Marvell’s AI business is experiencing a boom, prompting a substantial upward revision of the target price to $240.

Overview

This note is a post‑earnings commentary issued by J.P. Morgan following Marvell’s FY27 Q1 (April quarter) results. The key takeaway is that Marvell delivered a quarterly report that broadly exceeded expectations, with the robust growth momentum in the data center segment continuing to accelerate. Management’s guidance for the next few fiscal years also significantly outstripped prior market consensus. Citing clear evidence of synchronized growth across multiple business lines and the smooth progress of several large‑customer custom ASIC projects, J.P. Morgan has substantially raised both its earnings estimates and price target, reaffirmed an Overweight rating, and designated Marvell as one of its top picks in the semiconductor sector.

Core views

Both earnings and guidance exceeded expectations: In FY27 Q1, revenue came in at $2.42 billion, up 9% sequentially and 28% year over year—slightly above the market consensus of $2.40 billion. The key highlight was the data center segment, where all four product lines—optics, switches, ASICs, and storage—posted sequential growth, with data center revenue rising 11% quarter over quarter. Meanwhile, the communications and other segments grew 3% sequentially, reflecting the continued gradual recovery in both carrier and enterprise networking markets. For the July quarter, management guided revenue of $2.70 billion, implying another acceleration to 12% quarter over quarter (and 35% year over year), well above the consensus estimate of $2.60 billion. The primary drivers are projected sequential growth of 18–20% in the data center segment, led by optics, switches, and the early ramp of the next‑generation Trainium 3 XPU ASIC. Gross margin guidance stands at 58.75%, 40 basis points above consensus, while EPS guidance is $0.93, surpassing the $0.90 consensus. Data center growth has been significantly revised upward: Management raised its full‑year FY27 data center growth outlook from over 40% to 50%, and for FY28 from 50% to 55%. This “better‑than‑expected” momentum stems from three key factors: exceptionally strong demand for optical DSPs—growth now expected to surge from 50% to 70%, with a stable 60–65% share at the 1.6T end; switch chips at scale, which have more than doubled year over year thanks to robust capital spending; and custom XPUs, whose growth has been nudged up from an earlier 20% forecast to around 25%, albeit constrained by HBM and advanced substrate supply. Meanwhile, the often‑overlooked storage controller business—covering HDDs and eSSDs—is on track to exceed $1.0 billion in revenue this year, compared with just $700 million last year. Repositioning the full‑year revenue trajectory and long‑term earnings power: Based on a stronger April quarter baseline and the expectation of at least 10% sequential growth each subsequent quarter, the quarterly revenue run rate at the end of FY27 is now projected at approximately $3.4 billion (previously around $3.0 billion), lifting full‑year revenue to roughly $11.5 billion (from about $11.0 billion). Looking ahead to FY28, driven by more than doubling AI‑custom revenue—supported by 25+ XPU‑related design wins—and sustained strong demand for optical DSPs, alongside the gradual ramp of new projects in switching, AEC, UALink, and CPO, total revenue has been revised up to $16.5 billion (from $15.0 billion). Accordingly, J.P. Morgan estimates that annualized earnings per share could exceed $8 by the end of FY27, with the potential to climb even higher to $10 per share by the end of FY28. A clear roadmap for custom ASICs and major customer programs: As a critical indicator of Marvell’s long‑term growth sustainability, the report provides a detailed breakdown of its custom ASIC customer base and process‑node pipeline. The next‑generation Trainium 3 XPU ASIC has already begun initial shipments this quarter, with volume ramp-up expected to accelerate significantly in the second half. The Trainium 4 (2nm) project has entered the deep‑design phase and has secured additional design wins across CPO, NVLink I/O chiplets, and UALink switch chips, opening opportunities to co‑develop with Astera Labs. Meanwhile, Microsoft’s Maia 3nm XPU program is progressing smoothly, with plans to enter high‑volume production next year; however, J.P. Morgan views Marvell’s FY28 revenue assumptions as conservatively estimated, leaving considerable upside potential. In addition, the company has more than ten XPU‑related ASIC projects—including DPUs, CXL controllers, and networking solutions—set to ramp up over the coming year.

Analysis framework

This research report follows a core analytical framework characterized by “bottom-up drivers, validation against underlying data, and forward‑looking earnings power–anchored valuation,” which can be broken down into four distinct steps for clearer understanding. Step 1: Earnings Validation — The report systematically compares the prior quarter’s actual results and current‑quarter guidance with the company’s earlier forecasts, analyst estimates, and market consensus, assessing both the magnitude and distribution of any upside surprises. Crucially, the analysis goes beyond mere dollar‑level outperformance; it dissects how the positive deviation is allocated across individual product lines and evaluates the quality of that outperformance—for example, whether growth across all data‑center segments is sustained on a sequential basis, rather than driven by inventory restocking or one‑off factors. Step 2: Raising the Growth Outlook Ceiling — Leveraging the new revenue base (with Q4 revenues exceeding $180 million) and management’s commitment to sequential growth of at least 10% per quarter going forward, the report projects a full‑year run rate and derives a reasonable full‑year revenue forecast. This growth trajectory is cross‑checked against detailed segment‑level dynamics, including optical products posting 70% year over year, switching business expanding more than twofold, and custom XPU revenue growing by over 25%. Step 3: Segment‑Level Structural Decomposition and Multi‑Year Customer/Project Pipeline Analysis — This step is pivotal in determining whether the upward revision in growth is sustainable. The analyst devotes significant attention to dissecting the drivers behind each product category—optical DSPs, switches, storage, and custom ASICs—and then further disaggregates the custom ASIC segment along key dimensions: major cloud customers (AWS/Microsoft), process nodes (3nm vs. 2nm), and project types (mainstream XPU versus ancillary ASICs). This approach essentially amounts to a “project pipeline analysis.” Step 4: Forward‑Looking Earnings Power–Anchored Valuation — JP Morgan’s valuation methodology does not rely on applying a fixed multiple to current‑quarter EPS. Instead, it anchors its valuation on the annualized earnings power—$8.00 per share—as projected at the end of FY27 (i.e., in the quarter ending January 2028)—and applies a comparable multiple to AI‑related semiconductor and networking stocks (30x PE), arriving at a target price of $240. This forward‑looking valuation approach, which ties equity valuations to future earnings power at a specific point in time, is commonly employed in pricing high‑growth semiconductor and AI‑beneficiary equities.

Methodology notes

• Industry/ Sector Analysis FrameworkVolume-price decomposition

  Volume–Price Decomposition — This approach breaks down revenue growth into changes in shipment volume and average selling price, or further disaggregates growth rates by product line or customer segment. In this research report, Marvell’s revenue growth is attributed to the distinct growth rates across its product lines—optical, switching, custom ASICs, and storage—enabling an assessment of the quality and sustainability of that growth.

  Analysts do not merely focus on overall revenue growth; instead, they provide line‑by‑line growth forecasts for each product segment—e.g., optics at 70% year over year, switches more than doubling year over year, and custom solutions around 25% year over year—thereby illuminating the underlying “composition” of growth and the distinct drivers behind it. Some segments rely on volume gains bolstered by stable market share, while others depend on ASP expansion or product upgrades. This granular breakdown enables investors to distinguish between short‑term cyclical impulses and longer‑term structural growth trends.

• Industry/ Sector Analysis FrameworkPenetration Rate S-Curve

  The S-curve of market penetration and multi‑generation technology evolution—technology products typically experience rapid adoption and accelerating growth in the early stages of their lifecycle, with growth rates decelerating as penetration rises. Meanwhile, successive product iterations (e.g., from 3nm to 2nm) create a relay effect. This research report leverages the deployment of multi‑generation customer projects (Trainium 3 → Trainium 4; Maia 3nm → 2nm) to assess the sustained momentum driving mid‑term accelerated growth.

  By outlining how several major cloud‑service customers, including AWS and Microsoft, are advancing their custom ASICs from the current generation (3nm) to 2nm, and by expanding their portfolios from a primary XPU to more than ten ancillary ASICs—such as DPUs, CXL controllers, and networking chips—analysts seek to demonstrate that Marvell’s AI‑custom chip business is not a one‑off surge, but rather a sustainable growth pipeline spanning at least two to three generations, serving multiple customers and encompassing diverse product categories. This approach exemplifies the typical logic of leveraging penetration rates and an intergenerational iteration framework to explain medium- to long-term growth that outpaces expectations.

• Valuation MethodPE/PEG valuation

  Forward‑looking earnings‑power PE valuation—by multiplying the annualized run‑rate earnings power at a future point in time by a valuation multiple consistent with that of comparable companies, we derive the target price. In this report, we project an exit-year earnings power of $8.00 per share for FY27 and apply a 30x PE multiple, yielding a target price of $240.

  This is not the conventional “this year’s EPS multiplied by a fixed P/E ratio”; rather, we look further ahead—roughly to around January 2028, when the company has reached a higher revenue growth trajectory and most new projects have begun to contribute earnings—and project its annualized profitability at that point, applying a 30x multiple consistent with the AI computing/network semiconductor sector. The aim is to “discount” into the target price the value of firms currently enjoying rapid growth but whose full profit potential has yet to be realized, while keeping the valuation multiple in line with the peer group and allowing growth to justify the premium multiple.

• Competition and Strategic FrameworkEconomies of scale / learning curve

  Economies of scale and an expanding project pipeline are driving margin expansion— as revenue more than doubles, fixed‑cost amortization and the R&D expense ratio decline, enabling net profit margin and EPS growth to outpace top-line expansion. This research report incorporates this logic.

  As the model shows, revenue grows from $8.195 billion to $11.5 billion and then to $16.8 billion, while the profit margin expands rapidly (net profit margin rises from 30.1% to 31.9% and then to 35.0%), with EPS growth rates of +42% and +59%, outpacing the corresponding revenue expansion. Although the research report does not explicitly use the term “operating leverage,” it essentially attributes the acceleration in earnings per share—from $8 to $10—to the higher margins generated as numerous new product launches ramp up production.

• Company Fundamentals and Financial FrameworkFree cash flow analysis

  Tracking Free Cash Flow and Share Repurchase/Dividend Capacity — In FY27Q1, operating cash flow totaled $374 million, free cash flow reached $483 million, and the company maintained a cash balance of $3.8 billion while executing $200 million in share repurchases and $54 million in dividend payouts, underscoring its robust cash‑generation capability and commitment to shareholder returns.

  In the narrative surrounding high-growth stocks, maintaining positive free cash flow and a robust share-repurchase capacity serves as a “safety‑net” signal. By highlighting these metrics in their reports, analysts convey to readers that Marvell’s growth is not solely underpinned by heavy capital expenditure; rather, it achieves rapid expansion while simultaneously generating sustained cash flows and delivering returns to shareholders, thereby mitigating overall risk perception.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Marvell Technology Inc (MRVL.US)

  The core stocks covered in this research report are poised to benefit from the explosive growth in computing and networking demand for AI data centers. Multiple product lines—optical DSPs, Ethernet switches, custom ASICs/XPUs, and storage controllers—are experiencing synchronized acceleration, with the company securing a large number of design wins spanning multiple product generations.

  Strengths

  The company holds a market-leading position in data-center optical and connectivity solutions, with a 60%–65% share of the 1.6T DSP market, and maintains a robust pipeline of custom AI ASIC projects for multiple Tier-1 cloud customers, including AWS Trainium 3/4 and Microsoft Maia. Additionally, its switching chips and storage controllers serve as key differentiating growth drivers.

  Weaknesses

  Customized XPU growth is constrained by HBM and advanced substrate supply; while the communications and other segments are recovering slowly, they continue to weigh on short-term growth; historical valuations have exhibited significant volatility.

  Risks

  A sharp slowdown in data-center construction could weigh on multiple business lines; falling behind in next-generation HDD read-channel technology or being subject to OEM dual-sourcing strategies also poses risks to market share; weak 5G capital expenditure on the communications side may cap revenue growth in the communications segment; and underperformance in the integration or capabilities of custom ASICs, or weaker-than-expected synergy effects, could erode management credibility and prompt downward revisions to earnings forecasts.

Key data

• FY27 Q1 (April Quarter) Revenue$2.418 billionQuarter-over-quarter growth of 9%, year-over-year growth of 28%, slightly above the consensus estimate of $2.406 billion; all segments within the data center business posted gains, with optical components, switches, ASICs, and storage all recording quarter-over-quarter increases.
• FY27 Q2 (July Quarter) Revenue Guidance$2.7 billionQuarter-over-quarter growth of 12%, year-over-year growth of 35%, significantly above the consensus estimate of $2.6 billion; data center revenue is expected to increase by 18%–20% quarter over quarter.
• FY27 Data Center Growth Outlook (Latest vs. Previous)50% YoY (previously 40%+)The upward revision stems from stronger-than-expected demand for optical DSPs and switching chips, with custom XPU growth slightly exceeding 20%.
• Optical DSP FY27 Growth Outlook (Latest vs. Previous)70% year-over-year (previously 50%)Strong shipments of 800G/1.6T products are expected, with the 1.6T segment’s market share projected to remain at 60–65%.
• FY27 Exit-Quarter Revenue Run Rate (Latest vs. Previous)Approximately $3.4 billion (previously about $3.0 billion)Management expects that, going forward, quarterly revenue growth will not fall below 10% quarter-over-quarter.
• FY28 Total Revenue Outlook (Latest vs. Previous)Approximately $16.5 billion (previously about $15.0 billion)Data center revenue growth accelerated to 55%, with AI-customized chip revenues more than doubling year over year, and over ten proprietary ASICs are slated to ramp up next year.
• AI Customization (XPU) Revenue EstimateCY26 approximately $1.9 billion (+25% year-over-year), CY27 exceeding $4.0 billion (more than doubling)Trainium 3 sees volume expansion; Microsoft’s Maia 3nm begins ramp-up at year-end; and Trainium 4 enters the market in small volumes this year.
• JPMorgan’s Adjusted EPS Forecast (Latest vs. Previous)FY27E: $4.04 (previously $3.82); FY28E: $6.43 (previously $5.60)The upward revisions amount to approximately 5.8% and 14.9%, respectively, reflecting dual upgrades in both revenue and profit margins.
• Target Price and Valuation Anchor (Latest vs. Previous)$240 (previously $135)Based on FY27 exit, the year-end earnings per share are $8.00, valuing the stock at 30x PE—consistent with AI computing and network semiconductor stocks.

Impact & implications

The research report argues that Marvell’s results, which significantly exceeded expectations, coupled with a substantial upward revision to its guidance, have fundamentally reshaped the market’s perception of the company’s medium-term growth trajectory. The unexpected acceleration in data center revenue, the well-defined and comprehensive customer base and process‑node roadmap for its custom ASIC business, and the synchronized momentum across multiple product categories—optics, switching, XPU, ancillary ASICs, and storage controllers—signal that Marvell’s AI narrative has evolved from “solely benefiting from optical networking upgrades” to that of a full‑stack platform provider spanning AI computing, networking, and storage interconnects. From a valuation perspective, the target price has been doubled to $240, reflecting not a bubble premium typical of bull markets, but rather a natural recalibration of the P/E anchor as the company’s long‑term earnings power comes into sharper focus. Once annualized earnings of $8–$10 become both predictable and highly credible, the current valuation multiple relative to forward earnings will swiftly converge to an acceptable level (FY28E PE around 31x). This represents a quintessential three‑stage pricing dynamic within the broader AI semiconductor landscape: shifting from near‑term PE to forward‑looking PE, and ultimately anchoring valuations on underlying earnings power. In terms of market implications, while the report does not explicitly address the broader market environment, it clearly positions Marvell as one of the sector’s top picks. This suggests that, in the current AI CAPEX cycle, Marvell is viewed as one of the most direct beneficiaries of sustained AI spending, while also exhibiting complementary competitive dynamics with Broadcom and NVIDIA. Competitive overlap occurs primarily in switching and custom ASIC segments; however, Marvell’s status as an optics leader and its differentiated position in storage controllers endow it with strong portfolio‑wide appeal.

Risks

• Sustainability Risk of Data-Center Capital Expenditures: Should large cloud‑customer AI infrastructure deployment reverse or decelerate significantly, it would simultaneously weigh on multiple revenue streams, including optical modules, switching chips, and storage controllers.
• Technology Competition and Market Share Risk: Being outpaced by competitors in next-generation HDD read-channel technology or optical DSP, or the adoption of a dual-supplier strategy by major customers, could result in a loss of market share.
• Telecom and operator spending risk: Weak 5G wired and wireless capital expenditures by major telecom operators could continue to cap revenue growth in the communications sector.
• Execution Risk in Custom ASIC Projects: If projects such as Trainium and Maia fail to deliver on integration, performance ramp-up, or synergy effects as expected, they could result in damage to management’s reputation and downward revisions to earnings forecasts.

What to watch

• Can the quarterly growth of data centers maintain a quarter-over-quarter pace of over 10%—particularly as the upgrade to optical DSPs continues to drive robust shipments of 800G and 1.6T equipment?
• Trainium 3’s ramp-up speed and the actual pace of production capacity expansion in the second half of the year
• The timing of the high-volume manufacturing launch for Microsoft’s Maia 3nm XPU project, as well as whether Marvell’s order intake through FY28 is indeed as conservative as projected in the research report, remains to be seen.
• Advancement of the Trainium 4 (2nm) key design node and its share gains in emerging technologies such as CPO, NVLink, and U‑A‑Link.
• Whether the recovery trajectory of the communications sector, along with other segments such as operators and enterprise networking, is sustainable, and whether quarter-to-quarter volatility has emerged.
• Supply constraints—specifically, the pace at which supply of HBM and advanced packaging substrates is easing—will determine whether it can unlock stronger growth for the custom XPU business.