To The Moon

There was a time when reporting 221% growth in AI revenue would have been enough to send almost any semiconductor stock flying.

Broadcom just did exactly that.

The stock fell anyway.

And I think that tells us something much bigger about where the AI trade has reached.

Broadcom didn’t report a weak quarter.

It reported:

📈 Q3 revenue: US$29.6B, +86% YoY

🤖 AI semiconductor revenue: US$16.7B, +221% YoY

💰 Non-GAAP EPS: US$3.32, +96% YoY

💵 Free cash flow: US$13.7B

🚀 Q4 AI revenue guidance: US$21.7B, +236% YoY

Broadcom also expects AI semiconductor revenue to reach approximately US$115B in FY2027 and potentially US$230B in FY2028.

Read those numbers again.

Then ask yourself:

What exactly does an AI company have to do now to impress Wall Street?

Because I don’t think Broadcom’s biggest problem is growth.

I think its biggest problem might be expectations.

🧠 THE AI TRADE HAS ENTERED PHASE THREE

The way I see it, the AI trade has moved through three stages.

Phase One: Tell investors you have AI exposure.

That was enough.

Phase Two: Prove AI is generating serious revenue growth.

The winners separated themselves from the hype.

Now I think we’re entering:

Phase Three: Prove the growth can exceed expectations that have already become enormous.

That’s a completely different market.

Broadcom’s Q4 revenue guidance is approximately US$34.8B, representing around 93% YoY growth.

Sounds ridiculous.

But Wall Street’s consensus was around US$35.03B.

So a company can effectively tell investors:

“We’re going to grow quarterly revenue roughly 93%.”

And the market can respond:

“Yeah… we wanted more.”

Welcome to AI investing in 2026. 😂

🚨 THE NUMBER THAT MATTERS MORE THAN 221%

The headline number is Broadcom’s 221% AI growth.

But I’m watching something else.

Margins.

Broadcom’s Q3 gross margin was approximately 75%, down about 210 basis points sequentially as AI semiconductors became a larger proportion of revenue.

Management expects consolidated gross margin around 73% in Q4 as XPUs become an even larger part of the mix.

That’s fascinating.

AI is simultaneously becoming Broadcom’s greatest growth engine and changing the economics of the revenue mix.

So the next stage of the Broadcom story isn’t simply:

Can AI revenue keep growing?

It is:

How much profitable growth can Broadcom extract from every additional dollar of AI infrastructure spending?

That’s a much harder question.

💰 BUT THEN THERE’S US$13.7 BILLION OF FREE CASH FLOW

This is where the bear argument gets complicated.

Broadcom generated approximately US$13.7B in free cash flow in ONE QUARTER, equal to roughly 46% of revenue.

This isn’t an AI company burning mountains of cash while promising profitability somewhere over the horizon.

The machine is already producing enormous cash.

Operating income is growing.

AI semiconductor revenue is exploding.

Networking is expanding.

Custom accelerators are scaling.

And Broadcom isn’t dependent on one AI customer.

Its custom silicon ecosystem touches companies including Google, Meta, OpenAI and Anthropic. Broadcom says it has visibility into major infrastructure deployments extending through 2028.

That makes the post-earnings reaction even more interesting.

⚔️ NVIDIA VS BROADCOM MAY BE THE WRONG QUESTION

I keep seeing the AI semiconductor discussion reduced to:

NVIDIA or Broadcom?

I think that’s increasingly the wrong framework.

NVIDIA dominates general-purpose accelerated computing.

Broadcom is increasingly positioned around custom AI accelerators plus the networking infrastructure connecting enormous AI clusters.

Those markets can grow together.

In fact, the bigger AI infrastructure becomes, the more hyperscalers may want specialised silicon for specific workloads while still deploying huge quantities of GPUs.

The AI infrastructure pie could become large enough for multiple architectures to win.

And Broadcom’s numbers suggest that diversification is already happening.

🐻 THE BEAR CASE ISN’T STUPID

This doesn’t mean Broadcom automatically becomes a buy at any price.

There are legitimate risks.

AI expectations are enormous.

Gross margins are being pressured by the changing product mix.

Hyperscaler spending eventually needs to produce economic returns.

Competition in custom silicon is intensifying.

And when investors already expect extraordinary growth, even extraordinary execution can disappoint.

That last point matters most to me.

A brilliant company can still become a bad investment if the price assumes too much brilliance.

We’ve already watched this happen repeatedly across the AI trade.

🐂 BUT THE BULL CASE JUST GOT BIGGER TOO

Broadcom now expects approximately:

FY2026 AI revenue: US$58B

FY2027: ~US$115B

FY2028: ~US$230B

If Broadcom gets anywhere near that trajectory, we’re not talking about a side business anymore.

We’re watching the potential creation of one of the largest AI infrastructure businesses on Earth.

And management says demand currently exceeds its FY2027 outlook.

That deserves attention.

🎯 MY TAKE

Broadcom’s earnings reaction may tell us more about the AI market than Broadcom itself.

The first phase of the AI boom rewarded possibility.

The second rewarded growth.

The next phase may reward something much harder:

Growth relative to expectations, margins, cash generation and return on invested capital.

Broadcom grew AI revenue 221% and still couldn’t satisfy the market.

That’s not necessarily evidence that the AI boom is ending.

It might be evidence that the easy part of the AI trade is ending.

From here, investors may have to separate:

Great company.

Great business.

Great growth.

and

Great investment at today’s price.

Those four things are not always the same.

And perhaps that’s the real lesson from Broadcom.

The company didn’t fail the AI test.

The test changed.

⸻

👇 GAZADZ COMMUNITY QUESTION

If 221% AI revenue growth isn’t enough to impress the market, what matters most from here?

A) Faster AI revenue growth 🚀

B) Protecting margins 💰

C) Free cash flow and AI returns 📊

D) Valuation, because expectations have simply gone too far ⚠️

And the bigger question:

Would you rather own $AVGO, $NVDA, or both for the next stage of the AI infrastructure buildout?

$AVGO $NVDA $GOOGL $META $AMZN

Personal market analysis only. Not financial advice. Always do your own research.