To The Moon

Apple and AMD report this week.

Apple has the iPhone replacement cycle.

AMD has something potentially much bigger:

GIGAWATTS OF AI DEMAND.

OpenAI.

Meta.

Anthropic.

Microsoft.

These are not speculative customer names.

AMD has built one of the most impressive AI customer pipelines outside Nvidia.

Yet AMD around:

$516

creates a very different problem.

The market no longer needs proof that customers WANT AMD chips.

It needs proof that AMD can turn those commitments into:

REVENUE → MARGIN → CASH.

That is the real earnings test.

🔥 AMD'S AI STORY HAS QUIETLY CHANGED

A year ago, investors were asking:

Can AMD genuinely challenge Nvidia?

I think that question is becoming outdated.

Look at what has happened.

OpenAI intends to deploy up to:

6 GW

of AMD GPUs.

Meta intends to deploy up to:

6 GW.

Anthropic:

UP TO 2 GW.

Microsoft is deploying AMD Helios systems across Azure.

And just days ago, AMD disclosed that initial Helios rack-scale systems are already installed and being tested at Meta.

The question is therefore no longer:

“CAN AMD FIND AI CUSTOMERS?”

It clearly can.

The question has moved to:

“HOW FAST CAN AMD MONETISE THEM?”

That distinction matters enormously.

🧠 BECAUSE 14 GW IS NOT 14 GW OF REVENUE

This is where I think investors need discipline.

Gigawatt commitments sound enormous.

They are enormous.

But a deployment agreement is not the same thing as:

GPUs shipped

revenue recognised

gross profit earned

cash collected.

The market can price future revenue much faster than AMD can physically manufacture and deploy the systems.

That creates what I call:

THE AI CONVERSION GAP.

Customer commitment sits on one side.

Actual earnings sit on the other.

AMD's valuation increasingly depends on closing that gap.

🔥 TOMORROW'S HEADLINE NUMBERS ARE ALMOST SECONDARY

AMD guided Q3 revenue to approximately:

$13.0B ± $300M

with:

56% NON-GAAP GROSS MARGIN.

Consensus is roughly:

$13.1B revenue

$1.95 EPS.

Can AMD beat those?

Probably.

But a small beat would not impress me very much.

Why?

Because expectations have already moved.

AMD shares have had an extraordinary run.

When a stock rerates this aggressively, the market stops asking:

“Did you beat?”

It starts asking:

“HOW MUCH BIGGER DOES 2027 BECOME?”

That is where tomorrow's real battle sits.

🚨 THE MOST IMPORTANT NUMBER MAY BE 2027 DATA CENTER GROWTH

AMD previously said it expects Data Center revenue to:

MORE THAN DOUBLE IN 2027.

That is the number I want defended.

Better still?

Raised.

Because Q2 already showed what the operating leverage can look like.

Data Center revenue:

$6.7B

up:

107% YoY.

Data Center operating income:

$2.1B

versus a loss a year earlier.

Operating margin:

31%.

That is critical.

AMD isn't merely generating AI revenue.

Its Data Center economics are becoming substantially better.

And THIS is where my thesis becomes interesting.

🔵 AMD DOES NOT NEED TO BEAT NVIDIA

I think investors make a mistake when they frame AMD as:

“the company trying to dethrone Nvidia.”

It doesn't need to.

The AI accelerator market could become so large that AMD can create enormous shareholder value while remaining:

NUMBER TWO.

Suppose Nvidia remains dominant.

But hyperscalers increasingly want:

supply diversification

negotiating leverage

workload-specific alternatives

lower inference costs

open software ecosystems.

AMD doesn't need 50% market share.

It needs a sufficiently large share of a rapidly expanding market.

Think of it this way.

If the AI compute pie becomes:

MASSIVE

then:

15% of massive

can be worth far more than:

30% of a smaller market.

So my AMD thesis isn't:

AMD BEATS NVIDIA.

It is:

THE AI MARKET BECOMES LARGE ENOUGH FOR A VERY PROFITABLE NUMBER TWO.

🔥 BUT AMD HAS ONE PROBLEM NVIDIA DOESN'T

Nvidia has already proven its economics.

AMD is still proving them.

That's why:

GROSS MARGIN

may matter more than revenue tomorrow.

AMD guided:

56%.

For comparison, the strategic question is whether AI accelerators and rack-scale systems can steadily lift AMD's overall economics as volumes scale.

If revenue explodes but margins don't improve?

Then AMD may be buying market share.

If revenue explodes AND margins expand?

That tells us:

AMD IS BUILDING PRICING POWER.

That deserves a very different valuation.

So my earnings hierarchy is:

Revenue beat?

Good.

AI orders?

Already known.

MI450 / Helios ramp?

Important.

2027 Data Center outlook?

Very important.

But:

MARGIN EXPANSION?

Potentially the most important of all.

🧠 THE META DEVELOPMENT CHANGES TOMORROW'S CALL

Meta matters because we are moving from:

ANNOUNCEMENT

to:

DEPLOYMENT.

Initial Helios systems are already installed and being tested.

That's exactly what I want to see.

The AI conversion chain is beginning:

Agreement

↓

System qualification

↓

Deployment

↓

Volume shipments

↓

Revenue

↓

PROFIT

Every step AMD moves down that chain reduces execution risk.

So tomorrow I want Lisa Su to tell us:

HOW QUICKLY HELIOS MOVES FROM TESTING TO SCALE.

That's more useful to me than another giant customer announcement.

🔥 AND DON'T IGNORE EPYC

Everyone will talk about GPUs.

But AMD's CPU business could quietly be the earnings stabiliser.

Management expects server revenue to grow more than:

80% YoY

in the second half of 2026.

Agentic AI doesn't eliminate CPUs.

AI infrastructure needs:

accelerators

host CPUs

networking

memory

storage.

AMD participates in multiple layers.

That means its AI opportunity is broader than:

“SELL MORE GPUs.”

And if EPYC surprises positively while Instinct/Helios ramps?

AMD gets something I like:

TWO DATA CENTER ENGINES FIRING TOGETHER.

🚨 BUT HERE IS WHAT WOULD MAKE ME NERVOUS

At $516, I don't want:

“strong demand.”

I already know demand is strong.

I don't want:

“exciting customer conversations.”

We already have the customers.

I want:

EXECUTION.

Specifically:

🔴 Helios deployment pushed out

🔴 MI450 volume timing slips

🔴 2027 Data Center growth language softens

🔴 gross margin stalls despite AI mix increasing

🔴 operating expenses rise faster than earnings

🔴 cash generation fails to follow profit growth.

Any combination of these would tell me:

THE AI CONVERSION GAP IS NOT CLOSING FAST ENOUGH.

At this valuation?

The market would punish that.

📊 MY AMD EARNINGS SCORECARD

Q3 REVENUE

🟢 > $13.3B

Strong.

NON-GAAP GROSS MARGIN

🟢 > 56%

More important than a small revenue beat.

DATA CENTER

🟢 Continued acceleration.

EPYC

🟢 Server growth remains exceptional.

HELIOS / MI450

🟢 Volume ramp on or ahead of schedule.

2027 DATA CENTER OUTLOOK

🟢 More-than-double thesis reaffirmed or strengthened.

CASH CONVERSION

🟡 Needs to increasingly follow earnings growth.

This last one matters.

Because we just saw Microsoft and Google teach us the next rule of AI investing:

AI GROWTH EVENTUALLY HAS TO BECOME CASH.

AMD is no exception.

🔥 AND HERE IS THE MU / SNDK CONNECTION AGAIN

Look at what AMD's customers are building.

OpenAI.

Meta.

Anthropic.

Microsoft.

Gigawatts of AI infrastructure.

Those racks don't contain only AMD accelerators.

They need:

HBM

server DRAM

enterprise SSDs

networking

power.

So if AMD confirms that Helios deployment is accelerating?

I would treat it as another cross-check for:

MU

and:

SNDK.

AMD earnings therefore aren't merely about AMD.

They are another read on whether the broader AI infrastructure cycle remains intact.

📍 HOW I WOULD TREAT AMD AROUND $516

After this year's enormous run, I would NOT aggressively chase AMD into earnings.

The valuation already assumes substantial 2027 success.

So again, I want:

PROOF OR FEAR.

🟢 ABOVE $540-$550

BREAKOUT CONFIRMATION

If AMD beats, margins impress and management strengthens the 2027 AI outlook, a clean reclaim here would tell me institutions are willing to pay for the next leg.

Then:

$580-$600

becomes the next battlefield.

🟡 $490-$515

HOLD / SMALL ADD

Fine if earnings are strong but guidance merely meets elevated expectations.

🟢 $450-$480

FIRST PICK

This becomes interesting if the stock sells off on expectations while the 2027 thesis remains intact.

🟢 $410-$440

PREFERRED PICK

This is where risk/reward becomes substantially more attractive to me if:

Helios remains on schedule

Data Center growth remains strong

margins remain intact.

🔴 BELOW $400

THESIS CHECK

I would want to know whether something fundamental has changed in the 2027 deployment curve.

🔮 MY THREE SCENARIOS

🟢 BULL CASE: 30%

AMD beats.

Gross margin expands.

Helios is ahead of schedule.

2027 Data Center guidance strengthens.

EPYC remains exceptional.

Then:

$550-$600

becomes realistic.

🟡 BASE CASE: 50%

Good quarter.

Solid beat.

56%-ish margins.

Helios progressing.

But management doesn't materially raise 2027 expectations.

Then I expect:

$480-$550

as the stock digests how much future growth is already priced.

🔴 BEAR CASE: 20%

Revenue beats but margin disappoints.

Helios timing slips.

Or management becomes less aggressive about 2027.

Then the market may suddenly remember how much AMD has already risen.

$450-$480

could arrive quickly.

And a genuine execution problem could open:

$410-$440.

MY CALL

Tomorrow isn't really an earnings test.

It is:

AN AI CONVERSION TEST.

AMD has already won something extremely difficult:

CUSTOMER VALIDATION.

OpenAI.

Meta.

Anthropic.

Microsoft.

The next challenge is harder.

Turn:

GIGAWATTS

into:

SHIPMENTS.

Turn shipments into:

REVENUE.

Turn revenue into:

MARGIN.

Turn margin into:

CASH.

If AMD proves that chain tomorrow?

Then $516 may not be as expensive as it looks.

Because the market will stop valuing AMD primarily on today's earnings and begin underwriting:

THE 2027 AI REVENUE INFLECTION.

But if AMD gives us another quarter of exciting demand without faster economic conversion?

I would not chase.

At this stage of the AI cycle:

ORDERS ARE NO LONGER ENOUGH.

We already know customers want the chips.

Now?

SHOW ME THE MONEY.

That is AMD's watershed.

And interestingly, it is the same watershed we just identified across Microsoft and Google.

The AI trade is changing.

From:

WHO HAS AI?

to:

WHO CAN BUILD AI?

and finally:

WHO CAN TURN AI INTO CASH FASTEST?

AMD has the orders.

Tomorrow we find out whether the economics are catching up.

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

AMD investors: would you rather see a bigger revenue beat tomorrow, or a smaller beat accompanied by higher margins and stronger 2027 guidance?

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