What the filing says
SoFi Technologies (SOFI) began in student and personal lending and now holds a bank charter. It is widely held by retail investors.
Note 4 of the 10-Q it filed on August 6 shows unpaid principal on customer loans of $44.30bn at June 30, 2026. The balance sheet carries those loans at that amount plus $2.03bn — the "cumulative fair-value adjustment". Principal plus adjustment equals 104.6% of principal: lend $100, book $104.60.
To gauge the size of $2.03bn, compare it with tangible equity. Total equity of $11.08bn less goodwill of $1.43bn and intangibles of $0.23bn gives tangible equity of $9.42bn. The adjustment is 21.5% of that — a fifth of the company's tangible capital consists of money not yet received.
What the number means
A conventional bank records loans at principal and sets aside an allowance for expected losses. SoFi does the opposite for most of its personal, student and home loans: it books them at what they would fetch if sold today, an approach US GAAP permits under the fair value option.
Because SoFi's loans carry relatively high rates, the company judges that, counting future interest, they would sell above principal — hence a book value above 100%. How far above depends on three assumptions the company sets: how many borrowers default, the discount rate applied to future interest, and how quickly borrowers prepay. Note 12 gives the weighted averages for personal loans at quarter-end as a 4.8% annual default rate, a 5.0% discount rate and a 25.8% prepayment rate.
In short, the $4.60 is an estimate, not cash received, and it is adjusted as borrowers actually perform. On loans already 90 or more days delinquent, the company has written the fair-value adjustment down by $98.6m.
Why it matters
First, scale. $2.03bn is roughly 13 times SoFi's second-quarter net income of $157m. A modest change in the estimate can move a quarter's results materially.
Second, direction. The adjustment was $1.94bn at December 31, $2.04bn at March 31 and $2.03bn at June 30. Over the same period principal grew from $34.25bn to $44.30bn — up 29%. Loans expanded sharply while the adjustment stood still, so the premium per $100 of principal fell from $5.70 at year-end to $5.30 in March and $4.60 in June, two consecutive quarterly declines.
Third, the assumptions moved. The personal-loan default assumption rose from 4.5% at year-end to 4.8%, and the discount rate from 4.5% to 5.0%. Both push fair value down — the conservative direction — and help explain the lower ratio.
None of this appears in the earnings release or in most coverage, which lead with member growth and revenue. The fair-value adjustment lives only in the notes.
The company's explanation and context
SoFi states in the 10-Q that it elects the fair value option because it intends to sell or securitize the loans, making market value the more faithful measure. It sets assumptions from its own loan-performance data and has applied the approach for years.
Some short-seller research has argued that the adjustment is too optimistic; the company maintains that realized delinquency and recovery data support its assumptions. Future default and recovery rates will settle the question. The company's decision to raise both the default and discount-rate assumptions this quarter is a notable change.
What investors should watch
Three items in the next quarterly report (Q3 10-Q, expected early November):
- Loan book / principal — 104.6% in Q2, down two quarters in a row. Does it keep falling or turn back up?
- Adjustment / tangible equity — 21.5% in Q2 (22.2% in Q1). A rising ratio means a larger share of tangible equity rests on expected profit.
- Personal-loan assumptions — default 4.8%, discount rate 5.0%. Further increases would signal a more cautious stance; decreases the reverse.
These metrics accumulate on the ticker page each quarter.