Enova International, Inc. ENVA

Key data

The business

Enova is an online lender to the customers banks avoid: non-prime consumers and small businesses. Three brands do the work. CashNetUSA and NetCredit make short-term and installment loans and lines of credit to consumers with thin or damaged credit, and OnDeck lends to small businesses that cannot get a bank term loan. There are no branches. A customer applies on a phone, a machine-learning model prices the risk in seconds, and the money moves the same day. Enova keeps most of these loans on its own balance sheet at fair value, so its reported top line is interest and fees minus a running mark for expected losses.

The moat is the underwriting data: two decades of loan performance across cycles feeding models that price non-prime risk more accurately than a new entrant can, which is why the losses stay inside the rates charged. The thing to hold in your head is that this is a spread business on the riskiest borrowers, so the whole company lives or dies on whether the loss rate stays where the model says it should.

The numbers

Net revenue is gross revenue after the fair-value mark for expected losses, and it is the honest top line here. The last five quarters, oldest to newest (Q4 2025 derived from the audited full year):

Every line moves the right way and the pace is picking up. Q2 net revenue grew 29% against the year before, net income 38%, and GAAP EPS 40%. EPS runs ahead of profit because the share count keeps shrinking. On the adjusted basis analysts track, Q2 came in at $4.31 against $3.99 expected, the latest in a string of quarters ahead of forecast, broken only by a hair-thin first-quarter miss.

Net revenue has compounded about 16% a year since 2021, but earnings tell a two-act story. Profit fell for three straight years into 2023 as loss provisions and funding costs bit, then EPS jumped 35% in 2024 and 55% in 2025 as credit normalized and buybacks took hold. First-half 2026 EPS of $7.46 is up 34% on the same half last year, and the full-year consensus of about $17.34 asks for a heavy second half, which is normal for a lender whose book builds into the back half. The variant view: the market prices these earnings at 13 times forward because it doubts non-prime profit survives a downturn, and the one print that settles it is the fair-value discount holding as the book grows.

Management

David Fisher handed the CEO seat to Steve Cunningham at the end of 2025 and moved to executive chairman, an inside succession from the long-serving CFO rather than an outside gamble. The capital record is clean: the company has bought back stock every year, $214.6M in 2025 after $289.3M in 2024, steadily shrinking the count and lifting per-share earnings above net income growth. Return on equity runs near 26%. The discordant note is selling: insiders sold about $43.5M over the last twelve months with no open-market buys, Fisher himself the largest seller across two sales. Plan status is not disclosed in this run's filings, so read it as the people who know the book best trimming into a doubled stock, not as scheduled diversification you can wave off.

How it fails or surprises you

Credit normalization. The whole case rests on the loss rate. The fair-value discount narrowed to 38.8% of revenue from 42.2% a year ago, which is why net revenue margin widened to 61% and the earlier worry about deteriorating credit did not show up. But that is a cushion that reverses fast. A rise in non-prime unemployment that pushes the discount back toward 42% costs roughly $30M of net revenue a quarter and the multiple with it. Watch the change-in-fair-value line every quarter.

Funding is the hidden dependency. Loan-funding debt reached $5.0B, up 26% in a year, and it must roll through securitization and warehouse markets that can freeze in a panic regardless of how the loans perform. This is the fact the growth story explains least well: a healthy book funded on markets that can shut is still a book that can strand you. If spreads on new facilities widen sharply, earnings compress even with credit clean.

Operating leverage compounds (right tail). If credit holds through 2027, the machine surprises upward. Marketing is up 43% filling the funnel while operating margin held near 42%, so as those cohorts season, net revenue converts to profit and buybacks shrink the count another 5%. Analysts' $21.56 for 2027 is up 24%, and at 13 times forward a clean credit cycle re-rates the whole thing, not just the earnings.

Closing thoughts

The fair-value discount marks settle whether the credit bet works. Hold that discount flat as the book grows, and the earnings are real, 2027 EPS lands north of $21, and a 13-times multiple on a 26%-ROE compounder looks too cheap. Widen it back toward 42%, and net revenue margin, EPS and the multiple all give at once, because a non-prime lender re-rates hardest exactly when its losses turn. The left tail is fatter than the calm chart suggests, since credit and funding can break together in the same downturn, and the right tail is worth more time than the market gives it, because nothing in today's numbers says the cycle has turned. Judgment, not a percentage: the odds favor the compounder, and the damage if wrong is severe.

The bet is still that Enova keeps lending money online to people and small businesses that ordinary banks turn away, and keeps getting paid back. What breaks it is the loss rate, and the one pair of numbers that tells you first is the change-in-fair-value mark against revenue, 38.8% today, and the loss cushion inside it. If that discount climbs two quarters running while the book grows, the read is wrong.