Sam Altman JUST *LOST IT* | *Major* WARNING for AI

an extremely awkward confrontation of

the CEO of OpenAI, Sam Olman, and Satya

Nadella by podcaster, we'll call him Mr.

Brad. We'll talk about all of this.

There are critical takeaways from this

interview. We're going to talk about a

glut of chips, which could be a red flag

for chip companies, or is it? We're

going to talk about Sam Alman's goals

and why that could affect chips

massively. We'll talk about Microsoft

strategy and how it's very very

different from open AI. You might think

but Microsoft invests in Open AI. Satya

Nadella in this interview basically says

we could write down the OpenAI

investment to worthless and he'll still

show you how they're going to make

money. We're going to talk about that.

We'll also talk about how I think Nvidia

is protecting its moat and how different

the mindsets of these two founders uh or

these these two executives I should say

are. Now what's really important as well

is that there's a lot of drama going on

about how Samman responded. Sam Alman

has no shortage of haters. A lot of

people think he's like so evil that he's

a murderer. There are other people who

say he's just a liar. And then there are

other people who say like, "No, these

are people just hating on Sam Alman's

rise." I'm going to break all of us

down, but first we're going to start

with this awkward moment and then we're

going to go into the details. As you go

into this awkward moment, I want you to

have an analogy in the back of your mind

because I think it's useful to have

these numbers so you can make them

relatable. Let's say you are working

part-time at Jamba Juice, which I used

to do. Uh, and you're earning $2,500 a

year. Okay, that's not even enough money

after taxes for one month of rent, but

whatever. I guess it depends on what

you're renting, I suppose. But anyway,

let's say you're making $2,500 a year,

$2,500. It's pretty low, right?

But now, let's say that by the end of

the year or the end of next year, you're

making a commitment that to your parents

while you're making $2,500 a year that

you're going to spend $130,000,

you know, at the end of the year or the

end of next year. Rightfully so, people

are going to go, "Bro, dude, your

revenue is $2500.

your net is zero or negative. How the

hell are you gonna spend $130,000

by the end of the year? Okay, that's the

same proportion of what's going on with

Open AI where Open AAI is like, "Yeah,

boys, we going to spend $1.3 trillion

and we got revenue somewhere around 20

billion plus or minus. We don't know. We

know it's more than 13 and we know they

have a goal of getting to that's their

internal goal. The public thinks it'll

take longer of getting to hundred

billion by 2027. You know, so if their

revenue is 30 bill now, maybe they

double it in 26 and maybe they double it

again in 27 and they get to over 100

bill by the end of 27 if there's that

much of a doubling capacity. But going

back to anal the analogy, you know, how

are you saying, hey, we're going to

commit to $500 billion to Nvidia and

$300 billion to AMD and Oracle. You kind

of have to divide those. $250 billion to

Microsoft Azure over the next four to

five years. How are you going to spend

all that much all that money when your

revenue is so low? Right? That's where

the criticism comes from. And a lot of

people say that when the CEO reacts in a

frustrated manner of people questioning

him on these spending commitments, it's

a red flag. Other people say this guy's

just tired of being asked the same thing

over and over again, but it's worth

watching the interaction. So without

further ado, let's play it. Revenues are

still a reported 13 billion in 2025. And

Sam, on your live stream this week, you

talked about this massive commitment to

compute, right? 1.4 4 trillion over the

next four or five years with you know

big commitments 500 million to Nvidia

300 million to AMD and Oracle 250

billion to Azure. So I think the single

biggest question I've heard all week and

and hanging over the market is how you

know how can a company with 13 billion

in revenues make 1.4 4 trillion of spend

commitments, you know, and and and

you've heard the criticism, Sam.

>> First of all, we're doing well more

revenue than that. Second of all, Brad,

if you want to sell your shares, I'll

find you a buyer. [laughter]

>> I just

>> big moment here. This is where you see

Satya realize this is an awkward slam.

And Sadia doesn't really know the best

way to respond other than just by trying

to laugh it off to cool the tensions

because this is early into a podcast

that's about to get really awkward.

Maybe it's no surprise that soon after

this Sam Alman ends up dropping out. But

you even see Brad is kind of like, "Bro,

this is a little rude, man. Like, I'm

going to smile just to give us an exit

here." But clearly Sam Alman is pissed.

He's tired of hearing this. So rather

than really addressing it, he's lashing

out. Now, in fairness, he does end up

addressing it. But he the way he

addresses it makes it very clear that

the mindset that Sam Alman has at OpenAI

is way way different from the mindset of

a CEO of a business that actually makes

a ton of money like Microsoft. I want to

be clear about this. Sam Alman's

business generates maybe revenue of call

it $20 billion a year. They're netting

essentially zero. So they're making no

money, right? They've got a great

product basically making no money. This

is very different from the business

mindset of Satya Nadella. Satya

Nadella's point of view is listen, we

want to have a low RPO, high use

product. Okay, what does that mean? It

means low average revenue per user, but

we have a lot of users at a high margin.

So, we basically commoditize

intelligence by plugging it into Word,

uh, Excel, Outlook. We sell it to you

for a small monthly fee and you're using

it every single day. That's how we make

money, baby. And in fairness, they kill

it. I mean, Nvidia kills it more, but

Microsoft absolutely kills it.

Understand this is their annual, this is

a sort of a consolidated statement, a

summary of their operations. Revenue,

$281 billion. Okay? They take $11

billion to the bottom line after taxes.

That's a 36%

margin net. That is really, really good.

It is a lot of money. Microsoft has

mastered how to sell you Windows and

sell you monthly subscriptions or

storage subscriptions

and make a little bit of money off a lot

of people. A lot of people. And that's

how they end up with these 36% margins

and over a hundred billion dollars of

revenue. I'm sorry, net income, right?

They literally have probably 20x the

income that OpenAI has and they actually

bring infinitely more money to the

bottom line. Now, of course, the reason

people are so excited about OpenAI is

because they call it the growth story.

Satya Nadella doubles down and says,

"Listen, we don't really care so much

about this direction that Sam Alman is

going. Open AI could go bankrupt." He

essentially says this. He says, "We

could write down our $13 billion

investment into OpenAI to 0." And it

doesn't matter because we get

a free license essentially at that point

to utilize OpenAI in our co-pilot

products and in our other products. And

those give us the service that people

want functional AI essentially today at

a low cost. So SA very much taking the

business point of view here like we

don't need to worry about this AGI stuff

or make bets on that. We're just going

to make money today. In fact, there is a

fundamental difference and we get a

little bit mathy here. I'll explain it.

But there's such a fundamental

difference between the two that I want

you to think about this quote

and see the difference. And I'll explain

it. Sadia Nadella says, "Intelligence is

the log of compute." And so as costs for

compute come down through more efficient

chips or whatever, we're going to be

able to get more intelligence slowly.

Okay, that's, you know, simply put onto

a little basic chart. It looks like

this. When you first start with AI, you

get a lot of intelligence, apparent

intelligence. I'll talk about that in a

moment. But then for every doubling of

compute, you're only getting

incrementally more

intelligence. This is a very basic log

curve. Sam Oldman says, unfortunately,

it's more likely that the log of

intelligence actually equals the log of

compute.

Very complicated. If you break it down,

it actually gets very simple. He's

essentially saying we're going to cancel

out this logarithmic curve. We're not

actually going to see compute costs go

down and we're going to keep spending

like drunk sailors until we get

artificial general intelligence.

So when you take the log of intelligence

equals the log of compute and you

simplify this, you're basically saying

even if compute costs come down, we're

still going to spend exponentially more

in line with where we hope to grow

intelligence because our goal is AGI,

artificial general intelligence. And

this is going to be a gamecher for

chips. And I'll explain why in just a

moment. But let me break this down so

you could kind of maybe pick your side.

So as an investor, I look at this and

go, okay, Microsoft knows how to print

money on selling AI packaged down into

the most basic level to everybody.

OpenAI is going to keep burning every

bit of money they have because they're

trying to go for AGI. Now, we have to

understand the difference between the

two. And I'll give you a very simple

difference in terms of achieving it. I

think you already know artificial and

general intelligence would basically be

this super superhuman style of knowledge

that can come up with new discovery and

new content itself. Whereas AI we have

today isn't really that. Here's a basic

summary of what AI is today. Okay, let's

take this cup for example here. We'll

zoom in a little bit. All right, let's

take this cup. Let's say we fill this

cup with all written text, all audio

recorded speech and all video that

exists in the world. Okay? Then we stir

it up, you know, using hundreds of

thousands of Blackwell Nvidia chips and

H100s. We stir it all up together. Uh

and then we tell a computer go, hey,

understand

this cup. Which technically is a way of

understanding the world view of how

these things interplay. So, we're now

going to see a sample of all of this.

We're going to skim a sample of all of

it.

Got it. Okay. That's roughly the

bitterness. That's roughly how it

tastes. This is how things function.

This is the utility of it. Got it. Now

that I have a skimmed sample of all of

that content, I can probably generate

answers and pictures based on what I've

seen or video based on what I've seen in

the real world. Like I understand that

when a gun fires, when a little kitten

is holding a gun, even though we know

what a kitten looks like, we never have

videos of a kitten holding a gun, but we

know how kittens act and we know how

guns act and we know when the trigger is

pulled there's recoil. We could kind of

generate what that would roughly look

like together. Sort of like our

imagination, right? That's really what

it is. We take the world view of how we

understand the world and then we

generate answers based on probabilities

of this is likely how things are going

to be. That's AI today. Okay. It's not

actually generally intelligent yet where

it can generate its own solutions. This

is why Sam Alman talks about these

haters on X because when Sam Alman goes,

"Look, we just cracked this insane

scientific puzzle or this major science

problem." And then other people respond

an X. They're like, "Actually, you just

lifted all of the answers from this PhD

who posted it on this obscure blog over

here. Here are the receipts. You guys

are a scam."

Okay, so this is why people like Elon

Musk say it's scam old. Anyway, so now

you understand that that's what AI is

today. It's a a pattern-based and

probabilistic based worldview that has

skimmed the world and it can generate

answers from there based on roughly how

things should work. The more creative

you get, the more it hallucinates and

the more it breaks down. This is how we

get hallucinations in AI. The more

encyclopedic it is, the fewer

hallucinations we get. Okay, simple. Sam

Alman is taking the mindset that if we

just keep spending on computers, we will

get artificial general intelligence

eventually in our pocket. Well, he

specifically said in their laptop, but

presumably if you could have it on a

laptop, you'll be able to have it on a

phone in the future.

Sam Alman actually says at that point it

will actually become somewhat depressing

what you end up seeing with data centers

because data centers are going to have

to spend all this money on data centers

that you won't actually need because

people have their data center in their

pocket.

That's if you believe the Sam Alman

point of view. He argues there will be a

glut of H100s and Blackwell chips in the

future. Okay, fine. Say Nadella, man, we

don't really care about the future, bro.

Listen, dude. Dude, I'm a CEO. I'm just

going to sit here. I'm going to print

money. You guys go spend all the money

you want. We're just going to focus on

providing something to people right now.

We're going to make money. [laughter]

Which I get that. Okay. So, like side

tangent. Uh at my company reinvest.co or

househack.com. It's the same company and

just slightly different branding because

of the products we're launching uh

within the next few months. We are of

the mindset that you know we can

transform how people invest in real

estate whether they're investors,

they're flippers, they're home buyers,

how they renovate real estate, uh how

they get insights into how a purchase or

investment or a renovation changes their

net worth. And then we actually turn

real estate into a computational problem

rather than an emotional problem. Or at

least we augment people's emotions with

math, right? We think we could sell that

as a product or a software as a service

and make a lot of money. That's our

opinion. That's not relying on AGI.

That's relying on human intelligence

today, for example, like what I do in

real estate and training models to react

to new properties or renovation problems

with my mindset as an investor in real

estate. That's what we're doing. We

think we're taking very much the same

approach as Satya Nadella, which is

great. How can we take the technology we

have today and make it the most useful,

make as much money as possible with it?

I lean towards that direction. I support

what Sya Nadella is saying here. I think

it's a little bit

whimsical at this point to assume that

the AI we have today is capable of

getting to AGI. And unfortunately that

makes me concerned that Sam Alman is

probably overly optimistic on his

revenue projections.

And therefore at some point the Sam

Alman spending will slow. And when the

Sam Alman spending slows is probably

also when we're going to see spend

optimization at companies like Microsoft

and Amazon and we'll start seeing

layoffs. Oh, wait. We already are

because eventually

we will use AI to optimize our workflows

and a lot of people are going to get

fired. Not because one person uh can now

replace the work of 10. It's rather that

one person can optimize the work that

they're doing so much more rapidly and

there doesn't necessarily have to be

more work. Just we could do more with

fewer people. uh in this is more of a

reference to just because you have uh

the capability of now more time does not

necessarily mean you're going to

increase revenue more right it just

means the work that you need to get done

you can get done a little bit faster

with artificial intelligence great so

this gives you the world view of these

two individuals and it separates the two

here now Brad asks is there any chance

of a compute glut coming within the next

2 or 3 years and he asked Jensen Hang

this and talks a little bit or hang

this. He talks a little bit about it in

this podcast.

He says when he asked Jensen if there's

any chance of a chip glut in the next 2

or 3 years, Jensen's response was no

chance. Not a chance that we're going to

have too many chips in the next 2 or 3

years. So, we probably still have a bull

cycle here of spend spend spend on this

hope of AGI. So, this is where you kind

of pick your side during this hope of

AGI spend. Do you just park your money

as an investment with companies that are

trying to make artificial intelligence

revenue with practical applications like

either what we're doing at house hack

reinvest you know you can go invest

there if you want read the offering

circular uh reinvest.co cohes.com same

company or Microsoft or whatever or do

you believe in sort of the open AI point

of view and invest on those in those AGI

plays at the valuations that they have

now that's important to consider now why

do I did I mention earlier that Nvidia

has a a potential moat that might fade

so based on this idea that there's not

going to be a glut of chips in the next

2 or 3 years I I think Jensen is of the

mindset that their moat will stay strong

for the next two or three years because

we can't manufacture enough of these

chips. Intel got one of the first

shipments of one of the most advanced

lithography machines from ASML. And I

think one of the reasons Intel has been

doing so well is because people are

realizing we actually have a shortage of

being able to manufacture these chips.

And that's probably how Nvidia gets, you

know, a 72%

gross margin and a 56% net margin. you

know, they make way more money than

Microsoft on a net basis because they

have such a mode on chip manufacturing.

It's not just chip design that's good or

CUDA, but it's also I think they have a

mode on chip manufacturing. Think about

it like this. If Nvidia

before the AI boom tells TSMC, yeah, so

um you guys make 100 chips, 25 of the

chips you make percentage- wise, let's

just say 25 of the chips you make are

advanced AI chips. we want your

manufacturing capacity for 20 of those.

So, call it roughly 80%. Right? Well,

now there's an AI boom and other

companies are like, "Hey, we want some

of that 25 that you dedicate towards AI

chips. We want some of that." And TSMC

says, "Sorry, like we already have a

long-term commitment with Nvidia. You'll

have to wait for that other five to come

available, other 5% or whatever, or you

have to wait for us to open new fabs or

go to company like Intel."

That's I think one of the ways that

Nvidia has a really big secret moat is

it's not necessarily

just the quality or CUDA, but I think

there's an extra bonus mode of the fact

that they have the supply chain unlock

and it's just going to take time for

that supply chain to butter out which

eventually those Nvidia margins are

going to come down because it's those

margins are too juicy not to want to

compete it away. That's why Amazon makes

their tranium chips or Google makes

their tensor processors, their TPUs,

right? They want some of those margins.

It's much cheaper to make your own chip

and then run your own chip than pay

Nvidia a 56% net margin, 72% gross

margin. Way cheaper to just make your

own chip. But then when you call TSM and

you go, I want to make my own chip.

They're like, great, we'll get to you in

5 years when we open up more

manufacturing space cuz Nvidia already

has it all unlocked. And then what do

you do?

All right, Nvidia. will take some of

those chips. You like I have no choice

right now, right? So, and I'm biased

towards Nvidia here. I've I've got lots

of over seven figures of Nvidia shares.

So, don't get me wrong. I'm not trying

to be bearish here on Nvidia. I'm saying

for now that Nvidia mode is still

glorious. And that's why I have a $300

price target on Nvidia.

That said, I am strongly of the mindset

that the people like Saiya Adella,

they're going to be the winners in this

big revolution because they're focusing

on making money today by providing great

value with actual artificial

intelligence that we could use without

betting on well frankly AGI. I think Sam

Oldman is unfortunately he has seen so

many breakthroughs going from GPT2 to 3

to 35 to you know five. He's of this

impression that these breakthroughs are

going to be exponential. I actually

think the progression that we've seen is

actually much more logarithmic where you

know maybe we've saw this jump from two

to three or to three five and then you

know now we're you know then we got our

four or four five and five. We're on

this part of the curve and this is still

great for companies like Nvidia for now.

It's so it's going to continue to be

great for companies like Microsoft, but

I think Sam Alman thinks we're going to

have an exponential result here in

intelligence. And that's where I I can't

align with Sam Oldman. So Sadia argues

that

today we are still actually not even

worried about not having enough chips.

We're actually more worried about not

having enough power,

which indicates we're still relatively

early in the normalization of this

supply chain. So, there's a lot of money

still to flow into this. We're still, I

think, relatively early in this. Will it

all normalize when people realize, all

right, we're probably not going to get

to AGI anytime soon? Of course. In the

meantime, is Microsoft going to keep

cranking freaking money? Yeah. And

that's what I got out of this interview.

Even though it was really awkward when

Sam got drilled and this was sort of

Satya's face during it,

I think it helps establish that there

are two styles of investing that you

could take here. You take the certain

revenue that you could get now from AI

or you make a bet on AGI. That's my

point of view. And and that's actually

where I

I I think to some extent

we are going to get humanoid robots.

There's no question there. But I think

to some extent the amount of

breakthroughs that we're actually going

to need for AI robots to be functional

will will probably take uh you know

another decade. So this is where you

kind of have to evaluate all right do I

position my portfolio to where I can

make money now or am I going to bet on

that future in the decade because it's

inevitable. I actually kind of do think

it's inevitable. We're going to have

humanoid robots everywhere. Just think

it's going to take a whole lot longer

than people think. FSD in cars is one of

the reasons why we can think that

because well as we find everything takes

longer than we think. Anyway, this gives

a sort of an overview of my take on

artificial intelligence this interview

and hopefully you learned a lot. You

know, a lot of thinking went into this

video and hopefully it's a nice

consolidation uh for you. If you have

any questions uh leave a comment in the

comments down below, consider

subscribing to the video. consider

sharing the video and then of course um

this video is not a solicitation but if

you do want to invest in house hack or

reinvest with our uh real estate

artificial intelligence on its way make

sure you check out reinvest.co [music]

and read the offering circular. Thanks

so much and we'll see you in the next

one. Goodbye. Good luck.

>> 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 Praath there, financial analyst

[music] and YouTuber. Meet Kevin. And

always great to get your take.