JPMorgan Chase: Priced for the Best Decade to Last Forever

JPMorgan Chase reported a 20% return on tangible common equity for the third quarter of 2026, a fifth consecutive year at or above the 17% it sets itself as a through-the-cycle target, and cut its share repurchases by 84% because its chairman would rather deploy capital at "very good returns" than buy the stock at a 6% earnings yield. At 2.94 times tangible book the price requires that return — a little more, in fact — to continue in perpetuity. This note values the company on the return it has earned over 21 years, the deposit franchise that funds it, and three paths for the rate cycle the Federal Reserve has just restarted, and finds a fair value of $215 a share against $331, with the gap resting on one question: whether the best decade in the company's history is the steady state.

TON618 Capital Research · As of 14 October 2026 · Download PDF · Download model

On 13 October JPMorgan Chase reported net income of $15.3 billion for the third quarter, $5.57 a share, on managed revenue of $50.9 billion, up 8% on the year. Net interest income rose 13% to $27.1 billion; fees were half of revenue; the overhead ratio was 54%; net charge-offs fell to 0.59% of loans, the lowest in nine quarters; the common equity tier 1 ratio was 13.9% against a requirement of 11.5%; tangible book value per share was $112.64. The company earned 20% on tangible common equity, 21% excluding a $658 million loss on securities it chose to sell, and repurchased $1.1 billion of stock against $8.3 billion a year earlier. Its CFO guided net interest income excluding Markets to about $107 billion for 2027, and its chairman said the company has "probably $30 billion or something of excess capital" and would rather deploy it than buy shares. The stock closed the day at $331.43, 2.94 times tangible book and 14 times trailing earnings.

Terms used throughout, defined once. ROTCE: return on tangible common equity — net income to common shareholders over common equity less goodwill and intangibles; the return the company earns on the capital it actually holds. TBV, P/TBV: tangible book value per share, and the price as a multiple of it. NII, NII ex-Markets: net interest income; the company reports it with and without the Markets business, whose interest income is offset in trading revenue. Deposit beta: the share of a change in the federal funds rate that reaches the rate a bank pays on deposits. CET1: common equity tier 1, the core regulatory capital ratio. Cost of equity: the return an investor requires on the stock; the hurdle a return must clear to create value. Excess return: the return above the cost of equity on the capital held; the model values the company as tangible book plus the present value of that excess. Fade: the assumed path by which today's return converges on the long-run one. Holdco / bank-level: JPMorgan Chase & Co., the $5.1 trillion group, against JPMorgan Chase Bank NA, the $4.1 trillion insured bank whose Call Report the FDIC publishes; the two are labelled wherever they differ.

The print. Managed revenue $50.9 billion (+8%): net interest income $27.1 billion (+13%; excluding Markets $24.7 billion, +6%, "driven by higher deposit balances, higher balances in Card Services and higher wholesale loan balances, partially offset by the impact of lower rates"); fees $23.8 billion (+3%; excluding Markets, +7%, or +12% before the securities loss); Markets $10.3 billion (+15%, equities +49%, fixed income −4%). Expense $27.3 billion (+12% for a third quarter: front-office hiring, revenue-related compensation, marketing +30%, technology +17%). Credit costs $2.7 billion: charge-offs $2.2 billion, down $389 million on the year, and a $539 million reserve build, $452 million of it in the commercial and investment bank, "from new activity." Net income $15.3 billion (+6%), $5.57 a share ($5.75 before the $658 million securities loss, against a $5.94 consensus). ROTCE 20% (21% before the item), the lowest since the fourth quarter of 2025 and the same as a year earlier. Tangible book $112.64 (+7% on the year, −0.6% on the quarter as a −$12.1 billion move in accumulated other comprehensive loss outran retained earnings). CET1 13.9% standardized (−30 basis points: net income +70, risk-weighted assets −33, the securities mark −36, distributions −24, other −3); supplementary leverage 5.5%; liquidity coverage 113%. Dividend $1.65 (+10%); repurchases $1.1 billion gross (3.1 million shares at $342) against $6.7 billion in the second quarter and $8.3 billion a year earlier; net payout over twelve months 61%.

By business. Consumer banking earned $5.7 billion at a 36% return: deposits up 4% on the year after two years of outflow, a 2.78% deposit margin (2.70% in June), 650,000 net new checking accounts, card loans +7% at a 2.91% charge-off rate (3.34% in June). The investment and commercial bank earned $8.2 billion at 18%: investment-banking fees +14% and a 9.2% share of the global wallet, payments +14%, lending +20%, client deposits $1.2 trillion (+10%), charge-offs 0.14%. Asset and wealth management earned $1.8 billion at 45% on $5.2 trillion of assets under management (+13%), with $93 billion of inflows. Corporate lost $455 million: net interest income down $737 million on lower rates, and the $658 million loss on securities sold — the third negative quarter in five, after $395 million in the second quarter from selling agency mortgage bonds to buy Treasuries and foreign government paper.

The guide. Fourth-quarter net interest income excluding Markets about $25.5 billion (2026: ~$97.2 billion); total NII about $28 billion (~$106.2 billion); adjusted expense about $27.5 billion (~$108.6 billion); card charge-offs about 3.2% for the year. For 2027 a "central case" of about $107 billion of NII ex-Markets, decomposed by the CFO: annualize the fourth quarter and "you get something like 102. So we're essentially guiding to $5 billion of growth from that to the 107. The $5 billion, probably about a quarter of that, I would say, is related to the yield curve. So we remain asset sensitive and probably a little bit more asset sensitive than the EAR shows, just because of the lower than modeled consumer deposit betas. So about a quarter of that is the hikes that are currently in the forward curve. I think there's one in December and then two through the first half of the year." The rest is "a continuation of the balance sheet growth and mix trends that you've seen this year." The chairman added that "there's more uncertainty around that number than normal." On expense, the 2027 consensus of about $114 billion is one from which "our current preliminary read isn't too far."

What the morning taught, in the order it matters to the valuation. The buyback was cut by 84% by choice, not by constraint — "it's got nothing to do with AOCI or guessing about the environment or even Basel III endgame" — because management now believes it can deploy the $30 billion of excess at "very good returns over time"; the April letter said that same capital "is effectively earning a 4% after-tax return" (§11). The 2027 central case is $107 billion on a curve with three more hikes, three-quarters of the step from the fourth-quarter run-rate being balance growth, not rates (§5). The return was 20%, the lowest since the fourth quarter of 2025, in conditions the CFO called untypical (§3). The securities book moved — accumulated other comprehensive loss widened by $12.1 billion, the securities mark took 36 basis points off the capital ratio, and the company sold at a loss for a third quarter in five, calling the latest $658 million "discretionary" — and no analyst asked about any of it. Credit is at a nine-quarter best and management is "tightening up a little bit" anyway. The call's own time went to agents that move deposits and to where the capital goes next: the two questions the valuation turns on.

Before the arithmetic, the company. The company counts more than 1,200 predecessor institutions and dates itself to the Manhattan Company of 1799 and Drexel, Morgan & Co. of 1871; the Bank of the Manhattan Company and Chase National merged in 1955, and Chase Manhattan and J.P. Morgan & Co. in 2000 "to form today's JPMorgan Chase." Its modern shape is three acquisitions, two of them in crises. The Federal Reserve approved the merger with Bank One of Chicago in June 2004, which brought the retail branch system and the current chief executive, whose tenure the letter dates from Bank One in 2000. In March 2008 the Board approved the financing of Bear Stearns and opened the Primary Dealer Credit Facility the same day; in September 2008 the company took Washington Mutual's banking operations from the FDIC for about $1.9 billion, with $911 billion of combined deposits and 5,410 branches at the close — the balance sheet was $2.0 trillion then. In May 2023 it assumed First Republic's $229 billion of assets and $104 billion of deposits from the FDIC with a loss-share agreement, booking a $2.8 billion bargain-purchase gain. Assets went $2.2 trillion (2008), $2.7 trillion (2019), $3.9 trillion (2023), $4.4 trillion (2025), $5.1 trillion today.

Two laws made the universal bank possible and two crises set its rules: the Riegle-Neal Act of 1994 removed the interstate branching limits that dated from 1927, and Gramm-Leach-Bliley in 1999 repealed the Glass-Steagall separation of commercial and investment banking and created the financial holding company; Dodd-Frank in 2010 tightened the capital and liquidity regime and the 2018 act and 2019 tailoring loosened it; the 2023 failures produced a capital proposal within months and a retreat from it within two years. The result is the company the valuation is about: four businesses under one capital base — consumer banking (36% return on its allocated equity this quarter), the commercial and investment bank (18%), asset and wealth management (45%) and a corporate treasury that holds the securities book — with fees between 46% and 56% of revenue, the largest retail deposit base in the country, and a Markets book, two-fifths of interest-earning assets, funded and priced separately from the lending franchise. The industry backdrop — the sector's five-year ledger, the four regulatory tiers, the paper loss on the system's bond books — is the house's 10 October note, "US Banking Before the Print," and is not repeated here.

The record. The company's own series, from its 2026 letter, gives return on tangible common equity by calendar year from 2005: 15, 24, 22, 6, 10, 15, 15, 15, 11, 13, 13, 13, 12, 17, 19, 14, 23, 18, 21, 22, 20. The average over 21 years is 16.1%; over the last ten, 17.9%; the only years above 22% were 2006 and 2021. The company's through-the-cycle target has been 17% for years, and its chairman explained on the call why it has not been raised: "Times are pretty good now with high volume and high asset prices and fairly low credit losses and all that. If you could earn 17% consistently for a long period of time, that's a really good return for shareholders. That's our goal." Its CFO went further: "every single business is performing exceptionally well. The conditions have just been exceptionally favorable … the conditions that we've had recently where essentially every part of the franchise is beating records are not typical by historical standards." The letter's own worst case — a severe recession with rates cut to the floor, a 40% equity decline and credit losses doubled — is a return of about 10%.

The deposits. The company pays below the market for its funding — §4 sizes the gap. What the record says about how much of a hiking cycle reaches depositors: through the two completed cycles the cumulative deposit beta on interest-bearing deposits was 0.33 (2015–19: the cost rose from 0.14% to 0.81% as the funds rate rose from 0.13% to 2.16%) and 0.55 (2021–24: 0.03% to 2.82% on 0.08% to 5.14%); on the 2025 cuts it was 0.43. The non-interest-bearing share has fallen from 32% in 2015 to 24% today. The CFO's description of the book is the mechanism the bear case prices: "pretty barbelled between a bunch of very high beta deposits, like most of wholesale, as well as CDs and premium deposit and consumer, and then a stack of very low beta deposits. The question is probably less the pricing of any given product … and more the sort of migration behaviors across products in this cycle." Asked whether cash sorting is finished, he said: "some people believe, well, all the cash sorting that's going to happen has already happened … I don't personally believe that." His caveat on the history applies to the note's own beta figures as well — this cycle starts with the funds rate at 3.75% and the rate paid at 2.18%, a launch point for which, in his words, "we just don't have a lot of historical data."

The rest of the franchise. Retail deposit share first in the country for a sixth year; fees 46–56% of revenue across the nine quarters; asset-management fees +15%, card income +25%; $5.2 trillion of assets under management; Markets revenue $34 billion in nine months. Each claim in this section carries the number that would falsify it, listed in §11; the terminal returns the note uses — 17%, 15%, 12% — are argued from these claims, not from the quarter.

Against the system. The insured bank (CERT 628, Call Report, June) earned a 2.88% net interest margin against the system's 3.27% — at the holding company, where two-fifths of interest-earning assets are a Markets book yielding 0.6%, the lending-and-deposits franchise earns 3.70% — a return on equity of 18.6% against 13.1% for all insured banks, an efficiency ratio of 54.4%, five points worse than three years ago, charge-offs of 0.61% against 0.58%, and CET1 of 15.0% against the parent's 13.9%. Uninsured deposits are 49% of the bank's and, by the company's firmwide estimate, 64% of the group's at 30 June ($1.74 trillion, "primarily reflecting wholesale operating deposits"). Bank and holding company are different perimeters and are never mixed.

Where it sits: the five forces, with the company's own numbers. The franchise is a position in a market, and the market pushes back on five sides. Rivalry is high and specific: Goldman Sachs and Morgan Stanley in markets and advisory (the company's share of the global fee wallet is 9.2%), Bank of America and Wells Fargo in deposits and cards, the regionals in commercial lending; the chairman's own list of "very good competition" starts with Goldman Sachs and continues to "Stripe and PayPal and Cash and Block and Chime and SoFi and Revolut, and they're good." New entrants are the fintechs on that list and, in the letter's words, "a whole new set of competitors … based on blockchain, which includes stablecoins, smart contracts and other forms of tokenization"; the 10-K adds that "advocacy by non-banking competitors for exemptions from regulatory requirements could significantly disadvantage traditional financial institutions." Substitutes are priced every day: a three-month bill pays 4.23% against 2.18% on the company's interest-bearing deposits (§4), leveraged private credit at $1.8 trillion now exceeds the $1.7 trillion bank syndicated leveraged loan market, and stablecoins, at about $312 billion, are 1.6% of US deposits. Buyer power is the deposit beta — a third to just over half of each hike reaches depositors over a cycle — and, in wholesale, the CFO's "a little bit more competitive"; the consumer deposit margin of 2.78% says buyers have not used it. Supplier power sits with the regulators who set the 11.5% capital requirement, the 4.5% systemic surcharge and the capital rules the chairman wants recalculated ("If they think we should hold more capital, they should ask us to hold 10% more, and I'd be happy to do that. But I'm not happy to have these numbers falsely done"), and with labour: compensation rose 9% on the year and the front office is still hiring. The force the valuation turns on is the fourth: the rate the deposit buyer accepts is the margin the next section measures.

Yes, by the market's measure; whether that is a risk depends on which deposits, and the record answers it. The company paid 2.18% on $2.08 trillion of interest-bearing deposits in the quarter against a three-month bill at 4.23% and a funds rate of 3.88%; with another $655 billion paying nothing, the all-in cost of deposits was 1.66%, 257 basis points below the bill. On $2.73 trillion of average deposits that gap is about $70 billion a year — what the company saves against paying the bill rate on every deposit. Consumer banking's deposit margin, the company's own measure of what the retail book earns over what it pays, was 2.78%, up from 2.63% in March. The underpayment is the franchise; §3 argued why it persists, and §11 carries what would falsify it.

Whether it is a flight risk depends on which deposits. The record of the last cycle says the rate-sensitive money did leave: Banking & Wealth Management's average deposits fell from $1,112 billion in 2023 to $1,049 billion in 2024 and $1,041 billion in 2025 while short rates were above 4%, the CFO named "yield-seeking flows" as a risk in July, and the non-interest-bearing share of deposits fell from 32% in 2015 to 24% today. The record also says the franchise refilled: consumer deposits are up 4% on the year after two years of outflow, 650,000 net new checking accounts were opened in the quarter, firmwide deposits grew $232 billion in a year. The CFO's "barbelled" description is the answer to the flight question: wholesale, certificates of deposit and premium balances are already priced near market and move on price; the "stack of very low beta deposits" — operating and checking balances — moves on convenience and trust, and the question for it is migration into the first group, which costs margin, not a run, which costs the bank. The run gauge: $1.5 trillion of cash and marketable securities at 30 September against $1.74 trillion of estimated uninsured deposits at 30 June, 87% cover before Home Loan Bank and discount-window capacity; the chairman's own regulatory proposal is that banks hold "enough liquidity to pay off more than 50% of uninsured, nonoperating deposits," and the company does not publish its non-operating balance. What the note models is the price, not a run: the bear case carries a beta of 0.65 and $2–3 billion a year of migration cost. A beta above 0.60 by the fourth hike is the falsifier in §11.

The Federal Reserve raised the funds rate to 3.75–4.00% on 16 September. Its September projections put the median at 4.1% at the end of 2026 — one more hike, with 12 of 18 participants there — 4.1% through 2027, 3.9% in 2028, 3.6% in 2029 and 3.2% in the longer run. The company's central case assumes more: one hike in December and two in the first half of 2027. The note runs three paths, each with the deposit beta stated: the dot median (base), a flat 3.875% (no December hike), and one cut from mid-2027. Sensitivity: $1.6 billion of NII ex-Markets per 100 basis points for a full year, between the company's reported earnings-at-risk ($1.4 billion at 30 September, down from $2.1 billion at December; the company attributed the first-half decline to Treasury adding duration and gave no cause for the third-quarter move) and the CFO's own arithmetic (about a quarter of the $5 billion step is "related to the yield curve", for ~75 basis points partly in-year). The modeled beta is 0.50; the bull case uses 0.40, the bear 0.65 plus a migration cost on the existing deposit stock of $2–3 billion a year, the CFO's "internal migration out of checking and savings into CDs."

The three house paths are $0.6 billion apart in 2027, $0.4–0.6 billion apart through 2029 and about $1 billion apart by 2031, because the hikes are few and the company's rate sensitivity is small; the company's own three-hike curve sits $0.5–1.8 billion above them. The beta regime is worth ±$0.8 billion, wider than the spread between the three paths. The explicit forecast grows interest-earning assets by 4–5% a year fading to 3.5%, holds the 3.70% ex-Markets yield, grows fees by 5.5% fading to 4%, takes 2027 expense at the $114 billion consensus and grows it 4.5% fading to 3.5%, charges 0.65% of loans in 2027 and 0.70% through the cycle, taxes at 26% on the managed basis, and retains 13% of each year's growth in risk-weighted assets as capital — a buffer above the ~12.6% floor management's own $30 billion of excess implies, chosen as the conservative reading; the retained-capital line of the tornado (±$12 a share) bounds what it costs. On those inputs the base operating model earns 20.9% on tangible equity in 2027 and about 20% through 2036 if nothing fades — the model itself does not mean-revert; the house's judgement on persistence is in §6.

The method. A bank's equity is a claim on the return it earns above its cost of equity on the capital regulators let it run; its cash flow is funding, not value, so the note does not discount it. Value equals tangible book plus the present value of each year's excess return (ROTCE less cost of equity, on opening book) over ten years, plus a terminal value on the long-run excess return. The return fades from the 2027 operating-model figure to a terminal level on a half-life, so that it lands within a point of terminal by year ten; book grows by retained earnings; distributions are what is left; the terminal payout equals one minus growth over return. A dividend-discount model on the same distributions agrees within $5 in every case. The model passed the house's capital-consistency check with no failures and two warnings, described in §12.

The cost of equity. Ten-year Treasury 5.22% (8 October); beta 1.10 on two years of weekly returns (1.07 adjusted); equity premium 4.8%; CAPM 10.3%. Checks: the line the market draws through 21 banks pays tangible book at a 10.3% return; the FDIC system's return on equity has run 11–13%; the company's 21-year average return is 16.1%. The note uses 10.5% across scenarios — the CAPM reading and the market's line both say 10.3%, and the note rounds up toward the system's own return on equity, a choice worth $16 a share on the grid between 10.0% and 10.5% — and shows 9.5% and 11.5%.

The reverse valuation. At $331.43 the price requires a perpetual return on tangible equity of 24.1% at a 10.5% cost of equity, 21.2% at 9.5% and 27.0% at 11.5%. Holding the base case's explicit path and asking only what the terminal return must be: 23.2% at 10.5%, 19.8% at 9.5%. The base operating model run flat — about 20% forever, no fade at all — is worth $289, still 13% below the price. Across the cost-of-equity grid the base case is $272 at 9.0%, $250 at 9.5%, $231 at 10.0%, $215 at 10.5%, $201 at 11.0%, $189 at 11.5%, $178 at 12.0%. What moves the answer: the terminal return (±2 points, ±$28), the cost of equity (±1 point, −$26/+$35), the capital retained for growth (±$12); the rate path, the deposit beta, a 10-basis-point move in margin (∓$2.5) and in charge-offs (∓$1.5) together under $10.

The line. Across 21 US banks, price to tangible book rises 0.166× for every point of return on tangible equity (R² 0.88) and pays book at a 10.3% return — close to the industry note's 9 October line, which paid book at 10.7%, once every peer's tangible equity is put on the same basis (preferred stock deducted from six names whose filings carry no machine-readable preferred line; PNC and Capital One, whose preferred is carried at par inside surplus, are not adjusted). On that line a 20% return is worth 2.61× ($294); the company closed at 2.94×, 0.33× above it. On its cleanest current run-rate — 21.6% over twelve months, stripping the Visa and equity gains and the Apple Card reserve and adding back the securities losses — the line pays 2.88× ($324) and the premium is 0.07×. The industry note put the company 0.34× below its line on a first-half return of 23.4% that carried a $4.6 billion one-off and was measured on a tangible-equity figure that had not deducted preferred stock; on the company's own tangible equity that return is 25.6%, and on the corrected line the company sat about 0.6× below it at the first half. The price has not moved since; the return basis has. The mixed basis — the company at the third quarter, the peers at the first half — is labelled on the exhibit and re-run as the peers report this week. A bank-only line, excluding Schwab and BNY, whose tangible books are a tenth of their balance sheets, is flatter (book at 8.2%) and puts the company 0.53× above. Goldman Sachs sits 0.45× below the line on 23.2%, Morgan Stanley 0.27× above on 23.8%, Bank of America on it at 15.5%, Wells Fargo 0.20× below on 16.0%, Citigroup 0.22× below on 13.0%; the peer median excluding the company is 1.85× on 16.1%. Peers are measured on second-quarter book and first-half returns until each reports this week; the company is measured on the third quarter.

Own history. Year-end price to tangible book, on the company's own tangible book value per share: 1.37× (2015, on a 13% return), 1.68, 2.00, 1.79, 2.27 (2019, 19%), 1.92, 2.21 (2021, 23%), 1.83, 1.98, 2.46 (2024, 22%), 3.00 (2025, 20%). Median 1.98× at a median 18% return; the 13 October close 2.94×. Every year-end but the last printed below the current multiple. Control transactions: PNC paid 1.34× tangible book for BBVA USA in 2020 (PNC's own announcement), Truist's merger of equals was struck near 1.8× in 2019 (the BB&T registration statement against SunTrust's year-end tangible book), Capital One paid about 2.6× for Discover in 2024 (the merger proxy's exchange value against Discover's tangible book), and the company itself acquired First Republic from the FDIC in 2023 at a bargain-purchase gain. Control premiums for bank franchises have ranged from 1.3× to 2.6× in this cycle; the one above 2× was for a card issuer with its own payment network.

Exhibit 8 is the house's arithmetic. The Street's differs in two inputs, and the difference is worth stating in its own terms. Seventeen buys, thirteen holds and no sells; a mean target of $374 before the print, a high of $420 (Barclays), a low of $305. After the print, Goldman Sachs cut its target to $392 from $398 and kept its buy; RBC reiterated a buy and Jefferies a hold; overnight Barclays raised the high to $425, Baird the low to $312 (neutral), Morgan Stanley went to $377 (equal weight) and Truist to $361 — the range is now $312–425 and every change was a nudge of $5–9. The arguments, in a news site's same-day summary: "the NII guide is rising, loan growth is strong and credit is stable" for; "trading is soft, expenses are climbing and buybacks are shrinking" against. The fuller case has four parts, each already run in this note. One, the net interest income guide keeps rising on balances and the company is more rate-sensitive than its disclosures show — true, and worth about $2 a share across the rate paths in §5. Two, the return is a franchise fact and scale is compounding it — the steelman's first point; the reverse valuation in §6 is the test. Three, $30 billion of excess capital deployed at "very good returns" is worth more than buybacks — §11 shows both numbers management has given for what that capital earns. Four, the stock is cheap on earnings, 14 times trailing at the universe median — true, and also the model's warning: a 14.6 times peer multiple at a 10.5% cost of equity implies perpetual growth above any terminal ceiling, so the earnings multiple is the growth assumption restated.

The arithmetic of the targets: the $374 mean is 3.32 times tangible book, which on the steady-state formula is a perpetual return of 23.4% at a 9.5% cost of equity or 21.8% at 9.0%. The Street's price needs a 23.4% return held forever at a cost of equity a point below the note's, or today's 20% at a cost of equity near 8.5%; the difference between the house and the Street is not the facts of the quarter, on which they agree, but two assumptions — persistence and the discount rate — and the note has shown what each is worth (§6: ±2 points of terminal return is ±$28; ±1 point of cost of equity is −$26/+$35).

The company publishes three risk documents, and the note reads them against each other and then against its own list. The 10-K's risk factors are organised in twelve families — legal and regulatory, political, market, credit, liquidity, capital, operational, strategic, conduct, reputation, country, people — and read as every large bank's do, with two sentences that matter here: "security breaches due to cyber attacks … future breaches are inevitable," and the private-credit contagion channel, that "the interconnectivity across credit markets increases the risk that the significant expansion of private credit could worsen losses among non-bank lenders and their borrowers." The chairman's April letter lists the risks it calls "tectonic plates": "first and foremost" geopolitics; sovereign deficits and debt (US debt-to-GDP from 100% to 120% by 2036); "high asset prices and very low credit spreads" — household net worth at 560% of GDP against 460% at the 2006 housing peak, foreigners holding almost $30 trillion of US securities (and, in the letter's capital section, "many asset prices, including bank stocks, are fully valued"); Trade 2.0; US–China; private credit, at length ("losses on all leveraged lending in general will be higher than expected … actual losses right now are already a little higher than they should be"); private markets ("it's hard to imagine what will happen if and when we have an extended bear market"); and cyber, "one of our biggest risks." On the calls the risks are nearer: deposit repricing and "yield-seeking flows … probably a little bit of a risk" in July; this quarter, migration between deposit products in a hiking cycle, "tightening up a little bit on credit," AI exposures that are "quite modest" but not immune to "a major change in sentiment," and the agentic-AI question, which the CFO folded into the two old ones, yield-seeking and fintech competition.

What the letter and the calls share is that every risk they name is exogenous — geopolitics, deficits, asset prices, other lenders' credit, other people's agents; the 10-K's internal families (operational, conduct, strategic, people) are the boilerplate every large bank carries. None of them lists as a risk the one this note prices — that the company's own 20% return is the product of the conditions the letter calls fully valued rather than of the franchise — though two of them state it as a condition, and the unraised 17% target (§3) prices it. The closest the record comes is the chairman's "we just don't know how long it's going to last" and the CFO's "not typical by historical standards." Management's ranking and the note's are therefore different lists, and the reader should hold both: the letter's risks are what could end the conditions; the note's is the price's assumption that they persist.

The note's list — the risks that move value, in order of size: persistence — the base case is the fade itself, and the price minus the base case ($116) is its magnitude; a higher cost of equity as the ten-year rises further ("the risk of interest rates maybe moving even higher," as one analyst put it to the chairman, who "would never take that off the table") — $26 a point; deployment of the excess at 4% rather than 20% — the difference between the base and the bull; a deposit beta above 0.60 or migration out of checking, which costs little in the explicit years but is the mechanism for a lower terminal margin; crystallization of the securities mark — the company sold at a loss in three of the last five quarters and called the latest "discretionary"; expense growing 12–15% a year for three quarters against revenue growth of 8–10% in the first and third quarters (the second quarter's revenue, excluding the Visa gain, kept pace); a credit cycle, priced in the bear case at 1.2% charge-offs in 2028 with card losses near 5%, for which the data show nothing yet; and the AI and nonbank exposures the CFO described as "still quite modest … but that's not the same as saying that we would be immune to a major change in sentiment." Liquidity is a run gauge, not a solvency one (§4); the CFO's description of the 2023 failures — "an extremely concentrated and undiversified deposit base that was wholesale non-op and kind of monoculture" — does not describe this book.

Insiders and positioning, for the record. 122 Form 4 filings on the company's stock in twelve months; 43 open-market sales totalling $126 million and no purchases, 24 of them on three post-vesting dates, by eleven members of the operating committee, the chairman's $61 million among them at $306.56–307.19 against the $331 close — large in dollars, routine in pattern. Short interest was not checked; for an $880 billion company with no lock-up, convertible or issuance overhang it is not thesis-relevant.

The company's position is the incumbent's: adopt the rail, keep the deposit. Kinexys, its blockchain payments unit, has processed "more than $3 trillion in transactions since inception and [averages] more than $5 billion daily" (company, April 2026). In November 2025 it made JPM Coin — a deposit token, a claim on a JPMorgan deposit — available to institutional clients on Base, Coinbase's public Ethereum network, after test transactions with B2C2, Coinbase and Mastercard; the letter's instruction to itself is "we need to roll out our own blockchain technology," and its description of the competitive set is "a whole new set of competitors … based on blockchain, which includes stablecoins, smart contracts and other forms of tokenization." The company is not among the 21 institutions that announced a consortium stablecoin in September (a secondary report; the company has not said so itself), and it has announced no stablecoin of its own: a deposit token keeps the balance on the bank's balance sheet and inside deposit insurance, which is the distinction that matters to a company whose franchise is $655 billion of deposits paying nothing. On the call the retail product discussed was "Smart Cash … a targeted pilot type thing," and the threat discussed at length was agents that move money, which the CFO called "the intersection of two things that we've been talking about for a long time … yield-seeking flows [and] competition from fintechs … not new," while the chairman drew the line: "we're not going to allow just any outside agent to come in inside the company."

The future of the model is the question underneath the deposit question, and the chairman answered it when an analyst asked about deposit flight, adding "I'm not worried about over time": "banking services are necessary, credit cards and debit cards and movement of money and insured deposits and needed branches. And one of the ways you earn that is NII from money that sits there idle. But if you couldn't do it that way, over time, you'd do something different. Subscription, of course, you're getting investment accounts combined with checking accounts. So over time, it will adjust." That is the chairman's own statement of the franchise's long-run alternative — margin on idle balances replaced by a fee. The note does not model it inside ten years: stablecoins are 1.6% of US deposits, the company's own rails settle more than $5 billion a day, and the instruments the large banks have put live so far — deposit tokens and tokenised deposits — keep the dollar on a bank balance sheet; the consortium's stablecoin is announced, not launched. The industry note's reading was that the technology is being adopted by the banks, and nothing in the quarter changed it. The risk sits in the terminal return: a franchise charging a subscription for what it used to earn as idle-money margin could be a lower-return business than the 15% the base case assumes, or a higher one if the fee is priced right — a fee on balances the bank no longer holds capital against would, on the note's inference, raise the return on the capital that remains. Neither is knowable now.

What the retained dollar earns. The model distributes $45 billion in 2027 (69% of earnings) and $49 billion a year over 2027–29, retaining 13% of each year's growth in risk-weighted assets (the modelled ratio drifts from 13.9% to 13.6%) — above the 61% net payout of the last twelve months and the $22 billion annualized pace of the third quarter, below the $46 billion returned in 2025. Management's stated intention is the opposite: retain, hold "dry powder," and deploy the excess organically. The letter in April put it at $40 billion and said it "is effectively earning a 4% after-tax return"; the chairman on the call said "we can redeploy that capital at very good returns over time." Both numbers are the company's. If the marginal dollar earns the 20% the franchise earns today, the fade in §6 is too fast and value is closer to the bull case; if it earns the letter's 4% until deployed, the model is generous to the company by paying it out at once. The note cannot know which; it records the arithmetic. The accumulated other comprehensive loss of $19.8 billion — $7.5 a share of tangible book — accretes back as the securities pull to par and sits outside the model.

What the note cannot settle. Whether the next dollar of retained capital earns 20% or 4%. The third-quarter mark on the held-to-maturity book, which the 10-Q will show in November. How the deposit base behaves from a launch point the company itself says has no history. And whether a decade of returns above target is the franchise or the conditions — the note's base case treats it as the conditions and fades the return; the price treats it as the franchise; the company's own words describe both.