SoFi’s Next Loan: Keep It, Place It, or Pass?
Download case PDF (4 pages)Request teacher’s pack
Decision date: May 6, 2025
On May 6, 2025, SoFi Technologies filed its first-quarter report showing another record quarter for personal-loan originations. The report also showed how much the company’s lending model was changing: a growing share of the loans were being originated for institutional partners rather than held on SoFi’s own balance sheet.
SoFi’s executives had reasons to value both channels. Keeping a loan could generate interest income, use the bank’s growing deposit base, and preserve control over the borrower relationship. Originating or referring a borrower through the Loan Platform Business (LPB) could generate fees while using less of SoFi’s own balance sheet. Yet that route still left SoFi responsible for origination and servicing in many cases, with contractual repurchase exposure and dependence on a limited set of capital markets buyers.
The March and April agreements had already been announced, so the case does not ask whether SoFi should sign them. Instead, students take the role of outside advisers and assess a hypothetical allocation question: how should SoFi deploy the next increment of eligible personal-loan demand across retained lending, LPB activity, and other distribution or production limits? The public record does not establish a pending board or management vote on that allocation.
A bank with more than one way to lend
SoFi had grown from a student-loan refinancing business into a digital financial-services company with three reporting segments: Lending, Financial Services, and Technology Platform. It offered personal, student, and home loans, along with deposits, payments, investing, and other products. Management presented the breadth of products and growing member relationships as mutually reinforcing. Whether those relationships added enough value to justify retaining each marginal loan was difficult to observe from public results.
SoFi’s bank charter and deposits offered a funding source for loans held on balance sheet. At March 31, 2025, deposits totaled $27.3 billion. SoFi reported a company-wide consolidated net interest margin (NIM) of 6.01% for Q1, up from 5.91% a year earlier. NIM is an aggregate company measure: it does not show the incremental return on the next personal loan after its expected credit losses, funding, liquidity, capital, and servicing costs.
The LPB was not one single transaction type. It included loans SoFi originated and then sold to third parties, referrals of prequalified borrowers to separate origination partners, and servicing arrangements. For Q1, SoFi reported $1.6 billion of personal loans originated on behalf of third parties. The company’s filing says that these loans remain part of its lending relationship with borrowers and are included in its Lending segment product totals. SoFi also directly serviced its personal loans and provided servicing for LPB loans it originated for partners.
New institutional capacity
On October 14, 2024, SoFi announced a $2 billion LPB agreement with Fortress. On March 13, 2025, it announced a two-year Blue Owl agreement for up to $5 billion. On April 17, SoFi announced a $2 billion extension to the original Fortress agreement and a separate two-year $1.2 billion arrangement using a Fortress–Edge Focus joint venture. The March and April announcements added up to $8.2 billion in incremental headline capacity; including the October 2024 agreement, the headline total across these arrangements was approximately $10.2 billion. SoFi described the LPB as a way to meet borrower demand with less capital-intensive, more fee-based revenue.
Those public releases did not disclose every agreement’s eligibility rules, pricing, purchase timing, volume floors, termination rights, servicing requirements, or loss allocation. These headline amounts do not establish utilized capacity or total usable purchase capacity; eligibility, overlap, timing, pricing, volume floors, and other terms were not fully disclosed. They should not be treated as cash already received, a guaranteed takeout for every eligible loan, or complete risk transfer.
What Q1 showed
SoFi’s Q1 personal-loan originations reached $5.537 billion, up 69% from Q1 2024. The $1.6 billion originated for third parties was included in—not additional to—that total. Dividing the rounded $1.6 billion by the reported $5.537 billion yields an approximate 29% of personal originations through third-party LPB origination. Referrals are a separate LPB activity.
SoFi reported $92.750 million in Q1 loan-platform fees and $3.346 million in related servicing revenue, for a combined $96.096 million in noninterest income across Financial Services and Lending. The $92.750 million fee total included $19.700 million of “Referrals, loan platform business” and $73.050 million of “Loan platform business, other”; those components should not be added again to the $92.750 million. The company also described $96.1 million in consolidated adjusted net revenue from LPB activity across the two segments. Adjusted net revenue is a company-reported non-GAAP measure; it is not the same as channel-level profit.
The filing’s sale tables add detail. SoFi reported $1.488 billion in aggregate carrying amount and accrued interest for ordinary LPB personal loans originated and subsequently sold. It recognized $57.172 million in loan fees net of repurchase liabilities, $10.926 million in servicing assets, and $68.098 million in total LPB fees recognized. Separately, an LPB-originated securitization transfer included $39.134 million in retained securitization investments. That retained amount is distinct from the ordinary LPB whole-loan sales.
SoFi’s estimated accrued loan-repurchase obligation was $12.8 million at March 31, up from $11.9 million at December 31. The filing described personal, student, and non-GSE home-loan transfers as generally having no repurchase requirements beyond standard origination representations and warranties. A transferred loan could still be subject to repurchase for breaches of loan-transfer or servicing representations and warranties, and certain transfers included limited credit-related guarantees. The company also disclosed that it depended on deposits and third-party funding to originate loans, had historically sold loans to a limited pool of buyers, and could need its own deposits or capital to fund securitization residual interests.
Credit indicators offered another input, but not a complete forecast. SoFi reported a 46-basis-point 90-day delinquency rate on personal loans. Its annualized net charge-off rate was 3.31%, including the effects of asset sales, new originations, and delinquency sales. Management estimated that an annualized all-in net charge-off rate including recoveries would have been approximately 4.8% had late-stage delinquent loans not been sold. These are company-reported portfolio measures; they are not channel-specific lifetime loss estimates.
Alternatives and unresolved questions
SoFi could keep more loans, using its deposit base to seek interest earnings and retain greater control over customer and credit outcomes. That option would consume balance-sheet capacity, expose SoFi to credit and liquidity risk, and depend on deposit pricing and retention.
It could increase LPB originations and referrals, using partner capital to support more borrower demand and earning origination, referral, and servicing revenue. That option would depend on partner economics and performance, and would leave SoFi with servicing and contractual obligations. Concentration and screening incentives would need active management.
It could stage a mixed allocation, changing channel shares according to expected risk-adjusted returns and explicit limits for credit performance, deposit cost, partner concentration, and sale execution. A staged approach could preserve flexibility but might leave growth or fee revenue unrealized. SoFi could also sell loans or securitize them through other channels, or reduce production when risk-adjusted returns or funding conditions fell below its thresholds.
The relevant comparison was not “interest income versus fees” in isolation. Retained-loan economics depended on yield, funding cost, expected credit losses, servicing, liquidity, and capital. LPB economics depended on fee and referral terms, acquisition and servicing expense, repurchase exposure, partner capacity, and any continuing securitization interest. Public reporting did not disclose these costs in matched borrower-level or channel-level form.
LendingClub, an adjacent digital marketplace bank, reported that 34% of Q1 2025 originations were held for investment at amortized cost, up from 17% a year earlier; it said the share depended on risk-adjusted returns, liquidity, and regulatory capital. Upstart, an AI-powered marketplace lender with a different bank-partner structure, reported that 11% of its Q1 marketplace principal originations were held on its own balance sheet. These examples show that hold/distribute mixes can vary; they do not provide target percentages or directly comparable returns for SoFi.
Assignment
Assume you are advising SoFi’s senior leadership on personal-loan allocation after the May 6 filing. Prepare a two-page recommendation memo that:
- Chooses whether to accelerate LPB activity, stage a mixed allocation, retain more loans, or constrain production under specified conditions.
- Sets two or three funding, credit, partner-concentration, or execution guardrails.
- Names at least one measurable trigger that would change your recommendation.
- Uses the reported facts above and labels any derived calculation or teaching assumption.
You are not being asked to infer a pending board vote or to reconstruct confidential partner agreements. State what additional information you would request before committing to a larger channel shift.
Source notes
- SoFi Technologies, Inc., Q1 2025 Form 10-Q, filed May 6, 2025: printed pp. 16–19 (loan sales, LPB transfers, servicing and repurchase); pp. 44–45 (funding, buyer and residual-interest concentrations); pp. 71–72 (NIM and LPB fee/servicing); pp. 78–80 (originations); pp. 90 and 94 (liquidity and guarantee risks). All figures are as reported unless identified as approximate or derived.
- SoFi, Fortress agreement announcement, Oct. 14, 2024.
- SoFi, Blue Owl agreement announcement, Mar. 13, 2025.
- SoFi, Fortress and Edge Focus agreement announcement, Apr. 17, 2025.
- SoFi, Q1 2025 earnings release, Apr. 29, 2025.
- LendingClub, Q1 2025 Form 10-Q, May 1, 2025.
- Upstart, Q1 2025 Form 10-Q, filed May 6, 2025.