UPST · Forward model · Core personal (FICO under 720)
What has to happen in Core personal (FICO under 720)
Model as of
This page changes Core personal (FICO under 720) inside the complete UPST model. Every other vertical, corporate programme and valuation assumption stays in the selected portfolio case.
Shares this vertical and portfolio case. Slider and horizon edits stay in your browser.
Core personal (FICO under 720)
Upstart's own name for its engine: personal loans to borrowers below a 720 FICO, where an AI model that reads more than a credit score is supposed to have the biggest edge over a bank. $2,496M of originations in the basis quarter, 69% of unsecured volume, re-accelerating 27% sequentially after two flat quarters and a decline. Management calls this the superpower and credits it for the contribution margin recovering to 62%. It is the largest line in this model and the one that has to keep working.
Latest: $594M (2031Q2E)
| Period | Value |
|---|---|
| 2025Q1 | $122M |
| 2025Q2 | $164M |
| 2025Q3 | $177M |
| 2025Q4 | $177M |
| 2026Q1 | $172M |
| 2026Q2 | $224M |
| 2026Q3E | $235M |
| 2026Q4E | $247M |
| 2027Q1E | $259M |
| 2027Q2E | $272M |
| 2027Q3E | $286M |
| 2027Q4E | $300M |
| 2028Q1E | $315M |
| 2028Q2E | $331M |
| 2028Q3E | $347M |
| 2028Q4E | $365M |
| 2029Q1E | $383M |
| 2029Q2E | $402M |
| 2029Q3E | $422M |
| 2029Q4E | $443M |
| 2030Q1E | $465M |
| 2030Q2E | $489M |
| 2030Q3E | $513M |
| 2030Q4E | $539M |
| 2031Q1E | $566M |
| 2031Q2E | $594M |
Assumptions & reasoning
- Originations for this line are published on page 14 of the Q2 2026 deck; the revenue is not. It is the Unsecured Lending segment's disclosed fee revenue apportioned on that origination mix, which assumes Core earns the same fee per dollar as the other two unsecured lines.
- That assumption is the weak point and it cuts against this line. Upstart says super prime carries a LOWER take rate, so a pro-rata split understates Core's share of unsecured revenue and overstates super prime's. The contribution margins carry the difference instead.
- The sequential path is violent: +34.7%, +1.5%, +6.6%, -0.3%, +26.7%. A base case that extrapolates any single quarter of that is wrong, which is why the growth input sits well below the quarter just delivered.