starting to describe this as the largest uncontrolled corporate data leak in history.
14:04
When you talk to the people inside these companies who are actually responsible for
14:08
what’s going on, they know exactly what’s happening. They know the controls to stop
14:12
employees from pasting sensitive data into their AI tools aren’t in place.
14:16
They know the risk.
14:17
But they also know there isn’t an easy fix.
14:20
You can’t just spin up a secure, in-house version overnight. That takes millions of
14:24
dollars of hardware, specialized teams, and time most companies don’t have.
14:28
In the meantime, the pressure to move faster doesn’t go away. People will still smirk at their
14:33
company’s “enterprise” version of CoPilot which only allows them to polish emails and go Google
14:38
searches. They’ll sneakily use their consumer version of ChatGPT, and their older executive
14:44
bosses won’t care for the most part. They still see AI models as “glorified search engines.”
14:49
If there’s one thing people should have realized years ago,
14:52
it was never to trust Silicon Valley bros with your intellectual property. Today,
14:57
everyone is riding that productivity wave…and hoping it doesn’t come back to bite them.
15:02
Chapter 5: The Great Correction
15:05
At this point, we know things aren’t adding up.
15:08
The system is expensive. It’s resource intensive,
15:11
leaking data like a sieve, and still not making consistent-enough money for its investors.
15:16
So why is the industry still all-in on the AI boom?
15:20
The answer is because from the inside, none of this feels like a choice.
15:24
It feels like an arms race.
15:26
The constant warnings to not “fall behind” or “let
15:29
China catch up” has everyone moving at an urgent pace. It feels existential.
15:34
But it really isn’t.
15:36
The real bottleneck of today’s so-called “AI arms race” is semiconductors, or chips.
15:42
And the companies selling those chips, especially NVIDIA and AMD, are making out just fine.
15:47
The global semiconductor industry has been going gangbusters for years now. In 2026,
15:53
the industry is expected to earn almost $1 trillion in annual sales,
15:57
an all-time high. Analysts are predicting annual sales of $2 trillion by 2036.
16:03
According to the CEO of Taiwan Semiconductor Manufacturing Company
16:07
(TSMC), the single most important chip manufacturer in the world,
16:11
demand for advanced AI chips is currently running at three times the global supply.
16:16
Every major tech company is trying to lock in as much capacity as possible, years in advance. New
16:22
factories in Arizona and Japan won’t meaningfully ease the market pressure until 2027 or later.
16:28
Let’s be clear. Chip manufacturers want companies to believe there is not enough
16:33
compute, that there will never be enough compute. If you don’t buy everything now,
16:38
you’ll inevitably fall behind in this computational arms race.
16:41
That message doesn’t need to be false to be powerful.
16:44
It just needs to be repeated often enough.
16:46
And it is repeated, on annual earnings calls, at conferences, even in front of Congress.
16:51
This is how you end up with firms ordering billions of dollars of GPUs years in advance,
16:57
locking in supply they may not even be able to fully use yet.
17:00
That’s how you get a market where demand is running far in excess of supply. And while
17:05
chip companies end up delirious with cash-filled pockets, consumers are left footing the bill.
17:10
For months now, the AI arms race has bled over into a full blown shortage
17:14
of specialized memory those chips need to function. People call it “RAMageddon.”
17:19
The result is a squeeze on everything consumers
17:22
need. Laptops, phones, and even appliances cost more.
17:26
And somewhere out there, an ordinary guy just trying to build a new gaming PC is
17:30
wondering how a couple of sticks of DDR4 suddenly cost more than his memory card.
17:35
Today’s AI systems are built on two assumptions. First, that AI demand
17:40
will keep growing fast enough to absorb all the infrastructure it requires. And second,
17:44
that productivity gains will arrive quickly enough to justify the cost.
17:49
If either of those assumptions slip, the whole premise of an AI boom starts to wobble.
17:54
We’ve seen this before. In the late 1990s,
17:56
companies built out massive internet infrastructure with fiber servers and
18:00
networks filling entire buildings on the assumption that demand would catch up.
18:05
Eventually, it did, but not before the market corrected.
18:09
Hard.
18:09
The NASDAQ lost around 76% of its value. Companies like Cisco, Intel,
18:14
and Oracle saw stock prices tank overnight. Other companies like eBay and Amazon barely
18:20
managed to survive. It took the NASDAQ index fifteen years to reclaim its previous high.
18:25
The internet boom wasn’t fake, but the timeline for dramatic
18:29
overvaluation and hype was. The market had to go on life support as a result.
18:34
The so-called “dotcoms” just couldn’t turn the profit their investors had
18:38
dreamed of when they poured their cash into start-ups in the 1990s.
18:42
It’s not hard to imagine a similar correction coming today. This time
18:45
for the largest companies in the world. The money fueling today’s
18:49
AI boom isn’t just venture capital chasing upside; it touches everyone.
18:53
When these cycles turn, the people making the decisions rarely take the hit. Executives
18:58
still get paid and early investors find the exit. The losses spread outward.
19:03
So when you hear “AI arms race,” it’s worth asking: race to what? Because right now,
19:09
it looks less like a race to the future and more like a scramble to justify billions already spent.
19:14
If the returns don’t show up fast enough, the fallout won’t stay in tech. It'll land everywhere.
19:20
And if you think the biggest risk is money, power, or data…you might be underestimating
19:25
the real problem: What happens when these machines decide a life is expendable? Find
19:30
out in AI Just Tried To Murder A Human To Avoid Being Turned Off. Or watch this video.
