Financial Shenanigans

The Forensic Verdict

SanDisk lands at Elevated forensic risk (52/100). The headline complaint is not earnings manipulation - the income statement actually carries scars (a $1.83B goodwill impairment in Q3 FY2025 right after the spin) rather than cosmetics. The real issues are structural: a sixteen-month-old standalone company carved out of Western Digital using carve-out allocations, a $1.4B off-balance-sheet guarantee on Flash Ventures equipment leases plus $4.5B of multi-year Flash Ventures commitments, an explicit receivables-factoring program, multiple related-party ventures (Flash Ventures with Kioxia, SDSS with JCET, Unis Venture, residual WDC contracts), and a brand-new "New Business Model" (NBM) that has already booked $42B in remaining performance obligations from five customer contracts whose revenue-recognition geometry investors have not yet seen tested through a full cycle. The single data point that would most change the grade is the next 10-K disclosure of how NBM prepayments are classified between deferred revenue, customer deposits, and contract liabilities - if the $42B RPO materializes as revenue on the cadence the company suggests, the grade drops to Watch; if RPO is significantly back-end-weighted or contains take-or-pay accounting flexibility, the grade rises.

Shenanigans scorecard

Breeding Ground

The structural conditions are mixed, with one acute risk and several mitigants. SanDisk emerged in February 2025 from a Western Digital spin engineered by the same CEO (David Goeckeler) who continues to chair the SanDisk board - the leadership team that designed the carve-out is now reporting against the carve-out's economics. The auditor is KPMG, retained for FY2026 (its second standalone year), with no public auditor-change, late-filing, qualification, or material-weakness disclosure identified. Two transition directors from the WDC board (Massengill, Alexy) were not renominated, which is a normal governance refresh for a new standalone but reduces continuity. Compensation in the partial post-separation FY2025 period (Feb 21 - Jun 27, 2025) is reported as a stub, making the FY2026 say-on-pay the first cycle where pay-for-performance can be evaluated against AI-cycle earnings.

The single most important breeding-ground risk is comparability discontinuity. The FY2023 and FY2024 income statements are "Combined" carve-outs derived from WDC's books with allocated overhead; only the seven months from February to June 2025 are true standalone results. Any year-over-year ratio that crosses the spin date is partly an accounting artifact.

Earnings Quality

The first-pass earnings test is dominated by two distortions: a $1.83B goodwill impairment in Q3 FY2025 (a "big bath" immediately after the spin) and a depreciation schedule that collapsed from $525M in FY2022 to $163M in FY2025 even as the property base shrank only modestly. Both signals point in the same direction - reset the going-forward expense base lower right after standalone status was achieved.

The Q3 FY2025 impairment was triggered by SanDisk's own post-spin quantitative test against an observable market capitalization that fell short of carrying value. It is accounting-policy compliant, but the timing - taken on the third full quarter of standalone reporting, after the same management team controlled the spin valuation - fits the classic big-bath profile. The FY2025 10-K then confirmed the fourth-quarter qualitative test cleared. So the bath was concentrated, the post-bath income statement reset lower, and quarters since (Q4 FY2025 through Q3 FY2026) have shown rising gross margin off the cleaner base.

The revenue/receivable relationship is volatile but not deceptive. The FY2023 collapse in receivables to $605M reflects heavy use of factoring during the NAND downturn; the FY2024 rebound to $1,044M reflects lower factoring. The FY2025 receivables growth (+14%) outpaces revenue growth (+10%) by 4 percentage points - a yellow but not a red flag, and management explicitly attributes it to lower factoring. DSO at 54 days is below the FY2023 peak of 75 and matches the company's industry pattern. The forensic concern is that the variance in receivables is driven by management's discretionary factoring, not by collection patterns.

D&A fell 69% from FY2022 to FY2025 while PP&E only fell 40%. Three factors plausibly account for it: (1) the sale-leaseback of Milpitas in September 2023 moved owned assets off balance, (2) Flash Ventures (the Kioxia JV) holds wafer-fab equipment off SanDisk's books, and (3) carve-out allocations from WDC may have shifted in proportion. Even so, the D&A drop in FY2024 alone was $224M against a PP&E decline of only $142M - a 50% drop in expense against a 15% drop in the asset base. This deserves explicit disclosure of useful-life changes or asset class re-statements in the next 10-K, and is a yellow flag pending that explanation.

Cash Flow Quality

Operating cash flow has been on a four-year deterioration-then-recovery cycle. Looking through the recovery, the cumulative three-year (FY2023-FY2025) sum is -$938M of OCF on -$4,456M of net loss - a CFO/NI ratio of 21% during the downturn, which is structurally weak but consistent with a memory-cycle bottom. The Q3 FY2026 surge to $2.99B of FCF in a single quarter is real - the NAND industry is in shortage and pricing is up - but the same period also saw the first major NBM customer prepayments, which inflate working capital favorably.

The receivables-factoring disclosure is the single most important cash-flow item to track. The FY2025 MD&A explicitly states: "DSO increased 3 days when compared to the prior year, reflecting lower accounts receivable factoring." That phrase concedes a financing-to-operating shift - factored receivables are economically a financing transaction (proceeds from a financial institution against a customer obligation) but appear inside operating cash flow. The volume of factored receivables is not quantified in the data available here. A regular forensic adjustment subtracts factored receivable proceeds from CFO; without the quantum, the conservative read treats reported OCF as up to one DSO-quarter overstated relative to a no-factoring baseline.

SBC is a steady ~$170M add-back, equivalent to 2.5% of revenue in FY2025, which is moderate by tech-hardware standards and broadly aligned with peers. The bigger non-cash add-back is the FY2025 goodwill impairment ($1.83B), which converts -$1.6B of GAAP loss into +$84M of OCF on its own. Strip the impairment and the underlying business generated negative cash from operations.

For FY2025, including the SDSS divestiture proceeds ($401M) raises adjusted FCF to +$281M, but the divestiture is a once-only event. The investor-facing free cash flow improvement story is real on a 12-month basis only when the SDSS gain and the goodwill impairment are both excluded. A clean view of FY2025 FCF before non-recurring items is roughly -$120M, consistent with a memory cycle bottoming but not yet generating durable cash.

Metric Hygiene

Non-GAAP and GAAP reconciliations for the FY2026 quarters are unusually clean - the gap is dominated by stock-based compensation and a small amortization-of-intangibles line. The Q3 FY2026 reconciliation shows Non-GAAP EPS of $23.41 versus GAAP $23.03 - a 1.6% premium that is well-disciplined for a tech-hardware issuer. The historical reconciliations in Q3 FY2025 are wider, but the bridge is dominated by the $1.83B goodwill impairment, which is structurally one-time.

The standout metric to underwrite is NBM remaining performance obligations. Five contracts, $42B of contracted revenue, multi-year horizon, prepayments and third-party guarantees as customer backing. This is genuinely new accounting territory for a memory company - prior memory cycles ran on spot pricing and quarter-by-quarter purchase orders. The risk is asymmetric: if NBM contracts contain take-or-pay shortfall mechanics, revenue can be recognized on a smoothed basis even if customer drawdowns lag. If they are pure firm orders with delivery-based recognition, RPO disclosure becomes a high-quality forward indicator.

Excluding the on-balance debt and operating leases, the contractually committed off-balance-sheet outflow exceeds $11.9B, against an on-balance-sheet equity base of $9.2B and reported long-term debt of only $1.85B. The implied leverage when JV commitments are included is materially higher than the headline 0.2x debt/equity ratio suggests.

What to Underwrite Next

The forensic work does not point to a thesis breaker. It does point to specific items to verify in the next 10-Q and 10-K, and to a sizing posture that respects the structural uncertainty.

Top five items to track:

1. NBM contract-liability roll-forward in the next 10-Q (Q4 FY2026, expected August 2026). Look for a separate disclosure of customer deposits versus deferred revenue versus contract liabilities, the revenue conversion of $42B RPO by year, and whether any portion is recognized straight-line vs. delivery-based.
2. Receivables factoring quantification. The MD&A confirms factoring is used but does not quantify the FY2025 volume. Required input for any adjusted CFO calculation.
3. D&A useful-life or impairment disclosure. Reconcile the 69% drop in D&A from FY2022 to FY2025 against a 40% PP&E decline. Either a useful-life change was applied or the Flash Ventures off-balance treatment is doing more work than disclosed.
4. Related-party revenue exposure. Map FY2026 revenue to (a) products sold through SDSS-manufactured wafers, (b) any remaining WDC-aligned customers, (c) Unis Venture. Disclosure expected in Note 10 of the FY2026 10-K.
5. Goodwill carrying value at next annual test. With a stock price up over 50x from spin, the headroom over carrying value has expanded - but a single bad quarter could re-trigger the test. Watch the WACC sensitivity disclosure.

Signal that would downgrade to Watch (21-40): Q4 FY2026 10-Q shows clean NBM contract-liability roll-forward, factoring volume disclosed and modest (under 10% of revenue), and an explanation of the D&A decline that ties to specific asset-class useful-life changes.

Signal that would upgrade to High (61-80): Any one of - restatement, material-weakness disclosure, NBM contracts revealed to contain straight-line revenue recognition mechanics inconsistent with delivery-based GAAP, large new related-party revenue concentration, or a Flash Ventures covenant breach triggering the $1.4B guarantee.

Position sizing implication: This is a structurally complex situation with an extraordinary cyclical tailwind. The forensic work argues for a margin-of-safety haircut on multi-year forward earnings until at least two clean standalone fiscal years are reported and the NBM cohort produces a full revenue cycle. It is not a thesis-breaking accounting risk, but it is a position-sizing constraint - the off-balance-sheet $11.9B of JV and customer commitments warrants treating the underlying enterprise value as larger than the headline market cap and on-BS debt would suggest. The accounting is honest enough to size around; it is not yet honest enough to size into without reserve.