To The Moon

Marvell just announced one of the biggest technology deals ever.

$115 BILLION.

The target?

Lumentum.

One of the hottest companies sitting at the intersection of:

AI

optical networking

lasers

data centres.

The strategic story sounds almost perfect.

And the market's response?

SELL BOTH.

Marvell plunged.

Lumentum plunged even harder.

That sounds like investors believe the deal is terrible.

I think the truth is more interesting.

Because on the SAME DAY this merger was announced, Lumentum reported:

REVENUE +136.5% YoY

and guided its next quarter dramatically higher.

So I don't think the market is saying:

“OPTICAL AI IS DEAD.”

I think it is asking:

“HOW MUCH IS TOO MUCH TO PAY FOR THE FUTURE?”

And that makes this deal one of the most fascinating tests yet of the AI infrastructure boom.

🔥 FIRST, UNDERSTAND WHAT MARVELL ACTUALLY BOUGHT

AI data centres have a problem.

We keep building:

faster GPUs

more accelerators

bigger clusters.

But those processors need to communicate.

As clusters grow from thousands toward hundreds of thousands and eventually millions of accelerators, moving data becomes increasingly difficult.

Copper works beautifully over shorter distances.

But eventually?

PHYSICS INTERVENES.

Power consumption rises.

Signal integrity becomes harder.

Bandwidth requirements explode.

And increasingly:

LIGHT REPLACES ELECTRONS.

That is why optical connectivity is becoming one of AI's most important bottlenecks.

And suddenly Marvell's strategy makes sense.

Marvell brings:

switching

custom silicon

electrical interconnect

optical DSP

advanced semiconductor design.

Lumentum brings:

lasers

optical transceivers

optical circuit switching

indium phosphide manufacturing

specialised photonics.

Put them together and Marvell isn't merely buying another chip company.

It is attempting to own:

THE AI DATA HIGHWAY.

From electrical signals...

to optical signals...

to lasers...

to switches...

to connectivity across the AI cluster.

That is why management calls this a:

“ONCE-IN-A-GENERATION”

opportunity.

I understand the ambition.

🚨 BUT THEN I SAW THE PRICE

Lumentum shareholders receive:

4.5 MRVL SHARES

for every LITE share.

At announcement, that implied approximately:

$1,132 PER LITE SHARE.

Enterprise value:

$115B.

Premium to the unaffected October 28 price:

23.9%.

After the deal closes:

existing MRVL holders own ~67%

LITE holders own ~33%.

And this is where investors became nervous.

Because Marvell isn't paying cash.

It is paying with:

ITS OWN STOCK.

That means existing shareholders are giving Lumentum shareholders roughly:

ONE-THIRD OF THE COMBINED COMPANY.

This is not a bolt-on acquisition.

This is effectively:

A TRANSFORMATION OF MARVELL.

And the market immediately asked:

IS MARVELL BUYING A GREAT ASSET AT TOO GREAT A PRICE?

That is a legitimate question.

🧠 BUT HERE IS THE NUMBER THAT CHANGED MY VIEW

Forget the merger for one minute.

Look at what Lumentum itself just reported.

Q1 FY2027 revenue:

$1.263B

Up:

136.5% YoY.

Systems revenue:

+228.8% YoY.

Non-GAAP gross margin:

53.2%.

Non-GAAP operating margin:

42.3%.

Then management guided Q2 revenue to:

$1.575B-$1.625B.

Think about that.

The midpoint is approximately:

$1.6B.

That's another roughly:

27% SEQUENTIAL GROWTH

after revenue already more than doubled year over year.

And expected non-GAAP operating margin?

45%-46%.

This is not a struggling company being rescued.

This is:

A SCARCITY ASSET IN AN AI BOTTLENECK.

Suddenly Marvell's willingness to pay up becomes easier to understand.

🔥 AND LUMENTUM JUST DROPPED ANOTHER MONSTER NUMBER

This one deserves more attention.

Lumentum said it secured its:

SECOND MAJOR NPO/CPO WIN

backed by purchase commitments exceeding:

$5 BILLION.

Management also says optical circuit switching could become:

A MULTI-BILLION-DOLLAR ANNUAL REVENUE STREAM.

Now the deal begins looking different.

Marvell isn't paying $115B for:

TODAY'S LUMENTUM.

It is underwriting:

WHAT LUMENTUM COULD BECOME IF OPTICS BECOMES THE NEXT AI BOTTLENECK.

That is a huge difference.

🔵 THE DEAL IS REALLY A BET ON COPPER'S LIMITS

This is my differentiated thesis.

The merger is not primarily:

MARVELL + LUMENTUM.

It is a gigantic corporate bet that:

COPPER CANNOT SCALE WITH AI FOREVER.

If that assumption is right?

Optics moves closer and closer to the processor.

From:

rack-to-rack

to:

scale-up

to:

co-packaged optics

and eventually increasingly integrated optical architectures.

Every time optics moves closer to compute?

The addressable market expands.

That is why Marvell believes the combined opportunity could reach:

$550 BILLION.

This isn't simply about selling more transceivers.

It is about owning more of the:

CONNECTIVITY STACK.

🔥 AND THIS CONNECTS DIRECTLY TO NVIDIA

Everyone focuses on GPUs.

But AI performance increasingly depends on the entire system.

A $50,000 accelerator sitting idle because data cannot reach it efficiently is:

A VERY EXPENSIVE PAPERWEIGHT.

As compute becomes faster, connectivity becomes more valuable.

This is the same principle behind our memory thesis.

GPU performance increases.

Then the bottleneck moves.

First:

COMPUTE.

Then:

MEMORY.

Then:

NETWORKING.

Then:

OPTICS.

The AI trade doesn't disappear.

THE BOTTLENECK MIGRATES.

That is why I find this merger strategically fascinating.

🧠 MARVELL MAY BE TRYING TO GET AHEAD OF THE NEXT BOTTLENECK

The combined company would span:

custom XPU silicon

switching

electrical interconnect

optical DSP

silicon photonics

lasers

optical switching

compound semiconductors.

That creates something much broader than either company individually.

And management estimates:

$1B+ ANNUAL RUN-RATE COST SYNERGIES.

More interestingly?

The combined companies currently represent roughly:

$17.7B ANNUALISED REVENUE

with approximately:

39% NON-GAAP OPERATING MARGIN.

Using Lumentum's next-quarter guidance and current Marvell expectations?

Management says the combined annualised run rate could already exceed:

$21B

with operating margins in the:

LOW 40% RANGE.

That's important.

Because this isn't merely:

TAM STORYTELLING.

There are already serious economics underneath it.

🚨 SO WHY DID BOTH STOCKS COLLAPSE?

Because there are THREE separate risks.

1️⃣ PRICE RISK

Marvell is paying an enormous valuation for Lumentum after an extraordinary run.

Great company.

Great market.

Potentially:

GREAT PRICE TOO.

History is full of acquisitions where the strategic thesis was correct but shareholders still lost because the buyer overpaid.

2️⃣ DILUTION RISK

Lumentum shareholders receive 33% of the combined company.

Existing MRVL shareholders have to believe Lumentum's future growth is worth surrendering that ownership.

That is a very high bar.

3️⃣ TIME RISK

This is the one I think investors are underestimating.

The deal isn't expected to close until:

H1 2028.

That is a LONG time in AI.

Between now and then:

technology changes

competitors respond

regulators investigate

valuations move

customer architectures evolve.

A $115B transaction with an 18-month-plus closing runway deserves a:

RISK DISCOUNT.

🚨 AND HERE IS THE WEIRD PART ABOUT LITE

Because consideration is fixed at:

4.5 MRVL SHARES

the ultimate value Lumentum shareholders receive moves with:

MRVL'S SHARE PRICE.

This is crucial.

The $1,132 headline number is NOT guaranteed cash.

If MRVL falls?

The implied deal value falls.

So when investors sell MRVL because they dislike the acquisition...

they mechanically reduce the value of what LITE shareholders expect to receive.

Then LITE falls too.

That creates:

A NEGATIVE FEEDBACK LOOP.

MRVL ↓

↓

Deal consideration value ↓

↓

LITE ↓

That helps explain why BOTH stocks can fall even though Lumentum is technically the company being acquired.

🔥 THIS IS WHY I WOULD NOT CALL THE DEAL “FAILED”

Not even close.

The agreement has been signed.

Both boards unanimously approved it.

The transaction now needs:

shareholder approval

regulatory approval

customary closing conditions.

Expected close:

H1 2028.

So what failed Monday?

Not the merger.

THE MARKET'S WILLINGNESS TO PAY THE ANNOUNCEMENT PRICE.

Very different thing.

📍 SO WHICH STOCK WOULD I BUY?

This is where my answer may surprise you.

Between the two?

I PREFER MRVL.

Not LITE.

Why?

Because LITE is now partly:

A MERGER ARBITRAGE SECURITY.

Its upside becomes linked to:

MRVL's share price

regulatory approval

deal completion

time to close.

Marvell, meanwhile, gives me exposure to the underlying strategic thesis:

AI CONNECTIVITY BECOMES A BOTTLENECK.

And after the selloff?

I am at least getting a better price for taking that risk.

Based on the roughly $173 area shown now:

🟡 $170-$180

WATCH / SMALL STARTER

I would not go large immediately.

The market is still digesting the deal.

🟢 $155-$165

FIRST PICK

This becomes much more interesting if:

Marvell's core AI growth remains intact

Lumentum keeps delivering

the merger thesis remains unchanged.

🟢 $140-$150

PREFERRED FEAR ZONE

At this point I would seriously examine whether deal anxiety has overshot the deterioration in fundamental value.

But:

🔴 BELOW $140

THESIS CHECK

I would want to know whether we are seeing:

regulatory problems

customer issues

AI infrastructure slowdown

deteriorating Marvell fundamentals

rather than simple merger indigestion.

🔮 MY THREE SCENARIOS

🟢 BULL CASE: 30%

Lumentum continues extraordinary growth.

Marvell's AI business remains strong.

Investors begin appreciating the full-stack connectivity strategy.

The deal progresses smoothly.

MRVL could reclaim:

$190-$205

then:

$220+.

🟡 BASE CASE: 50%

The strategic thesis remains strong.

But investors continue applying a discount for:

dilution

valuation

execution

regulatory uncertainty.

MRVL spends time around:

$155-$190

while earnings gradually prove the deal.

This is my highest-probability outcome.

🔴 BEAR CASE: 20%

AI infrastructure spending slows.

Optical valuations compress.

Regulatory risk rises.

Lumentum's growth decelerates.

Or integration economics begin looking less attractive.

Then:

$140-$155

could arrive.

And below $140?

I reassess.

🔥 BUT HERE IS THE BIGGER INVESTMENT LESSON

We keep asking:

“IS THE AI BUBBLE ENDING?”

I think that question is becoming less useful.

Watch where the money is moving.

Microsoft is spending on infrastructure.

Google is spending on infrastructure.

Meta is spending on infrastructure.

OpenAI is committing gigawatts.

AMD is building rack-scale systems.

Micron is selling memory.

SanDisk is selling storage.

Nvidia is selling compute.

And now Marvell is willing to effectively reshape its entire company around:

CONNECTIVITY.

Companies don't make $115B strategic bets because AI infrastructure requirements are getting smaller.

They do it because they believe the bottleneck is moving.

And THAT is what I think investors should take from this deal.

AI'S NEXT WAR MAY NOT BE ABOUT WHO HAS THE FASTEST CHIP.

It may increasingly be about:

WHO CAN MOVE THE DATA FAST ENOUGH TO KEEP THOSE CHIPS BUSY.

That makes:

networking

memory

storage

optics

increasingly strategic.

Which brings us back to the investment framework I've been developing.

Don't simply buy:

“AI.”

Find:

THE BOTTLENECK.

Then ask:

WHO GETS PAID TO REMOVE IT?

Marvell just placed:

$115 BILLION

on its answer.

OPTICS.

The market hates the price.

I understand why.

But I am not ready to dismiss the thesis.

In fact?

After the selloff:

THE THESIS INTERESTS ME MORE THAN THE ANNOUNCEMENT DID.

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

Is Marvell paying far too much for Lumentum, or will $115B eventually look cheap if optics becomes AI's next major bottleneck?

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.

@TigerPM  @TigerStars  @Daily_Discussion  @TigerEvents  @Tiger_SG