HomeVideos

Why Greenland Won’t Fix America’s China AI Problem

Now Playing

Why Greenland Won’t Fix America’s China AI Problem

Transcript

358 segments

0:00

By the end of this video, you will

0:01

understand exactly why the US wants

0:03

Greenland, how it connects to AI, China,

0:06

[music] and Silicon Valley. Stick

0:08

around. But before we start, this

0:10

episode is brought to you by Href's

0:12

marketing platform, and their feature

0:14

called Brand Raider, [music] which helps

0:16

businesses grow in the AI era. You know

0:18

how businesses are obsessed with Google

0:19

rankings and SEO, but have you actually

0:22

checked if your business is showing up

0:24

in Chad GBT or Perplexity or Gemini? Is

0:28

it being mentioned in YouTube summaries

0:29

or Reddit threads? And if it doesn't or

0:32

you're not sure how it shows up in the

0:33

AI search, it's a problem because the

0:36

search behavior has changed forever.

0:38

People use AI to research everything

0:40

these days, whether they want to work

0:42

with you, who to buy from, what to

0:44

invest in, and if you're not in those

0:46

results, you're invisible. Brand Raider

0:48

shows you exactly how your brand appears

0:51

across AI search engines, what people

0:53

are seeing when they ask about you, what

0:56

percentage of AI chats mention your

0:57

brand, find out what AI says about your

1:00

brand, and this is the key part, what

1:03

you need to fix to actually show up. You

1:05

can spot AI visibility gaps, see what

1:07

content is being picked up by AI, track

1:09

AI answers, and position your brand

1:11

around prompts instead of keywords. If

1:14

you're running a business or a brand and

1:16

want AI to know about you, check out

1:18

HREFs. And by the way, Brand Raider is

1:21

available to all free and paid ATFre

1:23

users. All right, let's get back into

1:25

the video. It's 1946.

1:28

World War II just ended. President

1:31

Truman receives a classified memo from

1:33

his Secretary of State with the subject

1:34

line, Greenland.

1:37

The memo is blunt. We need to buy it

1:39

from Denmark. For hund00 million, all

1:42

cash. The Cold War is starting. The

1:45

Arctic is the shortest route between the

1:47

US and Soviet missiles. [music] And

1:49

Greenland sits directly in the flight

1:52

path. Whoever controls Greenland

1:54

controls one of the most critical pieces

1:56

of the warning system for nuclear war.

1:59

Denmark says no. Fast forward 80 years.

2:03

President Trump. Same island, same

2:06

[music] interest. Except this time, it's

2:08

not so much about the missiles. It's

2:11

about this, the magnets. [music]

2:14

The magnets that cool AI data centers.

2:18

Because somewhere between Truman's cold

2:19

war and today's AI boom, [music] the US

2:22

outsourced the entire supply chain for

2:25

the thing that keeps these machines

2:27

running. And now China controls it,

2:31

almost all of it. The only problem is

2:33

that even if the US took Greenland

2:35

tomorrow, it most likely would not fix

2:38

the China problem at all. [music] Today

2:41

is about Greenland in the context of

2:43

tech and AI, rare earths, and the next

2:47

US China tech warfront.

2:49

Let's dive in.

2:52

What happens when you generate an AI

2:54

image or ask Gemini a question? Your

2:57

words are turned into vectors. that are

3:00

fed into a giant probability machine.

3:02

But what else? You're activating a

3:05

pretty complex supply chain. And this

3:07

supply chain involves rare earth

3:10

permanent magnets. A single chip uses

3:14

about.3 W hours. [music] A typical

3:16

Google search is roughly a thousand

3:19

times less. All of that extra energy

3:22

turns into heat inside a data center.

3:25

And this heat doesn't dissipate on its

3:28

own. Every watt of compute generates

3:30

heat that must be actively [music]

3:32

removed and that removal depends on the

3:35

cooling systems. The cooling systems are

3:38

packed with neodymium iron boron

3:40

magnets. And this is what they look

3:42

like. The magnets are used in cooling

3:46

fans and high density racks for rapid

3:48

thermal response for basically cooling.

3:51

[music]

3:52

And the problem is that these very

3:54

magnets that cool down data centers are

3:57

manufactured almost entirely

4:00

in China [music]

4:01

and they're sitting under Chinese expert

4:04

controls that tightened pretty

4:06

drastically just last year in December.

4:09

Modern AI depends on models, chips,

4:11

[music] and data centers. For a data

4:14

center to function when it's got power

4:15

dense GPUs, it [music] needs a massive

4:18

cooling and power stack. rare earths are

4:22

inside that stack at multiple layers

4:25

from cooling fans to pumps to storage to

4:27

transformers to generators. So when you

4:29

hear new AI data center project in

4:32

Texas, [music] you also have to

4:34

understand that that means that the

4:36

demand for the magnets is scaling as

4:38

well. So what does Greenland have to do

4:41

with these magnets, cooling systems, and

4:43

AI? Well, just about everything.

4:49

It is important to understand the

4:50

historical context here. In 1985, the

4:54

United States controlled approximately

4:56

60% of global rare earth production and

4:59

most of it came from Mountain Pass mine

5:02

in California's Mojave Desert. China, on

5:05

the other hand, produced about 30%. The

5:08

United States was the world leader in

5:10

rare earth mining, processing, [music]

5:13

and magnet manufacturing. They dominated

5:16

the entire vertical from [music] ore to

5:19

finished product. But then the tables

5:22

have turned. Throughout the '9s and

5:24

2000s, China poured a lot of state money

5:26

[music] into dirty rare earth processing

5:30

as well as the magnet plants that

5:32

western companies were happy to shut

5:34

down and offshore to China. [music]

5:37

So by the time anyone noticed, nearly

5:39

the entire supply chain had migrated to

5:42

China. [music]

5:43

Starting 2006, for nine years in a row,

5:45

China weaponized its control, they

5:48

restricted exports and cut supply by 40%

5:50

in 2010. And then during a maritime

5:53

dispute with Japan over a detained

5:55

fishing boat captain, they halted rare

5:58

earth shipments to Japan entirely for 2

6:00

months. Japanese manufacturer panicked

6:03

because Japan was 90% dependent on

6:05

Chinese imports at the time. The Japan

6:08

embargo was a watershed moment because

6:11

for the first time [music]

6:13

the world saw China willing to use rare

6:16

earth supply as a geopolitical weapon.

6:18

[music]

6:19

In turn, Japan aggressively diversified

6:21

and reduced Chinese dependence from 90%

6:23

[music] to 60%. And cut total rare earth

6:27

consumption in half. The 2010 China

6:29

Japan embargo was a wakeup call for

6:32

everyone. And in 2014, the World Trade

6:35

Organization ruled that China's export

6:37

quotas violated trade rules because

6:39

China couldn't justify them as

6:41

conservation or environmental

6:42

protection. China had to abolish formal

6:45

quotas, but they retained export

6:47

licensing and fully controlled

6:50

processing of rare earths. And 8 years

6:53

later, as you know, the magnet

6:56

dependency collided with the AI boom.

6:59

2022 OpenAI launches Chad GBT that

7:02

became as you all know the fastest

7:04

growing consumer technology in history

7:07

and that triggered the current AI

7:09

infrastructure buildout. The AI boom

7:12

created demand for data center capacity

7:14

that we have never seen before and that

7:16

is how the rare earths re-entered the

7:19

chat because AI workloads generate

7:21

vastly more heat than traditional

7:24

computing. Server racks jumped from 10

7:26

kows to 50 kowatts, some exceeding 100

7:29

kow, where every watt of compute

7:31

requires cooling. And modern cooling

7:34

systems, high efficiency fans, liquid

7:36

circulation pumps rely on neodymium iron

7:39

boron magnets. [music] And they do so

7:41

for their size and energy efficiency.

7:44

Meaning that on top of the existing

7:46

demand, we're looking for a ton more

7:49

magnets, which means a ton more

7:52

dependency on China. And once again, I

7:55

will remind you that 90% of these

7:56

magnets are produced and processed in

7:59

China. And if you're the US and you are

8:01

in direct confrontation with China,

8:04

you're going to want to deal with this

8:05

sooner rather than later. In the context

8:07

of AI and tech, the US needs Greenland

8:10

because the rare earth dominance matters

8:13

for the US. [music]

8:14

And it matters because the current

8:16

situation gives China leverage over the

8:18

physical hardware that AI runs on. You

8:21

may ask why Greenland? Greenland holds

8:25

1.5 million metric tons of rare earth

8:27

reserves which is enough to supply

8:29

meaningful share of the US demand and

8:32

reduce dependence on China. But

8:35

Greenland alone can't fix this problem

8:38

[music]

8:38

because the problem isn't in the mining.

8:44

The common narrative frames rare earths

8:47

as the mining problem, [music] but it's

8:49

not. China produces most of global rare

8:52

earth ore, but that's the least

8:55

important part of the equation. They own

8:57

around 70% of global mining and they

9:00

process around 90. Processing being the

9:05

operative word. There are several

9:06

reasons why Greenland cannot fix this

9:08

problem. Reason number one, the

9:10

processing mode is not solvable by land.

9:14

Think of it this way. the US or Europe

9:16

can design its own servers, can design

9:18

their own data centers. But if the

9:19

cooling fans or pumps use Chinese-made

9:22

magnets, they're exposed to Chinese

9:24

export licensing by definition. And

9:26

Chinese licensing is now the default.

9:30

Trump's logic goes like this. If we

9:32

control more rare earth ore, we're less

9:34

dependent on China. But the problem is

9:37

that the ore isn't the bottleneck. The

9:40

biggest moat is processing of the ore.

9:43

[music] If the United States acquired

9:45

Greenland tomorrow, it would solve

9:47

exactly 0% of the rare earth problem.

9:50

Because Greenland's deposits won't

9:52

produce significant ore for 10 plus

9:54

years. And even after mining starts, the

9:56

ore would still need Chinese processing,

9:58

the very dependency Greenland was meant

10:00

to fix. This is why the bottleneck is

10:03

the processing capacity. [music]

10:04

Processing being a very specialized

10:06

chemical engineering expertise. For

10:08

example, rare earth separation

10:10

chemistry, which is taught at maybe a

10:12

dozen universities globally, hazardous

10:14

acid processing, radioactive waste

10:17

handling, toxic waste handling. There's

10:19

about 2,000 tons of toxic waste per ton

10:22

of rare earths. Now, in the interest of

10:24

being fair, the US wouldn't be so

10:26

interested in Greenland if there was

10:28

nothing more to it than just the ore.

10:30

So, why does the US want it anyway?

10:32

[music] A ton of land. It's a ton of

10:35

physical location for data inensive tech

10:38

and data centers and subsequently free

10:41

cooling. On top of it, raiders,

10:43

satellite links, ground stations, space

10:46

awareness, early warning and missile

10:47

defense. As space becomes more

10:50

contested, permanent control of key

10:52

ground sites becomes more valuable. But

10:55

the paradox is that companies can

10:57

already build and lease in Greenland.

11:00

the ownership of the land, the ownership

11:02

of the country would only change the

11:04

political risk for Washington. So in

11:07

this equation for the US, this is a lot

11:10

less about technology or or minds

11:13

themselves. It's a lot more about the

11:16

nervous system of the western tech and

11:18

American defense. Reason number two, the

11:21

timeline is unachievable even under

11:24

extremely favorable assumptions. If

11:26

Greenland gets acquired or annexed in

11:29

2026, uranium ban repealed, full funding

11:32

secured, environmental permits

11:34

fasttracked, commercial scale rare earth

11:37

production [music] lies 7 to 10 years

11:39

into the future, and processing capacity

11:41

to refine that ore is another 3 to 5

11:45

years beyond mining. The US is

11:47

attempting to build this capacity, but

11:49

it's going to take another decade and

11:51

even then they'll process just a

11:53

fraction of the [music] Chinese

11:55

capacity. Meanwhile, AI infrastructure

11:57

deployment happens in quarters and it's

12:00

happening right now. Microsoft Stargate

12:02

project is underway now. [music]

12:05

OpenAI is openly talking about another

12:07

order of 10 gawatts of power for future

12:09

models which is roughly a New York Citys

12:12

sized grid just for one company and they

12:15

need it now. Data centers will consume

12:18

around a,000 terowatt hour by the end of

12:21

the year. Greenland [music]

12:23

can't solve a problem that manifests in

12:25

2025 with production [music] that starts

12:28

in 2023. And finally the recycling

12:31

problem.

12:34

The recycling problem is the elephant in

12:36

the room that gets a fraction of

12:38

attention compared to Greenland. In

12:40

April 2025, several companies, including

12:42

Microsoft, piloted a hard disk recycling

12:45

project and announced the successful

12:47

recovery of rare earths from

12:48

approximately 50,000 pounds of shredded

12:51

and decommissioned hard disk drives, as

12:53

well as mounting hardware collected from

12:55

Microsoft's data centers across the US.

12:58

The project recovered around 90% of the

13:01

rare earth materials, recaptured about

13:03

80% of the feed stock mask, including

13:05

gold, copper, aluminum, and steel, and

13:08

tested an acidfree recycling process

13:10

that can cut greenhouse gas emissions by

13:13

around 90% compared to traditional

13:15

mining. But ironically, rare earth

13:18

recycling rates sit below 10% in the US.

13:21

[music] And if this program were to

13:22

scale globally, recycling could return

13:25

significant volumes of recovered rare

13:27

earths to the US without requiring new

13:30

ore extraction or new territories. And

13:33

the chemistry of this process is very

13:35

straightforward compared to mining. The

13:37

magnets are already separated from ore

13:39

and already in purified metallic form.

13:42

But instead of thinking about recycling,

13:44

we're choosing to think about

13:45

acquisition of new territories.

13:47

Conclusion.

13:49

So, is buying Greenland worth it from a

13:53

tech perspective? It depends. Does

13:56

buying the country solve the magnet

13:58

bottleneck problem? No. But does it

14:00

provide control over Greenland's

14:02

long-term tech advantage, including

14:04

space? Yes. Is it worth a geopolitical

14:07

blast? Not so sure. Technologically,

14:09

Greenland is absolutely valuable as a

14:12

critical mineral province, as a

14:14

potential green AI hub, and as a space

14:16

node. But the methods matter. By 2025,

14:20

China dominates rare earth mining as

14:22

well as the entire supply chain. The

14:26

ore, the processing, the chemistry, the

14:28

magnets themselves. And you can't buy

14:30

your way out of that with a real estate

14:32

deal. We hope this was helpful. Let us

14:35

know what you think in the comments.

14:36

We'll see you next time.

Interactive Summary

The video discusses the renewed US interest in Greenland, linking it to the burgeoning AI industry and the US-China tech rivalry. While the US historically sought Greenland for Cold War defense, today's interest stems from the critical role of rare-earth permanent magnets in cooling energy-intensive AI data centers. China currently dominates the global supply chain for these magnets, particularly in processing and manufacturing, having weaponized its control in the past. Greenland possesses significant rare earth reserves, but acquiring it would not solve the immediate dependency on China, primarily due to the complex and specialized processing bottleneck and the long timelines involved in developing new mining and processing capabilities. The video argues that the US's interest in Greenland is more about long-term strategic control for defense and future tech infrastructure than just rare earth ore. It also highlights the overlooked potential of rare earth recycling as a more immediate and environmentally friendly solution to supply chain issues.

Suggested questions

5 ready-made prompts