報告_Cantor_MRVL_20261102

PDF 原檔：報告_Cantor_MRVL_20261102_original.pdf

圖片清單（已驗證 2026-11-09）

原始內容

CANTOR

EQUITY RESEARCH

November 2, 2026

Price: $196.40

Price Target: $225.00 (From

$350.00)

Rating: Neutral

Key Statistics:

Research Analysts:

C.J. Muse

212-915-1935

CJ.Muse@cantor.com

Jamison Phillips-Crone, CFA

720-320-9242

jamison.phillips-crone@cantor.com

Matthew Prisco

212-610-2416

Matthew.Prisco@cantor.com

James Bowlin

541-948-9660

James.Bowlin@cantor.com

Galahad Caer

212-294-8015

Galahad.Caer@cantor.com

REV ($M)

Marvell Technology, Inc. (MRVL)

Company Update

Marvell Sees the LITE, The Market Doesn't

This morning, Marvell announced that it would acquire Lumentum for $115B in an all-stock transaction, creating a powerhouse in AI and data center connectivity. As for our take, we come away with mixed thoughts. As for positives, MRVL will now be a leader and able to co-design across the entire electrical/optical signal chain, gain supply security and scale, become a more credible competitor to Broadcom, and eventually gain margin benefit from vertical integration. On the other hand, we do worry about the timing of LITE selling itself and implications of where we are in today's optical cycle, see the potential for companies competing against MRVL and outside the NVDA ecosystem moving business away from MRVL/LITE to competitors, worry about timing of regulatory approval by China (mgmt. suggesting 1HCY28), highlight announced COGS synergies that appear more to be revenue-related, and MRVL now running fabs. Moreover, MRVL hosted an Analyst Day just 4 weeks ago with a story many investors felt quite compelling, led by XPU Attach and Connectivity, with a story now that has Marvell shares in deal purgatory for at least the next 5-7 quarters, and EPS that is only modestly accretive into CY28. Shares are -29% since Bloomberg discussed M&A potential less than a week ago highlighting these concerns. Amidst a backdrop of de-rating across all of Semis (particularly Compute where both NVDA/AVGO trade <10x CY28 earnings), we adjust our price target for MRVL to $225 - reflecting 18x proforma CY28 EPS of $12.50. As for shares, awaiting China approval, AI cyclical concerns, and premium multiple to other compute players keep us on the sidelines. We reiterate our Neutral rating.

Transaction Details. Marvell is acquiring Lumentum in an all-stock deal for an implied enterprise value of $115B (~18x EV/NTM sales multiple). Lumentum shareholders will receive 4.5 shares of MRVL for each LITE share, leading to LITE shareholders receiving $1,132/share (based on the unaffected value - before Bloomberg published article on potential M&A on 10/28) for a 23.9% premium. Post-closing, MRVL shareholders will own ~67% of the company, while LITE shareholders will own ~33%. The deal is expected to close in 1HCY28 and is subject to shareholder approval and regulatory approvals. We believe that China approval will be the biggest question mark. On breakup fees, LITE's breakup fee is $2.9B, while MRVL's is $5.8B. Additionally, if regulatory bodies do not approve the transaction, there is a $3.2B reverse regulatory termination fee that MRVL would pay (not stacked on top of MRVL's $5.8B fee).

Strategic Rationale. The combination of MRVL and LITE is primarily around creating a specialized, vertically integrated AI connectivity company that CANTOR°

can provide full stack solutions spanning interconnect, switching, compute, lasers, and CPO. Additionally, the acquisition will accelerate product time to market across connectivity through co-design and engineering synergies. Marvell believes the combination unlocks a TAM expansion to $550B from $400B standalone.

Financial Impact. Beyond the strategic rationale, the combination of the two companies is expected to lead to GM/OM accretion immediately postclose, with EPS accretion expected in year one post-close. Marvell discussed excitement around revenue synergies (though gave limited details here), instead noting that it sees $1B+ in cost synergies (mostly COGS, limited Opex), as MRVL is able to add LITE content into solutions in which it is not already present and vice versa, at cost (i.e., Marvell solutions would use Lumentum lasers and Lumentum modules would use Marvell DSPs, TIAs, and drivers). Additionally, the company highlighted that it believes it is quite conservative with the $1B+ synergy target outlined in the call, something that we agree with. At the end of this note, we outline our assumptions for a CY28 combined model, which illustrates modest accretion. Our initial model for CY28 combined results in revs/OM/EPS of $43.6B/47.0%/$12.55, which compares favorably to MRVL standalone of $32.0B/43.4%/$12.50.

Benefits. We see the following as the key benefits of the transaction. 1) Co-design across the entire signal chain. This is the largest benefit we see; as the industry moves from pluggable to NPO/CPO, the acquisition gives MRVL access to technology across the whole stack. 2) A more credible competitor to Broadcom. With Broadcom as the only large competitor that has technology spanning compute and connectivity (as well as a foundational CPO business), the combination gives Marvell additional scale and technology (largely InP and CPO) to compete with Broadcom. We note that LITE also disclosed a second NPO/CPO win today, which is expected to generate >$5B in revenues. 3) Supply security and scale. InP lasers remain one of the tightest bottlenecks in the industry, where LITE has highlighted it remains sold out through CY28. As a result, Marvell's ability to own the supply chain can help it control its own destiny in lasers. 4) Vertical integration and potential for margin expansion. As discussed above, vertical integration and subsequent internal sourcing are key benefits for MRVL. From a financial perspective, the effects of vertical integration should drive both GMs/OMs significantly higher, and we see additional room as the company scales past $100B in revenues.

Potential Risks/Challenges. We see the following as key risks and challenges. 1) Is LITE selling now telling us they don't believe in sustainability of the AI infrastructure cycle? While the all-stock offering could suggest the opposite, the question remains as to why LITE is choosing to sell now (with LITE noting difficulty in building full Semiconductor capabilities vs. combining now). 2) Companies competing against MRVL and/or outside the NVDA ecosystem could move business from LITE to competitors. The extent that revenue dis-synergies could take place remains a question. Given the longer time horizon to close, this leaves CANTOR°

time for customers to qualify COHR alternatives to LITE solutions, or potentially MRVL customers to move to AVGO or CRDO. 3) Regulatory Environment. We see potential hurdles for China approval. Additionally, this announcement puts both MRVL/LITE into deal purgatory for the next 5-7 quarters until the planned close in 1HCY28. We note a $3.2B reverse termination fee if regulatory approval fails. 4) Outlined COGS synergies appear more revenue- than cost-related. Beyond the benefit of sustained growth into CY29, it appears to us that much of the synergies are revenuedependent (even mgmt. specifically noted that this is not an Opex-driven synergy story). Yes, there is some margin capture on incremental shared content, but overall, it seems to us that to realize the full COGS benefits, the company will need to win designs where both companies' content is used, see volumes continue to grow into CY29 and beyond, and see customers accept bundled supply. We believe that the $1B is very conservative, so we expect upside to mgmt.'s estimates. 5) MRVL will now own fabs (through LITE's InP manufacturing and module assembly), something that it does not currently do, and creating a more capital-intensive business.

CANTOR

Exhibit 1: Illustration of CY28 Combined Company with Estimated Synergies

Source: Company Reports, FactSet, Cantor Fitzgerald CANTOR°

Valuation

Our $225 price target reflects ~18x CY28 combined MRVL/LITE EPS estimate of ~$12.50.

Risks

Where could we be wrong? Cyclical weakness across Storage, Networking, and Carrier Infrastructure could prove to be worse than expected and offset expected strength from the Data Center. Excess Custom Silicon strength without a cyclical recovery could weigh on GMs. Ramp of Custom Silicon could come in less than expected. Investors may be unwilling to put a higher multiple, particularly when reflecting on reported EPS (which excludes fairly aggressive stock-based compensation).

CANTOR

Company Description

Marvell Technology, Inc. is a leading supplier of data infrastructure Semiconductor solutions, spanning the data center core to network edge. Marvell is a fabless Semiconductor supplier of high-performance standard and semi-custom products, with core strengths in developing and scaling complex System-on-a-Chip architectures, integrating analog, mixed-signal, and digital signal processing functionality.

Companies Mentioned:

Broadcom Inc. (AVGO - NSQ) , $326.08, PT: $600.00: Overweight NVIDIA Corporation (NVDA - NSQ) , $204.64, PT: $350.00: Overweight

Prices are as of the close of Oct. 30, 2026

Disclosures Appendix