Phase Space AI

Financial Model Notes

Western Digital [WDC]

WDC — Financial Model Notes

Sources, transformations, and every place a vendor figure was rejected in favour of a primary filing. This name is the project's canonical post-corporate-action defect case; the provenance below is the main deliverable of this file.

Data provenance

Item Source used Why not the alternative
FQ3'26 revenue, margins, EPS; FQ4 guidance; 9M FY25/FY26 restated comparatives 8-K 0001628280-26-028878, EX-99.1 (2026-04-30) Primary, and it prints the restated continuing-ops comparatives that resolve the AV error
Quarterly revenue / gross profit / operating income EDGAR companyfacts XBRL, latest-filed value per period AV carries a pre-separation combined figure for the Dec-2024 quarter — rejected
Shares outstanding 10-Q XBRL CommonStockSharesOutstanding = 345m (2026-04-03); dei cover 344.7m (2026-04-23) AV commonStockSharesOutstanding = 376.0m is the diluted WASO — rejected
Diluted shares for valuation 385m (company FQ4 guidance; FQ3 actual 385–387m) Outstanding (345m) understates market cap by $18.1bn; the ~40m gap is the in-the-money convertible
Debt 10-Q XBRL: LongTermDebt = LongTermDebtCurrent = $1,581m; LongTermDebtNoncurrent = $0 The coordinator's "longTermDebt returns 0 in the newest quarter" defect DID fire here — but it is not a defect on this name: WDC genuinely has zero non-current debt because the entire $1.6bn convertible is classified current. Verified against the CFO's own statement of net cash and against DebtInstrumentCarryingAmount = $1,600m principal. A tool that treats a zero here as missing data would create a phantom $1.6bn of debt; a tool that treats it as total debt = 0 would create a $1.6bn EV understatement. Both are wrong; the answer is $1,581m and it required reading the 10-Q.
Net cash +$450m, company-stated (CFO, FQ3'26 call) Cash $2,050m − debt $1,581m = $469m; the $1,187m of short-term investments (very likely the retained 1.7m SanDisk shares) is excluded for conservatism
Exabytes Earnings-call transcripts, CFO prepared remarks Not in any filing
Opex lines for the bridge 8-K income statement: R&D $294m, SG&A $147m (FQ3'26) Per the coordinator's warning, AV sellingGeneralAndAdministrative was not used
Prices, realised vol Alpaca snapshots and bars (adjustment=all)
Splits / corporate actions AV SPLITS: one entry, factor 1.3230 effective 2025-02-24 This is the SanDisk spin-off price-adjustment factor, not a stock split. See below.

Vendor figures REJECTED, with magnitudes

Field Vendor value Correct value Error
AV INCOME_STATEMENT.totalRevenue (2024-12-27) $4,285m $2,409m +$1,876m / +77.9% — pre-separation combined basis inside an otherwise-restated series
AV INCOME_STATEMENT.operatingIncome (2024-12-27) $852m $560m +$292m / +52.1%
AV researchAndDevelopment (2024-12-27) $502m ~$250m restated ~+100%
AV commonStockSharesOutstanding (2026-03-31) 376.0m 345m +31m / +9.0%
AV OVERVIEW.MarketCapitalization $179,166m (implies a $519.80 price) $174,440m at 385m diluted × $453.09 +$4.7bn / +2.7%, and built on a stale price and the outstanding (not diluted) count — two errors partially cancelling
AV OVERVIEW.EPS $15.57 ~$9.88 FY2026E non-GAAP GAAP TTM inflated by the SanDisk monetisation gain; not earnings power
AV OVERVIEW.ProfitMargin 55.3% same GAAP artifact
AV OVERVIEW.OperatingMarginTTM 37.0% 30.3% (TTM GAAP $3,570m / $11,777m) +6.7pp — the AV ebit defect, confirming the coordinator's AMAT finding on a second name
AV OVERVIEW.EBITDA $3,928m not used AV ebitda is unusable by rule
AV INCOME_STATEMENT.operatingIncome (2026-01-02) $963m $908m +$55m / +6.1%
AV EARNINGS_ESTIMATES estimates: [] nothing returned
Alpaca adjusted bars × AV unadjusted reportedEPS, pre-2025-02-24 understated every multiple multiply pre-2025-02-24 adjusted prices by 1.3230 −24.4% on every pre-separation multiple. First-pass realised-forward P/E of 6.3x–7.2x was wrong; corrected 8.6x–17.1x. Same class as the KLA 10-for-1 split error.
Alpaca bars close for the current day (2026-07-29) $462.04 latestTrade $453.09 +2.0% — the daily bar for an in-progress session is not the last trade. Spot used is $453.09.

av_vs_edgar.py — re-run on the fixed version

Before fix After fix
STX cmp=0, disagree=0, trust=NO (null result reported as no disagreements) cmp=6, disagree=0, VERIFIED
WDC cmp=0, disagree=0, trust=NO cmp=8, disagree=1, worst=1.779x, FAILED

Both names need the fiscal-calendar tolerance to be checkable at all (max offset 4 days) — they are 52/53-week filers, which is exactly the bug class. The worst disagreement ratio of 1.779x is 4,285 / 2,409 = 1.7788 — i.e. the fixed tool independently reproduces the $1,876m Dec-2024 revenue error found by hand. The pre-fix version would have passed WDC as having nothing to resolve.

Model structure

FY2025A FY2026E FY2027E FY2028E
Revenue (continuing ops) $9,006m $12,822m ~$16.95bn ~$20.3bn
growth +42.4% +32% +20%
Non-GAAP gross margin ~40% ~47.6% ~51% ~52%
Non-GAAP opex ~$1,545m (12.0%) ~$1,650m (9.7%) ~$1,730m (8.5%)
Non-GAAP operating margin ~26% (FQ3'25) 36.3% ~42% ~44%
Non-GAAP diluted EPS $9.88 ~$16.05 ~$20.00
Diluted shares 359m ~382m ~380m ~375m

FY2026 = 9M actual ($9,172m revenue, EPS $1.78 + $2.13 + $2.72) plus the guided FQ4 ($3.65bn, GM 51–52%, opex $385–395m, EPS $3.25). FY2027–28 apply management's stated >25% exabyte CAGR with high-single-digit price in FY2027 decaying to low-single-digit in FY2028, and continued buyback at the recent pace (2.9m shares/qtr, partly offset by convertible dilution).

FY2025 revenue of $9,006m is the restated continuing-ops figure (9M $6,915m + FQ4 $2,605m − rounding), NOT the ~$13bn combined figure that appears in pre-separation sources.

Terminal-period assumptions

Parameter Value Derivation
m_gross,T 45.0% 5.5pp below FQ3'26 actual 50.5%, 6.5pp below FQ4 guide
R&D 6.5% FQ3'26 actual 8.8%; structurally above Seagate's because of the dual ePMR/HAMR roadmap
SG&A 3.5% FQ3'26 actual 4.4%
Other 0.5%
m_EBIT,T 34.5% bridge residual; 1.8pp below FY2026E actual, with three named causal reasons
t 16% company-guided and FQ3'26 actual
g 4.0% terminal
ROIC 35% current ~45% on ~$8.0bn invested capital, decayed
WACC 10.0% investment grade, net cash, high operating beta
Exit EV/EBIT 12.4x from the identity
FCF margin (interim) 26.0% FQ3'26 actual 29.3%, guided >30%

Known limitations

  1. reverse_dcf.py as shipped is terminal-only and overstates the required CAGR by 5.3pp here (38.6% vs 33.3%). Both reported; FCF-inclusive used.
  2. FQ4 FY2026 has not been reported. FY2026 figures are 9M actual + one guided quarter, and are labelled E throughout.
  3. Only ~5 quarters of same-basis (post-separation) history exist. Any margin mean, any multiple percentile, any volatility measure spanning 2025-02-21 mixes two different companies. The realised-forward P/E series is corrected for the price adjustment but the earnings basis still changes across that date, which is a second and non-correctable reason the multiple anchor is UNIDENTIFIED.
  4. Exabyte figures for FQ4'24–FQ3'25 are derived from disclosed y/y growth rates, not disclosed directly, and are marked as such.
  5. No options data priced — no IV−RV, no derivative vehicle admissible.
  6. SharesFloat (342.111m) ≤ SharesOutstanding (344.682m) ≤ diluted (385m); float × spot $155.0bn ≤ market cap $174.4bn. Invariant holds.