$40 BILLION AI Company JUST went INSOLVENT | The AI Bubble.

the AI bubble. We would be in quite the

[music] recession if it were not for AI.

I think AI completely saved the stock

market. It's kind of like Paul investing

$100 in John, who invests $100 in Bob,

who invests $100 in Paul.

>> The AI bubble popping. You might have

been watching your stocks today and

wondering [music] that

>> the threat of a bubble.

>> An AI bubble burst.

>> Reality check could be coming for AI

valuations. [music]

>> Another day of tech. What if I told you

a $25 trillion debt bubble was about to

burst? And even worse that every

investor in the world owns these stocks.

>> A massive artificial intelligence

company just released a memo in the

background for nobody to see that

suggests they face near certain

insolveny unless they get bailed out

from other players in the AI space. In

this video, we're going to break down

exactly what was just revealed in this

filing. We're going to talk about the

circular nature of artificial

intelligence, which we've talked about

before, but we're going to update it

with this filing that was just released.

We're also going to link it to what's

going on with Nvidia. Nvidia just

announced massively more efficient

Reuben chips. And we also have a lot to

say about the high bandwidth memory

cycle which is related to Reuben but

also makes the very problem about

potential AI insolveny

worse. Then we'll finish by talking

about what just happened with Elon Musk

because he told us that one thing was

false and it only turned out to be

substantially more true than he let on.

[laughter] We'll talk about all of it in

this video. So hey everyone, me Kevin

here. First things first, we got to

analyze the cycle. The coreweave cycle.

Yes, Coreweave just released an 8K and

Coreweave is really interesting because

as we know Coreweave, much like other AI

data center plays, whether it's iron,

NBS, Oracle, whatever we know that

companies like Cororeweave receive

investor money, money from people who,

you know, want to make money from

artificial intelligence. Coreweave goes

and buys artificial intelligence chips

like Nvidia chips or CPUs from AMD or

otherwise uh which of course are reliant

heavily even more so now thanks to an

increase in inference demand in high

bandwidth memory and then of course time

goes on because everybody knows

construction and a clock are basically

synonymous because it always takes

longer than we expect. Then we get data

centers and hopefully eventually we get

money and then hopefully eventually we

could repeat the cycle. This is, you

know, kind of just a happy data center

cycle. It really shouldn't even be

called the corewave cycle, but there's a

little bit of a problem. See, this chart

right here is the chart of Oracle's

credit default swaps, which a lot of

people track Oracle because they see it

as a safer version of Coreeave. And so,

if Oracle is sneezing, then Cororeweave

must really be having problems. But

anyway, back when we had this spike over

here in the price of credit default

swaps, basically insurance against the

collapse of Oracle, we were at the

highest level since 2008. Those have

since spiked up to about 155, settling

down into the low 140s, but then

recently ticking up again. And I suspect

they'll tick up even more tomorrow. Why?

Because of a filing that just came out

from Coree. And it's not my favorite

kind of filing. These AKs from

Cororeweave are pretty glorious at

telling us that there's probably a fire

cooking somewhere and nobody wants to

talk about it right now. Everybody's

just closing doors around the house and

like, "Well, no smoking here. What are

you talking about? We're good.

[laughter]

We're not good." Okay, so take a look at

this entry into a material definitive

agreement. Basically, coreweave here is

telling you, "Hey y'all, we got some

debt. Uh we might have a lot of debt.

our in fact our weighted you know

average interest is somewhere around

10%. So we spend a massive amount of

money on interest. We'll actually go

through the financials in just a moment.

Uh and we just want to let everybody

know that this this new set of debt that

we entered into in the summer of 2025

called our DT [music] uh or DDTL oo 30

credit agreement uh is actually now

being revised. And here's how we're

revising that agreement. So we got

together with all these different banks

including MUFG and US banks and we

decided that you know what we're going

to actually reduce and change a few

things. The first thing that we're going

to do is we're going to reduce the

minimum cash that we need to or are

required to have on hand to just $100

million. In other words, we need the

cash for other things. You'll see in

just a moment what those other things

are, but they're basically saying we

need some more cash. Now, they're

arguing that this is just to sort of

align with the timing of deliveries

because there have been some

construction delays. But what's

interesting is we actually go a little

deeper here. We're also now going to

postpone the initial testing date of the

DSCR, which is basically a ratio that

says, "Hey, like how much money are you

guys making compared to how much debt

you have?" And the banks are basically

saying, "Hey, you know what? Um, we're

just going to go monkey no evil until

Halloween of 2027. Okay, we just don't

even want to look at the financials. So,

here's another 18 months. Go f off.

Please just keep paying the bills.

Please, please. We'll we trust you, bro.

That's essentially what the banks are

saying here on billions of dollars, mind

you. Uh, got to keep the Ponzi going

somehow. But anyway, then they're also

going to postpone their initial testing.

So, like, hey, you know, before we go

like do a deep dive on these numbers,

can we just like see how many of your

contracts you're actually fulfilling?

So, rather than dividing like total

revenue and and debt into each other to

get a DSCR, they're going, "Hey, uh, how

how many of your contracts are actually

working out?" And the bankers are

basically saying, "Uh, yeah, you know

what? Before we even take a look at your

contracts, why don't we delay that to

February [clears throat]

28th? We're just going to close our eyes

on that one, too." And if for whatever

reason uh we end up looking on February

28th and if for whatever reason you all

still suck and you all still can't get

your act together,

you know, we'll just we'll just turn a

blind eye to any equity partner giving

you guys a money injection. And if they

inject money, we'll just say monkeys see

no evil. That's that's just what we'll

say, folks. They literally wrote it.

They wrote it right here. The First

Amendment also permits, you can't make

this up, an unlimited number of equity

cures for any failure to satisfy the

debt service coverage ratio and contract

realization ratio covenants prior to

October 2026.

[laughter] So, in other words, as long

as Daddy Jensen Hong from Nvidia comes

around and gives you a bailout, we don't

give 2 F that you're violating the

covenants of the billions of dollars of

debt we lent you cuz Daddy Jensen's

bailing you out. in doing this to some

extent some people actually look at this

with a really bearish lens and they're

like oh my gosh even the bankers realize

if they don't put up the sort of balloon

and like hey Jensen can you get ready to

bail us out the bankers might not care

[laughter]

that's but that's a jaded view you know

why do we want to be really jaded about

an unlimited right to cure violations in

your loan terms you know that it'd be

crazy to be jaded about that I mean come

on Nvidia has basically agreed to buy

6.3 billion dollar of Cororeweave's

unused data center capacity. So like if

if Nvidia has this agreement to buy $6.3

billion of unused data center capacity,

which Core has already asked Nvidia to

do by the way on some of their unused

data center capacity, which is odd

because it's sort of like how does

Cororeef have unused data center

capacity and they're still building a

bunch. More on what they're building in

just a moment. Uh spoiler alert, like

30% of their assets are not even built

yet. That's a lot of money. That's like

$6.9 billion of money. that's just sort

of like in the construction process

that's not going anywhere. But but yeah,

I mean like to actually get Nvidia to

fulfill the rest of the purchases,

Cororeef has to finish building. See,

that's all has to do with this material

services agreement which I have on the

screenshot right here. This is the

agreement uh an agreement that was made

on April 10th for an initial value of

$6.3 billion. It's basically Nvidia

guaranteeing that they'll buy unused

compute from Coree to try to help get

Coree a head start. So, Daddy Jensen's

already bailing them out on one hand

once the stuff is built, but now the

bankers are basically saying, "Hey, uh,

we know Daddy Jensen's going to bail you

out once your stuff is built. Can you

also get Daddy Jensen to maybe like

start hinting that he might have to bail

you out before the stuff is built cuz

you guys are falling behind?"

Yeah, that's not good. Now, now I know a

lot of us are like, "Bar Kevin, SanDisk

and and Micron are going to the moon."

Correct. They are. and their numbers are

great. We've done a lot of analysis on

them in the Meet Kevin membership. You

could see that in the stocks tab in the

courses. Uh we did a analysis on Bloom

Energy this morning as well, which was

phenomenal. They've actually got a

really good balance sheet. This is it.

And you know, when you jump in over on

the right side, you can see all of our

different analysis, whether it's on uh

Coreweave or we did SanDisk or Micron or

whatever. I mean, Sandis frankly has a

fortress fortress balance sheet and

massive cash flow. But more on them in

just a moment. Like I I know people like

Kevin Kevin, how how could this be bad?

Like memory is is doing so well. Okay,

spoiler alert. There's there's something

called a cycle and it is all built out

via innings. I'm going to explain all of

this on the screen in just a moment, but

I'll I'll point out exactly where we are

and how this works in just a moment. But

first, we have to understand the actual

numbers at Coref because I want you to

actually see what they're facing. See,

versus what they're telling us. Coref is

telling us, guys, everything is fine.

I'm going to pull it up here. We're only

blaming the weather. In their last

earnings call, if you went through their

last earnings call, you'd see that

they're just blaming the weather. It

rained a lot in Texas. And because of

that, here we go. We got a roughly, you

know, heavy rains and winds caused a

roughly 60-day delay in a city north of

Dallas, preventing contractors from

pouring concrete. So, hey, come on,

guys. We just need a little bit more

time and everything will be fine. That's

the idea here. As Cory was trying to

tell us, everything's fine. But when we

actually look at their financials, we

start facing,

let's just say, some red flags. So,

first that we should look at right here

is their cash flow statement. And what

we're going to find is that their free

cash flow is $1.6 billion per quarter.

If you subtract this out and divide it

by three, you get the quarterly number

since it's 9month. Basically, we are

bleeding $500 million every single

month. $500 million every month. 4.75

billion over 9 months. And these are

based on September financials. Now, if

you actually subtract out depreciation

off of this, you'd be losing another

$167

million per month. So, we're blowing a

lot of freaking money here. Obviously,

the cash flow situation sucks. But where

do we sit in a debt situation? Well,

that doesn't look good either. See, when

we look at their cash flows, we can see

that they've raised $7.5 billion of

debt. And their repayments are somewhere

around 2.9. So, so their net borrowings

here were just about $4.75 or so, which

happens to be exactly what they need

over a 9-month period. So, they're

borrowing because they need to. And then

they had IPO proceeds in there as well.

Fine, great. But what's the problem

here? The problem is, to me, it's really

a gamble on steroids. You know, it's not

like house hack in my opinion. This is

simple real estate where you're buying

homes like we don't have any debt on our

houses or our land or whatever because

we got a little land we're developing as

well. We're not buying homes on 30-year

mortgages even, right? But let's say we

were let's say we were buying houses in

desirable areas people could buy up and

live in and then rent out, right? Like

so we buy them, we fix them up, rent

them out, people live in them. Great.

30-year mortgage sustainable debt,

right? They're buying chips in a desert.

They're not even able to rent them out

right now because 30% of their assets

are sitting unused and they're not using

30-year fixed rate mortgages. They're

using very short-term debt. In fact,

$1.667

billion that they owe is due right here

in just $2.25 years. It's weird to say

that March of 2028 is only in 2 and a/4

years, but they have a 15% loan due in

March of 2028 at $1.6 billion. They got

another $5 billion right here due in

August of 2023. These are massive

numbers. That's just in four and a half

years. And they have to pay off all $14

billion of their debt in a 5-year

window. Their blended rate, their

blended interest rate is 10%.

So, no wonder they're blowing money.

This is a problem. Now, how much cash do

they actually have to be able to pay

this debt? And this is where you should

get nervous. So just like signpost for a

moment, Cororeweave is saying, "Daddy

Jensen, please bail us out. We have a

problem or we're going to." And they're

going to their bankers because they're

out of freaking money. They're

insolvent. Okay. Now, I just showed you

all the debt they have, how they're

financing more debt because they're

spending all this money and they got a

lot of debt coming due within the next 5

years. Okay. This is all going to, by

the way, relate to XAI, Jensen's Reuben

chips, and high bandwidth memory. All

right, we got to build. We got to

explain this all properly. People are

like, "Kevin, why can't you do this in a

twominut short video?" And I'm like,

"I'm sorry. I It's It's dense. It's a

lot." But when you look at the actual

facts, you're like, "Damn, I understand

why it takes a hot minute to explain it

all, cuz otherwise I'LL FALL ASLEEP. I

NEED SOMEBODY TO WAKE ME UP WHILE WE

TALK ABOUT THIS CRAP." BUT IT'S ALSO

really interesting. I mean, at least in

my opinion. I don't know, maybe I'm just

nerding out about this, but let's look

at their balance sheet, right? So, they

got a lot of bills to pay. We know that.

But the question is, how much money do

they have over the next year? I mean, I

started the video out by saying they're

essentially insolvent within the next 7

months. Why would I say that? Ah, okay.

Well, let's just use their own financial

filings to show you why I would say

that. Please, Cororeweave, don't sue me,

bro. So, current liabilities,

bills that they have on their desk. So

on their desk right now, they have bills

that they have to pay of $7.9 billion.

Okay? So imagine you had, you know, I'll

just use millions for an example.

Imagine you had $7.9 million sitting on

your desk right now in bills. And then I

told you you have another $10 million,

in their case, $10 billion of debt to

pay over, you know, the next five years.

Okay? You'd probably be like, "Oh my

gosh, Kevin, [laughter]

dude, what did I get myself into?" A and

then of course the the follow-up would

be like, "But Kevin, how much cash do I

got?" You know, if I if I got to pay

$7.9 billion or million dollars in my

case of debt, you know, how much cash do

I got to pay it all? I I can probably

pay it all with cash, right? No.

No.

>> No. [laughter]

Uh they only have $3.5 billion of cash

and accounts receivable. That's assuming

they get paid. Okay. assuming they

actually get paid on all their

receivables, which a lot of these

companies are delaying when they're

paying Coreweave, just part of the

agreements like Microsoft and all these

companies. They they get to negotiate

longer payment windows. So hopefully

that cash shows up for Cororeef. I was

generous and I assumed that they have

$3.5 billion of cash by adding those two

things together. And restricted cash is

really cash that's already spent likely

on new products and stuff, right? So

basically, if they're burning 500

million per month and they have or 500

Yeah. 500 million per month and they

have 3.5 billion of cash, they first of

all only have about 50 cents on the

dollar of all the bills that they have

to pay. That's a problem. You're way

upside down. You're sweating bullets.

And if I divide out 3.5 billion of cash

by their cash flow, you know that 500

mil burn, I've got seven months. That

means by April, given that this is a

September filing, by April, I'm out of

money. Well, here we are in January

begging the banks to not look at our

numbers and please enable Daddy Jensen

to bail us out. Like this, this novel is

writing itself. They are freaking

bankrupt basically. They're insolvent. I

shouldn't say they're bankrupt. They're

insolvent right now. Now, what could

make it worse or better? I mean, isn't

it good that we have new chips from, you

know, Jensen, Daddy Jensen coming out?

No. That could actually make it worse.

I'm going to explain that in just a

moment. It gets ridiculous here. But,

but first, understand that this is a

company as well that has about 35% of

their business, I said 30% earlier, it's

actually about 35% of their business

sitting idle. Now, why? Well, because a

lot of this is sitting as construction

in progress. of that they have about

$2.5 billion in acred purchases, part of

their their expenses, their liabilities,

right? But they just haven't been able

to turn on yet. So people say, "Oh,

magically as soon as they turn this 35%

of the business on, they'll make a

oodles and boodles and oodles of money."

But the problem with that is if you look

at the business and you actually take

the depreciation expenses, they don't

make money and they spend way more money

even when you add that depreciation back

in. The cash flow statement shows us

that because the cash flow statement

already adjusts for net income. So if

you jump over here, you take that net

loss of $715 million, add back in

depreciation, and then you go, "Oh,

great. They actually got 1.5 billion in

cash. Yeah, but they're spending 6.2. So

even if you get that next 35% up and

running and you go, you know, increase

your revenues by 35%. So maybe add, I

don't know, another another chunk of

revenue here. Even if you doubled that,

even if you doubled the net cash

provided by operating activities, they'd

still be upside down and they'd still

have massive debts coming due within the

next 5 years. They still have massive

debt obligations due in the next year

and a half, which is exactly why the

bankers are like, "Oh my gosh, we have

to not only bail out their revenues,

which Nvidia is already doing. We have

to bail them out during the construction

phase because they're literally

complaining about heavy rains and winds

delaying their projects, which might be

true, but everybody knows construction

gets delayed. The only people who

about it are people who got themselves

too deep into their eyeballs in debt and

they have no choice but to complain

about it to beg other people and point

the finger at other people. Okay. All

right. Enough of that for a moment.

Let's understand one last thing about

their financials and then let's get into

the future here. So the last thing I

want to show you about their financials

is just how much money every single

month goes to interest of their interest

expense in a quarter. They've got about

$310 million net going to interest,

which means a hundred million dollars,

more than a million dollars every single

day are going to debt, which sucks. So,

what's the big bet here? The bet here is

obvious. Get the servers online.

Everything will be fine. We're just

doing this paperwork to cushy over this

big cash liquidity squeeze we have.

Hence, this credit default swipe spike

that we've had. credit default swap

spike there. Let's get the words right.

Okay, that's why we've had this big

spike over at Oracle. You're seeing

similar things at Core as well. I'm

using Oracle just to show what's called

contagion. When all of a sudden you have

an issue at Coreweave, that contagion

shows up at Oracle as well because

Oracle is trying to compete with

Cororeweave. And if Corwe is going to go

bust, you're going to liquidate a lot of

those assets at bargain basement prices,

which then lowers potentially the actual

asset value of those chips at companies

like Iron and Bis and Oracle. Hence,

Contagion. Now, how is Daddy Jensen

already trying to bail out the cycle?

And what are these new chips that he's

been talking about?

Well, it turns out XAI just raised a

series E fund raise of $20 billion.

Now, why this is very interesting is

because when there were rumors about the

Elon Musk raising $15 billion and

everybody's like, "Oh, they must be out

of money." Elon Musk says, "Oh, that's

false."

But that's usually what he does. In

fact, I talked about it on the channel.

I go, Elon's going to say it's false and

then he's going to come out and be like,

uh, it was actually $16 billion. So,

[laughter] so jokes's on you. We

actually needed more money. [laughter]

And, and then it's like, wait, but who's

the joke actually on? Because Reuters

was correct. You were raising money.

Elon denied it was a $10 billion round.

Elon denied it was a $15 billion round.

And now they're cheering going,

[laughter] "Guys, everybody thought we

were raising 15. We actually managed to

raise $20 million and we were able to do

it with with Nvidia as a strategic

investor." Ah, yes, the famous Nvidia

strategic investor. Yes, this is how

venture capital and private equity

markets work. Okay, what people do is

you get somebody like imagine this.

Okay, imagine imagine Kathy Wood bestows

her holy arms uh on and wings on House

Hack and says, "I'm joining the board

and I will take a position in the

company." Doesn't disclose how much it

is just like Nvidia and then we're going

to raise money at $10 a share, you know,

10x what we raised money at previously

cuz Kathy is the strategic investor. And

then everybody's like, "Kathy has a

5year time horizon. Let's get in." You

know, it's kind of like that except it's

Nvidia as a strategic investor. Now,

don't get me wrong, I love Kathy. You

know, this isn't to make fun of Kathy.

I'm just saying you take a big name and

you anchor their name with a higher

round and that's how companies get

valuations that grow and grow and grow

and grow. You just get strategic

investors. So, what does that enable?

Well, Elon Musk, we know, is burning

cash hand over fist. And so remember the

beginning of the core reef cycle, it's

exactly the same for XA. They're

expected to burn $1 13 billion this

year. So really what they did is they

just fundraised. Well, that was actually

last year because we're in 26 now. So

$13 billion last year that might grow

this year, how much they actually burn.

And so what they really did is they just

financed enough burn for this year. So

they have enough money to get through

this year, right? But remember the

beginning of the cycle. I want you to

remember how this cycle works and then

we're going to show you the innings of

the AI wave. All right, so here's how

this works. The way this works is

investors put money in. Okay, that could

be you buying Core Weef stock or NBIS or

Oracle or or Core or whatever you want,

right? That's you buying stock. That's

you know senior retired seniors and

their hedge funds or whatever throwing

money into this XAI round because

everything Elon touches turns to gold.

So, you know, throwing money in. As long

as that money keeps coming in, we can

keep buying Nvidia chips and we can keep

buying HBM. Time passes, the data

centers eventually get built and then

hopefully we get money out in the

application layer. That's how the cycle

works. Now, remember the unlimited

equity cures part. All of this works and

Nvidia will probably keep bailing out

Coreweave

at some point. no guarantee that they

will as long as money keeps coming into

the cycle. So as long as the cycle keeps

going, Nvidia has the capacity to bail

out cororeweave. As soon as Nvidia

doesn't bail out cororeweave, the

bankers tighten the covenants,

cororeweave goes bankrupt, chip prices

collapse, memory prices collapse, data

center values collapse, and money stops

coming in. The cycle stops very quickly.

Okay, we already know that. It's just

we're starting to see some early warning

signs. Oh, is this like so bearish that

you have to go run out and sell

everything? Well, no. Because see, high

bandwidth memory is technically now sold

out through 2026.

Now, why is high bandwidth memory so

incredible? Well, a lot of it has to do

with our current needs for inference.

And that's where Nvidia has this like

gorgeous video right here where they

showcase what's inside of these racks.

Each of these being a little drawer. And

inside you see the various different

component chips. And all of these

various different component chips carry

what? Memory. See memory. DDR5. Look at

that. M juicy memory. That's just the

CPU, right? What about the GPU? We're

going to have memory over here. See,

there it is. High bandwidth memory four,

right? Uh

every chip. So when they make these

chips, we need high bandwidth memory.

Now who's buying all this memory? A lot

of it being the hyperscalers, Microsoft,

Meta, Amazon, Google, and obviously now

Elon's XAI. So we're obviously not at

peak cycle because people keep buying

these chips,

but we are somewhere on a cycle. And see

rising high bandwidth memory costs,

which are really good for companies like

uh uh what do we got here? SanDisk, you

know, when this is when we went through

their financials. We go through

fundamental analysis like every single

day in the course member liveream.

Remember, you can get lifetime access by

going to meet kekev.com and getting that

lifetime access uh to to this

fundamental analysis we do every single

day. And even if you don't watch it, you

could just click on our little tabs and

look at my analysis that I've done. And

I add to it every day the market is

open. But look at this good cash balance

sheet. Look at these numbers over here.

29.7%

margins. Revenues growing 22%. But

listen to this. Their operating income

uh is four times the annualized Q3 EPS

expectations. So they're exploding right

now relative to what Wall Street even

understands. Uh it's remarkable. Now

they've gotten a little bit puffy. Okay.

At 210, they were about 10% below what I

thought would be a fair value for a high

bandwidth memory company like Sandis.

The thing is because of these sellout

announcements, they actually moon way

past that. Why? Because of momentum. So

now we've gone to the point where we're,

you know, we're certainly overvalued.

We're trading at a three peg now, which

is a bit more expensive than where it

should be. should probably be somewhere

around 15 20% lower. So there will be

better entries in the future for stocks

like this. And you know we did that

analysis all the way back uh I mean at

this point it was like months ago. But

what matters now is where do we sit in

the broader cycle? Because remember

rising HBM costs isn't actually great

for the entire cycle. Why? Because it

reduces margins at AI companies. you

know, the margins that are already tight

on even uh coreweave installations or

Oracle installations get even more tight

when HBM memory goes to the moon in

terms of pricing. And that's where I

introduce what I call the AI capex

cycle. Okay, AI capex cycle. And where

do we sit today? Well, let's analyze

this together. So, my opinion is that

and I think we all agree this all

started with the GPT moment. Okay,

that's a basic one. That's like the

first inning of the cycle. Then we get

this infrastructure explosion.

Where we tend to start curving out is

usually in high bandwidth memory. High

bandwidth memory is a lagging indicator.

It is the last phase of the cycle to

tend to bubble. That's why we're seeing

the bubbling now. Now, that doesn't mean

we're at the end. We're at game over

stage. In fact, I would argue we're

probably over here at like maybe here,

the seventh inning. So, that means

there's still some room to go in this AI

capex movement. You know, JP Morgan says

the biggest mistake somebody could make

right now is sitting out the AI boom

that's coming, not being worried about

the bubble. Maybe they're right. Maybe

they're just shilling their book. I

don't know. But what always happens in

history is that capital expenditures

plummet and there's an infrastructure

crash once we get to an over supply

that's going to take out a lot of

companies in the application level as

well because if this happens the red

line happens we'll see some kind of

recessionary hit simultaneously which

will take out a lot of unprofitable

technology companies. This is normal.

Now that debris will actually pave the

way for the artificial intelligence

survivors to actually start making

money. And now you stop allocating money

to all the junk companies and you

allocate money to the survivors. That's

the long-term future that we really all

benefit from. And I believe that the AI

application era is actually still right

here. were really early innings like

maybe second inning of that application

level. Now Nvidia and Jensen suggests

right now that we are seeing such

massive benefits in Reuben technology.

He just had a big presser at CES where

he talks about Reubin should reduce the

cost of tokens by 10x. But I want you to

think about that for a moment. If you

reduce the cost of tokens by 10x, who

does that actually cost money, right?

Like, who loses on that bet? Well, it's

probably Coreweave. Now, I'm not saying

this because I've got like a bare

position on Coreweave. I don't have a

bare position on Coree. I don't really

care. Like, if you're in Coree, I'm not

trying to dump on Cororeweave. Okay. The

point is Reuben chips that cost 10x less

per uh chip basically or per token

rather and need four times fewer GPUs to

train parts of models. This is per

Jensen. What you're really doing is

you're saying all of that money that you

just spent on capex up here at this top

of the cycle may have been wasted on

those older chips because the next

generation chips are going to kick your

butt. Notice how even after that seventh

inning though, because remember if I

think we're right here at the seventh

inning, you still have HBM that's going

to keep going and memeing for a hot

minute probably because these Reubens

are going to need a lot of high

bandwidth memory. But the point is more

chip innovation is actually mega

deflationary for rental rates from

Coree. This is why Nvidia can actually

laugh its way to the bank.

any lowering of energy costs or

increases in efficiency make Nvidia sell

more chips even to people they've

already sold chips to because they're

like, "Hey, why are you using those old

chips? They're costing you more money.

You may as well use these new chips that

we have." So really, Nvidia is like

saying, "Hey, get with the program here.

Buy all the new junk and you too will be

good." Okay, great. That's really bad

though for again Cororeweave because if

Cororeweave invested in all the H100s or

early Blackwell chips, well then you got

some big problems. You got some big

pooper dupers coming and nobody wants a

big poopy dupy. Now, the hope is that

LLMs and AI keep progressing and as a

result of the continued progress, we'll

continue to need even the old

technology, the 81 H100s or older chips

that are still being used today. And

that's a great hope as long as investors

are patient with the profitability cycle

and obviously until we get to peak. So

my take, we're probably in the seventh

inning, late cycle of the LLM stage, but

we're actually quite early on the

application era. This is where I get

excited when Nvidia tells us they're

opensourcing Alpameo 1, their

self-driving thinking and reasoning

model. I actually think that their the

Nvidia FSD, I'm going to call it, just

full self-driving just to make it a

little bit more, you know, palatable or

understandable or comparable to Tesla

here. I actually think it is their most

undervalued asset. I actually think that

almost alone once they transition from

89% of their revenues is coming from

GPUs, but instead they start moving to

like full self-driving licensing or

whatever, especially since they they

take in all sorts of traditional

sensors, LAR sensors as well as vision

based sensors. The fact that their

models consolidate all of that, I think

makes them one of the most or makes this

one of the most underappreciated assets

at Nvidia. And so I actually think not

only is Nvidia playing the

infrastructure buildout game, but

they're playing the application era. Now

that's exactly why in my opinion they

bought Grock, not to be confused with

Grock, you know, the Twitter or X AI

bot, but rather the language processing

unit model company. That company that

makes or designs those chips is now

bought by has been bought by Nvidia for

or you know roughly licensed out for $20

billion. It's technically not a complete

buyout, but it's practically that. Uh we

made a separate video on that. But

anyway, Nvidia is riding both of these

curves. Nvidia will be a survivor when

this crash comes. We all know that capex

plummet is coming. It's a red flag where

Coreweave is showing us, you know, where

things sit right now. I mean, Amazon's

seeing it as well. People at Amazon are

complaining that Amazon relies on

anthropic LLMs too much and they should

have their own proprietary LLMs. But I

think that's because Amazon knows you

don't want to blow all your money on

LLMs because they're going to be a

commodity. And as soon as this cycle

peaks out, it's the actual productive

productive uses of AI that are going to

make people real money. Now, what's my

bias in all of this? Just to be clear,

because I don't have like calls or

shorts on like Nvidia or Coreweave or

whatever. My biases in all of this is I

am seeing the real profits from our

application layer. Now, we just raised

over $10 million for our startup in just

the last 3 weeks, which is insane the

amount of that's come in because, you

know, people want to invest in that next

cycle. Like these were our fundrais

stats for the last, you know, threeish

months here. And, uh, you could see it

on screen here. 14 October, 1 17

November. About 99% of this is, uh,

fundraising. Uh, then we announced the

closing of the fund raise uh, in

December, which, you know, we're we're

still closing it out. like we we stopped

telling people it's open but but like

people applied even as of you know

December 31st and they have to get

approved by a broker dealer like there's

a lot of SEC compliance involved and

stuff so people are still getting their

approvement emails and that's how we've

gotten like $3.4 additional million

dollars coming in in January and what

we're doing is we're building on that

application layer which I think is the

future. I got to update the website

because, you know, this has already been

released. Uh we released this as

promised in December. But the app's

already out, which is great. And now

we're in the improvement phase and the

the new iteration phase, which is really

exciting. But people are buying this and

uh they're buying the lifetime access to

it, which everybody keeps telling me is

a bad idea, that I should be doing

monthly recurring revenue, but for now,

you can still get lifetime access to it.

Uh and you can get that at houseack.com

or reinvest.co. But that's my bias in it

is real estate SAS. You can read more

about it by going to houseack.com. I

think it's a net worth sniping tool. You

know, I did a video the other day where

I saw a deal that was like model match

listed for $500,000

less

uh than what a comp was listed for. Same

exact model. And I'm like, this is

incredible. You're literally sniping net

worth with this kind of AI. Now,

obviously, you got to go renovate it and

there's work to do and there's risk with

every investment, but we're very excited

about that application error because we

think even if AI gets no better than it

is today, the AI boom can keep going. I

just think it shifts away from that

hardware stack and goes straight into

software. That's my opinion. I'm just

being transparent about the journey. If

you found this helpful or any of this

information helpful, consider sharing

the video. Consider subscribing. And

we'll see you in the next one. Goodbye

and good luck. Why not advertise these

things that you told us here? I feel

like [music] 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.