SoFi (NASDAQ: SOFI) published a quarter this morning in which nearly every metric hit a record. But the figure that explains the business model does not appear in the revenue line - and the question that matters most is not in the report at all.
What Was Reported
| Metric | Q2 2026 | Year ago | Change |
|---|---|---|---|
| Net revenue | $1,218.7 million | $854.9 million | +43% |
| Adjusted net revenue | $1,205.6 million | $858.2 million | +40% |
| Net income | $156.6 million | $97.3 million | +61% |
| Diluted EPS | $0.12 | $0.08 | +50% |
| Adjusted EBITDA | $357.8 million | $249.1 million | +44% |
| Adjusted net income | $160.4 million | $97.3 million | +65% |
And the operational figures:
| Metric | Value | Change |
|---|---|---|
| Loan originations | $14.8 billion | record |
| New members in the quarter | 1.1 million | record |
| Total members | 15.8 million | +35% (from 11.7 million) |
| Product additions in the quarter | 2.2 million | record |
| Total products | 24.4 million | +42% |
Note the two bolded rows against each other: 1.1 million new members against 2.2 million new products. That is exactly the two-to-one ratio - and it is the figure we will return to.
The Figure That Explains the Model
SoFi defines itself as an "everything app" for financial services - loans, savings, checking, investing and insurance in one app. And every model like that stands or falls on one question: does a customer who came for one product buy a second one.
Why that matters economically, not just in marketing terms
Acquiring a customer costs money. Advertising, promotions, sign-up bonuses - a cost recorded immediately. And an additional product that the same customer opens costs almost nothing. So the ratio between products and members measures the model's efficiency. When every member brings one product, the company has to keep acquiring more members in order to grow - and every unit of growth costs money. When every member brings two products, acquisition cost is spread over more revenue. That is the difference between growth that is bought and growth that feeds itself - growth where each unit makes the next one cheaper, similar to the effect of compound interest.
The Bank Charter - and What It Is Worth in Dollars
This may be the least-discussed part of the SoFi story, and it is worth $712 million a year.
Most fintech companies are not banks. They lend money, but they have to raise that money - usually through warehouse credit facilities from large banks. That is expensive funding, and it embeds a profit for the lender.
SoFi went a different way: it became a bank itself. And once you have a charter, you can take customer deposits - and lend them out.
And here is the number that quantifies it
According to the report, the rate SoFi paid on deposits in the quarter was 156 basis points lower than the rate it paid on warehouse facilities. In money: about $712.6 million of annualized interest expense savings. For scale, that is more than four times the company's quarterly net income. And deposits now comprise over 90% of average total liabilities - meaning the transition is nearly complete.
That turns the bank charter from a regulatory milestone into a profitability engine. The exact same loan, to the exact same customer, produces more profit - purely because of where the funding came from.
The Breakdown: Two Segments Surging, One Falling
| Segment | Net revenue | Change |
|---|---|---|
| Lending | $724.8 million | +63% |
| Financial Services | $466.3 million | +29% |
| Technology Platform | $84.5 million | -23% |
Lending is the engine: GAAP net revenue of $724.8 million (+63%), adjusted $711.7 million (+59%). Growth was driven by net interest income, up 54%.
Financial Services grew 29%, split almost evenly between $217.2 million of noninterest income (+28%) and $249.1 million of net interest income (+29%) - driven mainly by consumer deposit growth. Within the segment, the Loan Platform Business added $143.3 million to consolidated adjusted net revenue.
And Technology Platform - the weak spot. Revenue fell 23% year-over-year to $84.5 million, though it rose 13% sequentially. The reason: a large client that fully transitioned off the platform before the end of 2025. Accounts in the segment fell 16% year-over-year to 135 million for the same reason, though they rose 2 million versus the prior quarter.
And this is the most sensitive point in the SoFi story. The company presents itself as a financial technology company, not a digital bank. And the technology segment is precisely the part that should prove that definition - revenue from licensing its infrastructure to other institutions, with no credit risk and no rate dependence.
And right now it is the only segment contracting.
The Breakdown Not in the Headline: Where the Products Came From
| Metric | Value | Change |
|---|---|---|
| Financial Services products | 21.3 million | +43% |
| Lending products | 3.1 million | +36% |
| Technology Platform accounts | 135 million | -16% (but +2 million sequentially) |
And two facts from the report sharpen the picture:
- Financial Services products drove 89% of total product growth, and account for 87% of total products
- The growth came from SoFi Money, Relay and Invest - checking, financial management and investing
And the "loop" the company describes
The cross-buy figure is not static - it is accelerating:
| Quarter | Share of new products opened by existing members |
|---|---|
| Q2 2025 | 35% |
| Q1 2026 | 43% |
| Q2 2026 | 51% |
And within that, a figure that explains the mechanism: among existing members who signed up for SoFi Plus - the premium membership relaunched as a paid subscription - 25% added another product after joining.
In other words, the subscription is not only a revenue source - it is a mechanism that generates an additional sale.
The Gap: The Rule of 40 Cleared by Double, and a Falling Stock
And here is the real question.
In software there is a common rule of thumb called the "Rule of 40": growth rate plus operating margin should exceed 40%. A company that clears it is considered high quality.
SoFi this quarter: 40% adjusted revenue growth, and a 30% adjusted EBITDA margin. Total 70 - nearly double the bar.
And yet the stock is down about 46% from its 52-week high, about 40% year to date, and it traded lower after this report as well.
So why?
Four specific criticisms - not just "valuation"
First, earnings quality. Part of SoFi's profit comes from selling loans to third parties - and from the gain booked at the moment of sale. Analysts flagged that private credit fundraising and redemptions could reduce the fees SoFi earns on selling loans, and pointed to declining private-credit appetite for longer-duration consumer loans. Second, the interest spread. Market expectations were that the margin between loan rates and deposit rates would compress - this quarter and beyond. Third, and most concrete: credit quality. Pre-report coverage noted that some 2025 asset-backed securities had hit cumulative net loss triggers. That is a technical indication, but it touches exactly the sensitive question of whether the loans are performing as assumed. And fourth, concentration in personal loans while industry-wide origination decelerates, alongside intensifying competition from neobanks that raises customer acquisition costs.
And what explains it all: the Rule of 40 does not measure credit risk
This is, to me, the central answer to "why not a technology multiple."
The Rule of 40 measures two things: growth and profitability. It does not measure the risk behind them.
A software company growing 40% with a 30% margin charges a subscription and delivers a service. If the customer pays - the transaction is finished.
For a lender, the transaction begins with the payment. The revenue is booked today; whether the loan is repaid is decided over the next two years. And inside that profit are components that are estimates - the fair value of the loan portfolio, the expected gain on selling it - not cash already received.
So the market prices SoFi as a lender rather than as a software company. Not because it does not see the growth - but because it is pricing what can go wrong in it.
And that is exactly why the technology segment matters so much to this story. It is the only part where revenue carries no credit risk - and it is the part that shrank 23%. As long as it is contracting, the case for a technology multiple rests on a mix that does not support it - even with excellent results.
And what the CEO did with his own money
And within that decline, CEO Anthony Noto bought stock on the open market - repeatedly.
According to Form 4 filings with the SEC, in 2026 there were five separate purchases totaling about $2.25 million - roughly 130,000 shares at a weighted average price of about $17.29. Some of them in May, close to the stock's low for the year.
Why a purchase is distinguished from a grant
Senior executives receive shares as part of their pay - grants, options, restricted units. That signals nothing - it is compensation. An open-market purchase is something else: the executive spends their own money, at the market price, like any investor. That is the only kind of insider action considered a meaningful signal. And still - proportion: the CEO holds about 12 million shares, so buying 130,000 adds about one percent to his holding. It is a signal, not a bet. And a signal is not a promise - executives misjudge their own companies exactly like everyone else.
The Bull Thesis
Whoever reads it positively will point to the breadth: every major metric at a record - revenue, adjusted EBITDA, members, products and loan originations. Net income up 61% and revenue up 43% is a pace hard to find in financials.
Beyond that: the model is proving itself. Cross-buy accelerating for three straight quarters, and product additions double member additions, say acquisition cost is being spread over more revenue.
And guidance was raised - both on revenue and on the member target.
The Bear Thesis
Whoever reads it critically will note first the Technology Platform falling 23%. That is the segment meant to be independent of interest rates, and it is contracting.
Second, dependence on lending. The lending segment is about 60% of revenue, sensitive both to rates and to credit quality.
Third, earnings quality - the gain-on-sale and fair-value components discussed above.
And fourth, rates themselves. A model resting on the spread between loans and deposits is exposed to changes outside the company's control.
The debate in one line
The bulls see a record on every operational metric, profit up 61%, cross-buy accelerating three quarters running, a bank charter worth $712 million a year, and raised guidance. The bears see a technology segment contracting 23%, 60% of revenue dependent on lending, earnings containing estimates rather than cash, and credit indicators worth watching. Both sides are reading the same report.
My Angle
A personal opinion of Ilan Abramov - not advice, not a recommendation
The question I keep returning to in this report is not "was the quarter good" - it was. It is why that is not enough.
The company clears the Rule of 40 by nearly double - 40% growth plus a 30% adjusted EBITDA margin. In software, a number like that would earn a high multiple. Here it does not.
And the answer I arrived at is that the Rule of 40 measures growth and profitability - and does not measure risk.
A software company charging a subscription finished its transaction when the customer paid. A lender begins its transaction when the customer borrows. The revenue is booked today, and whether it was real is settled two years out. Same number in the report, different certainty behind it.
Which is why, to me, the technology segment is not "another segment" in this story - it is the whole argument. It is the only part where revenue carries no credit risk, and it is the only part contracting. As long as it is down 23%, the case for a technology multiple rests on a mix that does not support it - however good the results.
And what genuinely impresses me here is the cross-buy. 35% a year ago, 43% last quarter, 51% now. That is not a number that jumps - it is a number that climbs steadily. And a figure rising three quarters in a row describes a mechanism, not luck.
And the bank charter is worth more than all the rest combined: 156 basis points of funding difference, $712 million a year. Same loan, same customer, more profit - purely because the source of the money changed.
On the CEO's purchases: five open-market buys with his own money, some near the low. That is a signal that costs him something, so it is worth something. But it adds about one percent to his holding, and it is not a prophecy. Someone who knows the numbers from the inside saw no reason to run - and that is all it tells you.
And what I will follow are two numbers, not one: cross-buy, which will say whether the growth machine keeps feeding itself; and the technology segment, which will say whether the market ever agrees to price this company the way it defines itself.
Summary
SoFi delivered a record quarter on nearly every metric: revenue of $1.219 billion (+43%), net income of $156.6 million (+61%), a record 1.1 million new members and a record 2.2 million product additions. The bank charter is producing about $712.6 million of annualized interest savings, and cross-buy has accelerated for three straight quarters to 51%.
And the weak spot: the Technology Platform segment fell 23% year-over-year and its accounts 16% - the one part of the business meant to carry no credit risk.
And the real question: the company clears the Rule of 40 by nearly double, and the stock is still down about 46% from its high. The answer is not that the market missed the growth - it is that the Rule of 40 measures growth and margin, and not the risk behind them. Until the technology segment grows again, the case for a technology multiple rests on a mix that does not yet support it.
Sources: SoFi Technologies' official results announcement for the second quarter of 2026 (July 29, 2026), as filed with the U.S. Securities and Exchange Commission on Form 8-K, including GAAP and adjusted revenue, net and diluted income, adjusted EBITDA, member and product metrics, cross-buy figures, the deposit funding advantage, the three-segment breakdown and updated full-year guidance; Form 4 insider filings with the SEC for the CEO's open-market purchases; and analyst commentary ahead of the report as covered in the financial press, including TipRanks and Yahoo Finance. Stock performance figures fluctuate and are accurate as of the time of writing. The chart is shown in real time via TradingView. Nothing herein constitutes a forecast, recommendation or advice - see the full disclaimer at the bottom of the page.