Home
登录
注册
学习内容
Loading...
The Market Is Dead. AI is the Cartel - Video học tiếng Anh
听力练习
听力练习
/
Video
/
The Infographics Show
/
The Market Is Dead. AI is the Cartel
The Market Is Dead. AI is the Cartel
选择学习模式:
查看字幕
选词
重写单词
Highlight:
3000 Oxford Words
4000 IELTS Words
5000 Oxford Words
3000 Common Words
1000 TOEIC Words
5000 TOEFL Words
字幕 (153)
0:00
The average rent for a San Francisco apartment has hit $3,300. In some American neighborhoods
0:05
electricity costs have exploded by 267%. Even the RAM in your laptop has tripled.
0:11
And it’s all being driven by AI. So how can a software program in a server
0:16
room change the price of the dirt under your feet? To find out, you have to look at the land.
0:20
To run AI models like ChatGPT or Claude at a global scale, you don't just need
0:25
a few servers in a basement. You need a massive data center. These aren't office
0:29
buildings. They are vast facilities that can span hundreds of acres. It needs to be flat. It
0:34
needs to be close to high-voltage transmission lines. It needs access to fiber-optic trunks.
0:39
And that’s a problem. It’s the exact same land required for residential
0:42
housing. In the past, if a developer saw a plot of flat land near a city, it would be prime location
0:47
for real estate. They would build 300 units of starter housing. They would deal with zoning
0:52
boards, school districts, and traffic studies. Today, when a developer bids on that land,
0:56
they aren't competing with another local builder. They are competing with Amazon, Microsoft, Google,
1:01
and Meta - companies with cash reserves that dwarf the GDP of small nations. For the landowner,
1:06
it isn’t even a choice. And the consequences for
1:09
the local housing market are devastating. A residential developer needs financing.
1:13
They need distinct permits. They might back out. The tech giant? They pay cash. They pay
1:19
50 to 70% over market value. They don’t need a school system. They don’t need a sewer expansion
1:24
for 500 families. They just need the land. The result is visible in places like Loudoun County,
1:30
Virginia. They call it ‘Data Center Alley’ and the local government loves
1:33
it. They pull in nearly $900 million a year in tax revenue from the data centers.
1:39
But try buying a house there. A 2025 study from George Mason
1:43
University confirmed that these hubs drive prices house through the roof. It’s a paradox. Usually,
1:49
living next to an industrial noise factory lowers property values. Here, the influx of high-salaried
1:54
tech workers, combined with limited remaining land, drives intense competition. In New Albany,
2:00
Ohio, the arrival of Meta and Google didn’t just bring jobs; it brought million-dollar
2:04
listings that got snapped up and pushed the local middle class out of their own zip code.
2:09
The data center acts as a black hole for real estate. It doesn’t just occupy space - it
2:13
permanently removes land from residential use. Every acre covered in servers is an acre
2:18
where a starter home never gets built. That loss of land is bad enough. But
2:22
the way AI is reshaping the housing that does exist may be even worse.
2:27
Rent prices used to reflect supply and demand. A property manager would check the building
2:31
across the street, review vacancies, and set a price. If an apartment sat empty for two months,
2:36
that was money lost. The goal was full occupancy. But today, corporate landlords don't set prices.
2:42
AI-driven pricing engines do. These systems ingest millions of data points - competitor rates,
2:48
lease renewal dates, local income levels, and even tenant demographics. These algorithms
2:53
have discovered a cold, mathematical truth that a human landlord might have missed…
2:57
It’s often more profitable to have empty apartments.
3:00
If you have a 100 unit building and you raise rents by 10%,
3:05
you might lose 5 tenants. Normally that would be a problem. But AI crunches the
3:09
numbers and sees that the 95 tenants paying more outweigh the 5 who left.
3:14
So, the software tells the landlord to increase the price. It effectively creates an artificial
3:19
floor for rent prices across an entire city. When every major apartment complex in a region
3:24
uses the same software - and they do - it stops being a market and starts looking like a cartel.
3:30
Tenants face 5 or 6% annual hikes, regardless of whether the roof is leaking or the elevator
3:35
is broken. The price isn't based on the quality of the housing. It’s set by the
3:39
algorithm and the maximum it calculates the market can handle before breaking.
3:43
And this digital manipulation doesn't stop at your rent check. It’s actually changing
3:47
what you see before you even buy a home. This optimization extends to the sale process
3:52
itself. AI tools are now standard for "enhancing" property photos. It goes beyond brightening the
3:58
lighting. These tools can scrub power lines, generate lush green lawns over dead dirt,
4:03
and drop a sparkling virtual pool into a backyard that currently contains a rusted swing set.
4:09
This isn't just marketing. It forces buyers into bidding wars based on an illusion,
4:13
driving up perceived value and final sale prices. If the land grab slowly limits housing, the energy
4:19
crisis puts even more pressure on residents. And the numbers here are terrifying.
4:24
Five years ago, a standard server rack in a data center drew about 5 to 10 kilowatts of
4:29
power. It was like running a few hair dryers. A modern NVIDIA rack, designed specifically for
4:34
AI training, is a different beast. It requires over 100 kilowatts. That is a ten-fold increase
4:40
in power density. It is the thermal equivalent of parking a space heater in every square foot
4:45
of a warehouse and turning them up to maximum. The US electrical grid was not built for this.
4:50
It was built for a predictable world of air conditioners in the summer, lights at night.
4:55
It relies on “baseload” power - the steady, boring hum of electricity you can count on.
5:00
AI chips don’t sleep. They run at 100 % capacity, 24 hours a day, 7 days a week. In the
5:07
PJM Interconnection - the massive grid operator covering 13 states and 67 million people - this
5:13
new demand is eating away at safety margins. The grid operates on a simple principle:
5:18
supply must exactly match demand, every single second. To prevent this,
5:22
operators hold auctions to pay power plants to stay on standby. In 2024–2025, the cost of these
5:29
“capacity payments” jumped by over $9 billion. Who pays the $9 billion? It’s not Microsoft
5:36
or Google. It’s you.
5:37
The costs are passed directly to ratepayers as transmission and capacity charges.
5:42
In Maryland and Ohio, projections show monthly bills jumping by $16-18. That isn't because
5:48
someone left the lights on. It isn't because someone bought an electric car. Utilities must
5:53
build billions in new substations and transmission lines to feed the data centers. Regulations let
5:58
them spread that cost across all customers. It is a regressive tax. In Louisiana,
6:03
residents are already battling high bills from air conditioning and storm recovery. Now,
6:07
with a massive Meta facility drawing more power than the entire city of New Orleans, they are
6:12
effectively subsidizing the infrastructure for a trillion-dollar company’s chatbot.
6:16
Bloomberg’s analysis of wholesale prices in places like Baltimore confirms the link. The closer
6:22
you are to the data, the higher your rates. The International Energy Agency warns that by 2028,
6:27
data centers could consume between 6 and 12% percent of total US electricity. That sounds
6:32
small, but that’s the difference between a stable system and rolling blackouts.
6:37
The bottleneck isn’t power - it’s the hardware that delivers it.
6:41
Imagine you’re a city planner, ready to build a new subdivision. You have the land, the permits
6:46
and now you need a transformer. The transformer is the grey box that steps high-voltage power
6:51
down so your toaster doesn’t explode. A few years ago, you could order one and get it in months.
6:56
Today, the lead time is 3 years. Why? Because the tech giants have bought
7:01
them all. They are pre-ordering transformers by the thousands, effectively cornering the global
7:06
supply chain. They need them now, and they are willing to pay premiums that a municipal utility
7:11
or a housing developer simply cannot match. We are seeing the emergence of "ghost
7:16
neighborhoods” - rows of finished houses that sit empty for a year. Not because there are no buyers,
7:21
but because the developer cannot get the hardware to turn the lights on. The data center didn’t just
7:26
take the power; it stripped the industrial supply chain of the components needed to
7:30
expand the grid for anyone else. And it’s not just electricity.
7:34
You are now competing with servers for the very water in your tap.
7:38
High-density chips generate a large amount of heat. To keep them from melting,
7:41
data centers run massive cooling systems. A mid-sized facility can use 300,000 gallons
7:46
of water a day - the same as 1,000 households. In water-stressed regions like Arizona or Oregon,
7:52
this is a problem. Aquifers are draining. Farmers are seeing irrigation allocations cut. Meanwhile,
7:58
tech companies are securing long-term water rights, effectively privatizing the water
8:02
table. As water becomes scarcer, the utility cost rises for everyone. People are competing with a
8:07
server farm for the water in their tap. Eventually, it hits the grocery aisle.
8:12
This is where AI stops just using resources and starts driving prices up. Retailers are rolling
8:17
out electronic shelf labels - digital price tags updated instantly from a central server. Walmart
8:23
plans to have them in 2,000 stores by the end of 2026. On the surface, it’s about efficiency - no
8:29
more employees with sticker guns. In reality, it lays the groundwork
8:33
for price gouging in the physical world. We already see it with things like Uber. It rains,
8:38
the price goes up. Now, apply that to bread. Apply it to batteries during a hurricane warning.
8:43
AI sensors track foot traffic in the store. Say there’s a spike in demand on a Tuesday evening.
8:48
The price of pasta sauce goes up 30 cents instantly. It’s micro-inflation - invisible,
8:53
relentless, and automatic. This is yield management - maximizing revenue on every
8:58
item - applied to everyday essentials. McKinsey estimates these systems can
9:02
boost revenue by 15%. Where does that money come from? The consumer. Fixed prices are dying. Now,
9:09
prices are whatever the machine thinks you’ll pay. But it gets even more personal. The AI knows more
9:15
about your bank account than you might realize. Online, AI models analyze your browsing history,
9:20
your device type, your location, and your spending habits. They determine your "willingness to pay."
9:25
Two people can look at the same pair of shoes and see different prices. If the AI detects that you
9:30
are a high-spending shopper on a new device who rarely hesitates to buy, it shows you a higher
9:35
price. If it sees a bargain conscious user on an older device, you might get a discount. This
9:40
system erodes trust in the marketplace. You are no longer paying the market rate, but a personalized
9:45
price determined by your digital profile. And then there is the hardware itself.
9:49
You're paying a premium just for the scraps The devices we rely on to access AI are getting
9:55
more expensive. There is only so much capacity to make high-end chips, and building a semiconductor
10:00
factory takes 5 years and around $20 billion. You can’t just throw one up overnight. Right now,
10:06
chip makers are shifting away from the processors and memory used in everyday
10:09
devices to focus on the specialized chips AI needs - and where the profits are far higher.
10:15
And that affects the everyday consumer. Supply of standard RAM and SSDs is tightening.
10:20
Samsung has already hiked memory prices. The result is what some are calling “tech
10:24
stagflation.” The laptop you buy next year will cost more than the one you bought last year - but
10:30
it won’t be noticeably faster. You’re paying a premium just to get what’s left of the silicon
10:35
scraps after data centers have claimed the rest. And you’re not just competing with other companies
10:40
anymore - you’re competing with entire nations. Governments view computing power as a national
10:45
security asset, just like oil reserves or aircraft carriers. Nations like Saudi Arabia,
10:50
the UAE and China are pouring state funds into acquiring every available high-end chip. The
10:55
Saudi government is targeting $40 billion for AI investment. When a sovereign wealth fund enters
11:01
the market, price sensitivity disappears - they will pay any cost. This sets a permanent floor for
11:06
component prices, breaking the usual trickle-down effect where technology gets cheaper over time.
11:12
But perhaps the most insidious cost is the one you don’t see until it’s too late: insurance.
11:17
This is where the AI decides if your home is even allowed to have value.
11:21
For decades, insurance was about shared risk. Neighbors pooled resources to cover potential
11:26
disasters. A human agent might inspect your house, spot a problem, and give you a month
11:31
to fix it. Today, AI analyzes satellite imagery, scanning millions of properties in real time. It
11:37
can spot a single weathered shingle or measure a tree branch’s distance to a chimney within
11:41
inches - and it doesn’t issue warnings. Premiums spike or vanish without warning.
11:46
Your driveway, your roof, a single tree branch - suddenly, the AI decides you’re
11:51
too risky. In neighborhoods deemed “high risk,” homes become unbuyable. No insurance,
11:55
no mortgage. Without a mortgage, the house loses all value. The AI doesn’t see families or
12:01
communities. It sees numbers and probabilities - and it strips away the middle class’s most
12:05
important asset with relentless precision. Your rent will be set at the highest the
12:10
algorithm thinks you can pay. Your electricity bill helps power a server farm. Your insurance
12:15
could be canceled from a satellite scan. The systems of the future are here - but
12:19
right now, the cost falls on you. So the question is: how much more can the market
12:24
actually bear before it completely breaks? Now go check out Real Reason Humanity Is NOT
12:29
Ready for AI Superintelligence. Or click on this video instead.
