Toshiba (HDD)
Component manufacturer: designs and assembles nearline, client, and surveillance HDDs sold to hyperscalers, storage OEMs, and distribution; uniquely non-vertically-integrated (buys heads from TDK, media from Resonac); sits inside diversified Toshiba Corporation (energy, infrastructure, devices), privately owned by the Japan Industrial Partners consortium
Toshiba is a delisted former blue-chip, not a venture-backed startup - the trail is whole-company transaction and proposal marks (tender offers, buyout bids, analyst-implied EV), not primary funding rounds. The 2021 CVC proposal was never consummated; the 2023 JIP tender is the only completed transaction mark; the 2026 point is an analyst-implied estimate from disclosed EBITDA and comps, not a trade.
Earnings, margins, COGS & capex
Parent Toshiba posted record FY2025 (ended Mar 2026): net sales JPY 3,709.1B (+5.6%), operating income JPY 300.8B (+52%, ROS 8.1% - highest ever), net income JPY 1,967.3B (~7x YoY, driven by JPY 2,277.0B of Kioxia share-sale and valuation gains booked in non-operating income). Management explicitly credited the HDD business - higher operating income on higher sales, driven by data-center demand - as a core driver alongside Energy and Infrastructure. The HDD unit shipped an estimated 21.1M units and 182EB in CY2025 (~$3.5B revenue, ~17% unit share, ~11% exabyte share per Coughlin). Kioxia monetization funded major deleveraging and a leveraged-loan refinancing completed end-Mar 2026.
Revenue trend
Margins
improving - record high; FY2026 target 10%
improving
management states HDD operating income rose YoY on higher sales
COGS structure
HDD COGS is dominated by externally sourced heads (TDK) and media (Resonac) plus in-house assembly centered in the Philippines - Toshiba is the only HDD maker without vertical integration into heads/media, structurally thinning margin versus Seagate and Western Digital but lowering fixed-cost exposure in downturns. Parent-level COGS not broken out in the results deck.
Capex
Not disclosed at HDD-business level. HDD capacity expansion is moderate; the industry (all three makers) has kept unit capacity disciplined since the 2022-23 downturn, contributing to current tight supply and firm ASPs.
Latest earnings
n/a (no sell-side consensus; company is private). Management characterized FY2025 as record: highest-ever ROS and net income
Management targets ROS of 10% in FY2026 under the Revitalization Plan; no HDD-specific guidance disclosed
- Net sales (parent, FY2025)
- JPY 3,709.1B (~$24.7B)
- Operating income (parent, FY2025)
- JPY 300.8B, ROS 8.1% (record); JPY 394.0B / 10.6% before provisions
- Net income (parent, FY2025)
- JPY 1,967.3B (incl. JPY 2,277.0B Kioxia-related non-operating gains)
- EBITDA (parent, FY2025)
- JPY 481.8B (13.0% margin)
- HDD units / capacity shipped (CY2025, est)
- ~21.1M units, ~182EB, ~17% unit share / ~11% exabyte share
Growth drivers
- AI data-center nearline exabyte demand — industry effectively sold out through calendar 2026, with long-term agreements reaching into 2027-2028 and nearline lead times stretched past 52 weeks; ASPs at multi-year highs
- Capacity-per-drive roadmap — MG11 24TB CMR and MA11 28TB SMR launched; M12-platform 30-34TB SMR sampling since Mar 2026 with 28TB CMR samples due Q3 2026; first HAMR drives planned as 2026-2027 test vehicles, ~40TB targeted by 2027 and 55TB by ~2030
- Surveillance and client HDD franchises providing volume base
- Parent-level — Energy (transmission and distribution), Defense, Railways, Semiconductor Manufacturing Equipment all cited as FY2025 profit drivers
Bull & bear
A record-profitability, freshly deleveraged Toshiba with a sold-out HDD franchise in an AI-driven nearline supercycle, and a concrete relisting path (earliest FY2028) at a mark plausibly well above the ~JPY 2T take-private price.
- HDD industry is effectively sold out through 2026 with long-term agreements into 2027-2028 and nearline lead times past 52 weeks; Toshiba management named HDD a top FY2025 profit driver - the cycle is paying now, not hypothetically
- Parent fundamentals inflected: ROS 8.1% record, EBITDA JPY 481.8B (+42% YoY), FY2026 ROS target 10%, leveraged loans refinanced on better terms end-Mar 2026 with covenants relaxed
- Kioxia gains (JPY 2,277B) converted a balance-sheet overhang into deleveraging firepower - press reports buyout debt cut from over JPY 1.6T to ~JPY 1T; the classic PE playbook is ahead of schedule
- Oligopoly structure (3 players, disciplined capex) plus AI exabyte growth gives HDD better pricing durability than any prior cycle
- Relisting is a concrete catalyst: 2026 press reports a target as early as FY2028 with the capital structure already being simplified; an IPO into these numbers would likely re-rate well above the JPY 2T entry
A structurally #3 player, late on HAMR and hostage to a third-party supply chain, inside an opaque private conglomerate, at the top of a famously vicious cycle - with the Kioxia earnings tailwind already spent.
- FY2025 net income is mostly one-off: Kioxia-related gains of JPY 2,277B were ~85% of pre-tax income; strip them and the underlying earnings base is far smaller, and that lever is largely used up
- Toshiba is behind Seagate on HAMR - its first HAMR drives are 2026-2027 test vehicles, not volume products - and depends on TDK/Resonac to deliver HAMR heads and media at yield; if qualification slips, hyperscaler sockets go to STX/WDC for a full product generation
- ~17% unit share but only ~11% of exabytes - the weakest highest-capacity nearline mix means Toshiba captures the least of the AI upside and suffers most in a downturn
- HDD cycles end abruptly (2022-23 nearline glut); buying the story at record ASPs is buying the top unless AI exabyte demand proves secular
- No public HDD-unit financials, no independent governance track record post-privatization, and any exit depends on JIP's timing - liquidity and transparency risks stack on cycle risk
- eSSD substitution creep: every generation of cheaper QLC NAND (Kioxia, Micron, Samsung, SK Hynix, Sandisk) moves the HDD/SSD crossover closer in warm tiers
What it is worth
Take-private anchor + EV/EBITDA comps (private company - all values are analyst estimates, not marks). Anchor: the JIP consortium paid ~JPY 2.0T (~$13.5-14B) for all of Toshiba Corp in 2023 (JPY 4,620/share), when operating income was JPY 198.5B (FY2024) and the Kioxia stake was unmonetized. FY2025 EBITDA of JPY 481.8B (~$3.2B at 150) against HDD-peer and Japan-industrial EV/EBITDA ranges of ~5-8x implies a whole-company EV of roughly $16-26B, before crediting the Kioxia-funded debt paydown (press: buyout debt down from over JPY 1.6T to ~JPY 1T). The HDD unit alone (~$3.5B est revenue) would comp against Seagate/WDC at ~1.5-2.5x EV/S in the current upcycle, i.e. roughly $5-9B standalone - an estimate, not a disclosed figure.
~$12-14B whole-company EV (back to the take-private mark: HDD downcycle, HAMR lag costs share, one-off Kioxia gains not repeatable, exit delayed past FY2028)
~$18-22B whole-company EV (6-7x FY2025 EBITDA, HDD cycle normalizes but stays profitable, relisting lands FY2028-FY2029)
~$26B+ whole-company EV (8x+ record EBITDA sustained, ROS 10% achieved, FY2028 TSE relisting into a still-tight HDD market)
Reverse-anchor check: the 2023 take-private priced Toshiba at ~10x then-operating income; operating income has since risen ~52% and debt has been materially repaid, so intrinsic equity value is plausibly well above the entry price if FY2025 profitability holds. Key swing factors: HDD cycle durability, HAMR execution, relisting timing (press: earliest FY2028). Not financial advice; no public price exists.
SWOT
Strengths
- One of only three HDD makers globally — a rational oligopoly with firm pricing and industry supply sold out through 2026, with long-term agreements into 2027-2028
- Record parent profitability (ROS 8.1% FY2025) and sharply deleveraged balance sheet after Kioxia monetization and the Mar 2026 loan refinancing
- Proven capacity roadmap execution with partners — 24TB CMR / 28TB SMR shipping, 30-34TB SMR sampling, and 30TB+ HAMR/MAMR demonstrated with Resonac media and TDK heads
- Asset-light HDD model (external heads/media) cushions fixed-cost pain in downcycles
- Long-standing hyperscaler and OEM qualification relationships in nearline
Weaknesses
- Distant #3: ~17% unit share vs Western Digital ~42% and Seagate ~41%, and only ~11% of industry exabytes - the weakest exposure to the highest-capacity nearline mix
- Behind Seagate on HAMR — Seagate ships HAMR (Mozaic) at volume while Toshiba's first HAMR drives are 2026-2027 test vehicles, not volume products
- Dependence on TDK and Resonac for HAMR-critical components it does not control
- No standalone financial disclosure for the HDD unit - opacity for outside capital
- Parent conglomerate history of governance crises and restructuring; HDD competes internally for capital against energy/infrastructure priorities
Opportunities
- AI training and inference data lakes driving multi-year nearline exabyte growth; industry ASPs and margins at records
- HAMR transition (test vehicles 2026-2027, ~40TB by 2027, 55TB by 2030) resets the capacity race and could narrow the gap if execution is clean
- Toshiba relisting — 2026 press reports target as early as FY2028, with preferred shares consolidated into ~JPY 750B of bank loans - or a separate monetization/spin of the storage unit could crystallize value
- HDD supply tightness pushing hyperscalers to sign long-term agreements into 2027-2028, improving revenue visibility
Threats
- NAND/SSD substitution at the margin — QLC eSSDs (Kioxia, Micron, Samsung, SK Hynix, Sandisk) attacking warm-tier storage economics
- HDD is a deeply cyclical duopoly-plus-one; the 2022-23 glut showed how fast nearline demand can air-pocket
- HAMR execution risk via third-party supply chain (TDK heads, Resonac media) vs Seagate's vertical integration
- Yen volatility and US tariff regime (management flagged tariff costs in Retail & Printing; storage supply chains are also exposed)
- Hyperscaler buyer concentration compresses pricing power over time
Moats, dependencies & bottlenecks
Moats
Only 3 HDD makers remain after decades of consolidation; entry is economically irrational
Multi-quarter drive qualification creates switching friction, but hyperscalers deliberately multi-source
Real but shared - critical HAMR components come from TDK/Resonac, who also supply Western Digital
Toshiba is the only non-vertically-integrated maker and the smallest of three - a cost disadvantage vs STX/WDC
Dependencies
Sole external source of HDD heads incl. HAMR heads; TDK also supplies Western Digital
HDD media (ex-Showa Denko), incl. HAMR media co-development
Nearline demand concentration; long-term agreements improve visibility but concentrate pricing power
HDD capex competes with energy/infrastructure inside a PE-owned conglomerate; exit timing is JIP's call
USD-priced drives, yen cost base; FY2025 avg 150 JPY/USD
Advantages
- Rational 3-player market with record ASPs and disciplined supply
- AI data-center exabyte growth is the strongest HDD demand driver in a decade
- Deleveraged parent with record ROS gives the unit more investment headroom than at any point since 2015
- Asset-light component sourcing limits downside operating leverage in a bust
Weaknesses
- Smallest share (~17% units, ~11% exabytes) and weakest ultra-high-capacity mix of the three makers
- HAMR timing lag vs Seagate; dependent on suppliers for the catch-up
- No standalone disclosure; valuation and diligence opacity
- Cyclical end-market with brutal historical drawdowns
Bottlenecks
- HAMR head and media supply and yield from TDK/Resonac - the gating item for Toshiba's 40TB+ roadmap
- Nearline drive assembly capacity — industry sold out through 2026; capacity adds are deliberate and slow (12-18 month lead time to volume)
- Hyperscaler qualification slots for each new capacity point
- Parent-level capital allocation to storage vs other Toshiba segments
Top signals & trends
Top signals
The single biggest execution watch-item
FY2025 record results and the Mar 2026 refinancing look like IPO grooming; no filing as of 2026-07
Industry sold out through 2026 with LTAs into 2027-2028; watch for the first sign of hyperscaler order digestion
Management's stated goal under the Revitalization Plan
The structural substitution threat; watch Kioxia/Micron/Samsung eSSD pricing
Already biting Toshiba's Retail & Printing segment in FY2025
Trends
Training corpora, inference logs, and data lakes are growing exabytes faster than NAND can economically absorb
Resets the capacity race; opportunity if Toshiba/TDK/Resonac execute, share-loss risk if they lag Seagate further
Slow-moving but one-directional; HDD retains a large cost-per-TB advantage in cold/nearline for now
The JIP deal is the flagship case; a successful relisting would validate the model and reward patient holders
WDC and Seagate report LTAs into 2027-2028; improves revenue visibility for all three HDD makers
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
HDD recording heads incl. HAMR heads - sole external head source
HDD media platters incl. HAMR media co-development (ex-Showa Denko)
HDD SoC/controller and preamp silicon supplier to the drive industry
HDD controller/read-channel silicon
Dominant HDD spindle-motor supplier
HDD pivot assemblies and precision mechanical components
Hyperscale nearline buyer
Hyperscale nearline buyer
Hyperscale nearline buyer
Hyperscale nearline buyer
Server/storage OEM channel
Server/storage OEM channel
Storage-systems OEM
Nearline co-leader (~41% unit share), first to volume HAMR with the Mozaic platform
Unit-share leader (~42%) post-Sandisk separation, pure-play HDD; ePMR/UltraSMR now, HAMR next
NAND/eSSD substitution threat in warm tiers; also Toshiba's former affiliate whose share sales and valuation gains funded FY2025 deleveraging
High-capacity QLC data-center SSDs attacking the HDD/SSD boundary
NAND pure-play (ex-WDC flash, separated Feb 2025) pushing high-capacity eSSDs
Largest NAND maker; eSSD substitution pressure (context only)