SoFi's 3.6x Book Value Has a $36 Billion Blind Spot

SoFi Technologies reported $481 million in GAAP net income for FY2025 — eight consecutive quarters of profitability, $3.6 billion in total net revenue growing 35%, and a stock trading at 3.62x book value. The headline numbers say this is the strongest year in company history.

Every number in that sentence requires context. Members grew 35% to 13.6 million, each using an average of 1.49 products — up from roughly 1.1 two years ago, with product adoption growing faster than membership. Traditional banks earn $500+ per customer by cross-selling 5-7 products. SoFi earns $265 per member with 1.49. The cross-buy runway is the platform thesis in a single number.

But the FY2025 10-K, filed February 17, 2026, tells a more complicated story. SoFi trades at 3.62x book value — 2.5 times LendingClub's 1.46x — yet its FY2025 return on tangible common equity is approximately 7.7%, below LC's 10.2%. Lower capital returns at a dramatically higher multiple. The premium isn't for what SoFi is today. It's for what SoFi is becoming: a platform that earns fees on other people's capital rather than a lender that risks its own. The 10-K reveals why that transformation matters — and how far it has to go.

The Fee Engine That Changes Everything

The most important number in SoFi's 10-K isn't in the income statement headlines. It's buried in the segment footnotes: the Loan Platform Business generated $576 million in fee revenue in FY2025, up 234% from the prior year. Two years ago, this business barely existed.

The volume is compounding: $1.6 billion to $3.7 billion per quarter across four quarters. Fee rate compression from Q1 (6.0%) to Q3 (4.9%) reversed in Q4 (5.2%), suggesting pricing stabilization at scale. At Q4's run rate: $776 million annualized.

Here is why this changes the investment case. The Loan Platform Business is structurally the opposite of on-balance-sheet lending in every dimension that creates investor concern about SoFi:

Fee income — the platform revenue that doesn't depend on SoFi risking its own capital — grew to 38% of total revenue, up from approximately 30% in FY2024. Fee growth (+59%) is outpacing interest income growth (+42%). The transformation has a scoreboard, and it's moving.

The Financial Services segment — which houses the Loan Platform Business — went from $307 million in profit in FY2024 to $793 million in FY2025, nearly tripling in a single year. This is the lender-to-platform transformation measured in dollars.

But LPB revenue depends on a functioning secondary market for unsecured personal loans — the same market that historically contracts during credit stress, precisely when SoFi would need the fee income most. If fee-based revenue scales to match net interest income, SoFi's earnings become structurally less sensitive to model assumptions, less dependent on balance sheet growth, and more cash-generative. If the secondary market freezes, SoFi loses its fastest-growing and highest-quality revenue stream at exactly the wrong time.

The LPB's importance to SoFi's future also changes the goodwill impairment narrative. Every analyst covering SoFi has flagged the $1.39 billion in goodwill as an impairment risk tied to the Technology Platform (Galileo and Technisys, acquired for ~$2.3 billion). Enabled accounts are declining — 160.2 million to 157.9 million. Management warned of "potential impairment charges in future periods."

Except the 10-K's XBRL data tells a completely different story:

96% of the goodwill — $1,339 million — is allocated to the Loan Platform Business, not the Technology Platform. Technology Platform goodwill is $18 million — a complete write-off would be immaterial. The $1,339 million at the Loan Platform Business is supported by $576 million in annual fee revenue growing 234%. The impairment risk doesn't depend on whether Galileo's enabled accounts grow. It depends on whether the Loan Platform Business remains viable. At $576 million in annual fees and accelerating, that test looks very different from the declining-accounts narrative.

The Profit Behind the Profit

SoFi called FY2025 its "strongest financial performance in history." Net income was $481 million. But FY2024 net income was $499 million. The strongest year ever earned less than the year before it.

The explanation: in FY2024, SoFi released a $287 million deferred tax asset valuation allowance — a one-time tax benefit from recognizing accumulated losses that reduce future taxes. That single event inflated FY2024 net income by $287 million. Strip it out and FY2025 pretax income roughly doubled year-over-year. The operating improvement is real. The headline number hides it.

But the FY2025 10-K reveals a deeper issue with the $481 million figure. The cash flow statement shows $351 million in fair value changes on held-for-investment loans — unrealized gains that flowed through net income but generated zero cash. These aren't fraudulent or improper. They're standard accounting for loans carried at fair value. But they mean 73% of FY2025 net income is non-cash.

Strip out the fair value marks: $481M - $351M = approximately $130 million in "cash-adjusted" net income for FY2025. SoFi's GAAP net income was $481 million, but only $130 million was backed by cash generated during the year. The truth lies between $130M and $481M — the fair value gains will likely be realized as loans are sold or repaid, but the timing and amount depend on management's models being correct.

Why does this matter? Because as of December 31, 2025, 96% of SoFi's $38 billion loan portfolio is carried at fair value — and 93% of those fair value assets are Level 3, meaning they're valued using management's own models rather than market prices. Traditional banks typically hold single-digit percentages of assets at Level 3; even among specialty lenders, concentrations above 50% are rare. LendingClub will approach SoFi's profile in 2026 when it adopts fair value for held-for-investment loans.

As of year-end 2025, SoFi carries $2.15 billion in cumulative fair value premiums above what borrowers actually owe — $1,298 million on personal loans, $783 million on student loans, $72 million on home loans. This markup is the source of the $351 million in non-cash income. And it's entirely dependent on the models being right.

The charge-off sensitivity makes the concentration risk concrete. For every half-percentage-point increase in personal loan defaults, SoFi loses $108 million — 22% of its annual net income. The FY2025 personal loan charge-off rate is 2.80%. If defaults doubled to 5.6%, the incremental $606 million in credit losses above current levels would exceed FY2025 pretax income of $465 million. Current charge-offs are already embedded in the fair value marks — the stress scenario is the additional losses above what the models assume. The current 90-day delinquency rate of 0.08% suggests the book is clean today. But the sensitivity is not hypothetical.

The Scaling Machine

SoFi's FY2025 operating cash flow was negative $3.74 billion — which sounds alarming next to $481 million in net income. The entire gap comes from one line item: SoFi originated $5.3 billion more in loans than it sold during a year of 57% origination growth. Exclude that timing difference and core operations generated positive $655 million in cash flow. The negative OCF is a function of growth speed, not operational losses — LendingClub shows the same pattern.

Where does the cash come from to fund that growth? SoFi's customers fund SoFi's loans. As of December 31, 2025, the $37.5 billion deposit base — growing $11.2 billion in a single year — covers virtually the entire $38 billion loan book. The implied deposit cost of 2.88% (lower than the headline 4.6% APY because of rate cuts during 2025 and non-savings accounts) against loan yields of roughly 8.9% creates a 5.7% net interest margin spread. The funding model works. The question is whether credit losses and operating expenses consume the spread.

The efficiency ratio answers that question — and it's the single most important operational metric for SoFi's transformation. The ratio has compressed from 95% to 84% in eighteen months: SoFi is spending less per dollar of revenue each quarter. Each 5-percentage-point improvement adds $180 million to pretax income at current revenue scale. From 85% to 65% — bank-industry norms — is $720 million in annual pretax income, more than the entire Loan Platform Business generates today.

This is SoFi's largest single earnings lever, and it's a function of revenue growing faster (35%) than expenses (20%). JPMorgan runs at ~55%. Wells Fargo at ~60%. Even Ally Financial operates around 70%. At current rates, SoFi's efficiency ratio reaches bank norms around 2028-2029. Each year of 35% revenue growth with 20% expense growth compresses the ratio by roughly 5-7 points — and each point at SoFi's scale is worth ~$36 million in pretax income.

Revenue per member captures the remaining runway. SoFi earns $265 per member per year — roughly half what a traditional bank earns per retail customer. As cross-buy grows from 1.49 toward 3.0+ products per member, revenue per member grows without acquiring a single new customer. The $265 figure is the starting point, not the ceiling.

The Bank Charter: What You Can't See on the Earnings Call

The fee engine and the scaling margin are the bull case. But the bank charter that makes all of this possible — the deposits, the NIM, the regulatory moat — creates constraints the earnings release doesn't mention.

Net interest margin peaked at 6.01% in Q1 2025 and has compressed 29 basis points to 5.72% by Q4 — below LendingClub's 6.07% — as SoFi grew its lower-yield student and home loan books. Management guided to "a healthy NIM above 5% for the foreseeable future" — a floor, not a ceiling. The 6% era appears over.

As of Q4 2025, SoFi Bank holds CET1 capital of 22.8% — over three times the 6.5% well-capitalized minimum. This massive overcapitalization provides an enormous buffer for loan growth, but it suppresses return on equity. If the bank operated at 12% CET1 — still well above minimums — the freed capital of roughly $4 billion would approximately double bank-level ROE.

Instead, that capital sits inside a bank that has zero distributable funds available for dividend payments. The restriction reflects the bank's accumulated deficit — SoFi Technologies still carries -$998 million in consolidated retained earnings — combined with OCC capital adequacy requirements for a de novo bank in rapid growth. The bank makes the money. The parent can't access it.

This is why SoFi raised $3.2 billion in equity during FY2025 despite reporting profitability. The bank's profits are trapped inside by regulation. The parent — the entity whose shareholders provided the capital — cannot access them. External capital fills the gap. The December 2025 offering (54.5 million shares at $27.50, $1.5 billion gross) was the latest instance. Until the accumulated deficit clears (roughly two more years at the current improvement rate), expect continued equity raises.

The $428 million in convertible notes maturing October 15, 2026, adds a near-term decision point. At a conversion price of $22.41 and capped calls at $32.02, cash settlement is the most likely outcome and easily serviceable from SoFi's cash position. The resolution method signals management's view of intrinsic value: cash settlement says the stock is undervalued relative to $22.41; share conversion says dilution is preferable to spending cash.

What the Valuation Premium Requires

The 10-K data assembles into a valuation question that can be stated precisely: SoFi at 3.62x book value versus LendingClub at 1.46x. FY2025 ROTCE: ~7.7% versus 10.2%. Revenue growth: 35% versus 27%. SoFi earns lower capital returns at a higher multiple — a premium that requires growth to justify. What does it need?

The 2.5x P/B premium over LC prices in specific assumptions. Based on the 10-K data, five conditions need to hold:

1. The Loan Platform Business must continue scaling. At the Q4 run rate, annualized fee revenue reaches $776 million. If fee growth sustains its trajectory, LPB revenue crosses $1 billion within 15 months — enough to push fee income above 30% of total revenue and structurally change the earnings mix. But LPB revenue depends on a functioning secondary market for unsecured personal loans — the same market that historically contracts during credit stress. If fee rates compress below 3.8%, even continued volume growth produces less revenue than today.

2. The efficiency ratio must continue compressing. From 95% to 84% in eighteen months is real progress. Each 5-percentage-point improvement adds ~$180 million to pretax income. The premium assumes this operating leverage continues for 2-3 more years toward 65-70%.

3. Level 3 fair value marks must hold. A 1% adverse shift in fair value assumptions across $36.6 billion in Level 3 assets creates a $366 million equity reduction — roughly 4% of total equity. At 3.62x book, the P/B multiple amplifies the impact: a 4% book value decline produces a 4% market cap decline. LC's 1.46x P/B provides much smaller amplification of the same risk.

4. The convertible notes must be resolved cleanly. $428 million due October 2026, most likely settled in cash. Manageable but it reduces the cash position.

5. Deposit growth must continue funding the balance sheet. The $37.5 billion deposit base covers the loan book today. If deposit growth slows, SoFi needs more equity raises — further dilution at whatever the stock price happens to be.

The cross-filing picture crystallizes the choice. LC achieves higher ROTCE (10.2% vs ~7.7%) with zero debt, 3.4% SBC-to-revenue, and a straightforward single-segment model. The market charges 1.46x book. SoFi carries $36.6 billion in Level 3 assets, $1.82 billion in debt, 7.3% SBC-to-revenue, and an ~85% efficiency ratio — but also has the Loan Platform Business growing 234%, a $37.5 billion deposit base, 13.6 million members cross-buying at an accelerating rate, and an efficiency trajectory that implies $720 million in potential pretax income gains. The market charges 3.62x book.

On the data disclosed in the FY2025 10-K, that premium requires more from the Loan Platform Business — and more patience on the efficiency ratio — than most investors buying at 3.6x book have likely modeled. The 10-K data shows exactly what needs to happen for it to be right — and exactly how much it costs if it's wrong.

Frequently Asked Questions

Why does SoFi have negative free cash flow despite reporting $481M in profit?

SoFi's FY2025 operating cash flow was negative $3.74 billion, driven by one line item: "changes in loans held for sale, net" at negative $5.27 billion. This is the cash consumed by originating personal and home loans faster than they are sold. Excluding this timing difference, core operating cash flow was positive $655 million. The negative OCF is structural to the originate-and-sell model during growth — LendingClub shows the same pattern — not a sign of operational losses.

What does it mean that 93% of SoFi's fair value assets are Level 3?

Level 3 assets are valued using management's models rather than market prices. As of December 31, 2025, SoFi holds $36.6 billion in Level 3 assets — primarily personal loans ($21.5 billion) and student loans ($13.7 billion). The carrying values depend on assumptions about discount rates, default rates, and prepayment speeds that aren't directly verifiable. The $2.15 billion premium above unpaid principal balance (6.3%) represents management's estimate of future cash flows above par. A 1% shift in assumptions creates a $366 million swing in equity.

Can SoFi Bank pay dividends to the parent company?

No. The 10-Q disclosed that "the Bank would not have any funds free of restrictions that are available for dividend payments." Profits generated inside SoFi Bank — where the deposits, loans, and NIM reside — cannot be upstreamed to SoFi Technologies. The parent relies on the Technology Platform, its own cash, and external capital raises. This restriction exists despite the bank holding CET1 capital of 22.8%, which is 3.5 times the well-capitalized minimum.

What happens with the $428M convertible notes in October 2026?

The notes mature October 15, 2026, with a conversion price of $22.41 and 19.1 million shares issuable. SoFi purchased capped call options with a $32.02 cap. If the stock is below $22.41, the notes will likely be repaid in cash from SoFi's substantial cash position. If between $22.41 and $32.02, the capped call offsets dilution. Above $32.02, net dilution of approximately 1.5% would occur. The resolution method signals management's view of the stock's intrinsic value.

How does SoFi's stock-based compensation really compare to LendingClub?

Using total net revenue as the denominator (NII + noninterest income), SoFi's SBC is 7.3% of revenue ($262M on $3.6B total net revenue) — not the ~45% figure that uses only narrow fee revenue. LendingClub's is 3.4%. The ~45% number uses the "revenues" XBRL field that captures only fee income, excluding the $2.2 billion in net interest income that is the majority of SoFi's revenue. At 7.3%, SoFi's SBC is elevated but not alarming for a high-growth fintech. The ratio is declining as revenue scales at 35% while SBC grows only 4.4%.

What are Galileo's challenges and how much goodwill is really at risk?

The Technology Platform generated $361 million in third-party revenue (+14%) but enabled accounts declined 1.4%. However, only $18 million in goodwill is allocated to the Technology Platform — the $1,339 million (96%) sits at the Loan Platform Business. Even a complete Technology Platform write-off would be immaterial. The real goodwill test is whether the Loan Platform Business remains viable, and at $576 million in annual fees growing 234%, that test looks very different from the narrative about declining Galileo accounts.

Is SoFi going to be added to the S&P 500?

SoFi meets the published eligibility criteria: adequate market cap, eight consecutive quarters of GAAP profitability, trading liquidity, and US domicile. The $1.5 billion secondary offering in December 2025 increased the public float and institutional ownership base. However, the Index Committee has full discretion. Prediction markets have estimated approximately 47% odds for near-term inclusion.

What did SoFi change about its revenue reporting?

In Q2 2025, SoFi combined "noninterest income — loan origination, sales and securitizations" and "noninterest income — servicing" into a single line without restating prior periods. Investors can no longer independently track origination margins versus servicing economics — fundamentally different revenue streams with different cyclicality and rate sensitivity. The change occurred during rapid Loan Platform Business scaling, making the reduced comparability more consequential.

Methodology

Data Sources

This analysis uses MetricDuck's quantitative metrics (ROE, ROTCE, earnings quality, cash flow ratios, bank efficiency ratio) calculated from SoFi's XBRL-tagged SEC filings. For methodology on return on invested capital calculations, see our ROIC complete investor guide. Filing intelligence extracted via MetricDuck's 5-pass comprehensive analysis of the FY2025 10-K (filed February 17, 2026) and Q3 2025 10-Q (filed November 6, 2025). Loan portfolio composition, fair value hierarchy, cash flow reconciliation, deposit composition, goodwill allocation, and regulatory capital ratios extracted directly from the FY2025 10-K XBRL data. LendingClub (LC) data from MetricDuck's 5-pass analysis of the FY2025 10-K (filed February 12, 2026). Ally Financial (ALLY) data from Q3 2025 10-Q (filed October 30, 2025).

Limitations

• Valuation multiples (P/B, P/E) reflect MetricDuck's pipeline data as of Q3 2025 and may not reflect current market prices. Post-Q4 book value per share is higher due to the December 2025 equity offering ($1.5 billion) and Q4 earnings.
• The "cash-adjusted net income" of approximately $130 million is an illustrative calculation, not a GAAP or standard non-GAAP metric. Fair value gains will likely be partially or fully realized as loans are sold, repaid, or mature — the $130M figure represents a conservative floor, not a precise estimate of economic earnings.
• The implied deposit cost of 2.88% uses a simple average of Q3 and Q4 deposit balances as the denominator. Actual weighted average cost may differ based on intra-quarter balance fluctuations and rate changes.
• Charge-off sensitivity uses on-balance-sheet personal loan balances as of December 31, 2025. Actual exposure fluctuates with origination and sale timing.
• ALLY comparison uses Q3 2025 data. ALLY's FY2025 10-K was not filed at time of publication.
• SoFi's adjusted metrics (Adjusted Net Revenue, Adjusted EBITDA) exclude stock-based compensation. GAAP figures used throughout this analysis differ materially from management's adjusted figures.

Disclaimer

This analysis is for informational purposes only and does not constitute investment advice. The author does not hold positions in SOFI, LC, or ALLY. Past performance and current metrics do not guarantee future results. All data is derived from public SEC filings and may contain errors or omissions from the automated extraction process.