Big Tech Debt Is Exploding

Hey guys, it's Sasha. Amazon released

their results for 2025 a few days ago

and Amazon stocks started collapsing

just before those results came out

because there was a rumor that Amazon

would announce the biggest ever increase

in capital expenditures in corporate

history. And Amazon delivered on this

rumor. Amazon announced that they plan

to spend $200 billion on capital

expenditures in 2026. And I guess they

realize that this sounds like a really

really big number and maybe just maybe

investors are going to look at this

number and get a bit concerned about

whether this is the right move. So they

added this bit that says and anticipate

strong long-term return on invested

capital. If you look at Amazon's cash

flow statement, you'll see that the

company spent just under 132 billion on

capex in 2025. Like everybody else,

Amazon is buying a lot of Nvidia chips

and sticking them in these new giant

data centers. But they're saying that in

2026 that spend is going to increase by

over 50%

by around $70 billion to a total of $200

billion after already spiking last year.

And the reason this is a problem is

because if you look at the financials,

the metric that ultimately is the one

metric that matters to investors is free

cash flow. And free cash flow is the

operating income minus the capital

expenditures. In 2024, Amazon had $116

billion in cash from operating

activities and $83 billion in capex. So

they had a positive free cash flow of

$33 billion. In 2025, capex increased a

lot faster than profit. So free cash

flow dropped down from 33 to just $8

billion. And in 2026, if they go ahead

with a $200 billion spend as they are

saying they will, free cash flow will

then become negative because Amazon

profits are not going to increase by 50%

yearonear. And negative free cash flow

means that cash is being actively burnt

by the company on a net basis. more

money is leaving the company bank

account than is arriving. And not only

is Amazon burning money, but they're

also starting to borrow money so that

they can go and burn it. For years,

Amazon has been paying down its

long-term debts. All of these big tech

giants, as they became cash flow

positive, they began paying back the

debt, which is the right thing to do.

The debt for Amazon was down to just $50

billion 3 months ago. But in the most

recent quarter, to finance this data

center money incinerator, Amazon

borrowed a fresh $15 billion. The

company that was becoming known among

investors as a bit of a money printing

machine has very quickly changed to a

company that needs to borrow money to

finance expenditure on data centers that

they cannot otherwise afford. Talking

about being able to afford things,

Amazon's choice for buying 64 GB of RAM

now cost £875.

That's about $1,200.

According to Amazon's own price tracker,

that's approximately three times as

expensive as that same RAM cost just 3

months ago because data centers are

buying up all of the available virtual

memory inside of the chips that go to

train and to operate the LLMs, which

means that you and I have to pay more

for two sticks of RAM than an entire

computer would cost last year. You just

can't buy RAM or a graphics card without

selling a kidney or reorggaging your

house. You know what? You can buy stocks

on It Toro, who are the sponsors of

today's video. It Toro is one of the

best investing apps in the world and

lets you invest your money without

paying high fees. You can invest in

funds, you can invest in stocks. There's

over 6,000 different stocks to choose

from on the platform. And now, it Toro

offer ISIS in the UK. If you open the It

Toro Stocks and Shares Iser and deposit

at least £500 or transfer £15,000 from

another provider, it Toro will pay you

up to 3% cash back on that money. So

just for the privilege of closing that

old ISA that you don't like or maybe

where the fees went up, you can get an

immediate 3% bonus. Not bad. And then

there's also the cash ISER. The interest

rate is seriously good. If you sign up

for the cash is today, you'll get a

boosted rate for the first 12 months.

Right now, that rate is at 4.44% for the

whole year. That's way higher than the

Bank of England base rate. Just remember

that if you're watching this in the

future, hello from the future. That rate

might be different. So, if you are

wondering which new ISA to try, it quite

literally pays to check out E Toro.

Please go and click the link in the

description or in the pin comment and

give them a try. Try them out. They

published their quarterly results which

said to meet customer demand, which

customer is this? And capitalize on the

growing opportunities we have ahead of

us, our 2026 capex investments are

anticipated to be in the range of 175 to

$185 billion. And remember, Amazon's

capex increased by just over 50%.

Google's capex last year in 2025 was $91

billion. So Google has decided to more

than double their spend on data centers.

The reason $185 billion raised a few

eyebrows because last year Google's

total net income was $132 billion. So

the idea now for Google is to spend more

money on building out data centers for

AI than the entire company makes across

all of their businesses in an entire

year. And if you're wondering how they

are going to pay for it, h of course

last quarter the same Google's long-term

debt was sitting at 21.6 6 billion. This

quarter, that number has increased to

$46.5 billion. So, a cool $25 billion

worth of borrowing in just one quarter.

But the problem with borrowing money is

that you have to pay it back. So, the

genius accounting nerds over at Google

thought long and hard about this

problem, the fact that you have to

actually pay the money back after you

first get it. Because what happens if AI

does not immediately print you billions

and trillions of dollars of revenue?

just like it hasn't done for the last 3

years. Well, Google sold bonds in three

different currencies with repayment

terms of between 3 and 32 years. The

reason that you pick numbers like 32

years is because if you have to repay

the debt over a very long time, then you

don't have to pay it now. It's manñana

mñana. By that point, you'll be retired.

You maybe even will be dead. It doesn't

matter. And the one bond that really got

people talking was a bond that they sold

in pound sterling which has a 100year

term. The last company to offer a

100red-year bond was Motorola back in

1997. And investors took a big look at

that previous offering by Motorola, the

undisputed leader in mobile phones at

the time. That would definitely 100% for

sure would continue to be the leader in

that space for the next 100 years.

Right. and they thought, well, what

could possibly go wrong here? Why don't

we do exactly the same thing with

Google? So, according to Bloomberg,

there was 10 times as much interest in

buying these 100-year bonds as the total

amount of bonds that Google sold. Big

tech giants are now beginning to borrow

money at a rate that we have never seen

before to finance these projects. And at

the moment, it seems that people are

happy to put up the money because people

are seeing dollar signs in at some point

in the future. Morgan Stanley expects

borrowing by the massive cloud computing

companies to reach $400 billion this

year, but that could well turn out to be

a conservative view. Google share price

is down 11% after they announced to

their shareholders exactly how much they

are planning to burn this year. Similar

to Amazon, Oracle felt a bit left out.

So, they also went and raised another

$25 billion by selling guess what?

Bonds. And Oracle are a bit further

along this path of borrowing shitloads

of money than some of the others. Oracle

have been proud to announce the

construction of some of the largest data

centers out of all of these different

tech companies with money that they

don't have. And at the end of last year,

they already hit $und00 billion in

long-term debt, which has been going up

in a vertical line. And that's more than

the profits of the company over the past

10 years. I went and looked, and as the

debt pile has exploded, Oracle stock is

now down over 50% from its peak in

September last year. The obvious problem

with going into huge, impossibly massive

debt to pay for these giant data centers

is that you are banking on everything on

there being a commercial return on these

AI investments. You are putting

everything on the line. And if large

language models end up being largely

commoditized, not as useful in the

future as people maybe imagine, and

users, for example, choose not to pay

hundreds or thousands of dollars per

month to use these LLMs, well then the

hundreds of billions of dollars worth of

data centers will become redundant, but

the debts and all of the money

incinerated

won't disappear. The negative cash flow

would have already happened. Investors

typically value companies in one of two

ways. They either look at some sort of a

profitability metric like IBIDA and then

assign some kind of a multiple to that.

So for example 15 or 20 times I don't

know projected IBIDA at the end of

whatever forecasting period in 5 years

time in 6 years time and 10 years time

or investors look at free cash flow and

with free cash flow you can also use a

multiple some people do but for mature

companies it is more common to add up a

decaying free cash flow trend all the

way to perpetuity. And usually these two

different methods, these two different

ways of valuing a company would produce

very similar results if you're building

the model. But if you're trying to value

some of these tech giants today in the

current environment, you hit a problem

because capital expenditures don't

appear on the profit and loss statement.

So if you take the ibida of a company

like Google or whatever that's spending

$200 billion on capex, that 200 billion

doesn't count, right? And not only that,

but the depreciation

on the spend capex and the capex that

they will spend in the future during

your forecast period will also not count

either because eBid duh, right, is

earnings before depreciation. So using

that lens, you would go and look at

companies like Google, like Amazon

growing their bottom line profits and

you would get a very high valuation as a

result because you basically ignore all

of the money being spent on data

centers. But at the same time we are

entering a time when it is acceptable

now according to this new mantra to have

negative free cash flow sometimes zero

free cash flow when you used to be

printing money because you are spending

so many dollars on building out the data

centers. And the problem with data

centers is that whatever the coolest,

the best, the most powerful, most modern

technology is today, it will be old

defunct crap in 5 to 10 years time and

it will have to be replaced and you're

going to have to spend money again. In

their earnings score, Microsoft said

that roughly 2/3 of capital expenditures

were for short-lived assets, primarily

GPUs and CPUs, as well as continued

replacement for end-of- life server and

networking equipment. So the vast

majority of the money that these

companies are spending is not for the

long haul. It's not here for the next

hundred years. It's maybe here for the

next couple years, 3 years, four years

max some of it. The capex investment to

build our data centers is not a one-time

thing. It's not like you build the data

centers once, right? You spend the 200

billion and then you get to use them for

the next 50 years. No. So if these

companies are saying now they plan to

spend $200 billion in 2026,

the thing is it kind of means that they

also plan to spend $300 billion in 2027

and maybe 400 or 500 billion in 2028. I

don't know, make up a number.

>> $1 million.

>> Because the issue with this spending is

once you start spending at this level,

you can't stop because that would be

completely self-destructive. If you stop

spending money, investors are going to

ask why you spent that first 200 or 300

or whatever billion dollars in the first

place. Where are the profits from that

investment? Where are the returns? It's

fine to have side projects, you know,

small side projects where you invest a

bit of spare cash and see what happens.

Google has famously done this for many,

many years. They even have a separate

line item in their financial report

called other bets where all of this

high-risk but potentially highreward

projects sit and some of their best

business lines have originated from

being in this pile. But if the bet gets

so big, if it gets too big, then you can

enter a spiral where every year you have

to increase your bet because the sun

costs are so high, you cannot give up on

them. It's not money that you can afford

to lose. It's like in year one, you put

your house down as collateral on a bet,

but then you wait a year and the bet

just hasn't come through, right? You're

still waiting for the result. But you

have to then decide, do you stay in the

bet and put more money in or do you give

up on the house? So you have those two

choices and you admit right in choice

one that you your bet was a bad idea.

Maybe just [ __ ] the house. You get to go

and live under a bridge instead. Or

option number two, you go and max out

all of your credit cards, get as many

new credit cards as possible, maybe a

few loans, and you max all of those two.

And then you take all of that money and

you put all of that money into the same

bet. This seems to be roughly the

strategy that every big tech company is

pursuing right now. And one of the

things, one of the issues with this is

that one of two things has to happen at

the end of this. Either large language

models become so insanely good, so

useful, suddenly there is a giant leap

in performance that people genuinely

want to spend $1,000 a month to use them

to recoup all of these hundreds of

billions of dollars. or the amount of

debt and the cash burn gets so big that

in the end you just run out of time or

you run out of money or you run out of

both. Remember, go and check out it

Toro, the sponsor of this video. If you

live in the UK, you can get a fantastic

deal on an ISA. Just remember to use my

link below to get that deal. And outside

the UK, it is an amazing investing

platform for investing in the stock

market. Thank you so much for watching

and I'll see you