Major AI Data Center Warnings. Nvidia, Coreweave, SMCI

The question right now is, is Nvidia

orchestrating a Ponzi scheme with

Coreweave? Now, this is a really hard

opening for a stock and company that I

actually love. I'm a big fan of Jensen

Wong. I think Nvidia is actually selling

for a pretty reasonable valuation right

now. About a 1.3 peg, which is pretty

low given that they don't manufacture

their own chips. They just design them.

This is how they get to these 75% gross

margins which are amazing margins mind

you. You know they bring like 40 plus

percent to the bottom line. This is a

cashg generating machine. It's a

fantastic very rich wellpositioned moed

company with cuda. But what's

fascinating is there's some thought that

maybe is purposefully propping up

coreweave which we've made videos about.

I'll give you a little recap. to make

sure that the value of Nvidia chips

doesn't fall. Now, just a few days ago

in my Corey video, I showed you the clip

where Jensen Wong tells us that the

Hopper infrastructure H100 chips might

be difficult to give away after

Blackwell comes out, which is a nice

argument that Blackwell's going to be

really strong and good, but it's a

really bad sign for anybody wanting to

invest billions to hundreds of billions

of dollars into server infrastructure.

This is a potential very large red flag

that the market for artificial

intelligence is substantially slowing

down at least data center capex. Uh for

example, we heard the CEO uh or sorry

the chairman of Alibaba. We've already

covered this, but to reiterate the

chairperson of Alibaba telling us, hey,

we might be seeing a bubble in AI

infrastructure. Combine that with the TD

Cohen article where what do we hear? Oh,

Microsoft is scaling back some of their

investments in data centers. Now, that

wasn't exactly how they led their

article. TD Cohen's analysis actually

led by talking about Meta and Google

stepping in uh in place of Microsoft.

And Microsoft is still bullish LLMs.

They want to create their own LLM

software to compete with Open AI even

though they're investors in Open AI. Uh

but TD Cohen was bullish in that they

they actually see this this idea of a

data center pullbacks over at Microsoft

as maybe just a shift in priorities for

Microsoft. Now most people didn't talk

about that but the the article while it

did reiterate that yes Microsoft has one

walked away from 2 gawatt of capacity in

both the eur US and Europe in the last 6

months that were in process to be leased

it has also deferred and canceled

existing data centers in both the US and

in Europe last month. So in other words

this is why we got the news that there

were actually more

cancellations than we had first heard

because we first heard about this you

know a couple months ago and now we're

getting this news again. So it's not

just what's been happening in the last 6

months, it's also what's happening in

the last month and it's sort of getting

worse. Now their view is that the

pullback in new capacity leasing by

Microsoft was largely driven by the

decision not to support incremental

OpenAI training workloads. However, we

now now I'm going to throw in a note

here. At the same time, if Microsoft

doesn't want to support incremental

OpenAI workloads because they modified

their multi-year contract with OpenAI

and said, "Hey, go ahead and use other

companies like let's say a corewave and

which we'll talk about that at Nvidia

again in just a moment. This is very

important to it, although somewhat

tangential. Open AAI can now use other

companies other than Microsoft. This

thesis would make sense from TD Cohen,

but Microsoft has now announced that

they want their own LLMs rather than

just rely on open AI, which is

interesting because why are they

cancelling leases on data centers, old

and new, and walking away from deals if

they're also wanting to build out their

own LLMs? Well, it's likely because,

hey, they want the delicious data that

comes from having your own LLM, but they

don't necessarily need additional

capacity because capacity is becoming

cheap. And that's the problem of the

potential quote unquote Nvidia Ponzi. I

know it's a hard line. I'll explain why

I'm saying that in just a moment. So to

me, I don't think we can say AI demand

up and people use GPT or Grock or

whatever, therefore AI stock up. I think

the days of that are over. I think we

can all agree that we like using the

GPTs or the Grocks or whatever for

encyclopedic purposes, for

brainstorming, for ideiation, for legal

review, contract review, or just doing

stuff we don't want to do. This is

fantastic, but does it necessarily mean

we need even more data centers than

we've already built out? And the answer

here seems to be mostly and resoundingly

no. That we may have already hit some

form of capacity. That said, uh TD Cohen

actually think this is an thinks this is

an opportunity for Google to step in uh

or Meta fill in and backfill sort of

some of that canceled capacity. But

again, the point of that is, hey, wait a

second. If the other companies are

filling in, then then then you're still

building just like the chairperson of

Alibaba says, hey, it's still getting

built out even though we don't need it.

Somebody's picking it up. Somebody's

spending the money on it. It's probably

not the greatest ROI. So my argument is

while Google and Meta might pick up some

of these leases. I think Microsoft is

sending you a signal that the money

isn't in the data centers. The money is

in a collecting data from people. By the

way, who knows? Maybe Microsoft will buy

23 and me and get all your genetic data.

Which, by the way, in 2021, I got

interviewed by NASDAQ about 23 and me.

And I told them it was an overvalued

stock and you shouldn't buy it. You can

actually look that story up if you

Google NAS. I'll just do it right now.

NASDAQ meet Kevin Pafrath spa 23 and me.

I'm like no this this thing doesn't make

money. There it is. First line it is

right there. VG Acquisition Corp Spack

deal with 23 and me is two. Uh and then

it goes into like speculative is I think

what the full title is. You can go check

it out yourself. Uh but they did that

interview with me uh analyzing the spack

deal with Meet Kevin. But anyway, what's

really interesting

is we we have plenty of signals at the

moment that the data center buildout is

probably

overhyped. But now we have something

else interesting that's happening. We

have Nvidia saying they are going to buy

uh up to $250 million of coreweave stock

at IPO. Now, this is really interesting

because what that does is it creates a

floor. Just this morning, we heard that

Cororeweave is actually downsizing their

IPO from about $2.7 billion of expected

proceeds down to $1.5 billion. And in

addition to that, the price is going to

be lower than expected. Nvidia now, by

the way, coming in to backs stop this at

an anchor of $40 per share. Now, I find

this interesting because

Nvidia makes a lot of money selling to

companies like Coreweave, and this isn't

solely about Coreweave. It's about the

entire industry. Think about this, okay?

Let's just take this to the extreme for

a moment. I always think when you do an

analysis, you should assume the middle,

but study the extreme, right? So, in an

example, you lose your job and we go

into a recession and you can't make your

payments. What is the extreme solution

you're going to have? When you analyze

that before things go

poopy, you know how to prepare yourself,

right? Okay. That doesn't mean that's

going to happen. So, if Coree IPOs at

$40 and Nvidia doesn't back them and

Coreweave drops

to, I don't know, zero. Okay, let's just

go extreme. Go to zero. corewave goes to

zero. Well, what happens? Well, now all

of a sudden the value of every data

center in the country plummets, gets

written down because of a failed

Coreweave IPO. the industry cannot allow

Coreweave to fail, which in a weird way

is actually bullish for Coree's IPO

because you've got all these

institutions that have so much risk of

Nvidia stock going down. Think about any

VC who's investing in these private data

centers or even honestly XAI in their

Tennessee data center. If Coree went to

zero, the Tennessee data center that XAI

owns would be worth substantially less.

new data center projects would come into

question. And then guess what happens if

somebody's like, "Hey, Nvidia, I want to

buy 100,000 Blackwell chips to, you

know, invest in a data center." Oh crap.

But wait, the last company that did that

with Hopper chips. You guys bankrupted

them by coming out with too good of

chips too rapidly. Well, why the hell

would I buy Nvidia chips and go into

making a data center? You're basically

buying at peak bubble for data centers.

So it is industry critical that the

coreweave IPO goes well even if you

don't care about the core IPO if

Coreweave plummets. It's bad news for

Nvidia, AMD, and any of the relateds.

Super micro computer for the

infrastructure, Vertive, for water

cooling. I don't really care what it is.

Dell for the 4% of their business that's

exposed to AI data center servers

anyway. It's bad. So, this is where the

Ponzi thesis comes from. Let's take

other money and funnel it into Coreweave

to prop Coreweee up to make it seem

successful and then, you know, pay out

shareholders if the share price goes

down to make sure the whole system still

seems

sound. Meanwhile, it's propped up by

Fugazi just to prevent the entire thing

from from falling more substantially.

Now, I'm not here saying Cororee is

going to go to zero, okay? I'm just

saying even coreweave going down 20% and

let's say it IPOs. So usually what they

do is they'll IPO at 40 bucks, but

they'll probably be up at like $60 when

it first trades because they

purposefully restrict the shares. And we

know Nvidia is putting a $250 million

floor in at 40. So people will speculate

this up to $60 or $80 or whatever. This

is why you almost always see an IPO.

It's like it's up 50% on the first day

of trading. That's all part of the

marketing because the businesses want

the article that's like look at how

successful our IPO was. Even though

they'll raise less money because they're

not going to benefit from that run. It's

the people buying it who will benefit

from that run. Cororeeve will raise less

money. It'll still be good enough

because remember they've got about two

to that's like 2.8 to3 billion of bills

coming due this year. They don't have

the money. $ 1.5 billion is going to

help them and then they can always issue

more shares in six months. Although in

six months that's when your insiders are

going to have a lock up expiration. So

you probably are going to be like more

like 3 months when before they start to

raise money. And the issue with that is

how much is that stock potentially going

to fall and what kind of downstream

effects is that going to have to other

stocks. So, I'm not saying because I

think there's going to be a lot of

support for trying to keep corewave

propped up because it's so industry

critical. So, I'm not saying the stock

is definitely going to go down. I'm not

saying you should, you know, short at

day one because you could get totally

hosed by, you know, what I think is at

play here. But my point is what I would

encourage is watch the financials of

Coree and the price action because the

price action of Coreweave and the

financials of Cororeweave will give you

a really good barometer as to the future

of investing in chip plays like Nvidia.

So if I'm let's say a a future Nvidia

buyer, what am I looking for? Well, I'm

looking at, hey, if Coreweave

plummets, maybe I'm less interested in

Nvidia at today's price and I'll wait

for it to be $80 or $60 or whatever. If

Coree takes off, it stays up, they do

their secondary offering, it keeps

cranking. 6 months from now, they have

their insider transactions that unlock

and after those unlock, they're still

doing well and they're able to raise

money and invest more into data centers.

Great. that's supportive of Nvidia

stock. So, think about this as sort of a

uh a canary, if you will. The canary has

arrived to the proverbial coal mine. Why

not advertise these things that you told

us here? I feel like nobody else knows

about this. We'll we'll try a little

advertising and see how it goes.

Congratulations, man. You have done so

much. People love you. People look up to

you. Kevin Pra there, financial analyst

and YouTuber. Meet Kevin. Always great

to get your take.