Skip to content
MRVL
~5 min read · 1,123 words ·updated 2026-04-28 · confidence 50%

Executive Summary

Marvell Technology’s storage-controller business—historically the company’s cash engine via 88SS/88NV HDD and SSD controller families—is in secular decline despite pockets of enterprise cold-storage resilience. The segment’s estimated <$500M annual revenue (sub-6% of total) and >55% gross margins are being cannibalized by in-house solutions from Samsung, Broadcom, and specialized competitors like Phison and Silicon Motion. Marvell’s strategic pivot to AI infrastructure custom silicon (74% of FY26 revenue from data center) implicitly deprioritizes storage. The question is not if, but when storage becomes immaterial (<2% of revenue), triggering a modest GM consolidation headwind at the corporate level.

Current Segment Size & Trajectory

FY26 Estimated Storage Revenue: <$500M (declining YoY)

Marvell does not separately report storage as a distinct segment in current earnings disclosures. Based on the company’s stated data center focus (74% of revenue in Q4 FY26) and legacy storage business scale, the storage segment is estimated at <$500M of the $8.195B total FY26 revenue—roughly 6% or less. This contrasts sharply with the pre-Inphi era (2018-2020) when storage controllers contributed an estimated 15-20% of revenue.

Key metrics from recent filings:

  • FY26 total revenue: $8.195B (+42% YoY)
  • Data center revenue: $1.651B in Q4 FY26 (+21% YoY), representing 74% of total
  • Storage controllers increasingly bundled into “other” segments or written off as legacy

No specific storage segment gross margin is disclosed, but analyst estimates suggest the segment maintains 55-60% GM—well above the company-wide 59.7% non-GAAP GM, indicating its decline drags consolidated margins downward.

HDD Market Trajectory: Resilience, Not Recovery

Counter-intuitive finding: HDD demand rebounding, but Marvell’s share in question.

Industry data from 2025-2026 reveals the HDD market is not collapsing; instead:

  • Seagate and Western Digital market share: 39% and 37% globally respectively, shipping ~108M and ~103M units in 2025
  • Enterprise nearline demand: Strong and concentrated—58% of HDD demand in 2025 driven by cloud infrastructure, hyperscale storage, and long-term data archival
  • Supply constraint, not demand cliff: The industry faces a capacity crunch, with major manufacturers reporting fully allocated nearline capacity through 2026. This has driven price increases and extended lead times, suggesting demand resilience for enterprise cold storage and AI-era data lakes

However, Marvell’s role is diminishing:

  • Samsung, Intel (via older OEM relationships), and strategic ODMs increasingly design storage controllers in-house
  • Phison and Silicon Motion dominate client SSD controllers
  • Enterprise OEMs favor integrated solutions (NAND + controller bundles) from memory suppliers

Implication: Even if HDD unit shipments flatten or grow modestly, controller ASP compression and customer consolidation limit Marvell’s revenue uplift.

SSD Controller Competitive Moat: Eroding

Marvell’s legacy 88SS/88NV HDD controller families and enterprise SSD solutions (88NV, Denali OEM platforms) face intensifying competition:

SegmentCompetitorAdvantageMarvell Risk
Client SSDPhison, Silicon Motion, SamsungCost, integration, design-win velocityOut of favor; low-margin OEM business
Enterprise SSDSamsung, Intel (Optane exit re-entry), SK HynixVertical integration, NAND controlStandalone controller supplier losing relevance
HDDCustom/OEM controllers (Seagate, WD internal)Proprietary firmware, full-stack IPSuppliers consolidating to in-house

Market dynamics:

  • Global SSD controller market projected to grow 14.4% CAGR (2025-2032), but Marvell’s share is ceding to rivals with deeper supply-chain integration
  • ASP erosion: client SSD controllers <$2/unit; enterprise controllers $10-20/unit, but volumes constrained
  • Margin compression inevitable: historically Marvell achieved 55-60% on storage, but competitive loss of share pushes towards 40-50%

Historical Context: When Storage Was King

2018-2020: Storage contributed 15-20% of Marvell revenue

Before the transformational Cavium (2018, $6B) and Inphi (2021, $10B) acquisitions, Marvell was primarily a storage semiconductor company:

  • 88SS/88NV families (HDD controllers) supplied to Seagate, WD, Toshiba
  • Strong enterprise SSD presence via Denali OEM reference platforms
  • High gross margins (55-60%) and high absolute profits funded R&D
  • Vulnerability: customer concentration (three HDD vendors), cyclicality, and low growth

The acquisitions explicitly repositioned Marvell as a data center / cloud / 5G play, signaling management’s view that storage was mature and capital-inefficient.

Sunset Timeline: When Does Storage Hit <2% of Revenue?

Base case: 2-3 years (FY28-FY29)

Current state:

  • Storage estimated at <$500M (6% of $8.2B revenue)
  • On a CAGR of -15% to -20% (industry data + competitive losses)
  • Data center growing at +20%+ CAGR

Projection to FY28 (assuming $10B+ total revenue, +12% CAGR):

  • Storage revenue: $400M (post-sunset)
  • As % of total: ~4% → 2% by FY29

Immateriality threshold (<2%):

  • Expected by FY29-30 (fiscal 2029-2030, ending Feb 2030)
  • Enables Marvell to fully exit reporting, folding residual revenue into “other”

Key risks to timeline:

  • Enterprise cold-storage supercycle extends longer (positive for HDD, modest positive for Marvell)
  • Unexpected new NAND interface standard or PCIe evolution revitalizes controller market (unlikely)

Capital Implication: Margin Headwind

Storage’s decline pressures consolidated gross margin by 50-100 bps annually.

Mechanics:

  1. Current mix: Storage contributes <6% of revenue at 55-60% GM
  2. Company average: 59.7% non-GAAP GM (FY26)
  3. Replacement mix: Data center custom silicon at 55-58% GM (below average), but higher opex absorption
  4. Exit impact: Losing 55-60% GM products (storage) and replacing with 55-58% GM products (custom ASICs) actually neutral on headline margin, BUT

The real pressure: Storage decline forces R&D reallocation and leaves fixed infrastructure underutilized, driving SG&A deleveraging. Marvell must either:

  • Reinvest heavily in AI/optical IP (maintaining opex % constant)
  • Allow operating margin to decline 1-2% as storage winds down

Management has historically chosen the former (heavy R&D spending on 5G, optical, DPU), meaning operating leverage is muted despite gross margin stability.

Strategic Verdict

Marvell’s storage sunset is inevitable and welcomed by the board—it forces the company to allocate capital toward higher-growth, higher-IP-moat segments (optical interconnect, DPU, custom AI silicon). However, the transition is not painless:

  • Near term (FY26-27): Margins stable, but opex inefficiency rises as legacy support overhead persists
  • Medium term (FY28-29): Structural margin headwind as immateriality forces accounting changes and customer support withdrawal
  • Long term (FY30+): Margin floor rises as pure-play AI/optical gross margins (58-62%) exceed legacy storage exit-economics

Investor takeaway: Watch for an earnings rebase in FY29 as Marvell exits storage disclosures. Consolidated guidance may show GM compression (100-150 bps), but this masks improved underlying core margin quality.

Sources & Data

Confidence Level: ◐ (Medium) — Segment revenue is estimated; Marvell does not separately disclose storage. Timeline assumption relies on -15% to -20% annual decline, which is plausible but not publicly confirmed.

Cross-references