Marvell to Acquire Lumentum for $115B in All-Stock Deal

If your AI data center plan runs on Lumentum lasers, Lumentum optical circuit switches or Marvell DSPs, you have about 12 to 18 months to get your post-close supply terms in writing. Marvell announced on November 2 that it will buy Lumentum in an all-stock deal with an implied enterprise value of $115 billion. Lumentum told customers the same day that commercial agreements, orders and support processes remain in place until the transaction closes. The email does not say what happens to them afterward.

Treat your current contract as stable and your 2028 supply as open. The rest of this piece covers what the documents commit to, what Marvell said only on an analyst call, and what to ask for while you still deal with two separate companies.

---

What Marvell Is Buying and What It Is Paying

Marvell is buying the laser and optical switching side of the AI network to sit beside its chips. Lumentum brings high-speed and high-power lasers, optical transceivers, optical circuit switching and indium phosphide manufacturing. Marvell brings electrical and optical interconnect, network switching and custom AI processors. An optical circuit switch (OCS) redirects light between fibers without first converting it to an electrical signal. A DSP, or digital signal processor, is the chip in an optical module that corrects the signal so data survives the trip.

The terms are a fixed 4.500 Marvell shares for each Lumentum share, which Marvell puts at about $1,132 per share and a 23.9% premium to the October 28 unaffected close. Lumentum holders would own about 33% of the combined company. Marvell projects more than $1 billion in annual run-rate cost synergies, and the companies claim a $550 billion addressable market, which is their own estimate.

That $1,132 had already slipped by Monday morning. With Marvell at $190.43 in early trading, 4.5 shares were worth about $857 by our arithmetic, and Lumentum traded at $788.51, roughly 8% below that. Reuters later reported Marvell down more than 10% and Lumentum down more than 15% on the day. Reuters also calculated that the price values Lumentum at about 33 times projected core operating profit, against a peer median near 16. A target trading below its deal value reflects the long wait to close and some doubt about completion, so you should not build a capacity plan on the deal closing on schedule.

What Lumentum Told Customers, and Where That Promise Ends

Lumentum's promise to customers runs to the closing date and stops there. The customer email from its head of worldwide sales says "Until the transaction closes, Marvell and Lumentum will operate as separate companies," that "product roadmaps and delivery schedules remain on track," and that "There is no action you need to take." A separate note to suppliers calls this "business as usual."

The forward-looking language is about products. The press release promises "pre-validated optical solutions" and stronger strategic supply, and it names no customers and no supply agreements. The investor presentation in Marvell's filing lists "Complete Module Solutions" and "CPO / NPO light engines," and cites the "World's largest InP wafer fab capacity." Those lines point toward selling you an integrated optical system. That may suit you. It also changes how you would be quoted, and you have no written commitment that the separate parts stay available at comparable terms.

Question you need answered	What the documents say	What they leave open
Do current agreements hold?	Yes, until close	Anything after close
When does it close?	First half of 2028, or about 12 to 18 months	Which regulators beyond U.S. antitrust review
Will parts sell separately?	Not addressed in the release or email	Pricing and allocation for standalone parts
Do rivals keep their supply?	Yes, per executives on the call	Any contractual commitment

---

Why You Should Plan Around a Range of Closing Dates

The filings give you a window that runs from late 2027 to August 2028, and a merger agreement that sets a price on the deal failing. The release says the first half of calendar 2028, while Lumentum's emails say approximately 12 to 18 months, which would start in November 2027. The merger agreement's outside date is November 1, 2027, extendable to February 1, May 1 and August 1, 2028 if regulatory approvals are still pending.

Marvell has put money behind the regulatory risk. It owes Lumentum a $3.2 billion regulatory termination fee if the deal dies on antitrust or approval grounds, about 2.8% of the enterprise value. The filing names the Hart-Scott-Rodino waiting period and "certain other required regulatory approvals" without listing them. Wedbush analyst Matt Bryson questioned whether the deal can clear, according to one market write-up, which also flagged limited advanced optical capacity as a reason regulators may look at rivals' access and named China as another potential obstacle.

Reuters quoted IG Group's Axel Rudolph on the size of the wager: "This is an enormous bet and investors are right to ask if Marvell is paying too much and taking on too much at once." Whether it closes or not, your lasers still come from a supplier that says it is short of capacity. Lumentum's optoelectronic components were described as sold out through nearly 2029 in October coverage citing Bloomberg, and secondhand accounts of the CEO's interview put unmet demand at about 70% for some products through 2027 and about 30% for certain others through 2028.

The Neutrality Question Marvell Answered Only on a Call

The two CEOs say nothing will change for customers who also compete with them, and that assurance is verbal. JPMorgan's Harlan Sur asked whether Marvell moving further into modules would push module makers toward rival DSP suppliers. According to a third-party transcript of the call, which contains visible transcription errors, Lumentum CEO Michael Hurlston answered "We supply our lasers to almost all of our competitors," and Marvell CEO Matt Murphy said of the companies' component-supply strategy, "That strategy is not going to change."

The strongest case for the deal deserves a hearing. Both companies already sell to rivals, so coexistence is routine in this market. A larger balance sheet could fund more indium phosphide capacity while Lumentum guides fiscal second-quarter revenue to $1.58 billion to $1.63 billion, above the $1.48 billion analyst average. Murphy also said on the call that tight supply is not the reason for the deal, and that customers after the Cavium and Inphi acquisitions wanted to buy more of the broader portfolio.

The market is less sure. Coherent rose about 2% on a morning when Lumentum fell 6%. Stifel estimates Marvell holds 60% to 65% of PAM4 DSPs, according to one investor note, which argues the share is exposed if module makers qualify Broadcom instead. Nvidia, which put $2 billion into each company this year, did not respond to Reuters, though Murphy called the deal "a double down on supporting Nvidia."

---

Where Second-Source Supply Stands Today

Lasers and transceivers have named alternatives, and the optical circuit switching picture is less clear. Nvidia's own March deals show the pattern. Its Lumentum agreement and its Coherent agreement were announced the same day, each with $2 billion of investment, a multibillion-dollar purchase commitment, capacity rights and nonexclusive terms. Nvidia paid to keep two laser sources eight months before this deal.

Your options by component look like this:

• For lasers and transceivers, Coherent is the direct comparison, and Applied Optoelectronics also trades in the same group. China's Zhongji Innolight is a competitor too, but a proposed U.S. ban on new Chinese optical transceivers makes it a risky plan for U.S. deployments.
• For DSPs, Broadcom is the alternative named in investor analysis of this deal.
• For optical circuit switching, Lumentum's management has called itself "really the only merchant supplier today", with backlog "well beyond $400 million." That is a vendor's description of itself. Coherent lists its own OCS in configurations up to 512x512, and Cignal AI reports that Google has seeded purchase orders to Lumentum, Coherent and Huber+Suhner, so test the claim with an RFI. If it holds, OCS is where you have the least leverage after close.

Named alternatives do not guarantee delivery dates. A rival's capacity in a short market is something you find out from a quote. If you already have reserved capacity or a GPU commitment on a fixed schedule, match the optics delivery date to it before you assume the second source helps.

You may never sign with Lumentum directly. If you buy optics inside a switch or server order, or rent them through a GPU cloud, your exposure arrives through that provider's bill of materials. We saw the same mechanism with server DRAM repricing, where a component problem reached buyers as a line on someone else's quote.

---

What to Do While You Still Deal With Two Companies

This Week:

1. Have procurement list every purchase order, long-term agreement and OEM bill of materials containing a Lumentum laser, transceiver or OCS, or a Marvell DSP. Ask each OEM to confirm the parts in writing where you cannot see them.
2. Reply to your Lumentum account team and ask two questions in writing: do our price and allocation terms survive the close, and will standalone components stay on the price list after it?
3. Have counsel read the assignment and change-of-control language in those contracts, since a stock merger may not trigger clauses drafted for an asset sale. Our review of AI vendor exit clauses lists the terms worth checking.

This Month:

1. Request written quotes with delivery dates from Coherent and at least one other source for your highest-volume optical part, covering 2027 and 2028 deliveries.
2. Ask your switch and server vendors whether their current designs are qualified with a non-Marvell DSP and a non-Lumentum laser, and get the answer by part number.
3. Run your 2028 capacity plan under three dates: a close by November 2027, a close by August 2028 and a failed deal. If the sold-out reporting holds, Lumentum is short of capacity in all three, so the difference you are testing is who sets your terms.

Before Signing Anything That Runs Past November 2027:

1. Add a clause that price, allocation and delivery commitments bind any successor.
2. Require separate line-item pricing for components sold alongside any integrated module or optical engine.
3. Get written notice periods for any end-of-life or roadmap merger, and remove any exclusivity term that would stop you qualifying a second source.

Expect pushback if you are a small buyer in a sold-out market. Ask anyway, and record the answers, because a refusal tells you how much of your plan needs a second source.

The Bottom Line

This deal would be larger than Microsoft's $68.7 billion purchase of Activision Blizzard and Dell's $67 billion purchase of EMC, and more than ten times Marvell's $10 billion Inphi deal in 2021. Murphy cites Inphi as a precedent, saying customers then wanted to buy more of the portfolio. That can be good for you and still leave you with fewer ways to buy a part on its own. Marvell's August Google agreement, worth up to $120 billion in sales through fiscal 2033, is a fair guide to who gets served first when capacity is short, and a mid-sized enterprise should not assume it is near the front.

Send the two written questions to your Lumentum account team this week, and put the replies in your 2028 capacity file.

Continue Reading

• Google-Marvell AI Chips: The Inference Economics Shift
• Five AI Vendors Give Customers 0 to 90 Days to Get Data Out
• Reserved vs Spot GPUs: Commit Six Months, Not Twelve
• GPU Clouds Compared: Nebius on Price, CoreWeave at 3AM
• Nvidia Is Losing 3 Points to RAM. Re-Price Your Server BOM.
• Nvidia Unveils DSX POD and OpenShell at GTC Berlin