Marvell agrees to buy Lumentum for $115 billion in an all-stock bet on optical AI networks

What Marvell and Lumentum agreed to

Marvell Technology and Lumentum said on November 2 that they had signed a definitive agreement for Marvell to acquire Lumentum in an all-stock transaction, disclosed in a joint press release and confirmed in filings coverage from Reuters and CNBC.

Under the terms, each Lumentum share converts into a fixed 4.500 shares of Marvell common stock. The companies put the implied enterprise value at $115 billion, or approximately $1,132 per Lumentum share, a premium of about 23.9% to each company's unaffected closing price on October 28, 2026, the last trading day before speculation about a transaction began. Because the exchange ratio is fixed and there is no cash component, the dollar value moves with Marvell's share price until closing.

Ownership splits roughly two to one. Marvell's current stockholders will hold about 67% of the combined company on a fully diluted basis and Lumentum's about 33%. Both boards approved the deal unanimously. Matt Murphy stays on as chairman and chief executive, and three Lumentum directors, including chief executive Michael Hurlston, join a Marvell board that currently has seven members.

The companies expect to close in the first half of calendar 2028, subject to stockholder votes and regulatory approval. That is a long runway: roughly 18 months in which the AI capital cycle can turn, competitors can ship, and the two companies have to keep selling against each other's roadmaps while operating as separate firms.

The argument for optics is a bandwidth argument

Marvell's case, set out in the release, is that AI infrastructure has hit the physical limits of copper. Bandwidth requirements are rising faster than copper links can carry data over useful distances, so the industry is shifting from discrete, modular components toward integrated optical systems that require electrical and optical engineers to design together.

"AI infrastructure is undergoing a fundamental architectural shift toward optical connectivity, creating an opportunity that comes along once in a generation," Murphy said in the release. "The companies that move first and help define this transition will set the standard for decades to come."

That argument is not new. What is new is that Marvell is buying the laser and transceiver supplier rather than partnering with it. Marvell brings electrical and optical interconnect, network switching, custom XPU work and advanced packaging; Lumentum brings high-performance lasers, optical transceivers, optical circuit switching and compound semiconductor manufacturing. The stated aim is co-design across the whole signal chain, from links between data centers down to connections inside advanced packages.

Hurlston framed the size of the prize in the same release, describing "a market opportunity of $550 billion" and arguing that "surging AI data traffic requires rapid hardware innovation." Marvell repeated the $550 billion total addressable market figure in its own materials, according to SiliconANGLE, alongside an expectation of more than $1 billion in annual run-rate cost savings and accretion to adjusted earnings per share within a year of closing.

What Lumentum actually makes

The deal is easier to judge with the product list in hand. Lumentum sells indium phosphide lasers, a compound semiconductor approach that emits light more efficiently than silicon. Its line includes modulated lasers that physically connect servers to a data center's optical network and continuous wave lasers that drive longer-range links, according to SiliconANGLE.

It also makes single-mode optical amplifiers that boost laser signals travelling through long-haul carrier networks, indium gallium arsenide analog photodetectors used to monitor optical links for errors, and optical circuit switches that steer traffic with tiny mirrors instead of converting light back into electrical signals, which saves power.

There is a piece of Lumentum that does not fit the AI story at all: high-power lasers for cutting and welding machines. SiliconANGLE notes that it is unclear what happens to that industrial portfolio after closing, since acquirers sometimes spin off units that do not match the core focus. Anyone modelling the combined company should not simply add the two revenue lines together and assume the industrial business stays.

Investors marked both stocks down

The market did not treat the announcement as a win for either side on day one. Marvell shares fell about 13% and Lumentum shares nearly 19% in Monday trading, CNBC reported. Both had run hard into the deal: Marvell more than doubled over the prior year before the sell-off and Lumentum quadrupled over the same stretch.

Part of the reaction is arithmetic. A fixed-ratio all-stock deal transfers Marvell's own currency risk to Lumentum holders, and a 23.9% premium to an unaffected price is a real number when the buyer's stock drops double digits on the same day. Part of it is the broader question hanging over the sector: whether the infrastructure spending that produced those returns holds up.

Murphy, speaking on CNBC's "Squawk on the Street," argued the combination is about integration rather than consolidation for its own sake. "You need scale-up technology that needs semiconductors and optical working seamlessly together," he said. "We are going to create a new category of company here." He also described demand for the combined portfolio as sold out: "the reason that they're sold out is because the demand is exploding, and we don't see that changing anytime soon."

The Nvidia question

CNBC framed the deal against Nvidia's role as an investor across the AI supply chain, including stakes in companies that buy its chips or rent out compute, a pattern critics describe as a company effectively funding its own demand. Murphy answered that directly, saying the deal is "a double down on supporting Nvidia, who is the ecosystem leader in this area," and that bringing two key suppliers together should benefit that ecosystem.

The strategic logic cuts both ways. Consolidating two suppliers into one bigger supplier gives Nvidia and the hyperscalers a more integrated partner, and it also gives them one fewer independent vendor to play off against the other. Marvell's stated ambition is to rival Broadcom in AI connectivity, which makes the deal a direct challenge to the largest merchant supplier of custom AI silicon and switching.

For customers, the honest summary is that the deal concentrates optical and electrical supply in fewer hands at exactly the moment demand is least price-sensitive. That is good for the merged company's pricing power and a reason for procurement teams at the hyperscalers to care about the regulatory review.

What to watch before 2028

Four things decide whether this works. First, approval: a $115 billion tie-up between a networking silicon vendor and a laser supplier will draw scrutiny in the US, China and Europe, and 18 months is long enough for conditions to be attached.

Second, the cost savings. More than $1 billion in annual run-rate savings is a promise about overlapping functions and manufacturing, not a number anyone has verified. Third, the industrial laser business, which either gets integrated or gets sold, and either path produces a different company than the one described in the release.

Fourth, the cycle. Marvell narrowed its third-quarter fiscal 2027 revenue range to plus or minus 2% from plus or minus 5% while keeping the midpoint of prior guidance, a small detail that says the near term is stable. The deal's value depends on the years after that: if optical interconnect keeps displacing copper in AI clusters, Marvell has bought a position at the chokepoint; if the buildout plateaus, it has paid a 23.9% premium in stock for a supplier tied to the same cycle.