KLAR · Forward model · Bull case

The Bull case, 20 quarters out

Each segment is projected from its own operating driver, rolled up into consolidated cash flow, and discounted back to a fair value per share. The assumptions below are editable — change them and every number on this page moves with them.

WHAT IS DISCLOSED AND WHAT IS NOT. Klarna states under IFRS 8 that it 'operates as one operating segment and has one reportable segment'. The US / Global ex-US split used here is the company's own supplementary management disclosure, published quarterly in the SEC-filed EX-99.5 for ten quarters (2024 Q1 to 2026 Q2), and the US revenue figure is corroborated to the dollar by the IFRS geographic note ($376m in both). Every actual in both verticals is a DIRECTLY REPORTED figure - none is estimated, apportioned or derived. Disclosed by region: GMV, revenue, each transaction-cost line and transaction margin dollars. NOT disclosed by region and therefore NOT split: operating expenses, depreciation, share-based payments, other income, tax, net income and non-controlling interests. The model puts the regional transaction margin inside the verticals and the ENTIRE non-transaction cost base into corporate overhead. Nothing below the transaction margin is apportioned to a region, and no split Klarna does not publish has been invented. THE LABELS ARE WRONG AND THE PAGE CANNOT CHANGE THEM. This is an EQUITY-level DCF. Where the page says 'Enterprise value' it means equity value; where it says 'Net cash' it means a deliberate zero. $11.7bn of consumer deposits fund 88% of the funding base and $1.7bn of notes payable sit alongside them; those are funding at a licensed EEA bank, not financing to be netted against the $2.7bn of cash and $2.6bn of debt securities held as the liquidity buffer. Same convention as BAC, GS and JPM here. A DISCONTINUITY BETWEEN HISTORY AND PROJECTION, ON PURPOSE. From H2 2026 Klarna moves US and German Fair Financing to fair value through profit or loss, which cuts reported revenue AND reported transaction costs each by about 10bp of GMV. Prior periods are NOT restated. The reported FY2026 take rate is guided at 2.74-2.75% against a comparable 2.84-2.85%, and the guidance table prints the ratio twice precisely because the two are not comparable. The unit prices and margins in this spec are on the POST-change reported basis (the disclosed pre-change ratios divided or multiplied by 0.9323), so that the projection is comparable with guidance; the ten quarters of actuals are left exactly as reported, on the pre-change basis. Transaction margin DOLLARS and pre-tax profit are identical on either basis - only the ratios move - but a reader comparing the 2026 Q2 actual of $1,042m with the modelled 2026 Q3 of $969m is looking at a 7% guided decline plus a roughly 3% presentation change, not a 10% collapse. THE ONE PLACE THE MODEL DISAGREES WITH GUIDANCE, STATED HERE RATHER THAN LEFT TO BE FOUND. The model's Q3 2026 transaction margin dollars are $447m against a guided $340-360m - 24% above the top of the range. It is deliberate. The model starts from the DELIVERED basis-quarter regional margins because those are what Klarna discloses by region; the guided Q3 trough is not published by region, and haircutting both verticals uniformly to reach $350m would break the calibration gate and substitute a modelled ratio for a disclosed one. Klarna's own quarterly guides also run conservative: it guided Q2 2026 TMD at $375-395m and delivered $446m, and guided adjusted operating income at $30-50m and delivered $91m. And Q3 is a disclosed one-quarter investment trough carrying 'the highest level of share-based payments in 2026', not a level - the FY guide implies Q4 2026 TMD back near $450m. The engine cannot seasonalise a margin, so it cannot reproduce a one-quarter dip. Guidance-implied Q3 pre-tax profit is about -$59m (adjusted operating income midpoint $10m less roughly $23m of D&A and roughly $45m of share-based payments); the model prints $29m. That $88m gap is one quarter of twenty. NON-CONTROLLING INTERESTS ARE A REAL LEAK. The engine has no NCI concept, so its free cash flow lands on TOTAL net profit ($9m in the basis quarter), not on the $4m attributable to shareholders of Klarna Group plc. Over H1 2026 non-controlling interests took $11m while attributable profit was NEGATIVE $1m - more than the whole of the group's profit. The mitigating fact is that these are Employee Equity Program interests in Klarna subsidiaries whose instruments exchange into Klarna Group plc ordinary shares (1,948,166 were exchanged in April 2025 alone), so they are better read as future dilution of a flat share count than as a permanent third-party claim. Calibrating to attributable profit instead would require an 85.2% 'tax' rate, which is not a tax rate. SHARE COUNT: 379,160,066 PERIOD-END BASIC, and it points the OPPOSITE way from GS and BAC. Those two shrink their counts through buybacks, so a flat count understates their terminal EPS. Klarna has NO buyback authorisation and NO dividend - it retains everything to fund loan growth - and its count GROWS: 1,652,156 ordinary shares were issued in the six months to June 2026 on 377,507,910 at the start of the year, about 0.88% a year, through RSU vesting, subsidiary share exchanges and the Employee Equity Program. Held over twenty quarters that is roughly 4.5% of dilution the model does not carry: terminal EPS of $0.380 on a flat count becomes $0.364, and the $4.45 base fair value becomes $4.26. The DILUTED alternative is immaterial here, unlike GS where it flipped the headline sign: weighted-average diluted shares of 378,889,915 are 0.07% BELOW the period-end basic count and would move fair value from $4.4507 to $4.4539. The 27.1m warrants and options at a $60.6 weighted-average exercise price and 20.4m Class C options at $42.0 are all deeply out of the money at $14.20 and are correctly excluded. CAPITAL IS NOT THE BINDING CONSTRAINT AND NO BUFFER IS BEING SPENT DOWN - the opposite of Goldman. CET1 ROSE from 15.1% to 17.4% over twelve months against a 12.5% overall capital requirement, while total risk exposure FELL from SEK 92,654m to SEK 84,787m over six, helped by receivable offloading and, after the period, a $518m significant risk transfer completed on 16 July 2026. Surplus CET1 is about SEK 4,155m, roughly $428m. The leverage ratio at 9.6% against a 3.0% requirement binds far less, and the liquidity coverage ratio is 970.7%. So the BAC and GS payout question does not arise at all: the honest statement is not 'earnings are distributable' but 'there are barely any earnings yet, and every dollar is retained'. These ratios are disclosed in SEK on what the EU KM1 template indicates is the Swedish regulated bank's consolidated situation, while the group reports in USD; they are context, not model inputs. NO CASH-FLOW SIGNAL EXISTS FOR THIS COMPANY. Klarna's own release warns that operating cash flow 'primarily represent[s] the net flows of money coming in and going out from the bank's consumer lending and retail deposit activities' and 'is not a reflection of the bank's net operating result for the period'. Reported free cash flow was -$269m in a quarter the company earned $9m, and -$3,046m in 2025 Q4, because deposits fell $1.3bn over the half. Any R40 score computed from Klarna's reported cash flow is a deposit gauge. The free cash flow on this page is an earnings construct, calibrated to net profit, and does not equal the reported line. THE PRICERUNNER JUDGMENT IS NOT IN THIS MODEL, and it is 39% of the market capitalisation. A first-instance Stockholm Patent and Market Court decision on 1 July 2026 ordered Google to pay Klarna's subsidiary approximately $2.1bn in damages and accrued interest. Both parties appealed on 22 July 2026, the claim is financed under a third-party litigation-funding arrangement, any recovery 'would be reduced by' payments to litigation funders, ATE insurers and PriceRunner's former shareholders and by taxation, Klarna publishes no net-to-Klarna figure, and 'the criteria for recognition had not been met'. There is no honest way to size it with this engine's levers, so it is excluded from netCash and from every scenario and recorded here as an unmodelled contingent asset. CURRENCY IS ON ONE BASIS THROUGHOUT. Klarna reports IFRS in US DOLLARS and its NYSE ordinary shares trade in US dollars, so spec.currency is USD and no exchange rate touches this model - unlike Nokia, which is modelled in euros against a dollar ADS quote. The $14.20 reference price is the 28 August 2026 close carried in data/companies/klar/profile.json; KLAR is not yet in data/prices-close.json, so the render layer has no live quote to substitute and falls back to this price. SEASONALITY. Both verticals carry factors derived by ratio to a CENTRED four-quarter moving average of their own disclosed revenue, normalised to mean 1.0, with the basis quarter deseasonalised before the driver runs - which is why the 'units' inputs are $8,233.6m and $28,491.3m rather than the disclosed $7,887m and $28,762m of GMV. The US line passes the signal-to-spread test decisively (9.37:1 in Q3, 4.20:1 in Q4). The Global ex-US line does NOT pass comfortably: 1.59:1 in Q3 and 1.99:1 in Q4 with two Q4 windows that trend, which is the shape for which Goldman's Global Banking & Markets seasonality was rejected here. It is applied anyway, for the reasons in that vertical's notes, and the cost of the call is visible: leaving it aseasonal leaves the base fair value unchanged at $4.45 and only flattens the modelled Q4 2026 lift from 9.8% to 6.4% sequentially, against a guidance-implied Q4 about 15% above Q3. TWO AND A HALF YEARS OF HISTORY, FIVE YEARS OF PROJECTION. Klarna has filed one 20-F and four quarterly 6-Ks. The ten quarters used here exist only because it republishes pre-IPO periods in a supplementary exhibit, and there is no pre-2024 comparable line-item history in any filed document. Seasonality rests on six centred windows and two of the four factors per vertical rest on a single window each. This is the model's largest structural weakness and no better input fixes it.

Where each case comes from

From cash flow to fair value

The projected path

Every row is projected. A year-over-year change is shown only where the quarter it compares against exists — an em dash means there is no comparable quarter, not a flat year.

Model revisions

Assumptions are marked to reality as each quarter prints. Every change is appended here, with the fair value the model produced at the time, so the model's own history stays visible.