TRANSCRIPTEnglish

Dangerous Catalyst for Nvidia & Chip Stocks [CRWV, NBIS, IREN] - as a Shareholder.

16m 22s2,949 words432 segmentsEnglish

FULL TRANSCRIPT

0:00

There's a whole lot of confusion going

0:01

on about what's going on with Nvidia and

0:05

renting back chips. No, they're not

0:08

buying back chips, but renting back

0:10

chips. What's going on with DGX Cloud

0:12

and what's going on with this material

0:14

service agreement with Nvidia? Because,

0:18

see, yesterday I made a video on this

0:20

topic and there were a lot of people

0:21

that seem to really misunderstand what's

0:24

going on. And I I think there's a chance

0:26

maybe they only watched 2 minutes of the

0:27

video, came up with a conclusion, and

0:29

then left, which is really their loss

0:31

because not a lot of people are actually

0:33

talking about this. I feel like once

0:35

again, I'm early and I'm one of the few

0:37

people talking about this. Like for

0:39

example, there's this one person said

0:40

yesterday, what are you even on about?

0:42

Are we reading the same filing? This

0:44

isn't proof of over supply. It's just a

0:47

contract clause where Nvidia shares some

0:49

downside risk if utilization drops.

0:52

Okay. Then I said, "The filing literally

0:55

says, Cory, we've entered into a new

0:57

order demanding Nvidia buy unused

1:00

compute." And the person replies and

1:02

says, "Yes, the filing says Nvidia has

1:05

to cover unused capacity." That part is

1:07

true, but it doesn't prove over supply

1:10

today. It just shifts some risk back to

1:12

Nvidia. Yes, but you have to realize

1:16

this is the first time we are hearing

1:17

about this with Nvidia. We never knew

1:20

that these agreements existed to the

1:23

point that in 2023 apparently we have

1:27

this material services agreement from

1:31

2023

1:32

which is now being activated. That's the

1:35

key, right? You have to understand it's

1:37

now being activated. Floreave isn't just

1:40

reminding you, hey, we have a material

1:42

service agreement where Nvidia is going

1:44

to buy our unused capacity. Clore is

1:47

literally saying, "Hey, Nvidia, we can't

1:50

sell this. We are potentially overs

1:52

supplied," which is implied when they

1:55

can't sell capacity. They have like,

1:57

"Hey, we we have too much. We can't find

2:00

buyers for this server technology. You

2:02

need to buy it." So, we are entering the

2:04

order for you to buy it. That is by

2:07

definition either a sign that Chlorine

2:10

sucks at selling, which is possible, or

2:12

there's over supply. though yes it

2:14

doesn't guarantee there's over supply

2:16

but it's an early canary it's an early

2:19

red flag and we'll get the full material

2:21

service agreement when the next earnings

2:23

call from coreweave comes out which is

2:26

expected to be between November 12th and

2:28

14th Baron says the 12th so I'm going to

2:30

go with the 12th to be safer on that

2:32

side now why is this such a big deal

2:34

well it's such a big deal because it's

2:35

also coming at the same time as Nvidia

2:37

is also agreeing to rent back from

2:40

Lambda uh over 18,000 of its own chips,

2:45

which you think, okay, well, that's

2:46

probably because they're building out

2:47

their own data centers, right? But

2:49

they've been scaling back their own data

2:51

centers. Now, this is so fascinating

2:54

because think about this. Nvidia has

2:56

been pitching the DGX cloud that, hey,

3:00

in August of 2023, Jensen tells us that

3:03

DGX cloud is an enormous success. Then

3:05

in late 2024, we hear that this enormous

3:08

success generates $2 billion of

3:10

annualized revenue. Well, that's less

3:12

than 2% of the money they make from

3:14

chips. Then in 2025, we're not going to

3:17

spend any more money building out our

3:19

own cloud infrastructure, which remember

3:21

cloud infrastructure is warehouse,

3:23

utility, power, infiniband, network

3:26

connections. It's a lot of

3:27

infrastructure, chip stacks, uh you

3:29

know, that could be partly their own. Uh

3:31

and then of course server rice. There's

3:32

a lot that goes into this and then you

3:33

have to manage it all, right? It's a lot

3:35

of work.

3:36

But now all of a sudden September of

3:38

2025 comes around and we hear from

3:40

Nvidia that they're no longer

3:41

prioritizing DGX for external use at

3:43

all. Instead, we'll just go lease from

3:46

other cloud service providers. And this

3:49

is where things and people start

3:50

scratching their heads and gohm this

3:52

feels a little circular because on one

3:53

hand you were telling us that DGX cloud

3:55

was such an enormous success. your own

3:57

NVIDIA data centers were such an

3:59

enormous success and now you're telling

4:01

us yeah we're not like that's that was

4:04

an enormous success and it was good

4:06

revenue but it's less than 2% we're not

4:08

going to focus on it you know we'll just

4:09

use it internally that's the argument

4:11

now is what we're just using our own

4:13

servers internally to design chips and

4:15

and to work on CUDA and in case you're

4:18

wondering why our commitments to leasing

4:20

chips are skyrocketing where we're

4:23

leasing our own AI chips back from the

4:25

likes of Microsoft Oracle and for reef

4:27

as reported by the information. Uh don't

4:29

worry, this is totally normal. We are

4:31

doing this because we need the

4:32

technology.

4:34

Okay? Or it's potentially because it

4:37

makes more sense to recognize the

4:39

revenue and then be able to buy back

4:40

your own stock because it makes people

4:42

inside and holders of the stock very

4:44

very rich. But it does create serious

4:46

concerns that this start of over supply

4:51

has begun. Now consider this when it

4:53

comes to this lambda deal. I want you to

4:55

think about this.

4:56

Nvidia agreed to rent 10,000 of its own

4:58

chips from Lambda for $1.3 billion over

5:01

four years and another $200 million deal

5:04

to rent another 8,000 chips. So

5:06

recognize how this works and why this

5:08

makes so much money. And this is why

5:10

people get pissed off when you call this

5:12

stuff out. It makes massive money

5:14

because Nvidia gets to recognize the

5:16

chip sales immediately. If you enter

5:19

into a chip purchase with Nvidia, Nvidia

5:22

gets to recognize the revenue

5:24

immediately. They recognize the Nvidia

5:26

and they re the revenue right away and

5:28

they recognize the margins right away.

5:30

So, it's kind of like you selling

5:31

lemonade for $100 and then promising to

5:35

buy back one cup of lemonade for five

5:37

bucks every year for the next seven

5:38

years. Okay. Well, you still only paid

5:41

back 45, but you got a hundred. You got

5:43

a 100 today and you gave back 45 divided

5:46

over the next seven years. It's a great

5:48

deal for Nvidia because Nvidia gets the

5:50

revenue today, recognizes the margins

5:52

today. It gives some of that back to

5:54

Lambda. Lambda goes public. When Lambda,

5:57

like Corewave, goes public, what do they

5:59

do? They raise money and then they can

6:00

use that money to buy more Nvidia chips.

6:03

It's extremely circular, right? That's

6:07

that's why people start wondering is

6:09

this is this bubbly? Because look at

6:11

what in Lambda is forecasting. Lambda is

6:14

forecasting that well, they're telling

6:16

us they had 114 million of revenue in

6:18

Q2. That's a little under half a billion

6:21

dollars of annualized revenue. They're

6:23

estimating to double to a billion

6:25

dollars of revenue by 2026. That could

6:27

be reasonable. Could be very, very

6:29

reasonable. But would it be reasonable

6:32

then to say we're going to 20x after

6:35

that? I don't know. Lambda is literally

6:39

estimating that they're going to 20x by

6:42

2030. But wait a minute. In order to

6:45

double, we're already relying on Nvidia

6:49

leasing back the very chips that we are

6:51

buying. That's to get to a double. Now,

6:54

we're supposed to expect that Lambda is

6:56

going to 20x from here. I thought this

6:59

random comment put it very well. When

7:01

there's no killer app, money needs to be

7:03

cycled around. And that unfortunately

7:06

seems to be a little bit of what's

7:07

happening here. And it's pissing off a

7:09

lot of people who have big allocations

7:11

to the chip place. And mind you, I have

7:13

a bunch of exposure to Nvidia as well. I

7:15

promise course members that when I sell

7:17

Nvidia, they're going to be the first to

7:18

hear. I will send out an alert to my

7:20

course members in our alpha report just

7:22

like we do when I buy stocks in the top

7:25

10 stocks to buy list. For example, you

7:27

know, I've been sending out alerts on

7:29

like, you know, there was a company that

7:31

just recently IPOed. I almost never buy

7:33

IPO companies. We bought this IPO

7:35

company. We did fundamental analysis

7:37

together. We sent out the alert to buy

7:39

it. That stock today that it's been like

7:42

a week now since it's IPOed. That

7:44

company today is up another 13%. This is

7:48

based on the fundamental analysis that

7:50

we do and the technical analysis and

7:52

those are the kind of calls that we like

7:54

to make. And then when I send an alert,

7:55

I send it to members in the alpha report

7:58

last week. You got I'm buying this

8:00

stock. I plan to buy more. I'm buying

8:02

this allocation now. And then I even

8:04

bought the dip on it soon after that.

8:07

And now it's way up from there.

8:09

because it, you know, was expected to be

8:11

volatile within the 24 hours after IPO.

8:13

Very normal. Now it's way up, which is

8:15

exactly what we expected. Anyway, my

8:17

point is like I'm not trying to be

8:20

negative on people making money. I want

8:22

people to make money. I want you to make

8:24

money. And as an Nvidia shareholder, I

8:27

look at this and I'm like, some of this

8:29

stuff makes me nervous. It's like you

8:31

guys are expecting me to believe that

8:33

you can 20x your data center, but you're

8:36

already forcing Nvidia to buy your

8:38

excess capacity at Lambda and at

8:41

Coreweave

8:43

and you're walking back from running

8:46

your own data center for external use

8:48

because you're realizing that's not

8:49

where the money is being made. And yeah,

8:51

we'll just use that internally, which is

8:53

a nice copout. But we also don't know

8:56

how much Nvidia uses themselves versus

8:59

rents out. Again, I'll give you an

9:00

example. Let's say Nvidia rents, you

9:03

know, back 100%. Uh, so whatever they

9:05

rent back, they rent back $10 billion

9:07

worth, right? Let's just make up, let's

9:09

call $10 billion 100%. Well, maybe they

9:11

rent out 30% of that to somebody else,

9:14

like they rerent it basically, and maybe

9:16

they use 10% themselves. They're still

9:19

losing, you know, 60 to 70% there. We

9:22

don't know what that number is, but it's

9:24

a red flag because it's likely to lead

9:27

pricing to go down over time. Now,

9:30

hopefully not, but it is likely to lead

9:32

pricing to go down. And that's the

9:34

concern that Nvidia actually went from,

9:37

in my opinion, being a company that

9:39

which is what made me so excited about

9:41

them. They went from a company that

9:43

could recognize the revenue and not be a

9:45

bag holder to literally committing to

9:48

being the bag holder. and this this

9:52

coreweave 8K. I think people are

9:54

misunderstanding it. Like they're

9:56

they're being like blind here to this

9:58

being a red flag. Like I can be bullish

9:59

Nvidia and I could love the margins of

10:01

Nvidia and I could love what Nvidia does

10:02

as a company, but I can also recognize

10:05

that this is a red flag and I can also

10:07

recognize that people are, you know,

10:09

shilling

10:11

without letting their brains realize

10:14

this is bad. You know, like somebody in

10:15

the chat earlier, this jackass in the

10:18

chat earlier, he's like, "Oh, Kevin, you

10:20

sure like it when they buy back their

10:21

own stuff, don't you?" And like, first

10:23

of all, you're you're acting like I'm

10:25

trying to be like sinister or like do

10:28

something greedy or whatever. I'm

10:29

providing you perspective as I usually

10:32

do on this channel. That hurts me,

10:34

right? As an Nvidia shareholder, me

10:36

pointing out a red flag of Nvidia hurts

10:37

me. And I don't shear the buybacks. I

10:40

actually see the buybacks as a red flag.

10:42

The company's so good. Go do buybacks.

10:44

Invest more into better equipment or

10:46

infrastructure, whatever. If the data

10:48

centers are doing so good, instead of

10:49

buying back the stock, go build data

10:51

centers. Do something productive with

10:53

the money. When you're buying back

10:54

stock, what you're doing is you're

10:55

trying to prop up the stock valuation so

10:57

insiders can get rich or shareholders

11:00

can get rich or you just end up using

11:02

that money to conduct equity investments

11:05

into the smaller data center place to

11:07

keep the ponds going. That's what makes

11:10

me concerned is that there are early

11:11

warning signs that there's a pawn. Now,

11:14

how do you play that? Well, in my

11:16

opinion was a trailing stop. Like, I'm

11:18

not I'm not like rushing to the exit on

11:20

the video, but I'm going to be aware of

11:21

this because where there's smoke,

11:22

there's fire. And and when it starts, it

11:25

usually doesn't get better over time.

11:27

Hopefully, it does.

11:30

But again, I want you to recognize that

11:32

sometimes you get these sort of like

11:34

bots. I mean, the comments that are just

11:36

like blind to real issues. Like there

11:40

was another one yesterday. Who was it?

11:41

Uh like I personally think this is a

11:43

pretty important inflection point to pay

11:45

attention to. Uh and then and then this

11:48

person I think they legitimately didn't

11:51

listen to the video at all, but they

11:53

make it sound like they're smart people.

11:55

Right. This video is theoretical at

11:58

best. Nvidia has not bought back any

12:00

chips. They're just insurance for

12:01

coreweave in a hits the fan

12:04

scenario. No, they're not. This is not

12:09

in a shitty hits the fan scenario.

12:10

Because if it's a shitty hits the band

12:12

scenario, guess what? It's happening

12:14

right now because Coreweave in their

12:17

filing says we are pulling the trigger

12:20

on demanding Nvidia purchase unused

12:22

capacity now. This is not some contract

12:26

clause for maybe. It is. We are doing

12:28

it. A B I never said Nvidia is buying

12:32

back chips. You didn't pay attention,

12:34

homie. Homie Nate. So your comment is

12:38

what's theoretical cuz you didn't even

12:40

pay attention to the content, which is

12:43

sad. And and I think this is where

12:45

people don't realize uh there are and

12:49

can be red flags. That doesn't mean that

12:53

we're going to fall off a cliff, but it

12:55

does mean we should pay attention to

12:57

these things. Now, Baron says that we

13:00

are going to get the copy of this MSA on

13:02

November 12th. I previously said that

13:05

was November 14th or 15th. Uh that was

13:08

based on when I thought the Cory Vern to

13:10

be. Baron says it's actually going to be

13:13

uh on the 12th. So, I highlighted that

13:15

right here. You can see that on screen

13:17

right here. So I I adjusted my calendar

13:19

and I encourage you to adjust your

13:20

calendar as well. But again my POV is

13:23

let me provide information you know to

13:25

to give perspective on on what could be

13:28

a real issue. We don't know if it is uh

13:32

you know everything could be just fine

13:34

but uh it it is something where what

13:38

this sets up is a precedent that anybody

13:41

in the future who is going to buy chips

13:42

from Nvidia deserves a guarantee from

13:46

Nvidia. But eventually you can only

13:48

provide so many guarantees, right? The

13:51

like the bar to provide guarantees is

13:53

going to get higher and higher and

13:54

higher. Soon your next data center plays

13:57

are going to go, "Hey, I don't want an

13:58

agreement that you're going to buy my,

13:59

you know, unused capacity through 2032.

14:02

If I'm going to spend $10 billion on a

14:04

new data center, I actually want you to

14:07

buy back my unused equipment all the way

14:10

through 2040." You know that that that's

14:14

where it's an issue. Jiro here says why

14:17

is coreweave buying more if they have

14:19

unused capacity. That's the thing.

14:22

Coreweave gets to be frivolous with

14:26

their investment because Nvidia

14:28

guarantees it. So why is Nvidia buying

14:31

more or why is Coree building out more?

14:34

Because they just dump it all on Nvidia

14:36

in a circular motion. So it doesn't

14:39

matter if there's unused capacity. That

14:41

is the definition of malinvestment.

14:45

That's crazy.

14:48

That's crazy.

14:49

So again, I'm not saying this is a

14:53

nearterm pooper dupers, okay? I'm saying

14:56

this is a longterm

14:59

uh potential issue. Not near-term. It's

15:01

a long-term

15:02

>> poopy dupee.

15:04

>> For now, it might very well continue to

15:07

be

15:07

>> bullish catalyst.

15:08

>> A bullish catalyst.

15:09

>> Bullish catalyst. And we like that. But

15:12

remember there's there there are some

15:14

things to remember about Kevin.

15:16

>> Kevin's somebody we consider Kevin is

15:19

fantastic too.

15:20

>> Kevin is very talented. But I don't know

15:22

it's going to be him, but he's a very

15:24

talented guy.

15:25

>> You know,

15:25

>> Kevin is much more interested than most

15:28

people by the way in the balance sheet.

15:31

>> I'm just saying. All right. So these are

15:35

things to pay attention to. Uh if you

15:37

want to know what I sell, I will be very

15:40

very clear about when I sell my Nvidia

15:42

stock in the M Kevin membership. Come

15:44

join us this way in the Me Kevin

15:47

membership. Every trade alert, all eight

15:48

courses, every private live stream,

15:50

every alpha report, you get it all uh in

15:53

the meet Kevin membership.

15:58

[Music]

16:03

>> You want to talk about love making,

16:04

right? Why not advertise these things

16:06

that you told us here? I feel like

16:07

nobody else knows about this.

16:08

>> We'll we'll try a little advertising and

16:10

see how it goes.

16:11

>> Congratulations, man. You have done so

16:12

much. People love you. People look up to

16:14

you.

16:14

>> Kevin Praath there, financial analyst

16:16

and YouTuber. Meet Kevin. Always great

16:18

to get your take.

UNLOCK MORE

Sign up free to access premium features

INTERACTIVE VIEWER

Watch the video with synced subtitles, adjustable overlay, and full playback control.

SIGN UP FREE TO UNLOCK

AI SUMMARY

Get an instant AI-generated summary of the video content, key points, and takeaways.

SIGN UP FREE TO UNLOCK

TRANSLATE

Translate the transcript to 100+ languages with one click. Download in any format.

SIGN UP FREE TO UNLOCK

MIND MAP

Visualize the transcript as an interactive mind map. Understand structure at a glance.

SIGN UP FREE TO UNLOCK

CHAT WITH TRANSCRIPT

Ask questions about the video content. Get answers powered by AI directly from the transcript.

SIGN UP FREE TO UNLOCK

GET MORE FROM YOUR TRANSCRIPTS

Sign up for free and unlock interactive viewer, AI summaries, translations, mind maps, and more. No credit card required.