Financial Shenanigans

Financial Shenanigans

Figures converted from KRW at historical FX rates — see data/company.json.fx_rates for the rate table. Ratios, margins, and multiples are unitless and unchanged.

Auros earns a Forensic Risk Score of 38/100 — Watch (upper end). There is no evidence of restatements, auditor changes, regulatory action, material weakness disclosures, or short-seller allegations, and stock-based compensation has normalised post-IPO. The two material concerns are (1) a striking earnings-versus-cash divergence in FY2024, when reported net income of $4.0M coexisted with operating cash flow of negative $3.6M as receivables and inventory soaked up $13.0M of working capital, and (2) a structural related-party footprint — the main Dongtan-1 factory is leased from the 33.5% controlling shareholder FST, and a $3.7M asset disposal to 17.3% affiliate CM Technology was board-approved in 2025. The single data point that would most change the grade either way is whether the FY2024 receivables and inventory build either collects to cash in FY2026 (downgrade to Clean–Watch) or sits stale and forces a write-down (upgrade to Elevated).

1. The Forensic Verdict

The five-year accumulated picture is the clearest forensic frame. Auros has reported $1.4M of cumulative net income between FY2021 and FY2025 but generated negative $26.8M of cumulative free cash flow over the same window. The gap is not the result of aggressive revenue recognition — Auros is selling capital equipment to two highly creditworthy memory IDMs — but of (i) chronic working-capital absorption inside a long-lead-time capital-equipment business and (ii) heavy ongoing capex into Dongtan-2 thin-film R&D, ahead of revenue. The accounting is not the problem; the economics of the reported earnings are the problem, and that is what the verdict has to grade.

2. Breeding Ground

The breeding-ground profile is structural KOSDAQ small-cap — multiple amplifiers that are sector-normal individually but, taken together, give a controlling-shareholder block meaningful latitude in any contested situation. None of it rises to a disqualifier; the question is how much weight an institutional investor puts on each.

The audit-committee absence and 1-of-4 independent representation are the most cited governance items, but the single most consequential structural fact is the FST relationship. FST is the largest shareholder, the landlord of the main production factory, and itself a loss-making KOSDAQ-listed semiconductor materials company. Chairman of FST also concurrently serves as CEO of Korea Lam Research Inc., one of the largest deposition/etch sub-systems suppliers into Korean fabs — an industry-relationship benefit, but also a related-party node that has not been disclosed as fully arm's length. None of these facts indicate misconduct; together they raise the prior on aggressive accounting from "remote" to "non-trivial," which is why the breeding ground reads yellow not green.

3. Earnings Quality

Reported earnings are volatile rather than smoothed, which is itself a quality signal — the company is not hiding cycle losses, and the FY2025 swing into a $4.4M net loss matches the underlying revenue decline (−15% YoY) and gross-margin compression (FY24 59.0% → FY25 48.6%; FY23 66.8% peak for reference). The single forensic question that needs underwriting is whether FY2024's profit was overstated by capital tied up at year-end.

Only FY2021 shows a clean ladder of net income → CFO → FCF. Since IPO the pattern has been CFO consistently below or close to NI, and FCF persistently negative as capex absorbs more cash than the business generates. FY2024 is the standout outlier on the wrong side: profit was reported, but the cash never showed up.

Receivables jumped from $1.1M (FY23) to $6.8M (FY24) while revenue grew 35% — receivables expansion of 589% against revenue growth of 35%. That is exactly the canonical pattern Lev/Thiagarajan and Sloan flagged: outsized receivables expansion in a strong year. Through the FY24→FY25 downcycle receivables stayed roughly flat at $7.1M while revenue fell 15% — DSO ratcheted from 60 to 72 days. This is consistent with two innocent explanations (Q4-FY24 was a back-end-loaded shipping quarter — $19.7M of revenue in Q4 alone, 47% of annual; FY25 collections lengthened as customer-CapEx tightened) and one less benign one (channel stretching at quarter-end). The next pertinent test is whether FY2026 Q1 collections normalise the AR balance back toward 30–45 days.

Inventory days remain extreme but appropriate for the business — Auros builds capital tools with 3-month lead time at ~$1.5-2M per system, so 300-500 days of inventory cost translates to ~10-15 tools-in-progress for a 20-30/year capacity line. The forensic question is not the level but the vintage: $16.2M of FY25 inventory needs to clear at a sustainable margin to validate the carrying value. No inventory write-down has been disclosed despite the 15% revenue decline and FY25 gross-margin collapse to 48.6%.

4. Cash Flow Quality

Cash flow quality is the weakest dimension of the forensic profile. The cumulative five-year picture is unequivocally negative.

FY2024 working capital absorbed $13.0M of operating cash flow — receivables and inventory together. FY2025 working capital gave back $3.9M, primarily through the inventory drawdown, and that single mechanism is what flipped CFO from negative $3.6M to positive $1.2M. Excluding working-capital movement, FY25 CFO from earnings + D&A + SBC would have been roughly negative $0.6M. The bald reading is: FY25 cash generation came from balance-sheet unwind, not from operations.

Capex/D&A averages 2.5x over five years, with FY25 at 2.9x ($10.1M against $3.5M D&A). The capex spending reconciles cleanly to disclosed projects — Dongtan-2 thin-film R&D centre buildout, HQ relocation planned March 2026, and capacity expansion. PP&E grew from $12.8M to $23.3M over five years; intangibles from $0.6M to $1.4M. There is no evidence of operating-cost capitalisation — the historian R&D-to-revenue ratio (36.7% in FY25) confirms most discretionary spend is run through SG&A.

The financing pattern is consistent with a cyclical-down-year-needing-a-bridge: short-term debt rose from zero (FY22-FY23) to $6.9M (FY24) to $9.8M (FY25), with a new $1.4M long-term loan in FY25. Debt-to-equity rose from 38.5% to 45.2%. There is no receivables-factoring, securitisation, or supplier-finance arrangement disclosed — the financing is plain bank debt, properly classified in the financing section of the cash-flow statement.

5. Metric Hygiene

Korean filings under K-IFRS do not lend themselves to non-GAAP gymnastics — there is no adjusted-EBITDA / adjusted-EPS culture in Auros's disclosure stack, no organic-growth/same-store-sales construct, and no headline metric that diverges materially from the consolidated statement. This is a structural plus.

The deferred-tax-asset accumulation deserves one sentence of commentary: deferred tax assets rose from $1.3M (FY24) to $2.9M (FY25) and produced the $1.8M tax-benefit cushion that cut FY25's pretax loss of $6.2M down to a reported net loss of $4.4M (a 29% optical improvement). The mechanics are correct — net operating loss carryforward in a tax jurisdiction (Korea) where carryforwards are usable — but the realisability of the asset depends on returning to taxable profit. If FY26 stays loss-making, expect a partial-realisability assessment that could reverse part of the benefit.

Net FY2025 related-party cash position: roughly $2.8M inflow ($3.75M CM Technology asset sale less $1.0M aggregate lease/welfare outflows), or ~0.8% of revenue but ~43% of FY25 operating cash flow. The CM Technology asset sale is the only line item that materially shifts a forensic ratio — the absence of detail on what asset was sold, at what carrying value, and against what comparable third-party benchmark is the single most asked-for disclosure missing from the FY25 filing.

6. What to Underwrite Next

The forensic verdict is Watch — upper end, not Elevated, because every red and yellow flag has an alternative innocent explanation supported by either disclosed business circumstances or sector-normal practice, and there is no external corroboration of accounting concern (no restatement, no auditor change, no regulatory probe, no short report, no class action). But several specific items need to clear in the FY26 reporting cycle to keep the grade where it is.

The signal that would downgrade the forensic grade toward Clean–Watch is a clean FY26 Q1 disclosure: AR collected back to ~$4.1-4.8M (DSO 45-50 days), inventory unwound by another $1.4-2.1M without write-down, and an explicit related-party-transaction note quantifying the CM Technology asset disposal against third-party benchmarks. The signal that would upgrade the grade toward Elevated is FY26 first-half disclosure of an inventory write-down, AR aging past 90 days, or a restated comparative for FY24-FY25 working-capital lines.

For position sizing, the accounting risk here is best treated as a valuation haircut, not a thesis breaker. The two structural facts — 5-year cumulative FCF of negative $26.8M against cumulative net income of positive $1.4M, and a related-party-rich governance posture — argue for a discount-rate add-on of perhaps 100-200 basis points relative to comparable Korean small-cap semicap names that do not carry the same FST-affiliate complex, and for sizing such that an FY24-era inventory or receivable charge of $2-3.5M would not be thesis-defining. Auros is not Wirecard. It is also not yet a company whose reported earnings the market should fully trust as cash-equivalent — and the discount applied today should reflect that asymmetry.