Marvell Stock Price Forecast - MRVL Surges 9% to $95.86 as Nvidia Bets $2B — Record $8.2B Revenue and a $94B TAM

Marvell Stock Price Forecast - MRVL Surges 9% to $95.86 as Nvidia Bets $2B — Record $8.2B Revenue and a $94B TAM

FY2027 Revenue Approaching $11 Billion, 43% EPS Growth Projected for FY2028, and an AI-RAN Market Worth $200B by 2030 Not Yet in the Models | That's TradingNEWS

TradingNEWS Archive 3/31/2026 12:12:05 PM

Key Points

  • Nvidia invested $2B in MRVL to accelerate AI factory infrastructure, sending shares up 9% to $95.86 — the stock is now up over 50% in the past year.
  • Marvell posted record FY2026 revenue of $8.195B, up 42% year-over-year, with FY2027 revenue approaching $11B and data center growth projected at 40%.
  • At 16x FY2028 EPS and a PEG of 0.55x versus a sector median of 1.23x, MRVL remains undervalued relative to 43% projected EPS growth — a strong buy.

Marvell Technology (MRVL) opened Tuesday at $93.05, hit a session high of $97.49, and was trading near $95.86 — up 9.17% or $8.05 on the day — against a previous close of $87.81. The market cap expanded to $82.49 billion. The year range tells the full story of what this company has been building toward: $47.09 at the low, $102.77 at the high, with the stock now sitting in the upper half of that range on a single-session catalyst that fundamentally rewrites the growth narrative. This is not a relief rally in a weak market. This is a strategic inflection being priced in real time.

Nvidia (NVDA) announced a $2 billion investment in MRVL as part of a strategic partnership centered on NVLink Fusion-based scale-up networking, silicon photonics, co-packaged optics, and custom AI accelerator chips. Nvidia CEO Jensen Huang did not mince words: "The inference inflection has arrived. Token generation demand is surging, and the world is racing to build AI factories." Marvell CEO Matt Murphy framed his company's role with equal precision: the partnership centers on the "growing importance of high-speed connectivity" as AI models scale to sizes that stress every layer of the data center infrastructure stack. This is not a partnership about GPUs. It is a partnership about everything that connects GPUs to each other and to the rest of the rack — which is exactly the market Marvell has been positioning into for the past eighteen months through acquisitions, product pivots, and design wins that are now at all-time record levels.

Why Nvidia's Investment Completely Changes Marvell's Competitive Position

To understand what Tuesday's announcement actually means, the competitive context matters. Marvell spent the better part of 2024 and the first half of 2025 navigating a painful transition. Its largest XPU customer, Amazon (AMZN) AWS, began diversifying its XPU vendor base, opening the door for Taiwanese players like Alchip to win ASIC programs that had previously belonged to Marvell. Simultaneously, Broadcom (AVGO) was robustly growing its custom silicon order book while MRVL was caught in a market where margins were thin, competitive pressure was intensifying, and the XPU business was becoming a structural drag rather than a growth engine.

Management made a pivotal decision: exit the low-margin XPU headache and pivot aggressively into AI connectivity — specifically co-packaged optics and silicon photonics for scale-up networks. Two transformative acquisitions executed that pivot. The Celestial AI takeover gave Marvell competitive capability in CPO and SiPho. The XConn Tech acquisition advanced Marvell's position in CXL and PCIe-based solutions for scale-up networks. Both together, combined with the XPU attach strategy — covering PCIe retimers, coprocessors, CPO components, and CXL-based memory controllers — expanded Marvell's total addressable market to approximately $94 billion. Nvidia's $2 billion investment on Tuesday is the market's most credible outside validator that Marvell executed that pivot correctly.

The investment creates three distinct growth ramps that the market has not yet fully priced. First, Marvell's participation in Nvidia's NVLink Fusion-based scale-up networks, leveraging the Celestial AI and XConn acquisitions to provide CXL-based memory controllers, optical light engines, and optical pluggables deployable on NVLink Fusion platforms. Second, Marvell gets a renewed shot at the XPU market — expanding beyond its current AWS and Microsoft (MSFT) Azure customer base to additional hyperscalers building on Nvidia's architecture. Third, and most underappreciated, Nvidia is pulling Marvell into the AI-RAN market — radio access network infrastructure where artificial intelligence is creating an entirely new demand category expected to exceed $200 billion by 2030. That $200 billion AI-RAN TAM is not currently factored into any of Marvell's forward revenue projections of 33% growth in FY2027 and 37% in FY2028. It is pure upside optionality sitting outside the modeled numbers.

Nvidia has been executing this investment playbook consistently. In December 2025, it invested $2 billion in Synopsys (SNPS). In January 2026, it purchased $2 billion worth of CoreWeave stock. The pattern is deliberate — Nvidia is colonizing the AI infrastructure stack layer by layer, ensuring its architecture remains the foundational platform even as hyperscalers attempt to design competing custom silicon. Every company Nvidia backs becomes embedded in the Nvidia ecosystem in a way that is structurally difficult to displace. MRVL is now inside that ecosystem with a $2 billion institutional endorsement from the dominant AI hardware platform company in the world.

The Financial Foundation: $8.195 Billion in Record Revenue, 42% Growth, and Numbers That Wall Street Just Had to Revise

The Nvidia investment lands on top of an already exceptional fundamental trajectory. MRVL delivered fiscal year 2026 revenue of $8.195 billion — a 42% year-over-year increase and a record for the company. Data center revenue for Q4 FY2026 alone came in at $1.651 billion, up 21% quarter-over-quarter and 9% year-over-year. Full-year data center revenue hit $6.1 billion, a 46% year-over-year gain. Management guided Q1 FY2027 revenue to $2.4 billion at the midpoint — with a 5% tolerance band — and projected full fiscal year 2027 revenue approaching $11 billion. Data center revenue is expected to grow approximately 40% in FY2027, with interconnect revenue specifically expected to exceed 50% growth.

The January 2026 quarterly financials available confirm the trajectory: revenue of $2.22 billion in the quarter, up 22.08% year-over-year. Net income hit $396.10 million — a 97.85% year-over-year increase. Net profit margin expanded to 17.85%, up 61.98% on a year-over-year basis. Earnings per share came in at $0.80, a 33.33% jump. EBITDA reached $730.90 million, up 32.96%. The balance sheet shows $2.64 billion in cash and short-term investments — up 178.27% year-over-year — providing the financial flexibility to pursue further strategic moves without diluting shareholders. Total assets stand at $22.29 billion against total liabilities of $7.98 billion, leaving total equity at $14.31 billion.

The only metric that gives pause is free cash flow, which came in at $72.74 million — down 87.54% year-over-year — reflecting the capital intensity of the acquisition strategy and the investment required to execute the silicon photonics pivot. Cash from operations also softened to $373.70 million, down 27.30%. These are not structural red flags; they are the natural byproduct of a company in aggressive investment mode, deploying capital into acquisitions and R&D to capture a $94 billion TAM. When revenue is scaling 42% annually and net income is growing 97.85%, a temporary compression in free cash flow is a trade-off that makes mathematical sense.

Analyst projections were already moving sharply higher before Tuesday's Nvidia announcement. Revenue growth expectations for FY2027 were revised to approximately 33% — FY2028 projections jumped to approximately 37%. EPS growth is outpacing top-line expansion, with 35% EPS growth projected for FY2027 and 43% for FY2028, compared to the 33% and 37% revenue growth rates respectively. That EPS acceleration relative to revenue is the margin expansion story compounding at the earnings line — and it is exactly the kind of operating leverage that justifies a premium multiple.

The Valuation Argument: 29x Forward P/E Looks Cheap Against 35% EPS Growth

MRVL trades at a forward non-GAAP P/E of approximately 29x. The sector median is 20.64x — implying a roughly 20% premium on a nominal basis. The forward EV/EBITDA stands at 21.71x against a sector median of 12.52x. On raw multiples, the stock looks expensive relative to peers. On a growth-adjusted basis, it looks cheap. The forward non-GAAP PEG ratio is 0.55x against a sector median of 1.23x — a 51% discount to peers on growth-adjusted valuation. Revenue growth at 37.18% forward compares to a sector median of 10.67%. MRVL carries an A- growth grade. At 16x FY2028 EPS, the stock is priced at a level that reflects genuine earnings growth rather than multiple expansion speculation.

The market cap at $82.49 billion against a $94 billion total addressable market, combined with FY2028 EPS trajectory and a PEG of 0.55x, makes the current valuation argument straightforward: the market is paying for quality but not for the full rate of growth available. Nvidia's $2 billion investment, the AI-RAN opportunity that is not yet in the models, and the Celestial AI and XConn revenue ramps that are just beginning to show in forward guidance all represent earnings catalysts that are not yet reflected in the 29x forward multiple. The stock at $95.86 is a buy.

Design Wins at All-Time Records and Bookings That Eliminate the Visibility Problem

One of the most persistent concerns about MRVL through 2025 was revenue visibility — the question of whether design wins would actually convert to shipment volumes. The FY2026 results address that concern directly. The company reported that 2026 design wins hit a record high. Bookings are growing at a record pace. Management guided Q1 FY2027 explicitly to $2.4 billion at the midpoint — not a range that reflects uncertainty, but a number with direct line of sight from existing purchase orders. For the custom silicon business specifically, management confirmed purchase orders covering the entirety of the next-generation large customer program that supports 20% segment growth. They described no expectation of air pockets.

The XPU attach segment is the piece most underappreciated by the market. Marvell now has more than 15 XPU attach wins, with two primary use cases scaling across hyperscalers: smart NICs and CXL-based memory expansion. Both are already contracted — management has line of sight to a $2 billion revenue opportunity in this segment by 2029. That figure does not include adjacent use cases that are still being scoped. The Celestial AI integration is already included in Q1 FY2027 guidance, and CEO Matt Murphy characterized the initial revenue build as being driven by the PF chiplet — a product that is the entry point for a much larger optical franchise inside AI racks, not a single product bolt-on.

Marvell President and COO Chris Koopmans described hyperscaler spending as "still growing massively" — a characterization consistent with McKinsey data showing global data center capacity demand expected to expand from 82 gigawatts to 219 gigawatts by 2030, with AI workloads making up approximately 70% of that total by decade's end. The infrastructure that connects all of that AI compute — the optical interconnects, the switching layers, the custom silicon plumbing — is precisely what Marvell sells. As AI clusters scale from large to absurdly large, the connective tissue becomes more valuable, not less.

How MRVL Compares to Broadcom, Micron, Credo, and Astera Labs

Understanding MRVL's position requires mapping it accurately within the semiconductor competitive landscape. Broadcom (AVGO) is the heavyweight — deep custom silicon capability, massive switching market share, and a forecast of more than $100 billion in AI chip sales in the coming year. Broadcom competes directly with Marvell for hyperscaler custom ASIC business, and it is winning more of it. Micron Technology (MU) owns the memory side of the AI rack — specifically high-bandwidth memory — and wins when HBM demand surges. Credo Technology (CRDO) excels at connectivity but lacks the breadth to compete across the full rack. Astera Labs (ALAB) is strong in server interconnect and memory-related infrastructure.

Marvell sits in a distinct lane: broad enough to participate across the entire AI rack through custom ASICs, PAM and coherent optical DSPs, switches, PCIe retimers, CXL controllers, NICs, and scale-up photonics — yet focused enough in its pivot toward connectivity and optical solutions that it is not trying to out-Broadcom Broadcom on custom silicon volume. CEO Matt Murphy's characterization is accurate: "We are everywhere in the AI rack. And we are just getting started." NVDA's $2 billion investment is the most credible outside confirmation that this positioning is real rather than aspirational.

The Nvidia Ecosystem Flywheel — Why This Investment Is Structural, Not Transactional

The strategic dimension of this investment deserves more analytical attention than the headline dollar figure. Nvidia is not making a financial bet on Marvell's stock price. It is making a platform bet on ensuring that its NVLink Fusion architecture has the best possible optical and custom silicon ecosystem surrounding it. Every dollar Nvidia puts into Marvell's silicon photonics and CPO capabilities strengthens the NVLink Fusion platform relative to competing architectures. Every Marvell product that integrates into NVLink Fusion deepens the switching cost for hyperscalers who build on Nvidia's foundation.

This is the Nvidia ecosystem playbook: invest in the companies whose products are essential to your platform, lock in their roadmaps to your architecture, and create a moat that is not just about GPU performance but about the entire fabric of AI compute infrastructure. Synopsys received the same $2 billion treatment. CoreWeave received $2 billion. Coherent (COHR) and Lumentum (LITE) — both optical component makers — are also in Nvidia's investment orbit. The pattern is visible, deliberate, and extraordinarily effective. By the time a hyperscaler tries to build an alternative to Nvidia's architecture, it finds that the entire supply chain of critical components has been pre-aligned to the Nvidia ecosystem through strategic capital relationships.

For MRVL, membership in that ecosystem is not just a revenue opportunity — it is protection against the hyperscaler capex digestion risk that remains the primary bear case for the stock. If a large cloud provider delays a rack deployment or digests custom silicon inventory for a quarter or two, a standalone Marvell suffers directly. A Marvell embedded in the Nvidia AI factory ecosystem has a structural demand anchor that partially insulates it from single-hyperscaler timing risk.

September 2025 Spooked Investors — Tuesday's Move Is the Redemption

MRVL spooked the market in September 2025 when its sales forecast came in below estimates and data center revenue disappointed. Shares dropped sharply on that miss, and the stock spent the better part of H2 2025 rebuilding investor confidence. The recovery from the $47.09 year low to the current $95.86 reflects both the fundamental execution — 42% revenue growth, record design wins, improving margin profile — and the beginning of the market's re-rating of what Marvell's optical networking and silicon photonics pivot is actually worth.

Tuesday's Nvidia investment is the definitive answer to the September 2025 trust deficit. Nvidia does not put $2 billion into companies whose strategic direction is uncertain. The investment is both capital and validation — and Marvell CEO Murphy's statement that the company is positioned at the "growing importance of high-speed connectivity" is now backed by the most credible capital allocator in the AI infrastructure space.

The stock's current P/E of 31.23 on a trailing basis and the forward 29x multiple will look conservative if FY2027 delivers the 33% revenue growth that analysts now project. At 16x FY2028 EPS — a year in which 43% earnings growth is the base case — the medium-term valuation argument is among the most compelling in the semiconductor sector.

The Risk Picture: Hyperscaler Concentration and Capital Intensity

The primary risk for MRVL is the same one that has defined its history: concentration in a small number of hyperscaler customers. Amazon AWS remains the anchor for the XPU business. Microsoft Azure is the second significant customer. The Nvidia partnership expands that customer universe through NVLink Fusion deployments, but the fundamental dynamic of being dependent on large cloud providers' capex cycles is not eliminated — it is mitigated. If a major hyperscaler pauses a rack deployment or pushes out a node transition, MRVL stock reacts first and asks questions later. That September 2025 episode proved it.

Free cash flow at $72.74 million against $396 million in net income reflects aggressive capital deployment into acquisitions and R&D. The Celestial AI and XConn acquisitions are not yet generating revenue at scale — they are in the investment phase, consuming capital to build the product stack that will drive FY2028's 37% revenue growth. Cash from operations down 27.30% year-over-year is the near-term cost of the long-term strategy. For those with a shorter time horizon, this cash flow compression is a genuine concern. For those positioned for the FY2027-FY2028 revenue acceleration, it is the correct capital allocation.

The broader market volatility — the S&P 500 (SPX) is down 7.8% in March, the Iran war is in its fifth week, and the Fed is frozen at 3.5%-3.75% — creates the geopolitical risk backdrop that could push MRVL shares lower even as the fundamental story improves. That is not a reason to avoid the stock. That is the buying opportunity that the volatility creates for those who understand what the Nvidia investment structurally changes.

The Verdict: MRVL Is a Strong Buy — The AI-RAN Upside Alone Justifies Current Levels

MRVL at $95.86 is a strong buy. The Nvidia investment eliminates the question of whether Marvell's silicon photonics pivot was the right strategic bet — it was. The $94 billion TAM, record design wins, $11 billion in FY2027 revenue approaching, 43% EPS growth projected for FY2028, and a PEG ratio of 0.55x against a sector median of 1.23x create one of the most asymmetric risk-reward setups in the semiconductor sector right now. The AI-RAN market at $200 billion by 2030 is not yet in the models. Celestial AI's optical franchise is just beginning to generate revenue. The XPU attach segment has $2 billion in line-of-sight revenue by 2029 before adjacent use cases are credited.

Wall Street rates MRVL a Strong Buy with a score of 4.53. SA analysts rate it Buy at 4.08. The quant model is at Hold — which is the signal that value-oriented quantitative frameworks are not yet capturing the strategic transformation that fundamental analysis clearly identifies. The quant Hold is not a red flag. It is the gap between what the models know and what the Nvidia investment confirms.

At 16x FY2028 EPS, this is not an expensive stock relative to its growth trajectory. It is a cheap stock in a volatile market that hasn't fully processed what the Nvidia ecosystem endorsement actually means for Marvell's competitive position from 2027 onward.

That's TradingNEWS