SoFi Technologies Maintains Strong Credit Quality: Can It Persist?

SoFi's Q2 2026 loan originations rose 69% year over year to $14.8 billion, while reported personal-loan net charge-offs fell to 2.62% from 3.03% sequentially and 90-day delinquencies declined to 40 bps. Its core personal-loan portfolio generated a 6.1% risk-adjusted margin, and lending-adjusted net revenue increased 59% to $712 million. Credit performance is improving, but the rapid expansion of unsecured personal lending and management's 7-8% maximum cumulative-loss assumption remain key risks; SOFI also trades at 22.45x forward earnings versus 15.63x for the industry.

Analysis

The relevant signal is not the reported loss rate but the gap between it and the economics of delinquent-loan sales. SOFI is expanding unsecured credit fastest while retaining limited seasoning evidence on a meaningful portion of recent vintages; a benign labor market can make this look like underwriting alpha until 12-24 months after origination. At a premium earnings multiple, even a modest reserve-build or loss normalization would disproportionately pressure both EPS estimates and the valuation multiple.

SOFI's funding-cost advantage is the key offset: deposit-funded lending can preserve spread economics better than marketplace-originators when wholesale funding tightens. That creates a relative advantage versus unsecured-credit peers reliant on securitization or bank partners, but it also makes SOFI more rate-sensitive than the market appears to price: lower policy rates aid borrowers and funding costs, while a sharp risk-off deposit beta increase compresses the spread. ALLY is a cleaner defensive consumer-credit expression because its auto exposure and scale diversify the unsecured-credit tail, although used-car residual values remain its separate risk.

Consensus is likely extrapolating unusually low early delinquencies rather than underwriting the stated cumulative-loss ceiling. The differentiated catalyst is not another quarter of low charge-offs; it is vintage roll-rate disclosure, reserve coverage, and whether incremental originations maintain yield without relaxing borrower mix. A sustained deterioration in unemployment or a 50-100 bp increase in personal-loan charge-offs would likely matter more to SOFI's equity than continued revenue outperformance over the next 1-3 quarters.

This is not a standalone directional catalyst today: the evidence supports monitoring credit normalization rather than chasing a premium-valued lender on an early-cycle credit print. Over 6-18 months, successful seasoning below the modeled loss ceiling would justify durable ROE expansion and multiple support; failure would expose the stock as a high-growth consumer-credit beta rather than a technology-platform compounder.

Key Decisions for Investors

• Maintain a neutral-to-underweight SOFI position into the next two earnings reports; do not add on headline charge-off improvement unless management provides cohort-level cumulative losses, roll rates, and reserve coverage. Falsify the cautious view if seasoned vintages remain below the modeled loss ceiling while originations sustain without yield compression.
• For a relative credit-quality trade over 3-6 months, favor long ALLY / short SOFI in equal dollar risk, not equal dollar notional. The thesis is SOFI multiple compression if unsecured-loss normalization emerges; exit if SOFI's risk-adjusted lending margin holds above 6% while ALLY's auto losses or used-car values deteriorate materially.
• Set an alert for the next SOFI quarterly filing: a 50 bp sequential rise in underlying personal-loan charge-offs, rising 30-89 day delinquencies, or reserve growth lagging receivable growth warrants initiating a SOFI short or put spread; absent those data, options carry is not justified.
• Use macro as the timing filter: a meaningful upward surprise in unemployment or consumer-delinquency data is a near-term catalyst for the SOFI/ALLY relative trade. Conversely, easing funding conditions with stable labor data would favor covering the SOFI short leg quickly.