To The Moon

$Amazon.com(AMZN)$  $Meta Platforms, Inc.(META)$  $Microsoft(MSFT)$  

For two years, Wall Street rewarded one sentence:

“WE ARE SPENDING MORE ON AI.”

More GPUs.

More data centres.

More HBM.

More networking.

More power.

More capex.

Investors cheered because spending meant demand.

This week, something changed.

Amazon's AWS grew:

39%

its fastest growth in nearly five years.

Yet Amazon fell sharply after earnings.

Why?

Because trailing 12-month free cash flow went from:

+$14.8B

to

-$42.4B.

Think about that.

AWS accelerated dramatically.

Operating cash flow actually increased 29% to $168.3B.

Yet free cash flow collapsed.

Amazon did not suddenly become a bad business.

The market simply changed the question.

Yesterday:

“How fast is your AI business growing?”

Today:

“SHOW ME WHAT I GET BACK FOR EVERY DOLLAR YOU SPEND.”

That is a completely different AI market.

And I think it gives us a new way to rank Microsoft, Meta and Amazon.

🔵 THE AI TRADE HAS ENTERED PHASE 3

I think the AI investment cycle can now be divided into three phases.

PHASE 1: PROVE THE TECHNOLOGY

ChatGPT explodes.

Generative AI works.

GPUs become scarce.

Nvidia wins.

PHASE 2: BUILD EVERYTHING

Hyperscalers race to secure:

GPUs.

HBM.

Data centres.

Networking.

Power.

Land.

Capex becomes bullish because scarcity is the problem.

PHASE 3: PROVE THE RETURN

We are entering this phase now.

The market knows companies CAN spend $100B.

It wants to know:

WHAT DID THE $100B PRODUCE?

This changes how I analyse every hyperscaler.

Revenue growth alone is no longer enough.

Cloud growth alone is no longer enough.

AI adoption alone is no longer enough.

My new scorecard is:

AI PAYBACK PERIOD.

How quickly can today's AI investment create tomorrow's incremental cash?

🔥 AMAZON JUST SHOWED US WHY THIS MATTERS

Amazon's quarter looks extraordinary until you follow the cash.

Revenue:

$201.9B

AWS revenue:

$46.0B

AWS growth:

+39.3%

AWS operating income:

$17.8B

Operating cash flow, TTM:

$168.3B

Sounds fantastic.

Then comes the number that changed the story.

Property and equipment purchases, net of proceeds and incentives, TTM:

$210.7B.

Free cash flow:

-$42.4B.

And Amazon expects approximately:

$220B

of capital spending this year.

Andy Jassy says 2027 spending will be:

“MEANINGFULLY MORE.”

That is astonishing.

Amazon is not suffering from weak AI demand.

Amazon is suffering from:

THE COST OF SATISFYING AI DEMAND.

There is a huge difference.

🧠 AMAZON'S PROBLEM IS ACTUALLY A GOOD PROBLEM

This is why I would not simply become bearish on AMZN.

AWS growth accelerated to its fastest pace in almost five years.

Amazon says demand exceeds available capacity.

The company expects AWS growth above 40% in both 2027 and 2028.

So the problem is not:

“Nobody wants Amazon's AI infrastructure.”

It is closer to:

“Everyone wants it, and Amazon has to spend enormous amounts before it can serve them.”

That can become incredibly profitable.

But there is a timing mismatch:

CAPEX NOW

↓

CAPACITY LATER

↓

REVENUE LATER

↓

CASH FLOW EVEN LATER

And Wall Street has suddenly become much less patient with that gap.

That gap is:

THE AI PAYBACK PERIOD.

🔵 NOW COMPARE MICROSOFT

Microsoft just reported:

$92.3B revenue

up 19%.

Azure:

+48%

Microsoft Cloud:

$63.9B

Commercial RPO:

$677B

Operating cash flow for the quarter:

$58.0B

Additions to property and equipment:

$39.4B

Approximate cash remaining after those property additions:

$18.6B.

And Microsoft still returned:

$10.2B

to shareholders through dividends and buybacks.

That is the difference.

Microsoft is spending aggressively.

But the existing machine is generating enough cash to absorb much of the investment.

So Microsoft's equation currently looks more like:

AI CAPEX ↑

AZURE ↑

CLOUD REVENUE ↑

OPERATING INCOME ↑

CASH STILL POSITIVE

That is exactly what the market wants.

🥇 MY CURRENT CASH-CONVERSION WINNER: MICROSOFT

Not necessarily because Microsoft has the best AI technology.

Not because Azure must beat AWS.

But because Microsoft currently has the cleanest evidence that:

AI SPENDING AND CASH GENERATION CAN COEXIST.

Azure +48%.

Operating income +21%.

Operating cash flow $58B.

And roughly $18.6B remains after quarterly property additions.

That is what I mean by:

DEFENDABLE AI CAPEX.

The market can see both sides of the bridge.

Money goes in.

Growth comes out.

And cash remains.

🧠 META IS THE MOST INTERESTING ONE

Meta is different.

It does not primarily sell AI infrastructure.

It uses AI to make its existing advertising machine more valuable.

That means its AI return pathway is much shorter.

AI

↓

BETTER RECOMMENDATIONS

↓

MORE ENGAGEMENT

↓

MORE AD IMPRESSIONS

↓

BETTER TARGETING

↓

BETTER ADVERTISER ROI

↓

HIGHER AD PRICES

That is powerful because Meta does not necessarily need customers to buy an AI product.

AI simply needs to improve the product billions of people already use.

And look at the latest quarter.

Revenue:

$64.6B, +26%

Ad impressions:

+10%

Average price per ad:

+14%

Capex:

$34.9B

That tells me AI is already interacting with the monetisation engine.

But Meta's quarterly FCF was:

-$2.1B.

So Meta has not escaped the cash question either.

This is why I put it between Microsoft and Amazon.

The monetisation loop is extremely attractive.

But spending has become enormous.

🥈 META: FASTEST AI MONETISATION LOOP

Microsoft currently has the best cash defence.

But Meta may have the:

SHORTEST AI PAYBACK LOOP.

Why?

Microsoft builds AI infrastructure and sells compute/software.

Amazon builds infrastructure and sells compute.

Meta uses AI to improve an advertising system that already serves billions of users.

The distance between:

AI INVESTMENT

and

REVENUE

may therefore be shorter.

This is why Meta deserves very close attention.

If Meta can keep:

ad impressions ↑

price per ad ↑

while eventually slowing capex growth,

its free cash flow could snap back extremely quickly.

🥉 AMAZON: BIGGEST CASH-FLOW GAP, BIGGEST OPERATING LEVERAGE?

Amazon ranks third today.

But that does NOT mean I think Amazon has the weakest long-term opportunity.

It means Amazon currently has the:

LONGEST PAYBACK PERIOD.

Look at the numbers again.

Operating cash flow:

$168.3B

Infrastructure purchases:

$210.7B

That difference produces the ugly FCF number.

But imagine AWS keeps growing near 40%.

Eventually the capacity Amazon is building today starts generating revenue.

At that point, if capex growth merely slows rather than collapses, the cash-flow mathematics can reverse dramatically.

That creates a fascinating setup.

Amazon may currently have:

🔴 the ugliest cash conversion

but potentially:

🟢 the greatest future FCF operating leverage.

That is why I would not confuse:

NEGATIVE FCF

with

BAD BUSINESS.

The question is whether negative FCF is:

temporary investment

or

permanently poor economics.

We do not know yet.

🔥 HERE IS THE METRIC I THINK WALL STREET WILL START USING

I would call it:

AI CASH CONVERSION.

Very simply:

INCREMENTAL AI-RELATED OPERATING CASH FLOW

divided by

INCREMENTAL AI INFRASTRUCTURE INVESTMENT.

We cannot calculate this perfectly because hyperscalers do not separately disclose every AI dollar.

But directionally, it is incredibly useful.

I want to see:

cloud/advertising growth ↑

operating cash flow ↑

capex growth eventually ↓

free cash flow ↑↑

The crossover is what matters.

Because once:

INCREMENTAL CASH GENERATION > INCREMENTAL CAPEX

the AI investment narrative changes completely.

The infrastructure starts financing itself.

🚦 MY CURRENT AI CASH SCORECARD

🥇 MICROSOFT: BEST CASH DEFENCE

Azure +48%.

$58B quarterly operating cash flow.

Capex huge, but absorbable.

Existing enterprise software ecosystem monetises AI quickly.

🥈 META: BEST MONETISATION LOOP

Revenue +26%.

Ad impressions +10%.

Price per ad +14%.

AI directly improves the advertising machine.

But quarterly FCF has slipped negative as infrastructure spending explodes.

🥉 AMAZON: BEST CAPACITY OPTIONALITY, WORST CURRENT CASH CONVERSION

AWS +39%.

AWS operating income $17.8B.

Operating cash flow extremely strong.

But TTM infrastructure spending exceeds operating cash generation.

The market now wants proof that today's $220B capex becomes tomorrow's AWS cash machine.

🔥 BUT HERE IS THE PART THAT MATTERS FOR NVDA, MU AND SNDK

This hyperscaler cash-flow debate has created fear across semiconductors.

I wrote about this with Micron.

But remember:

bad cash flow for AI buyers

does NOT automatically mean

bad revenue for AI suppliers.

In fact, today's negative hyperscaler FCF exists partly because:

THEY ARE BUYING SO MUCH AI INFRASTRUCTURE.

Amazon's infrastructure purchases reached:

$210.7B TTM.

Someone is receiving that money.

GPU suppliers.

Memory suppliers.

Networking suppliers.

Storage suppliers.

Power infrastructure.

Construction.

So for NVDA, MU and SNDK investors, the signal I am watching is NOT:

“Is Amazon FCF negative?”

It is:

“DOES AMAZON RESPOND BY CUTTING CAPEX?”

So far?

Amazon is doing the opposite.

It says spending will be:

MEANINGFULLY HIGHER IN 2027.

That matters enormously.

🧠 THIS CREATES A STRANGE MARKET

Hyperscaler investors may hate higher capex.

AI supplier investors should love it.

That creates opposing incentives.

For Microsoft, Meta and Amazon:

MORE CAPEX = MORE PRESSURE ON FCF.

For Nvidia, Micron, SanDisk and the AI infrastructure ecosystem:

MORE CAPEX = MORE ORDERS.

So the next phase of the AI trade may create a divergence between:

THE SPENDERS

and

THE SELLERS.

Eventually the spenders need returns.

But until spending actually slows, the sellers can continue getting paid.

🔮 WHO DO I THINK PROVES AI RETURNS FIRST?

My ranking today:

🥇 MICROSOFT

Highest probability.

Azure growth is accelerating.

Cash generation remains enormous.

Enterprise AI monetisation is already embedded across the stack.

Probability of convincing the market first: 45%

🥈 META

Potentially the fastest monetisation pathway.

If AI continues improving engagement and ad economics, Meta could surprise dramatically.

Probability: 35%

🥉 AMAZON

AWS demand may be the strongest signal of all.

But the infrastructure bill is enormous.

Amazon needs more time for today's capex to become tomorrow's cash.

Probability: 20%

These are not probabilities of the stocks rising.

They are my estimate of who first produces a sufficiently convincing AI cash-return narrative.

🚨 WHAT WOULD CHANGE MY RANKING?

Microsoft loses first place if:

capex accelerates much faster than Azure

while

operating cash conversion deteriorates.

Meta moves into first if:

ad pricing and impressions remain strong

while

capex growth peaks

and FCF rebounds sharply.

Amazon moves dramatically higher if:

AWS stays around 40% growth

while

capex growth begins slowing.

That combination would be explosive.

Because Amazon would move from:

BUILDING CAPACITY

to

HARVESTING CAPACITY.

That is the crossover investors should watch.

MY CALL

The AI bubble debate is asking the wrong question.

The question is not:

“Are companies spending too much on AI?”

Nobody knows yet.

The better question is:

“HOW LONG UNTIL EACH DOLLAR OF AI CAPEX PRODUCES MORE THAN A DOLLAR OF ECONOMIC VALUE?”

That is measurable.

And the companies are already separating.

Today:

MICROSOFT

looks closest to proving that AI investment and enormous cash generation can coexist.

META

may have the shortest route between AI intelligence and incremental revenue.

AMAZON

has the biggest gap between today's infrastructure bill and tomorrow's potential cash flow.

But Amazon also gives us perhaps the most important signal for the wider AI trade:

IT IS NOT CUTTING SPENDING.

It plans to spend even more.

So my AI framework has changed.

I no longer ask:

Who has the biggest AI model?

Who owns the most GPUs?

Who spends the most?

I ask:

WHO GETS THEIR MONEY BACK FIRST?

Because the next trillion dollars of AI market value may not go to the company that spends the most.

It may go to the company that proves:

AI CAN PAY FOR ITSELF.

I am not a financial advisor. Trade wisely, Comrades!

Who gets there first: Microsoft, Meta or Amazon? And would you rather own the AI spender or the suppliers getting paid by them?

If you find this post interesting, give it wings! 🪽 Repost and share the insights.

Do consider ‘Follow me’ and get firsthand read of my daily new post. Thank you.

@Tiger_SG  @TigerEvents  @Daily_Discussion  @TigerStars  @TigerPM