What is the five-pillar scorecard?
Moderate-to-good fit: 2 of 5 pillars pass, 3 partial. Score 68/100, Selective Positive.
| PASS | Sunrise industry |
| PASS | Leadership |
| PARTIAL | Moat, Iron Fortress, Free Cash Flow |
| 68/100 | Overall DalalBytes score |
| Pillar | Grade | One-line evidence |
|---|---|---|
| 1. Sunrise industry | PASS | Digital banking and embedded finance are in a long adoption S-curve; SoFi grows members 35% and deposits faster |
| 2. Leadership | PASS | Dated promises on profitability and the consumer loop were made and kept; the CEO buys the declines with his own money |
| 3. Moat | PARTIAL | Charter plus deposit funding plus 51% cross-buy are real; the funding is bought at promo rates and the charter premium is converging |
| 4. Iron Fortress | PARTIAL | Capital ratios are strong and tangible book compounds; fair-value marks and serial equity raises weaken the fortress claim |
| 5. Free Cash Flow | PARTIAL | GAAP profitable with $1.05B of adjusted EBITDA; growth consumes cash through the loan book, and SBC runs $262M a year |
Pillar 1, Sunrise industry: why does it pass?
The sunrise test asks whether the industry itself is rising. Consumer finance is moving, irreversibly, from branch-and-fee banking to app-first, deposit-led, multi-product platforms, and the regulatory door that mattered, the national charter for a fintech, is now open to a queue of applicants (Upstart conditionally approved July 2026, Affirm applied January 2026, Chime buying Stride Bank). A rising industry with new entrants at the gate is the definition of a sunrise, not a sunset.
SoFi's own numbers say it is riding the front of the wave rather than the back: members +35% year over year to 15.81M, products +42% to 24.38M, deposits from $7.3B (2022) to $45.5B (June 2026), and adjacent sunrise options in bank-issued stablecoins (SoFiUSD settlement live on Mastercard rails since September 22, 2026) and Lightning-network remittance across 30+ countries. The counter-evidence is maturity risk at the edges: personal-loan credit is a cyclical, not secular, business, and the Technology Platform's market is consolidating around independence (Marqeta) and in-house builds (Chime). The core, deposits and members, is sunrise. The pillar passes.
Pillar 2, Leadership: why does it pass?
The leadership test is whether they do what they said, in order, on dates. Anthony Noto's record on the commitments that were his to keep:
- Profitability, promised and delivered on schedule. Quarterly GAAP net income by Q4 2023, committed in January 2023, delivered at +$48M in Q4 2023, and sustained for eleven consecutive quarters.
- The student-loan call. Management guided the moratorium end and the refi recovery timing correctly through 2023, a forecasting hit rate few consumer lenders matched.
- The consumer loop. Financial Services went from -$0.3M of contribution (FY2023) to +$792.9M (FY2025), and cross-buy reached 51% of new products in Q2 2026. The 2026 medium-term EPS guide was raised across its life ($0.55-0.80 to $0.56-0.81 vintage, FY2026 guided at $0.60 with the revenue guide raised in July 2026).
- Skin in the game, verified. Five open-market purchases in 2026 totaling about $2.25M at $15.73 to $18.06, including 28,900 shares on the day the Muddy Waters report published. He holds about 11.96M shares, the largest individual stake.
The debits are real and belong in this section, not a footnote. The Technology Platform was promised as a growth platform (the "AWS of fintech") and has shrunk since Chime left, a guidance failure by omission across 2023-2025. Funding growth with about $3.3B of 2025 equity, including a December raise at $27.50 weeks after the peak, taxed the very holders the leadership courted. Q2 2026 beat and did not raise, with the stock down 9% on the print.
Borrowing conviction is fine for starting a position. But to go big on a ticker, you need to spend a ton of hours on it.
Anthony Noto
Net: a proven scaler (Twitter operations, the NFL's media deals, now a $45B deposit bank built in four years) whose dated promises hold and whose platform promise did not. The pillar passes on the weight of the kept promises, with the platform miss recorded as the standing deduction. CFO continuity (Chris Lapointe since April 2020) supports the grade.
Pillar 3, Moat: why only partial?
Three moat layers are verifiable. The charter: a national bank charter held since February 2022, one of very few among consumer fintechs, with OCC supervision that itself becomes a credibility asset in securitization markets (LPB collateral pricing at 86-91bp spreads in Q2 2026). The funding: deposits at 3.07% average cost against 4.63% warehouse money, worth about $712.6M a year, on a base that grew $5.3B in one quarter. The loop: 51% cross-buy, 1.54 products per member, direct-deposit-linked accounts, and a paid tier (206,000 Plus subscribers) building switching costs that show up in retention economics rather than in marketing claims.
Why partial, not pass. The funding is rented, not owned: SoFi pays more for deposits than 96% of banks, its non-interest-bearing base is 0.3% against Bank of America's 28.7%, and the 4.40% APY promo is the price of the flow. The charter premium is converging as Upstart, Affirm, and Chime each acquire the same regulatory asset. And the one moat that was supposed to compound outside the bank, Galileo's processing network, is demonstrably breachable: its largest client built in-house and left, taking accounts from 167.7M to 134.8M. A moat that holds on the bank side and leaks on the platform side grades partial.
Pillar 4, Iron Fortress: why only partial?
The fortress evidence: holdco CET1 of 18.7% and total risk-based capital of 18.8% at June 30, 2026 (company-estimated), against a 10.5% requirement including buffers; tangible book value per share up 56% year over year to $7.34; total equity of $11.08B; warehouse lines repaid in full in Q4 2025; deposits, the stable funding, at more than 90% of average liabilities.
The cracks: the loan book is carried at fair value and marked above par ($46.6B on $44.3B of principal), so the fortress wall is partly a mark that a rate shock can reprice in one quarter, which is precisely KBW's above-$1B Q3 2026 headwind estimate. Balance-sheet growth is consuming capital fast enough that management raised about $3.3B of equity in 2025 to fund it, dilutive fortress-building. Credit card charge-offs run at 7.65%. And an unresolved short-seller dispute over the charge-off numerator hangs over the capital story: if the all-in loss content were materially higher than disclosed, both earnings and capital would be restated in the same motion. Strong ratios, procyclical accounting, and a funding model that returns to the equity market too often: partial.
Pillar 5, Free Cash Flow: why only partial?
The earnings are real: FY2025 adjusted EBITDA of $1,053.9M, GAAP net income of $481.3M, H1 2026 adjusted EBITDA of $697.7M, and an FY2026 guide of about $1.6B of adjusted EBITDA on $4.75-4.85B of adjusted revenue. Tangible book compounding at this pace is cash generation expressed through a bank's balance sheet.
The deduction is conversion and leakage. A growing bank consumes cash through its loan book: bears cite trailing operating cash burn near $8.5B against the EBITDA figure (single-sourced critical commentary, but directionally the arithmetic of a loan book growing $10B+ a year). Stock-based compensation of $262.1M in FY2025, 7.3% of revenue, is a real cost paid in shares. And the two 2025 equity raises mean the enterprise generated cash while the per-share claim on it was diluted by roughly 7% in a year. A mature SoFi that stopped growing the loan book would gush cash; the SoFi that exists chooses growth and funds part of it with issuance. Partial, trending toward pass if the raise cadence stops.
What would change the grades?
| Pillar | Upgrade trigger | Downgrade trigger |
|---|---|---|
| Sunrise | Stablecoin and remittance volumes become disclosed, material revenue | Charter applicants stall in supervision, signaling the door is closing |
| Leadership | Technology Platform returns to growth under the bundle, or is honestly restructured | Another dated consumer promise missed, or a raise below $15 without a credit event |
| Moat | Non-interest-bearing deposit share rises materially; Plus passes 1M paid members | Deposit cost stays at promo levels while growth slows |
| Iron Fortress | A rate shock quarter absorbed without a capital raise | Any restatement, or CET1 drifting toward the low teens |
| Free Cash Flow | A full year of per-share tangible book growth above 25% with no issuance | SBC re-accelerating above 10% of revenue |
The five-pillar read and the report verdict agree: SoFi is a good company file, 68/100, Selective Positive, held back from a higher grade by a platform that shrank, a funding base that is bought rather than owned, and a credit dispute that only time and clean prints can close.