SoFi Technologies
A bank that lives in a phone app: checking, investing and loans behind one login, with the loans paid for by its own members' deposits.
The scores
Growth4/4 ↑
Members, products and revenue are all accelerating, broad-based across three segments, and growing far faster than the banking system around them (C21, C15).
Business quality3/4 →
The funding-cost edge over Chime and Affirm is real and quantified, but untested through a cycle and dented by a lost platform client.
Financial position3/4 →
A convertible-note maturity sits inside the horizon but is well covered by cash and a 17.0% capital ratio against a 10.5% floor (C28, C62).
Execution record4/4 ↑
Guidance was raised mid-year, 8 of 8 recent quarters beat EPS estimates, and Q3 2026 was a 20th straight Rule-of-40 quarter (C18, C61, C43).
Per share capture2/4 →
Diluted shares rose 14.3% in a year - heavy - and stock has already retired debt once. Growth outruns it today, not the need to.
Potential2/4 →
The deposit pool is enormous but does not compose into a figure; disclosed committed demand is a small, conditional slice of a discretionary case.
Risk profile3/4 →
One disclosed, already-realized concentration (the Galileo client loss) plus heavy bank-holding-company regulatory dependency, but diversified elsewhere across three segments and a broad consumer base.
Valuation2/4 →
The price already implies about 20% sustainable return on net worth, the bottom of management's own target and about twice the ~10% earned today.
Evidence quality3/4 →
Central claims are filed facts or confirmed against two competitors' own filings, but the return target rests on assumptions graded, not verified.
If it works 0.7× · if it fails you keep about 67%.
The Verdict
We rate SoFi neutral. It is a bank that lives in a phone app, funding its loans with members' deposits, and the business is solid: members grew by about a third in a year. The price is full: it only makes sense if SoFi earns about 21% a year on its net worth for good, while today it earns roughly 10%. The view hinges on that return climbing toward 15%, or the price falling. As a bet it is priced for success: if its new crypto, business-banking and small-business lines work the shares are worth less than today, and if they fail shareholders keep about two-thirds.
The long caseMembers who move their main account to SoFi keep adding products, so each new one, from crypto to business accounts, sells to people already there.
The evidence under this - 8 claims
- 2026-10-28editedThe Q2 2026 member and product counts are closed; the Q3 2026 counts are C61.
- Q2 2026 GAAP net income of $156.6 million annualized against $9.5 billion of tangible book value implies a current run-rate return on tangible common equity of roughly 6.6% - well under a third of the low end of the stated target.
- Lending segment contribution margin, the segment funding most of net interest income, fell from 60% to 55% year over year even as revenue grew (C06).
- Technology Platform - SoFi's most capital-light segment, though its 32% contribution margin in FY2025 was the lowest of the three (C36) - had total accounts down 16% year over year at the same quarter (C08), meaning the path to materially higher blended returns still depends on a segment currently shrinking, not growing.
- 2026-09-26editedCounter-evidence corrected: it called Technology Platform SoFi's highest-margin segment; the FY2025 10-K segment note puts its contribution margin at 32%, the lowest of the three (C36). The claim itself is unchanged.
- 2026-09-26editedRe-struck with the research pack refreshed on 2026-09-26: $18.06 and 2.46 times at the 2026-08-28 close to $16.58 and 2.26 times at the 2026-09-25 close; tangible book per share unchanged at $7.34 until the Q3 print.
- 2026-10-28editedRe-struck at the 2026-10-27 close: $16.40 and 2.19 times on the Q3 tangible book of $7.48 per share (was $16.58 and 2.26 times at the 2026-09-25 close on $7.34).
- 2026-09-26editedCorrected: the old 'about 94% of FY2025 total net revenue' divided Lending plus Financial Services ($3,391.0M) by the consolidated $3,613.4M, which nets a $227.8M Corporate/Other loss, while the thesis's Technology Platform 'about 12%' is a share of the $3,841.2M segment total, so the two added to more than 100%. On the segment total, from the 10-K segment note, they are 88% and 12%. Added the three segments' FY2025 contribution margins from the same table, because the thesis called Technology Platform the highest-margin segment and it is the lowest.
The same arithmetic run at a 10% rather than 12% cost of equity, or 7% rather than 5% growth, moves the implied sustainable ROTCE by several points either way - the conclusion is sensitive to inputs this thesis chose rather than inputs the market disclosed (A01, A02 are both flagged sensitivity: high/medium).
- 2026-09-26editedRe-struck from C25 at the 2026-09-25 close: 0.12 + (2.26 - 1) x 0.07 = 20.8%, roughly 21%, was roughly 22% at 2.46 times. Still inside management's 20-30% target, now at its bottom.
- 2026-10-28editedRe-struck with C25: at 2.19 times tangible book the implied sustainable return is about 20% (was about 21% at 2.26 times).
Annualizing one quarter overstates precision, and the 2026 guide of $840.0 million of net income against $9.7 billion of tangible book is a similar roughly 9%.
- 2026-10-27editedRe-struck on Q3 2026: about 10%, up from about 7% in Q2; still short of the 15% route P2 needs.
The company
Most people bank at one firm, invest through a broker and borrow from a third lender. SoFi puts all three behind one login: checking and savings, an investing app, and personal, student and home loans. It also rents out Galileo, the software other banks and fintechs use to issue cards.
The bundle is working: in the June quarter, 51% of new products were opened by existing members, up from 35% a year earlier.
The evidence under this - 2 claims
- 2026-10-28editedThe Q2 2026 cross-buy share is closed; the Q3 2026 share is C61.
How it makes money
Mostly as a bank does. It lends to members and keeps the gap between what borrowers pay and what depositors are paid, which in the June quarter came to 5.98% a year of everything it lends and invests.
Fees make up the rest. The accounts, investing and card business grew 88% to $1.54B in 2025.
The evidence under this - 2 claims
- 2026-10-27editedRe-struck on the Q3 2026 release; the Q2 figures were $712.6 million and 5.98%.
- 2026-09-26editedCorrected: the old 'about 94% of FY2025 total net revenue' divided Lending plus Financial Services ($3,391.0M) by the consolidated $3,613.4M, which nets a $227.8M Corporate/Other loss, while the thesis's Technology Platform 'about 12%' is a share of the $3,841.2M segment total, so the two added to more than 100%. On the segment total, from the 10-K segment note, they are 88% and 12%. Added the three segments' FY2025 contribution margins from the same table, because the thesis called Technology Platform the highest-margin segment and it is the lowest.
Where the money comes from, stream by stream
| Stream | Share of sales (Q2 2026) | How it earns it |
|---|---|---|
| Lending | 56.8% | Interest on personal, student and home loans held on the balance sheet, plus fees for originating loans sold to partners through the Loan Platform Business. |
| Financial Services | 36.6% | Interest earned on members' deposits, card interchange, investing and other fees, and referral fees from the accounts, card and investing products. |
| Technology Platform | 6.6% | Account and processing fees paid by banks and fintechs that run their cards and core banking on Galileo and Technisys. |
Who runs it
Anthony Noto has run SoFi since 2018 and bought the bank licence that lowered its funding costs; Christopher Lapointe has run finance since 2020. Insiders own a small slice, about one share in forty, and almost half of Noto's stake is options. His bonus rides on revenue, an adjusted profit measure, return on equity and new members; his performance shares pay on growth in net worth, not counting money raised by selling shares.
The evidence under this - 3 claims
Nearly half of Noto's 1.9% is options rather than shares he holds, so his direct stake is about 1%; his five open-market purchases in 2026 (C32) add to it only at the margin.
- Seventy per cent of the bonus rides on adjusted measures of revenue and EBITDA, and the 2025 return-on-equity target of 6.8% sits far below the 20%-30% management says it will reach (C23).
- Until 2026 target pay was earned at 80% of goal, which the company changed only after shareholder feedback.
Why it wins, and who could beat it
Why customers come back
In 2022 SoFi became a national bank by buying one. The charter lets it fund loans with members' insured deposits instead of borrowing from other lenders. In the June quarter those deposits cost 1.56 percentage points less, which management puts at about $712.6M a year of interest saved.
Rivals confirm the edge in their own filings: Chime says it "is a technology company, not a bank", and Affirm says a charter of its own would bring "lower funding costs".
The evidence under this - 4 claims
- 2026-10-27editedRe-struck on the Q3 2026 release; the Q2 figures were $712.6 million and 5.98%.
The threat
The weak spot is the software business. Its accounts fell 23% in 2025, to 128.5 million, after, in SoFi's words, "a significant Galileo client recently moved to a competitor". Marqeta names Galileo as a rival, but neither company says who won the client.
The evidence under this - 2 claims
Who else does this, rival by rival
| Ticker | Rival | Where it competes | How far along | Priced against it |
|---|---|---|---|---|
| CHYM | Chime Financial, Inc.Priced against it: One segment only | SoFi's three reportable segments: Lending, Financial Services and Technology Platform | On sale | One segment only |
| MQ | Marqeta, Inc.Priced against it: One segment only | SoFi's three reportable segments: Lending, Financial Services and Technology Platform | On sale | One segment only |
| COIN | Coinbase and the crypto exchangesPriced against it: One segment only | The 2026 launches: SoFi Crypto, SoFiUSD, Big Business Banking and small-business loans | On sale | One segment only |
How big it could get
How big this could get
SoFi's bank competes for every dollar deposited at an insured American bank: $20.7 trillion in June 2026. SoFi held $45.5B of it, and its deposits grew 13.2% that quarter while the industry's grew 0.8%.
Nobody publishes how much of that pool an app-only bank can win, so this is a size, not a target.
The evidence under this - 2 claims
- 2026-10-28editedThe 30 June 2026 deposit figure is a closed fact and stays the basis of the industry comparison in C17; the September figure is C61.
Who has signed
Buyers for its new small-business loans. Basepoint Capital, a private lender, agreed to take $3B over three years; a second, unnamed partner is smaller and gave no term. Both depend on conditions rather than guaranteeing a purchase.
The evidence under this - 2 claims
The partnership is disclosed only on the earnings call, in no filing or press release, and Basepoint Capital is private and confirms nothing independently; 'binding: conditional' reflects that forward-flow arrangements in this market are typically volume targets rather than take-or-pay commitments.
The counterparty is not even named and no term was disclosed, which is itself evidence this is an early-stage, non-binding arrangement rather than a firm contract.
The results
The numbers
Sales grew 41% over the twelve months to Jun 2026, faster than the 35% in the year to Dec 2025.
It keeps 15% of sales as profit, up from −14% in the year to Dec 2023.
Yearly sales growth has ranged from +26% to +74% over the last 5 years.
Shareholders' money earned 7.1% a year, up from −23% in the year to Dec 2021.
The share count rose 157% in 4.5 years, and pay in shares costs 6.6% of sales.
All the numbers
Filed figures from the company's 10-K and 10-Q, read 2026-10-04. Fiscal years end in December.
| Dec 2021 | Dec 2022 | Dec 2023 | Dec 2024 | Dec 2025 | 12 mo to Jun 2026 | |
|---|---|---|---|---|---|---|
| Size and growth | ||||||
| Sales | $985M | $1.57B | $2.12B | $2.67B | $3.61B | $4.31B |
| Sales growth | +74% | +60% | +35% | +26% | +35% | +41% |
| Profit | ||||||
| Net margin | −49% | −20% | −14% | 19% | 13% | 15% |
| What growth costs, per sales dollar | ||||||
| Research and development | 28% | 26% | 24% | 21% | 18% | 17% |
| Sales and marketing | 43% | 39% | 34% | 30% | 30% | 31% |
| General and admin | 51% | 32% | 24% | 22% | 19% | 19% |
| Pay in shares | 24% | 19% | 13% | 9.2% | 7.3% | 6.6% |
| Capital spending | 5.3% | 5.9% | 5.2% | 5.8% | 6.7% | 6.7% |
| Cash | ||||||
| Cash from operations | −$1.35B | −$7.26B | −$7.23B | −$1.12B | −$3.74B | −$8.50B |
| Free cash flow | −$1.40B | −$7.35B | −$7.34B | −$1.27B | −$3.98B | −$8.79B |
| Balance sheet | ||||||
| Cash and short-term investments | $495M | $1.42B | $3.09B | $2.54B | $4.93B | $3.13B |
| Debt | $3.95B | $5.49B | - | - | - | - |
| Net cash | −$3.45B | −$4.06B | - | - | - | - |
| Returns and shares | ||||||
| Shares (diluted) | 527M | 901M | 945M | 1.10B | 1.25B | 1.35B |
What growth costs
Growth is making SoFi more profitable, but not everywhere. Technology, admin and stock pay cost less for each new dollar of revenue. Marketing to win new members costs about the same share of revenue as two years ago, and management has chosen to spend extra revenue on more growth.
The evidence under this - 2 claims
Sales and marketing, the largest cost line, has not fallen per revenue dollar since 2024, and management chose in 2026 to put revenue above guidance back into growth spending rather than profit (C31).
An analyst asked this exact question on the call; management could have banked the beat and raised EBITDA guidance and chose not to - a discretionary, reversible capital-allocation choice, not a cost forced on the business.
- 2026-10-28editedWhat the CFO said on the Q2 call is closed; Q3 then raised EBITDA and EPS guidance (C18), so the reinvestment explanation was followed by a raise.
The last report (29 Jul)
Quarter ended June 2026, filed figures.
The September quarter of 2026 was a record: revenue up 44% to $1.4 billion, profit of $241 million, and 17.1 million members. SoFi raised its 2026 revenue forecast to about $5.04 billion and said 2027 should bring about $6.4 billion.
On the call it said 2027 needs no new capital. Its return on shareholders' money rose to about 10%, and even its 2027 profit guide is only about 13% of net worth, short of the 15% we need.
What it did to the thesisIt showed growth and funding holding, which supports the business. But even the 2027 profit guide stays under the return on net worth our view waits for, so the view holds.
The quarter in detail
Sales growth from the previous quarter, filed figures.
Earnings per share $0.12 against $0.11 expected; $0.12 as filed, with no adjustments.
The evidence under this - 6 claims
Working the guidance through by hand (C20) shows it implies second-half growth decelerating to about 24%-29% from H1's 41% pace - a real slowdown embedded in the guidance itself, not merely a beatable floor.
- 2026-10-27editedRe-struck on the Q3 2026 release: the 2026 range of $4.75B-$4.85B became about $5.04B, and a 2027 outlook was introduced.
Management's own framing on the call was that the guidance raise reflects strong demand and 'a cushion... so we can deliver regardless of the environment', i.e. the implied deceleration may be conservatism rather than an expected slowdown.
- 2026-10-27editedQ3 2026 printed at 44% growth, above the 24%-29% the earlier guidance implied for the second half; re-struck on the raised guidance.
Annualizing one quarter overstates precision, and the 2026 guide of $840.0 million of net income against $9.7 billion of tangible book is a similar roughly 9%.
- 2026-10-27editedRe-struck on Q3 2026: about 10%, up from about 7% in Q2; still short of the 15% route P2 needs.
- 2026-10-27editedRe-struck on the Q3 2026 release: 140 million accounts, up 6 million on Q2 and down 11% on the year.
The 2027 net income guide of $1.28 billion is about 13% of today's $9.7 billion of tangible book value even before book grows, so the guide itself sits under the 15% return the thesis's first criterion and route P2 require.
What management did with the money
It issued shares rather than buying them back: the diluted count rose 14.3% in a year. Net worth per share reached $7.48 in September, up from $7.34 in June. The capital cushion is 17.0% against the 10.5% regulators require, and management says it can grow without raising outside money.
Insiders in 2026Q3: 28 purchases and 32 sales, from their Form 4 filings.
The evidence under this - 4 claims
- 2026-10-28editedRe-sourced to the Q3 2026 release: $9.7 billion and $7.48 per share (was $9.5 billion and $7.34 at Q2).
Reaching the above-target ratio (18.8% at Q2 2026, 17.0% at Q3 2026) required raising equity capital, not only retaining earnings, so the pace of decline toward the stated low-to-mid-teens target depends on deploying capital into growth SoFi does not fully control - loan demand and Loan Platform Business buyer appetite - rather than on time alone.
- 2026-10-27editedRe-sourced on the Q3 2026 call: management said no capital raise is assumed or needed for 2027.
- 2026-10-28editedCounter-evidence worded against both quarters now that Q3 reads 17.0%.
The price, and where we disagree
In short. Full. Fair only for a bank already earning about 20% a year on its net worth. SoFi earns about 10%, and the gap is a promise not yet kept.
The next answer comes with the December-quarter results, expected late January: the forecast implies sales about flat on the September quarter, so a clear beat says the guide is caution and a miss says the slowdown is real.
The price eased to about 2 times net worth by 26 October, below the 2.14 times line route P2 names, then rose with results to 2.19 times at the 27 October close, above it. P2 also needs the return on net worth at 15% and it is about 10%.
The evidence under this - 4 claims
Working the guidance through by hand (C20) shows it implies second-half growth decelerating to about 24%-29% from H1's 41% pace - a real slowdown embedded in the guidance itself, not merely a beatable floor.
- 2026-10-27editedRe-struck on the Q3 2026 release: the 2026 range of $4.75B-$4.85B became about $5.04B, and a 2027 outlook was introduced.
Management's own framing on the call was that the guidance raise reflects strong demand and 'a cushion... so we can deliver regardless of the environment', i.e. the implied deceleration may be conservatism rather than an expected slowdown.
- 2026-10-27editedQ3 2026 printed at 44% growth, above the 24%-29% the earlier guidance implied for the second half; re-struck on the raised guidance.
Annualizing one quarter overstates precision, and the 2026 guide of $840.0 million of net income against $9.7 billion of tangible book is a similar roughly 9%.
- 2026-10-27editedRe-struck on Q3 2026: about 10%, up from about 7% in Q2; still short of the 15% route P2 needs.
- 2026-09-26editedRe-struck with the research pack refreshed on 2026-09-26: $18.06 and 2.46 times at the 2026-08-28 close to $16.58 and 2.26 times at the 2026-09-25 close; tangible book per share unchanged at $7.34 until the Q3 print.
- 2026-10-28editedRe-struck at the 2026-10-27 close: $16.40 and 2.19 times on the Q3 tangible book of $7.48 per share (was $16.58 and 2.26 times at the 2026-09-25 close on $7.34).
How we read the multiple
A bank is priced against its net worth: what it owns minus what it owes, not counting what it overpaid for firms it bought. Buyers pay about $2.19 for each $1 of SoFi's. It earns about 10% a year on it.
A standard formula shows what that price assumes:
- Assume buyers want 12% a year and net worth grows 5% a year.
- Then $2.19 per $1 needs about 20% a year, for good.
- That is the bottom of management's promised 20% to 30%: possible, but already in the price.
Against its own history. Dearer than its own normal, and partly deservedly: revenue and net worth per share are both growing faster than they were.
The evidence under this - 4 claims
Annualizing one quarter overstates precision, and the 2026 guide of $840.0 million of net income against $9.7 billion of tangible book is a similar roughly 9%.
- 2026-10-27editedRe-struck on Q3 2026: about 10%, up from about 7% in Q2; still short of the 15% route P2 needs.
- 2026-09-26editedRe-struck with the research pack refreshed on 2026-09-26: $18.06 and 2.46 times at the 2026-08-28 close to $16.58 and 2.26 times at the 2026-09-25 close; tangible book per share unchanged at $7.34 until the Q3 print.
- 2026-10-28editedRe-struck at the 2026-10-27 close: $16.40 and 2.19 times on the Q3 tangible book of $7.48 per share (was $16.58 and 2.26 times at the 2026-09-25 close on $7.34).
The same arithmetic run at a 10% rather than 12% cost of equity, or 7% rather than 5% growth, moves the implied sustainable ROTCE by several points either way - the conclusion is sensitive to inputs this thesis chose rather than inputs the market disclosed (A01, A02 are both flagged sensitivity: high/medium).
- 2026-09-26editedRe-struck from C25 at the 2026-09-25 close: 0.12 + (2.26 - 1) x 0.07 = 20.8%, roughly 21%, was roughly 22% at 2.46 times. Still inside management's 20-30% target, now at its bottom.
- 2026-10-28editedRe-struck with C25: at 2.19 times tangible book the implied sustainable return is about 20% (was about 21% at 2.26 times).
- Q2 2026 GAAP net income of $156.6 million annualized against $9.5 billion of tangible book value implies a current run-rate return on tangible common equity of roughly 6.6% - well under a third of the low end of the stated target.
- Lending segment contribution margin, the segment funding most of net interest income, fell from 60% to 55% year over year even as revenue grew (C06).
- Technology Platform - SoFi's most capital-light segment, though its 32% contribution margin in FY2025 was the lowest of the three (C36) - had total accounts down 16% year over year at the same quarter (C08), meaning the path to materially higher blended returns still depends on a segment currently shrinking, not growing.
- 2026-09-26editedCounter-evidence corrected: it called Technology Platform SoFi's highest-margin segment; the FY2025 10-K segment note puts its contribution margin at 32%, the lowest of the three (C36). The claim itself is unchanged.
Priced against its rivals
None of the listed rivals above sells the same thing to the same buyers as a large part of what it does. A price comparison needs at least two, so the price is judged against its own history instead.
Also checked and set aside: Affirm Holdings (a substitute, not the same product).
The bet: what it pays if it works
The bet is on SoFi's newest products: crypto trading, its own dollar coin, accounts for larger businesses and loans to small firms. None discloses any revenue yet, so the bank is counted separately, at the price this thesis would call fair.
Only the small-business loans have a size so far. Even if they fill, the shares are worth less than today, because the price already pays far more for the new products. If they never earn, shareholders keep about two-thirds.
| The 2026 launches: SoFi Crypto, SoFiUSD, Big Business Banking and small-business loans revenue in two to three years | Billions of dollars a year of loans sold needed | At the stated multiple |
|---|---|---|
| $50.00M (the big win) | about 1 at 100% kept | 0.7× |
| $150.00M | about 3 at 100% kept | 0.7× |
To pay 2× (a Fair bet) it would need about $6.37B of The 2026 launches: SoFi Crypto, SoFiUSD, Big Business Banking and small-business loans revenue (about 138 billions of dollars a year of loans sold); to pay 5× (a Cheap bet), about $20.77B of The 2026 launches: SoFi Crypto, SoFiUSD, Big Business Banking and small-business loans revenue (about 452 billions of dollars a year of loans sold), at the same multiple.
Each row is The 2026 launches: SoFi Crypto, SoFiUSD, Big Business Banking and small-business loans alone, plus the rest of the company (SoFi's three reportable segments: Lending, Financial Services and Technology Platform) counted at about $13.59B.
The steps it has to take
- Done SoFi Crypto and the SoFiUSD stablecoin launched
- Done Big Business Banking processing payments for its first clients
- Done Small-business loans launched, with a $3B buyer agreement signed
- In progress Basepoint's agreement filled at about $1B of loans a year due 2027-07
- Not started Revenue from the launches disclosed due 2027-03
The detail behind the label
The evidence under this - 6 claims
The chief executive said in April that crypto had not yet been marketed aggressively, and the Technology Platform segment the launches are built on lost its largest client and shrank in 2025 (C08), so the engineering they draw on is already stretched.
The partnership is disclosed only on the earnings call, in no filing or press release, and Basepoint Capital is private and confirms nothing independently; 'binding: conditional' reflects that forward-flow arrangements in this market are typically volume targets rather than take-or-pay commitments.
The counterparty is not even named and no term was disclosed, which is itself evidence this is an early-stage, non-binding arrangement rather than a firm contract.
- 2026-10-28editedRe-sourced to the Q3 2026 release: $9.7 billion and $7.48 per share (was $9.5 billion and $7.34 at Q2).
The same arithmetic run at a 10% rather than 12% cost of equity, or 7% rather than 5% growth, moves the implied sustainable ROTCE by several points either way - the conclusion is sensitive to inputs this thesis chose rather than inputs the market disclosed (A01, A02 are both flagged sensitivity: high/medium).
- 2026-09-26editedRe-struck from C25 at the 2026-09-25 close: 0.12 + (2.26 - 1) x 0.07 = 20.8%, roughly 21%, was roughly 22% at 2.46 times. Still inside management's 20-30% target, now at its bottom.
- 2026-10-28editedRe-struck with C25: at 2.19 times tangible book the implied sustainable return is about 20% (was about 21% at 2.26 times).
The figure is fee revenue on loans sold rather than held, so it depends on buyers' appetite for SoFi's personal loans; management says it chose not to fill all the demand, which is a choice that can reverse with funding markets.
Where we disagree with the market
Unconvinced: analysts split 10 buy, 13 hold and 4 sell, and the shares fell after a quarter that beat forecasts and raised them.
The price is not doubting management's return target. It has already paid for it in full.
The doubt is real: 15.0% of the tradable shares were sold short in mid-September. But run the standard bank formula backwards and the price pays for the return management promised, not a discount to it.
We are wrong if second-half growth holds near the first half's 41% and the software business grows its accounts again.
What analysts say, and the evidence
The evidence under this - 8 claims
- 2026-09-26editedRe-read against the research pack refreshed on 2026-09-26: the split is unchanged at 10 buy, 13 hold, 4 sell.
- 2026-10-01editedIts valid_through of 2026-09-30 passed; re-read against the research pack refreshed on 2026-10-01: the split is unchanged at 10 buy, 13 hold, 4 sell.
- 2026-10-19editedIts valid_through of 2026-10-08 passed; re-read against the vendor's consensus endpoint on 2026-10-19: the split is unchanged at 10 buy, 13 hold, 4 sell.
- 2026-10-28editedIts valid_through passed; re-read on 2026-10-28, the day after Q3 results: unchanged at 10 buy, 13 hold, 4 sell.
- 2026-09-24editedRe-sourced to the FINRA 2026-08-31 settlement after the claim's refresh_by (the next settlement) fired: 172,515,246 shares, 13.7% of float and 3.41 days to cover at 2026-08-14 became 177,480,044 shares, 14.1% and 4.08 days at 2026-08-31, up 2.9%. Same float denominator, still the Q2 2026 10-Q cover page. The reading - a sceptical minority that rose after the Q2 selloff and has not left - is unchanged.
- 2026-09-26editedRe-sourced to the FINRA 2026-09-15 settlement, published after the 2026-09-24 review: 177,480,044 shares, 14.1% of float and 4.08 days at 2026-08-31 became 188,982,888, 15.0% and 5.23 days, up 6.5%. Same float denominator.
The same arithmetic run at a 10% rather than 12% cost of equity, or 7% rather than 5% growth, moves the implied sustainable ROTCE by several points either way - the conclusion is sensitive to inputs this thesis chose rather than inputs the market disclosed (A01, A02 are both flagged sensitivity: high/medium).
- 2026-09-26editedRe-struck from C25 at the 2026-09-25 close: 0.12 + (2.26 - 1) x 0.07 = 20.8%, roughly 21%, was roughly 22% at 2.46 times. Still inside management's 20-30% target, now at its bottom.
- 2026-10-28editedRe-struck with C25: at 2.19 times tangible book the implied sustainable return is about 20% (was about 21% at 2.26 times).
- Q2 2026 GAAP net income of $156.6 million annualized against $9.5 billion of tangible book value implies a current run-rate return on tangible common equity of roughly 6.6% - well under a third of the low end of the stated target.
- Lending segment contribution margin, the segment funding most of net interest income, fell from 60% to 55% year over year even as revenue grew (C06).
- Technology Platform - SoFi's most capital-light segment, though its 32% contribution margin in FY2025 was the lowest of the three (C36) - had total accounts down 16% year over year at the same quarter (C08), meaning the path to materially higher blended returns still depends on a segment currently shrinking, not growing.
- 2026-09-26editedCounter-evidence corrected: it called Technology Platform SoFi's highest-margin segment; the FY2025 10-K segment note puts its contribution margin at 32%, the lowest of the three (C36). The claim itself is unchanged.
Management's own framing on the call was that the guidance raise reflects strong demand and 'a cushion... so we can deliver regardless of the environment', i.e. the implied deceleration may be conservatism rather than an expected slowdown.
- 2026-10-27editedQ3 2026 printed at 44% growth, above the 24%-29% the earlier guidance implied for the second half; re-struck on the raised guidance.
- 2026-10-27editedRe-struck on the Q3 2026 release: 140 million accounts, up 6 million on Q2 and down 11% on the year.
What could go wrong, and how we would know
The risks
- A shrinking software business. Galileo's accounts were still 16% below a year earlier in June, at 135 million, after a large client left.
- Loans earning less. Lending kept 55 cents of each revenue dollar in 2025, down from 60 cents, even as its revenue grew 24%.
- A downturn it has not seen. Every loan is exposed to American consumers, and SoFi Bank, chartered in 2022, has never been through a full credit cycle.
- Rivals getting charters. Affirm has applied for a bank charter of its own. If approved, the funding-cost gap would start to close.
The evidence under this - 4 claims
- 2026-10-27editedRe-struck on the Q3 2026 release: 140 million accounts, up 6 million on Q2 and down 11% on the year.
What we still do not know
- Where the departed Galileo client went. Neither SoFi's filing nor Marqeta's says.
- How much of the short selling is a bet against SoFi, and how much hedges its bonds that convert into shares. The count cannot tell them apart.
The evidence under this - 4 claims
- 2026-09-24editedRe-sourced to the FINRA 2026-08-31 settlement after the claim's refresh_by (the next settlement) fired: 172,515,246 shares, 13.7% of float and 3.41 days to cover at 2026-08-14 became 177,480,044 shares, 14.1% and 4.08 days at 2026-08-31, up 2.9%. Same float denominator, still the Q2 2026 10-Q cover page. The reading - a sceptical minority that rose after the Q2 selloff and has not left - is unchanged.
- 2026-09-26editedRe-sourced to the FINRA 2026-09-15 settlement, published after the 2026-09-24 review: 177,480,044 shares, 14.1% of float and 4.08 days at 2026-08-31 became 188,982,888, 15.0% and 5.23 days, up 6.5%. Same float denominator.
What would change our view
- Neutral→AttractiveThe multiple implies a return that is actually in reach
It needs:
- ○The price falls to about 1.43 times net worth, where it pays for a 15% return: half as much again as the 10% earned today.about $10.71 today, 32% below the last close
- Neutral→AttractiveThe return shows up, at a price short of the full target
It needs all of:
- ○SoFi's return on its net worth reaches 15% a year, up from roughly 10% in the September quarter.checked at the next results, due by 27 Jan
- ○The price is at or below about 2.14 times net worth, where it pays for 20%, the bottom of management's range.about $16.03 today, -2% below the last close
- Reopens the researchThe platform leg stops shrinking
It needs:
- ○The software business grows its accounts from a year earlier again, and someone names where the departed Galileo client went.checked at the next results, due by 27 Jan
We would also be wrong if
- The lending business keeps less than 55 cents of each revenue dollar for two more quarters.
- Losses on personal, student or home loans rise well above today's.
- SoFi sells new shares to raise money, outside its own funding plan.
What is coming
- 15 October 2026 2026 convertible notes mature: Whether $428.0M is repaid in cash as elected, with no new shares issued while the stock sits below the conversion price.
Every dated event ahead
- 15 October 2026 2026 convertible notes mature: Whether $428.0M is repaid in cash as elected, with no new shares issued while the stock sits below the conversion price.
- 27 October 2026 Q3 2026 results: Adjusted revenue growth against the guided slowdown, Technology Platform accounts and revenue, and whether run-rate return on tangible equity keeps rising from about 10%.
- 31 December 2026 CEO's 2024-2026 performance-share period ends: Whether tangible book grew enough without capital raises to vest the CEO's 2024 performance shares, a filed read on organic capital build.
- January 2027 Q4 2026 results and 2027 guidance: Whether the at-least-30% revenue growth path to 2028 is kept, and whether sales and marketing per revenue dollar finally falls.
Since we wrote this
- 28 OctPrice movePrice re-struck at 2.19 times net worth after the resultsThe 13% rise on results put price to net worth back above route P2's 2.14 line, on net worth of $7.48 per share. The return is about 10%, so the view holds.
- 27 OctEarningsSales grew 44% and the return rose to about 10%; view holdsGrowth beat the 24-29% second-half pace and guidance rose, but return on tangible equity is about 10% against the 15% route P2 needs.
- 27 OctEarningsCall: no capital raise needed for 2027, return still under 15%The call added segment growth for 2027 and a no-raise statement, but even the 2027 profit guide sits under the 15% return route P2 needs.
- 19 OctPrice movePrice reached the route's tangible-book line; the return leg did notPrice to tangible book fell to about 1.98 times, below route P2's 2.14. P2 also needs a 15% return on tangible equity, about 7% in Q2, so the view holds until the 27 October report.
- 11 OctThesis reviewSoFi's 2026 launches declared an early business, with a betThe owner's rule of 2026-10-11: every thesis names its lines, and an early line owes a bet. Management is spending its revenue beat on four launches that disclose nothing.
- 5 OctThesis reviewThe long case now sits under the verdictThe owner asked to pilot the long case on five theses without a bet before deciding on the rest.
- 4 OctThesis reviewOwner's checklist: revenue streams, cost of growth, management, catalystsThe owner's checklist found the method silent on who runs the company, what growth costs, where the money comes from and what is dated ahead; this pass fills those four blocks from the proxy and filings.
- 1 OctPrice movePrice reached P2's level; its return half is not met. View heldThe share price fell to about 2.14 times tangible book, but the route also needs a reported 15% return on tangible equity against about 7% now, and no new quarter has been filed.
- 26 SepThesis reviewSegment shares corrected; price re-measured at 2.26 times net worthLending and everyday banking are 88% of segment revenue and the software business 12%, once measured on one base; the old 94% was not. The price now implies about a 21% return. View held.
- 24 SepPrice movePrice routes re-measured at 2.25 times net worth. View heldThe price slipped from 2.46 to 2.25 times net worth and short selling was re-read at 14.1%. Nothing the view turns on is answered before the October 27 results.
- 30 AugThesis writtenNeutral, medium convictionA well-run, fast-growing bank whose price already pays for management's own return target. Two questions stay open: the forecast second-half slowdown, and the shrinking software business.
- 30 AugThesis reviewInsider trades re-read: seven purchases, no real sales. View heldA data fix showed the chief executive bought shares five times with his own money. Supportive, but it answers neither question the view turns on.
- 30 AugThesis reviewShort-selling data added. View heldReal money is betting against the shares, which fits a sceptical market. Some of it may hedge SoFi's convertible bonds, and it settles neither open question.
Also checked 23 times with nothing new to act on.
Sources 33
Sources - 33
Every claim above cites one of these, or states the reasoning that derived it. The quote is frozen at the date it was read: a link can rot, and an agent fetch returns a reading of a page rather than the page. Backs counts the whole claim ledger, so it can name a claim this page does not surface.
- S01primarycompany filing2026-02-17Segment descriptions, competition risk factors, MD&A revenue/margin tables by segment, member and product KPIs, income-tax valuation-allowance note, debt/convertible-notes note, net-interest-income table, Technology Platform account decline and the Galileo client-loss risk factor.backsC01C02C03C04C05C06C07C13C14C28C36C39C50C53C54
- S02primarycompany filing2026-07-29Q2 2026 GAAP and adjusted results, segment results (Lending, Financial Services, Technology Platform), balance sheet, deposit growth, tangible book value, share counts, FY2026 guidance, Technology Platform account figures for the quarter.backsC08C12C16C18bC19C21C22C26C27C30C37C40C42
- S03primarytranscript2026-07-29SoFi Technologies Q2 2026 earnings call transcript (2026-07-29)Management commentary on Loan Platform Business partnerships (Basepoint Capital and an unnamed SMB partner), the ROTCE target, capital ratio strategy, competitive positioning, and the CFO's explanation for not raising EBITDA guidance despite the revenue beat. The Q&A is a distinct origin from the prepared remarks even though both share this filing's independence group.
I believe we will achieve the returns that Chris walked through before of 20% to 30% return on tangible common equity, and it's just a matter of when, not if the market can connect the dots.
backsC23C29aC29bC31C34C35C38C56C57 - S04primarycompany filing2026-08-27Confirmation, from a competitor's own filing, that Affirm relies on third-party originating banks and is separately applying for its own bank charter to reduce that reliance and lower funding costs.backsC11
- S05primarycompany filing2026-03-06Confirmation that Chime holds no bank charter of its own, and its Active Member count.backsC10
- S06primarycompany filing2026-02-24Marqeta's own naming of Galileo as a direct card-issuing/core-banking competitor.backsC09
- S07primarygovernment data2026-08-30Total deposits across all FDIC-insured U.S. institutions as of June 30, 2026, and the count/change in the number of insured institutions in the quarter.
All Insured Institutions ... Total deposits (in billions) ............................ 20,722.3 ... Second Quarter 2026, All FDIC-Insured Institutions.
backsC15 - S08secondaryvendor data2026-09-26SOFI research pack (market tier) - price, volatility, analyst split, insider filings, earnings historyLast close price and market cap, 20-day volatility, analyst consensus split, Form 4 insider activity, congressional disclosures, 8/8 EPS-beat history, and the measured price reaction to the Q2 2026 print and other recent news.backsC30C32C41C43
- S10primarycompany filing2026-01-30FY2026 guidance as first issued in January (before the Q2 2026 raise), and the 2025-2028 medium-term growth algorithm: at least 30% compounded annual adjusted-net-revenue growth and 38%-42% compounded annual adjusted-EPS growth.backsC46
- S11primarycompany filing2026-06-17aff10b5One = 0 - this and the CEO's four other 2026 purchases carry no Rule 10b5-1 plan footnote, confirming discretionary open-market buying.backsC32
- S12primarycompany filing2026-08-24Footnote: 'completed pursuant to a Rule 10b5-1 Trading Plan adopted by the Reporting Person on July 30, 2025' - this and Keough's other eleven 2025-2026 sales in the window carry the same plan footnote.backsC32
- S13primarycompany filing2026-06-19Footnote: 'completed pursuant to a Rule 10b5-1 Trading Plan adopted by the Reporting Person on June 2, 2025' - this and Rishel's other three sales in the window carry the same plan footnote.backsC32
- S14primarycompany filing2025-11-24Footnote: 'completed pursuant to a Rule 10b5-1 Trading Plan adopted by the Reporting Person on June 11, 2025' - Pinto's only sale in the window.backsC32
- S15primarycompany filing2026-02-09Footnote: 'Sale of shares to cover the cost of exercise, withholding tax obligations, and pay broker fees and commissions, arising from the stock option exercise reported above' - the vendor's 'S-Sale' code covers a mandatory sell-to-cover, not a discretionary sale.backsC32
- S16primarycompany filing2026-06-22Footnote: 'Shares sold to satisfy tax withholding obligation applicable to the vesting of stock-settled RSUs' - the vendor's 'S-Sale' code covers mandatory tax withholding, not a discretionary sale.backsC32
- S17primarycompany filing2026-02-09No Rule 10b5-1 footnote; only footnote concerns trust ownership - a discretionary open-market purchase.backsC32
- S18primarycompany filing2026-02-09The Rule 10b5-1 checkbox is unchecked and no plan footnote appears - a discretionary open-market purchase.backsC32
- S19primarygovernment data2026-08-30SOFI's short interest at each FINRA settlement date from 2026-04-30 through 2026-08-14.
currentShortPositionQuantity: 172515246, previousShortPositionQuantity: 190758849, changePercent: -9.56, daysToCoverQuantity: 3.41, settlementDate: 2026-08-14
backsC48 - S20secondaryvendor data2026-08-30Float shares (1,259,508,295) and free-float % (98.19%), sourced by the vendor to the Q2 2026 10-Q cover page - used as the denominator for short interest as a percentage of float.backsC48
- S21primarygovernment data2026-09-24SOFI's short interest at the 2026-08-31 settlement, the latest FINRA had published when fetched on 2026-09-24; the 2026-09-15 settlement was not yet out.
currentShortPositionQuantity: 177480044, previousShortPositionQuantity: 172515246, changePercent: 2.88, daysToCoverQuantity: 4.08, settlementDate: 2026-08-31
backsC48 - S22primarygovernment data2026-09-26SOFI's short interest at the 2026-09-15 settlement, the latest FINRA had published when fetched on 2026-09-26.
currentShortPositionQuantity: 188982888, previousShortPositionQuantity: 177480044, changePercent: 6.48, daysToCoverQuantity: 5.23, settlementDate: 2026-09-15
backsC48 - S23primarycompany filing2026-09-23Footnote: 'completed pursuant to a Rule 10b5-1 Trading Plan adopted by the Reporting Person on July 30, 2025' - the 10,203-share sale of 2026-09-21.backsC32
- S24primarycompany filing2026-09-23Footnote: 'completed pursuant to a Rule 10b5-1 Trading Plan adopted by the Reporting Person on June 9, 2026' - the 18,624-share sale of 2026-09-21.backsC32
- S25primarycompany filing2026-08-06Segment note for Q2 and H1 2026 (net revenue and contribution profit by Lending, Technology Platform, Financial Services, Corporate/Other), the debt note on the 2026 convertible notes' maturity, settlement election and earlier exchanges for shares, and the H1 2026 income statement.backsC49C53C54C55
- S26primarycompany filing2026-04-30Beneficial ownership as of April 20, 2026, director and executive officer biographies and tenure, the CD&A: 2025 bonus metrics, goals and funding, PSU design and goals, the 2021 stock-price PSUs, the 2026 design changes, and the performance-measure definitions.backsC50C51C52
- S27primarytranscript2026-04-29SoFi Technologies Q1 2026 earnings call transcript (2026-04-29)management's account of SoFi Crypto, SoFiUSD and Big Business Banking: launched, not yet marketed aggressively, and expected to start generating revenue in 2026
We made the decision last year to build out crypto buy sell and hold, made the decision to launch SoFiUSD. Those products are launched and they will start to generate revenue in 2026. That will be incremental
backsC57 - S28primarycompany filing2026-02-12Coinbase Global, Inc. Form 10-K, fiscal year ended 31 December 2025 Coinbase Global, Inc.the largest US crypto exchange's own account of its competition, naming traditional financial services and fintech companies offering crypto among its main competitors
Further, more traditional financial and non-financial services businesses may choose to offer crypto-based services in the future as the industry gains adoption and barriers to entry lower.
backsC58 - S29secondaryvendor data2026-10-19FMP analyst grades consensus for SOFI, retrieved 2026-10-19The current count of buy, hold and sell ratings: 10 buy, 13 hold, 4 sell, no strong ratings.backsC33
- S30primarycompany filing2026-10-27Q3 2026 revenue, net income, members, segment margins, charge-offs, tangible book, Technology Platform accounts, raised 2026 guidance and the 2027 outlook.backsC08C12C18C20C24C25C43C45C59C60C61C62
- S31primarytranscript2026-10-27SoFi Q3 2026 earnings call transcript (FMP), held 2026-10-27data/cache/transcripts/SOFI-2026Q3.jsonThe 2027 segment growth rates, the no-capital-raise statement, the rate assumptions, SoFi Plus shortfall and the revenue-per-product outlook.
our 2027 outlook is going to be funded organically through two levers.
backsC29bC60 - S32secondaryvendor data2026-10-28FMP end-of-day prices, SOFI, close of 2026-10-27, retrieved 2026-10-28The 2026-10-27 close of $16.40 (the session of the Q3 results), against $14.46 the session before.backsC25
- S33primarycompany filing2026-10-22Footnote: 'completed pursuant to a Rule 10b5-1 Trading Plan adopted by the Reporting Person on July 30, 2025' - the 10,203-share sale of 2026-10-20 at a weighted $14.7226.backsC32
- S34secondaryvendor data2026-10-28FMP analyst grades consensus for SOFI, retrieved 2026-10-28The current count of buy, hold and sell ratings: 10 buy, 13 hold, 4 sell, no strong ratings.backsC33