Financial Shenanigans

Financial Shenanigans — AUTO1 Group SE (AG1)

Figures converted from euros at historical FX rates — see data/company.json.fx_rates for the rate table (FY-end EUR→USD: 2022 1.0666, 2023 1.105, 2024 1.0389, 2025 1.175). Ratios, margins, multiples, share counts, dates and names are unitless and unchanged.

AUTO1's audited numbers are, line by line, a faithful IFRS representation — KPMG signs clean opinions, the related-party notes are empty of related-party revenue, and there is no restatement, regulator, or auditor exit anywhere in the record. The forensic risk does not live in the ledger; it lives in the gap between the audited cash flows and the investor-facing narrative. In FY2025 AUTO1 reported its best-ever $91.6m net profit while operating activities consumed $544.1m of cash [1]. Management bridges that hole with non-recourse asset-backed securitizations (ABS) it presents as "no corporate debt" and "over half a billion free cash," and reports "record" results on an Adjusted EBITDA that adds back the same costs every year. None of this is fraud. All of it requires a buy-side reader to underwrite the framing rather than accept it.

Forensic verdict — Elevated (53 / 100)

Source: derived from reported financials — FY2025 net profit $91.6m vs operating cash flow −$544.1m [1]; company filings, as reported.

The two red flags that matter. First, cash-flow quality: AUTO1 has burned operating cash every year for at least four years — a cumulative −$1.25bn of operating cash flow against cumulative net losses of −$278m across FY2022–FY2025 — yet management frames the business as "self-funding" because ABS facilities plug the gap [1]. Second, off-balance-sheet leverage and non-GAAP framing: $1.9bn of working-capital debt is presented as "no corporate debt" because it is non-recourse [12], while headline "record profitability" rests on an Adjusted EBITDA that adds back recurring share-based pay and "other non-operating" costs year after year [17].

The cleanest offsetting evidence. The audited cash-flow statement is conservatively prepared: inventory and captive-finance receivable growth sit in operating outflow, and ABS proceeds sit in financing — AUTO1 does not flatter operating cash flow by reclassifying the financing inflows into it [1]. Related-party notes disclose only board remuneration — there is no related-party revenue — and KPMG has issued unqualified opinions with revenue recognition transparently flagged as a Key Audit Matter [36].

What would change the grade. Downgrade (toward High) if an ABS market freeze or covenant event forced the captive-finance and inventory book back onto on-balance-sheet corporate funding — management already concedes liquidity "increasingly depends on access to external structured funding rather than on-balance-sheet cash" [7]. Upgrade (toward Watch) if operating cash flow turns durably positive after funding captive-finance growth, proving the model self-funds without perpetual new ABS issuance.

The core question: positive profit, deeply negative cash (CF1, CF4)

The single most important forensic fact about AUTO1 is that net income and operating cash flow point in opposite directions, and the divergence is widening as the company turns "profitable." Net result improved from −$262.8m (FY2022) to +$91.6m (FY2025); over the same window operating cash flow went the other way, from −$418.5m to −$544.1m [1]. The FY2025 accrual ratio — net income less operating cash flow, over average assets — is +21.3%, an extreme reading that means reported earnings are almost entirely non-cash this year.

Source: FY2025 Annual Report, Consolidated Statement of Cash Flows (FY2024–FY2025 columns) [1]; earlier years derived from reported financials.

Name the mechanism. The cash drain is not an earnings problem — it is a deliberate balance-sheet build. The FY2025 operating outflow is driven by an $828.4m increase in operating assets [2], which management attributes to a $424.1m increase in inventories plus growth of $215.5m in instalment (consumer-finance) receivables and $104.1m in merchant-financing receivables [3]. AUTO1 is buying cars and lending to its customers; that consumes cash regardless of accounting profit.

Source: FY2025 Annual Report, Financial Performance — inventory increase of KEUR 360,923 plus instalment and merchant-finance receivable growth [3].

Where the cash comes from. The $544.1m operating hole is filled by $560.4m of financing inflow, led by $1,502.2m of proceeds from new borrowings — the ABS drawdowns [5]. AUTO1 states the design plainly: inventory and receivables are refinanced via ABS and "the resulting cash flows are reported in cash flow from financing activities" [4]. This split is correct under IFRS — and it is the cleanest thing about the cash story. The forensic concern is not the statement; it is what management says the statement means.

Management's own admissions undercut the "self-funding" narrative. On the Q1 2026 call, the CFO conceded the period's cash inflow was a timing effect: "we expect normal working capital patterns to reverse some of Q1's overall EUR 48 million" cash inflow [9]. On the Q4 2025 call, asked about 2026 free cash flow, management said positive "free trading operating cash flow" assumes "we continue to use ABS structures to fund the inventory" and the receivables [10]. The cash generation is real only so long as the securitization market stays open — a conditionality the FY2025 risk section itself acknowledges [7].

Off-balance-sheet optics: "no corporate debt" (KM2)

AUTO1's balance sheet carries $1,883.3m of financial liabilities against $831.3m of equity — roughly 2.3x financial-liability leverage and a $1,558.5m accumulated deficit [15] [16]. To investors, that debt is presented as "All of our long-term debt is non-recourse; No Corporate Debt," with the explicit claim that "non-recourse re-financing does not count towards net debt" [12].

Source: FY2025 Annual Report, Consolidated Statement of Financial Position [15]; financial-liabilities note (FY2025 total KEUR 1,602,842, "recourse only to the assets pledged as collateral") [6]; earlier years derived from reported financials.

The "non-recourse" framing is real but partial. AUTO1 tells investors the financing SPVs are "legally NOT part of AUTO1 Group," then concedes on the very next slide that "we therefore fully consolidate the assets and liabilities" because the group bears the SPVs' economics [14]. The debt is on the consolidated balance sheet; the slide simply asks the reader to discount it. A skeptical underwriter should treat the $1.9bn as economic leverage that is cheap and ring-fenced — not as something that "does not count."

This is also where genuine derecognition risk lives. In September 2025 AUTO1 carved a closed consumer sub-portfolio out of its ABS facility and placed public ABS notes — its second public securitization — and management has said it is "open to… whether or not there are other structures to have it off balance sheet" [13]. Receivable derecognition is the mechanism that could one day move cash from financing into operating; it has not done so yet, but it is the line item to watch.

Earnings quality: a thin margin, propped two ways (EM3, EM4, EM5)

AUTO1's reported gross margin is 12.1% and operating margin 1.7% — there is very little cushion, so small accounting choices move the headline. Two choices flatter that thin margin, and one reserve looks light.

Capitalization into inventory and intangibles (EM4). FY2023 was the inflection year: AUTO1 began capitalizing development costs as internally generated intangibles under IAS 38 [31]. It also capitalizes vehicle refurbishment costs into inventory rather than expensing them [28], so refurbishment work lifts gross profit until the car sells. Management quantified the refurbishment value-add capitalized in Q4 2025 alone at roughly $83m [29], and concedes its reported Autohero GPU "differs to the reported GPU in previously issued financials, which was adjusted for effects from capitalisation of internal refurbishment costs" [30]. Reported capex ($30.1m, 0.3% of revenue) is less than depreciation ($65.1m) — so the headline capex line looks conservative precisely because the real investment is routed through inventory and intangibles instead.

Source: FY2025 Annual Report — own-work-capitalised intangibles credit KEUR 3,285 (2024: 4,218) [26]; internal refurbishment capitalisation KEUR 3,309 (2024: 1,102) [27]; capex and D&A from Consolidated Statement of Cash Flows [1].

The amounts booked through the P&L are modest in absolute terms (single-digit millions), so this is a yellow flag, not a red one — but on a $91.6m net profit and a 1.7% operating margin, single-digit millions are not trivial, and the policy is the right one to track as production centres scale.

Reserves that lag the book (EM5). The inventory write-down to net realisable value rose to $50.8m (FY2024: $34.7m; FY2023: $23.9m) — but that is roughly 4% of a $1,242.8m inventory book, and in cash-flow terms the period inventory value-adjustment actually fell to $11.5m from $12.3m even as inventory grew 52% [32] [33]. The warranty ("provisions for vehicles") balance grew only to $21.6m from $18.0m — about 6.5% — even though Autohero retail units grew far faster and the shareholder letter flags that "almost every second Autohero Retail customer added an additional longer term warranty" [34] [35]. Rising warranty attach rates and longer terms against a barely-moving provision is a classic under-reserving pattern in a business already exposed to falling used-car prices.

One-time income is not the problem (EM3, clean-ish). Crucially, FY2025 profit was not manufactured by a tax credit — tax expense actually rose $14.9m and the financial result worsened $7.3m, both drags on profit [17]. The only recurring "one-time-flavoured" income items are the own-work-capitalization credits above and recurring severance add-backs — small, and disclosed.

Metric hygiene: "record profitability," adjusted (KM1)

AUTO1 runs the equity story on Adjusted EBITDA and GPU, not on IFRS profit. For FY2025 the wedge is large: EBITDA of $203.8m becomes Adjusted EBITDA of $232.1m, against net profit of $91.6m [17]. The $28.3m of add-backs to reach Adjusted EBITDA equal 31% of net profit, and the same two lines — share-based payments and "other non-operating expenses" — are added back every year.

Source: FY2025 Annual Report, Development of EBITDA and Adjusted EBITDA (FY2025 adj. EBITDA KEUR 197,542; net profit KEUR 77,949) [17]; FY2024 report (FY2024 adj. EBITDA KEUR 109,240; 2023: KEUR -43,948) [20]; FY2022 report (adj. EBITDA KEUR -165,578) [21].

The add-backs are not "one-time." Share-based payment expense has been added back every year — $8.5m (FY2022), $18.5m (FY2024), $18.6m (FY2025) — and "other non-operating expenses" recur too ($7.6m FY2022, severance-led $3.9m FY2023, $9.8m FY2025) [19] [21] [22].

Source: FY2025 Annual Report, 6.11 Share-Based Payment (FY2025 SBC KEUR 15,822) [19]; FY2024 report (FY2024 SBC KEUR 17,843) [20]; FY2022 report [21].

Three further hygiene flags. The Adjusted EBITDA definition has quietly narrowed since the IPO — the prospectus version carried a third add-back, "correction of payables," that the current two-line definition has dropped, with no reconciliation of the change [23]. The glossary defines GPU as "gross profit divided by units sold," yet the segment footnotes admit "GPU is not equal to gross profit divided by the number of sold cars" because capitalized refurbishment is excluded — two different definitions in one report [24]. And a second non-GAAP profitability metric, "Adjusted Net Income," was introduced in Q1 2024 — exactly as GAAP profitability first turned positive — framed as "best ever profitability" [25]. None of this is prohibited; all of it nudges the reader toward the flattering number.

Breeding ground: founder-controlled, with the chairman on the audit committee

The governance set-up amplifies rather than dampens the framing risk. AUTO1 is founder-controlled: co-founder Hakan Koç chairs the Supervisory Board and sits on the Audit and Risk Committee that oversees the financial reporting he is meant to challenge [44]; co-founder Christian Bertermann is CEO. At IPO, SoftBank (via SVF Midgard) was the largest holder at 19.99%, with the two founders' vehicles (BM Digital, HKVV) at roughly 15.7% each [41]. Incentives are equity-skewed through large discrete grants — the LTIP 2025 alone awarded Bertermann 7,500,000 share options with a $4.7m grant-date fair value [40] — which sharpens the incentive to showcase Adjusted EBITDA and GPU, the metrics that move the share price.

Two structural mitigants keep this from becoming a red flag. The audit relationship is young, not entrenched — KPMG has been group auditor only since FY2021 [37] — and the related-party note is genuinely clean: it discloses only board remuneration, with no related-party revenue, sales, or circular dealing [39]. The one historical related-party item of note — a $289m convertible loan at IPO involving the CEO's BM Digital entity — was disclosed and has rolled off [42]. The IPO-era admission that "our internal controls may not be sufficient" is the kind of risk-factor boilerplate every newly listed company carries [43], but five years on there is no material-weakness disclosure to corroborate it.

The 13-category scorecard

Coverage of every shenanigan category. Only three rows are red, and they cluster on the same theme — cash-flow sustainability and the non-GAAP/leverage framing built on top of it. Most earnings-quality tests are yellow (capitalization, reserves) or clean (bogus revenue, big baths, smoothing, CFO classification).

Sources: cash-flow classification and operating-asset build, FY2025 Annual Report Statement of Cash Flows and Financial Performance [1] [3]; Adjusted EBITDA bridge [17]; "no corporate debt" framing [12]; revenue-recognition Key Audit Matter [36]; related-party note [39].

What to underwrite next

The accounting risk here is a valuation-and-position-sizing question, not a thesis-breaker. The audited numbers are reliable; what is fragile is the narrative of a profitable, self-funding, "no corporate debt" business. Five things to track, in order of value:

1. Operating cash flow ex-ABS. The decisive test of the "self-funding" claim. If operating cash flow stays deeply negative while net income rises, the profit is an accounting event, not cash — exactly the FY2025 pattern (CFO −$544.1m on $91.6m profit) [1]. Watch whether the Q1 2026 working-capital inflow reverses as management warned [9].
2. ABS facility headroom, drawdown rate and pricing. Liquidity now "increasingly depends on access to external structured funding" [7]. A widening in ABS spreads or a covenant trip is the fastest route to a forensic downgrade.
3. Warranty provision and inventory NRV coverage versus retail-unit growth, attach rates and used-car price indices — the line where a thin margin is most exposed if reserves keep lagging [34].
4. Gross profit / GPU ex-refurbishment-capitalization. Management has now restated GPU to strip the capitalization effect [30]; compare the two GPU bases each quarter to size how much margin is policy.
5. Receivable derecognition. If a future securitization moves consumer receivables genuinely off-balance-sheet, watch whether the gain or the freed cash gets routed through operating cash flow [13].

Bottom line. AUTO1's forensic risk is a valuation haircut and a position-sizing limiter, not a fraud flag. A PM should underwrite this name on IFRS net income and ABS-adjusted cash flow — never on Adjusted EBITDA, GPU, or a "no corporate debt" balance sheet — and should size the position for the reality that the entire working-capital engine runs on continuous access to the securitization market. If that market stays open, the model works and the grade stays Elevated; if it tightens, this becomes a High-risk, liquidity-dependent story very quickly.