Marvell-Lumentum Talks: Tea Leaves For Photonics Stocks

 Late Wednesday, Bloomberg reported that Marvell Marvell Technology (MRVL) is in advanced talks to buy Lumentum $LITE. The talk values Lumentum at roughly $83 billion, and a deal could be announced as soon as next week. Nothing is signed. No terms are public. The tape traded it like a done deal anyway. $LITE ran up double digits after hours, $MRVL dropped about 9%, & every laser, transceiver and photonic-chip name in the book got marked up on takeout math. We've been on this trade since Nvidia's $96 billion quarter repriced it. Here's our read on what the biggest bid the sector has ever seen says about where photonics goes next. The earnings gauntlet opens Monday, and POET POET Technologies (POET) goes last on Nov 16.

The chip side is buying the light

Start with the pattern, because the pattern is the whole story. On March 2, Nvidia Nvidia (NVDA) put $2 billion into Lumentum and another $2 billion into Coherent $COHR on the same day. Both deals came with multibillion-dollar purchase commitments and rights to future laser capacity. A month before that, Marvell closed its purchase of Celestial AI, a photonic-fabric startup. Credo $CRDO bought Dust Photonics for its silicon photonics chips. Now Marvell wants the whole laser factory.

Why? Because the supplier is sold out. Lumentum CEO Michael Hurlston said on Oct 8 that the company can't meet an estimated 70% of demand for some products through 2027. He's described its optical parts as sold out through early 2029. When the guy who makes your lasers tells you he's booked for three years, you don't haggle. You buy the company. Nvidia put a deposit on both houses in March. Marvell wants the deed.

Think about what $83 billion means in a sector this size. Lumentum did $1.01 billion in revenue last quarter, up 109% from a year ago. That bid is about 20 times annualized sales. The chip companies have decided that owning light is worth more than renting it. That's not a trade. That's a land grab.

Buyer pays, seller gets paid

The reaction split the way M&A reactions always split. $LITE jumped 11.8% after hours to about $1,021 and was still up 3% at $941 on Thursday. $MRVL fell 9% after hours and was down 11.4% at $223 on Thursday. Bernstein pointed out that Marvell has about $3.92 billion of cash against an $83 billion target, which means stock, debt or both. Stifel said even a signed deal would take a while to clear U.S. and Chinese regulators. So the market priced the dilution into the buyer & the premium into the seller, and it did it in about forty minutes.

Here's the part that matters for the rest of the group. Coherent $COHR added about 1% to $242. Applied Optoelectronics $AAOI sat flat at $82. The market did not mark every optics name as a target. It paid the premium to the one company with a bid on the table and told everyone else to wait for a price. And it did that on a red tape. $QQQ was down 1.2% on the Alphabet hangover, iShares Semiconductor ETF (SOXX) down 0.9%, Micron Micron Technology (MU) down 3.5%, and the optics names were green or flat anyway. The group is trading its own story now, not the index. That's the discipline to carry into earnings. A disclosed deal price becomes the comp sheet for every laser and transceiver stock in the book. Until that number prints, it's rumor premium, and rumor premium leaves faster than it shows up.

The earnings gauntlet starts Monday

Fabrinet $FN opens it on Nov 2. Coherent reports Nov 4 after the close. Lumentum and AOI both report Nov 5 after the close, $LITE on a 5:00 p.m. ET call & $AAOI at 4:30. POET closes it on Nov 16 with a 7:00 a.m. pre-market release and no call.

The demand side already showed up this week. Microsoft $MSFT grew Azure 48% and doubled cash capex to $39.4 billion. Alphabet $GOOGL spent $51.4 billion in the quarter, took its 2026 guide to $195 to $205 billion, and said 2027 goes up significantly from there. Amazon $AMZN reports tonight. The money that buys lasers is not slowing down.

What we're guarding against is the May setup. In May, Lumentum, Coherent & Fabrinet all beat, all posted record revenue, and all three sold off after hours anyway. The prints didn't move the stocks. The guides did, and the capacity commentary did. Lumentum's own guide for this quarter is $1.225 to $1.275 billion with non-GAAP operating margin near 40%. Beating that is table stakes. The number that moves the stock is the December-quarter guide, plus whatever Hurlston says about 2027 capacity, if he can say anything at all with Marvell in the room.

For AOI the question is simpler. Second-quarter revenue was $192 million, up 86% from a year ago. The Street wants the hyperscaler ramp to keep stacking in the Q4 guide. If it does, $AAOI stops trading as the everyone-but-the-target name and earns a premium of its own. If the guide is flat, it stays parked at $82.

Where POET sits in this

Let's be clear about what we're not saying. We're not calling POET a takeout target. Nobody is bidding, and we don't write fiction. What we're doing is reading what an $83 billion bid for a laser company says about the lane POET plays in, with the stock flat at $5.83 on a red tape and the print 11 trading days out. Four tea leaves.

One: light sources are the choke point, and somebody just paid up to own one. POET's products are optical engines and light sources built on its optical interposer. Blazar, its external light source, is the product POET's own AGM put on an at-scale timeline of 2028. The Marvell bid says the biggest chip designers in the world think the light source is worth more than the procurement budget can rent. POET is a light-source company. That's the lane.

Two: POET already has a Marvell chapter, and it's closed. On April 27, POET disclosed that Marvell, which had acquired Celestial AI, cancelled all of POET's Celestial AI purchase orders and cited confidentiality. The stock got cut nearly in half intraday. A Marvell that owns Lumentum's laser fabs has even less reason to shop outside for light. POET's path doesn't run through Marvell. It runs through its own customers. Lumilens placed a $50 million initial order under a framework worth up to $500 million over five years, then came out of stealth with a $700 million raise. Mitsubishi is on 1.6T engines. Lessengers is on the list too.

Three: consolidation cuts both ways for an independent. When Nvidia and Marvell lock up the incumbent laser supply, the top of the market goes captive. Fewer open doors. But every customer that isn't Nvidia or Marvell, the hyperscalers building their own stacks & the startups like Lumilens, now needs a light source nobody else owns. That's the opening an independent lives on, and it's the question POET's Nov 16 release has to answer with shipments and revenue, not slides.

Four: the market has flipped from paying for cash to demanding revenue. We wrote that at $5.38. At $5.83 it still reads the same. The whole company is worth about $1.0 billion at this price, and $796.3 million of it was cash and short-term investments at June 30. On 173 million basic shares that's about $4.60 a share in cash. It leaves roughly $1.20 a share for the business on gross cash, and about $1.40 on the net-cash math we used last week. Warrants, options and RSUs take the fully diluted count to 264 million, so run your own dilution math. We don't publish a price target, and we won't start now. We'll just note the gap: a buck and change for the business, in the same week the sector's biggest buyer is paying 20 times sales for light. That gap is the whole debate. Nov 16 is where POET gets to argue its side.

What we're watching

1. If Marvell and Lumentum sign next week, the deal price becomes the comp for every laser and transceiver stock, and $COHR gets re-marked first. If the talks collapse, $LITE gives the premium back and the group follows it down.

2. If Lumentum guides the December quarter above its own $1.275 billion top end and talks 2027 capacity, the shortage trade holds. If the guide only matches, we expect a May-style sell-the-news, deal or no deal.

3. If AOI's Q4 guide shows the hyperscaler ramp stacking on Q2's $192 million, $AAOI breaks out of its $82 holding pattern. A flat guide leaves it in the pack.

4. If Amazon raises capex again tonight, the demand side of the optics trade gets its third confirmation in 24 hours. The Nov 4 and Nov 5 prints then walk in with a tailwind.

5. On Nov 16, POET's release needs to show the ramp converting into shipments and revenue. If it does, the cash-versus-business split we keep writing about finally has a numerator. If it doesn't, the stock trades on the balance sheet again until the first call in March.

Subscribe to StockAlpha.ai for the earnings-week coverage, the gamma maps and the POET Nov 16 read as it happens.