JPMorgan Chase

IThe business

A bank takes deposits, lends them out, and keeps the difference. JPMorgan Chase does that on a scale no other US bank matches, and then does four other things with the same balance sheet: it trades stocks, bonds and currencies for clients, advises companies on deals, moves payments, and manages money for rich people and institutions. At 30 September 2026 it had $5.1T of assets, $2.78T of deposits, $1.58T of loans and 322,964 employees.³ Shareholders own $372.4B of equity, of which $351.3B belongs to common holders and $299.1B is tangible, meaning what is left after $52.7B of goodwill and $2.4B of other intangibles.³

The firm splits itself into three businesses and a corner called Corporate. Here is what each earned in 2025, on the firm’s managed basis.²

Three facts in that table matter more than the rest. First, the Commercial & Investment Bank is the biggest earner, 49% of net income, and it uses 49% of the equity, so the firm’s profit leans on how well Wall Street does C (calculated).² Second, the consumer bank earns 32% on its equity, and 86.6M customers are the base for everything else.² Third, the line I find most telling is the last but one. Corporate holds more than a quarter of the equity and earns a low return on it, about 3.7% in 2025 C (calculated). That is capital the firm has not yet found a use for, and Jamie Dimon’s letter puts about $40B of it in that class, earning about 4% after tax.²⁶ I come back to it in the valuation, because it can add to the return or drag it, depending on what the firm does with it.

What it sells

Net interest income, the difference between what the firm earns on loans and securities and what it pays for deposits and debt, was $95.4B in 2025, 52% of revenue C (calculated).¹⁴² Fees, trading, and other income made up the rest at $87.0B.² Management guided in July to about $105.5B of net interest income for 2026, of which $96.5B excludes Markets, the trading business, whose interest income swings with its inventory.⁵ In the first nine months it was $77.9B.³

The fee side is spread widely. In the nine months to September 2026 asset management fees were $17.1B, up 17%, principal transactions $23.5B, investment banking fees $9.0B, up 24%, card income $4.0B, and lending and deposit fees $7.5B.³ Markets revenue, which includes both trading gains and the interest the desk earns, was $34.0B in nine months, up 23% and 21% of the firm’s managed revenue C (calculated).³ That is a strong pace, and I do not treat it as normal. My starting point leaves it in, and my base lets the return fade from there, because trading profits depend on clients being active, and activity moderated in the third quarter, as Dimon himself noted.⁴

Revenue grew 7.4% a year from 2016 to 2025, from $96.3B to $182.4B C (calculated).¹⁶² Two years in that run include one-time items. In 2024 a $7.9B gain on Visa shares sits in revenue, and in the second quarter of 2026 a further $4.6B Visa gain and $1.0B of other equity gains did the same.²⁵ I remove the 2026 gains from my starting point. The firm tendered 18.6M Visa Class B-2 shares in an exchange offer, received a mix of Visa Class B-3 and Class C shares, and booked the gain when Visa accepted the offer in May.⁵

Where the customers are

Chase had 86.6M consumer customers and 7.4M small business customers at the end of 2025, with 5,083 branches, up from 4,897 two years before.² Active mobile users reached 61.7M. Debit and credit card sales volume was $1.94T, and the firm says it ranks first in US retail deposits for the sixth year running.²⁴ In the third quarter it added over 650,000 net new checking accounts.⁴ The card business is the part that shows both the strength and the risk: 116.5M cards in force, $252.7B of card loans at 30 September 2026, up 7%, and a net charge-off rate of 2.90% in the quarter, against 3.15% a year before.²³ Chase will also become the issuer of Apple Card. The firm committed on 30 December 2025 to buy the portfolio, with closing about 24 months later, and booked a $2.2B provision on the deal in the fourth quarter of 2025.²³

On the corporate side, it ranks first in global investment banking fees with 9.2% of the wallet, according to Dealogic data cited in the Q3 release.⁴ Payments revenue was $5.6B in the third quarter, up 14%, and securities services $1.6B, up 14%.⁴ In wealth, client assets were $7.9T, up 15%, and Dimon said net inflows were $263B in the quarter.⁴

The cost of being wrong is in that chart. In 2020, the pandemic year, the firm set aside $17.5B for credit losses. In 2021 it released $9.3B of those reserves back into profit. In 2025 it provisioned $14.2B, which included the $2.2B for Apple Card.² Over ten years the provision averaged $7.0B a year, 0.62% of year-end loans C (calculated), and I use more than that, because 2021 was a release and 2020 a spike.¹⁶¹⁵² At 30 September 2026 the firm held an allowance of 1.75% of retained loans, including 6.16% against card loans.³

The balance sheet

This is where a bank differs from every other company on this site. Loans were 57% of deposits, a low ratio, and the firm estimates it holds $1.5T of cash and marketable securities.³⁴ It also holds $821.8B of investment securities, of which $547.2B are available for sale.³ The accounting for those securities matters now that rates have risen. The ten-year Treasury yielded 4.94% on 30 October 2026, and the firm’s accumulated other comprehensive loss, mostly unrealized losses on those securities, was $19.8B at 30 September, against $4.3B at the end of 2025.¹⁷³ That loss is part of equity and therefore of tangible book. It has cut book value by $15.5B in nine months, and about $10B of that counts in regulatory capital.³ It is not a cash loss unless the bonds are sold, and the firm’s securities mature toward par, so it reverses with time, but today it sits in the numbers.

The operating cash flow in a bank’s statements is not a usable number. For 2025 it was negative $147.8B, because the firm grew its trading assets.¹⁴ A free-cash-flow model would call JPMorgan a company that burns $148B a year. That is why I do not use one on this page, and why Section VI explains a different method.

IIThe moat

A durable advantage is a cost or a habit that a rival with money and talent cannot copy in a reasonable time. A bank’s edge is often narrower than it looks, because loans and trades are commodities and the customer can usually go elsewhere. Where JPMorgan is hard to copy is in funding, scale and breadth, and each of those can be tested against numbers.

The test of an advantage is whether it has widened. Here is the evidence both ways.

Evidence it is widening

• Return on tangible equity went from 12% in 2017 to 20% in 2025, and 24% in the first nine months of 2026 as reported.¹⁶²³
• The overhead ratio, expenses over revenue, fell from 59% in 2016 to 52% in 2025, so scale is paying for itself.¹⁶²
• Over the past ten years the firm’s proxy says it reached a return of 18% or better six times, while ten rivals together did it eight times.¹¹
• Customers rose from 82.1M to 86.6M in two years, and wealth client assets rose 15% in a year.²⁴

Evidence it is narrowing

• Expenses are growing faster than revenue. In the third quarter expenses rose 12% and revenue 8%, and the overhead ratio went from 52% to 54%.⁴³
• Card income in the consumer bank fell from $3.1B to $2.2B in 2025, because rewards and partner payments rose faster than interchange.²
• Capital rules have been ratcheting. The GSIB surcharge is set to reach 5.5% in 2028 under the current rule, and the Basel III re-proposal would add about 6% to required capital, which the firm says would net out to about 4% with the surcharge change.⁵
• Competition in payments is changing. The 2025 letter warns of risks from AI and agentic commerce, and a firm that sits between buyers and sellers has the most to lose if software routes around it.⁶

On balance I think the advantage is wide in funding and breadth, and flat to slightly narrower in returns, since the largest drivers of the past five years, rates and a boom in deal and trading activity, are cyclical. The advantage is real. The question for the valuation is how much of the 20% return it will keep.

IIIThe competitive advantage

The moat is about why a rival cannot copy the firm. The advantage is what the firm earns because of it. For a bank, the measure that matters most is how much profit it makes on the tangible equity it holds, and the cleanest test is to put it beside the others that are as big or bigger in what they do.

JPMorgan earns more than any of the three universal banks on its tangible equity, and the market pays more for it, in rough proportion. Bank of America earns 12.6%, or 14.9% before a $1.2B regulatory settlement charge, and trades at 1.7 times tangible book. Wells Fargo earns 16.3% and trades at 1.9 times. JPMorgan earns 20% and trades at 3.0.⁷⁸⁴ Morgan Stanley earns 22.6% and is priced a little higher at 3.4. Citigroup earns 11.5% and trades near book at 1.2. The pattern is the one the theory predicts: the multiple tracks the return. It also says that the market is already paying for JPMorgan’s edge. I cannot call the shares cheap because the bank is better. I can only ask whether they are priced for a better bank than the one I expect.

Why it earns more

Three things stand out when I take the firm apart. The first is cost. The overhead ratio, expenses over revenue, was 52% in 2025 and 54% in the third quarter of 2026, against 62% at Bank of America in the same quarter, before a $1.2B settlement charge that took it to 66%, and 63% at Wells Fargo.²⁴⁷⁸ A point of overhead is about $2B of profit at JPMorgan’s revenue C (calculated). A gap of eight points, 54% against Bank of America’s adjusted 62%, is worth about $16B before tax, or four points of return on tangible equity C (calculated).

The second is funding. Interest-bearing deposits cost 2.18% in the third quarter and the average rate on all interest-bearing liabilities was 2.88%, while loans earned 6.59% and the net yield on all interest-earning assets was 2.48% C (calculated).³ The firm’s earnings-at-risk table shows how little a normal rate move changes that. A 100 basis point rise adds $1.4B to annual net interest income and a 100 basis point fall takes away $2.2B, which is 1.3% and 2.1% of the roughly $105B base.³⁵ So the rate cycle is not what makes JPMorgan’s returns, though it did help in 2023 and 2024.

The third is credit selection. Over the ten years the net charge-off rate stayed between 0.27% and 0.74%, and the firm made money in every year.¹⁶¹⁵² Its card book is the highest risk, at a 2.90% loss rate in the third quarter, and the allowance on cards is 6.16% of balances.³

The ROTCE chart is the picture of the advantage. In every year of the decade the firm earned more than my 9.5% cost of equity, by a margin that ranged from 2.5 points in 2017 to 13.5 in 2021 C (calculated).¹⁶¹⁵² The average was 17.9%. The 2021 to 2025 average was 20.8%. The firm’s own through-the-cycle target is 17%, and the 2025 letter says that competitors have exceeded that return only about 9% of the time in the last ten years.⁶ I take the last ten years at face value as evidence of an edge. I take the last five as partly a good stretch for the industry: the firm itself shows a worst-case recession return of about 10%, so the range from 10% to 24% is wide.⁶³

The advantage is therefore real and measurable, and it is largely in the price. The valuation asks which return to expect for the next ten years: the 20% of today, the 17% the firm itself says it expects over a cycle, or the 12 to 14% that was common in 2016 and 2017.

IVManagement

Jamie Dimon has run the firm since December 2005 and has chaired it since December 2006. He was 69 at the end of 2025.² Every senior officer named in the 10-K held a senior role at the firm throughout the last five years, and the bench to succeed him is the best-defined part of the story. In June 2026 the board named Doug Petno and Troy Rohrbaugh Co-Presidents. Petno became sole head of the Commercial & Investment Bank and Rohrbaugh head of Consumer & Community Banking. Marianne Lake, who ran consumer banking, announced her retirement.²⁰

The bench is deep and it is also expensive to keep. On 24 June 2026 the independent members of the compensation committee approved one-time retention awards of $30M each for Petno and Rohrbaugh, and $20M each for Erdoes and Piepszak, $100M in all.²⁰ They are restricted stock units that vest in full after three years, only if the firm’s average return on tangible equity for 2026, 2027 and 2028 is at least 12%. The net shares then have to be held for two more years.²⁰ The company says the awards exist to keep the likely successors from leaving.²⁰ I think that is a sensible use of money. I also note the hurdle. 12% is below the 17% the firm calls its through-the-cycle target, and well below the 20% it earns today. It is a floor against disaster, not a reward for excellence.

Pay, and what it pays them to do

The board set Dimon’s 2025 pay at $43.0M, up from $39.0M. That was a $1.5M salary, $5.0M in cash and $36.5M of performance share units.¹⁰ The units are 88% of his variable pay, which the proxy says is a higher share of at-risk pay than any peer chief executive gets.¹¹ What the units pay for is return on tangible equity, three-year average, absolute and relative to ten rivals. The payout runs from nothing below 6% to the maximum of 150% at 18% or more.¹¹ The proxy notes that 18% was rare: the firm has reached it six times in ten years and ten rivals together eight.¹¹

Follow that incentive through. The measure is return on tangible equity, which rewards two things: earning more, and holding less equity. A manager can raise the ratio by lending more, by taking more trading risk, or by sending capital back to shareholders. The scale has guard rails. The proxy says up to a third of unvested units can be cut by the board for each year-end the CET1 ratio is under 8%, and awards have clawbacks.¹¹ But the bottom of the scale is 6%, below the cost of equity, so units still pay for a result that destroys value. And the top is 18%, so past that point more return buys no more pay, which, if anything, tempers the urge to chase. I think the structure fits a firm that has to meet a ratio for regulators and shareholders at once. What it does not do is tell the CEO what price to pay for the firm’s own shares.

Other facts on pay. Say on pay won 91% of votes in 2025.¹¹ The pay ratio and perquisites are disclosed in the proxy; for Dimon they include aircraft, cars and security paid by the firm, about $1.6M in 2025.¹¹ Stock ownership guidelines apply to the operating committee, and they cannot hedge or pledge unvested shares.¹¹

Their own money

Dimon owned 6,266,647 shares outright in the proxy, plus 653,426 units, 6,920,073 in all. At $337.04 the shares are worth about $2.1B, and with the units $2.3B C (calculated).¹¹¹ Directors and executive officers as a group, 21 people, held 8,573,082 shares, about $2,889M C (calculated), and under 0.4% of the company.¹¹ In 2026 Form 4 filings show Dimon sold 69,512 shares on 19 February at about $307.19 and 130,488 on 15 April at about $306.56, 200,000 shares and about $61M in all, roughly 3% of his holding C (calculated).¹²¹³ Both filings tick the box for a Rule 10b5-1 plan, so the sales were set in advance.¹²¹³ Others sold too: Rohrbaugh 50,000 shares in February, and smaller sales by Petno, Erdoes, Piepszak and Lake. In the Form 4 filings I read for 2025 and 2026 I found no open-market purchase by any officer.¹²¹³ I cannot see why the others sold, and I do not read much into it, since these are small slices of very large holdings. The letter itself says bank stocks are fully valued.⁶

What they did with the money, over a full cycle

The ten years from 2016 to 2025 cover a rate collapse, a pandemic, a reserve release, a rate surge and the 2023 bank failures. In them the firm earned $398B, paid $124B in dividends, including preferred, and bought back $157B of stock C (calculated).¹⁴ That is 70% of earnings returned. The share count fell 24%, from 3,561M to 2,696M, and tangible book per share more than doubled from $51.44 to $107.56 in the nine years to 2025 even after those payouts C (calculated).¹⁶²

Three things stand out. First, the firm did not buy anything large. The big acquisitions were First Republic in 2023, bought out of FDIC receivership with a $50B secured note and a $2.8B bargain gain, and now the Apple Card portfolio, with closing expected in about two years.² Second, the firm raised capital when it needed to and slowed buybacks when it had to. In 2022 it bought back only $3.2B, and in 2025 it bought $31.6B.¹⁴ Third, the buybacks have been made at prices far above book. In the first nine months of 2026 the average price was $307.36, about 2.7 times tangible book.³ Buying a stock at 2.7 times what it holds only adds to value if the firm earns more than the cost of equity on the premium, and it has, but the margin shrinks as the multiple rises.

Management seems to know it. In the third quarter of 2026 the firm bought back only $1.0B, 3.1M shares at $342.33, down from $6.7B in the second quarter at $308.21, and net of employee issuance only $664M.³⁴ Dimon said the firm cut repurchases because its order of priorities puts investing in the franchise and serving clients first, and that it can deploy the capital over time at very attractive returns.⁴ The 2025 letter said it more bluntly: the firm is clear-eyed that many asset prices, bank stocks included, are fully valued, it prefers to deploy capital where it can, and repurchases have to enhance value for the continuing owner.⁶ That is the right instinct. It is also a hint for the model: if the firm holds capital rather than return it, my payout figures in the early years are too high and the return on tangible equity too low, and the value depends on whether that capital finds a good use.

The record on capital is good. The record on succession, the one thing a bank run by a long-tenured chief cannot show until it happens, is not in yet.

VFinancials

I use calendar years, since the firm reports on them. The ten-year table has the figures the rest of this page leans on, with nine months of 2026 added in the last column. Everything is as the firm reports it, with no adjustment of mine.

Revenue grew 7.4% a year and earnings per share 13.9% a year over 2016 to 2025, helped by a 24% fall in the share count C (calculated).¹⁶² Tangible book per share grew 8.5% a year C (calculated). That last figure is the number to anchor on for a bank. It is what the business adds to the owners’ capital per share after paying out most of what it earns. Even so, tangible book per share is lower than it would have been, because rising rates cut it. The accumulated other comprehensive loss grew by $15.5B in nine months of 2026, about $5.85 a share C (calculated), and $10.0B of that, $3.76 a share, counts in regulatory capital.³

Reading 2026

The first nine months of 2026 brought net income of $53.0B, up 20% on the same months of 2025.³ But the figure has noise in it. The second quarter included the $4.6B Visa gain and $1.0B of other gains. The third quarter included $658M of investment securities losses, which the release lists as a significant item. I remove the gains and leave the losses in, because losses on a securities book are part of running one. After that the quarterly run-rate is close to the third quarter’s $15.3B.⁴

Two things in the normalization are judgement calls. I leave in a very strong year for the trading desks. And I leave in credit costs that are low: the net charge-off rate was 0.59% in the third quarter and 0.64% in the nine months, against 0.74% for 2025 and a ten-year average closer to 0.5%.³² A ROTCE of 21% on that base is what the firm earns in a good year.

Capital, and the capital the owner can reach

The firm’s Standardized CET1 ratio was 13.9% at 30 September 2026, against a requirement of 11.5% as of 30 June.³⁵ The gap of 2.4 points is about $52B of capital over the requirement C (calculated). The requirement has three parts: a 4.5% minimum, a 4.5% GSIB surcharge and a 2.5% stress buffer that stays at that level until 30 September 2027.⁵ The surcharge is set to reach 5.5% on 1 January 2028 under the rules as they stand, which would lift the requirement to 12.5%. The agencies have proposed to change the surcharge and also to revise the risk-weighting rules in the Basel III re-proposal. The firm estimates the two together would add about 4% to its required CET1 capital against the 2028 level under today’s rules.⁵ Dimon’s letter argues that the surcharge method is flawed and that its result is out of proportion to the firm’s resolvability.⁶

The letter also says the firm’s excess capital, under its assumptions about regulatory reform, is about $40B. Held as it is, it earns about 4% after tax. He says it may take several years to deploy at the returns the firm wants.⁶ The 2.4-point gap I compute is larger, $52B, because it uses the current requirement and not the reformed one the firm assumes. The two figures bracket the question: $40B to $52B of capital that earns little today, 13% to 18% of tangible common equity C (calculated).

Reported earnings and owner cash

Reported earnings are not what the owner can take out. A bank has to keep its capital ratio as its balance sheet grows, so some of every year’s profit is spoken for before it is free. I call what is left owner cash. Stock pay is already inside net income as an expense, so it is deducted and not added back: it was $3.6B in 2025.¹⁴ The capital to stand still is the equity the firm must add so its ratio does not slip when the balance sheet grows only with inflation. I take that growth as 3% a year.

So the owner can take out about 86% of reported earnings if the firm grows only with inflation, about 6.0% of the market value. But if the firm grows its risk-weighted assets as fast as it did in the last year, up 13%, it would keep about $34B of capital at the same ratio, and owner cash falls to about 45% of earnings C (calculated).³ The cash I put in the valuation is the owner cash after growth, not this standstill figure.

VIValuation

Start with what would go wrong

Before any numbers, I ask what would most likely make this a bad investment from here, and work back from that. The business is excellent. These are the three ways it goes badly wrong, in the order I think most likely.

1. The multiple falls back to where it usually is. The market pays 2.99 times tangible book. In the nine year-ends from 2016 to 2024 it paid between 1.7 and 2.5 times, and 2.0 on average C (calculated).¹²¹⁵¹⁶ If JPMorgan does what I expect and the market pays 2.0 times tangible book in 2036, the return from $337.04 is about 6.8% a year, with the cash paid out on the way, which is less than two points above the 4.94% Treasury E (estimate).¹⁷ Nothing in the business has to break for that to happen.
2. A credit cycle and a markets slump arrive together while capital rules tighten. The firm’s own worst case, with front-end rates cut to the floor, stocks down 40% and credit losses doubled, is a return on tangible equity of about 10%.⁶ My bear case is close to it for one year: revenue down 5% in 2028, credit costs at 1.9% of loans, which is higher than the 1.7% of 2020, and a return of 11% E (estimate). After that it never gets back above 15%. Under those terms the value is $189.
3. The spare capital stays spare, and the people who allocate it change. Corporate holds $98B of equity and earns about 4% on the part the firm calls excess.²⁶ If it is used well, it lifts the return. If it is used on a large acquisition at a high price, or sits idle while rules ratchet up, it drags. Dimon was 69 at the end of 2025, and the board says it is working toward an orderly transition in the medium term.¹¹ A successor has no record yet at this scale. I cannot put a number on this one, and I do not try.

I can show what the first does, and I do below. The second is my bear case. The third I flag and move on.

What the price implies

Over the ten years the price rose 4.87 times, from $69.26, while tangible book per share rose 2.19 times from the end of 2016 to September 2026 C (calculated).¹¹⁶⁴ The multiple supplied the rest, about 1.78 times of it. That part of the record cannot repeat without the multiple going to 5 times book.

How I value a bank

For most businesses on this site I discount free cash flow. A bank cannot be valued that way. Loans and deposits are its raw material, so the cash flow statement moves with the size of the balance sheet, and in 2025 JPMorgan’s operating cash flow was negative $147.8B.¹⁴ The right question is how much a shareholder can take out and still leave the bank as strong. So I work on tangible equity.

The idea is simple. A dollar of tangible equity that earns exactly the cost of equity is worth a dollar. One that earns more is worth more than a dollar, and the more it earns above the cost, and the longer it can, the more it is worth. So the value of the shares is tangible book plus the present value of the profit above what equity costs. That gives the same number as discounting the cash the owner can take out each year, after keeping the capital ratio at a target. I do both, and they agree to the cent, which is a check on my arithmetic.

The last row is the one that surprises people. About 59% of the value sits beyond 2036. That is not unusual for a business expected to last. But it means that the value is sensitive to what I assume the firm earns in ten years, and I show that sensitivity next.

Reverse the question: what return does the price need?

I take my base growth and my base capital ratios and ask a different question. What return on tangible equity, held for ten years and for ever after, makes the shares worth $337.04? I solve it.

The answer is a return on tangible equity of 19.2% every year C (calculated). With the faster growth of my bull case it is 18.5%, and with the slower growth of my bear case, 21.6% C (calculated). There is a shorter route. If the firm holds a return R for ever and grows at g, the multiple of tangible book is (R minus g) over (cost of equity minus g). At today’s 2.99 times, R has to be 20.5% if growth is 4%, 22.4% if it is 3% and 18.5% if it is 5% C (calculated). Either way the price needs the firm to keep earning close to its recent best, for a decade or more, not the 17% it describes as its through-the-cycle target.

I find it useful to put this beside the history. A flat 19% for ten years is below the last five years’ average of 20.8% and above the ten-year average of 17.9%. The price is therefore not crazy. It prices the recent run as the norm. I cannot say it won’t be, and that is what I mean by the honest range being wide.

The discount rate

I discount at 9.5%. The capital asset pricing model with the 4.94% ten-year Treasury and Damodaran’s 4.20% implied equity risk premium gives 9.0% on JPMorgan’s beta against the market over 36 months of 0.97, 9.2% over 60 months at 1.01 and 9.5% over 119 months at 1.08 C (calculated).¹⁷²¹²² I take the highest of the three, because a bank’s equity is the thin top layer of a very leveraged balance sheet and its beta tends to read low in calm years. The betas use price returns, without dividends.²²

At 9.0%, the rate on my Visa and Mastercard pages, the base is $320. At 8.5% it is $352, above the price E (estimate). So the price is not out of reach at a lower required return. I use 9.5% because a bank’s earnings depend on credit in ways I cannot see from outside.

Three cases

My base case is the firm keeping most of what it has but giving some back. Revenue grows 6% in 2027 and eases to 4% by 2033, about 4.6% a year on average, a little under the 7.4% of the last decade E (estimate).¹⁶² The overhead ratio drifts up from 53.9% to 58.5% as technology, pay and the cost of the branch push take more of each dollar. Credit costs run at 0.85% of loans, above the 0.69% I estimate for 2026, above the 0.62% ten-year average of provision over year-end loans, and a little above the 0.82% of 2023 to 2025, since card balances and the Apple Card are growing C (calculated).¹⁵² Risk-weighted assets grow one and a half points faster than revenue, because loans have been growing faster. The result is a return on tangible equity of 20.6% in 2027, 18.1% in 2031 and 16.2% in 2036, an average of 18.1%, a little above the ten-year average of 17.9% and about a point above the firm’s own 17% target C (calculated). I let the fade run toward 16.5% because every large bank’s return drifts toward the cost of capital over time, and because capital rules are rising.

The bull case has the firm repeat the past five years for ten: revenue growth near 6%, an overhead ratio flat at 53 to 54%, credit costs at 0.75%, and the excess capital put to use at high returns so the capital ratio falls to 12.4% of risk-weighted assets E (estimate). That gives an average return of 21.8%, above even the 20.8% average of the best five years on record. The bear case has revenue down 5% in 2028 as markets fall and rates drop, credit costs at 1.9% of loans, a higher capital target of 14.1%, and, afterwards, an overhead ratio stuck near 61% and a return of 14 to 15% that fades to 12.5% for ever E (estimate). Even the bear is a profitable bank. It earns more than the cost of equity every year, and is worth less than its price because the price pays for a lot more.

What I can’t judge

Some inputs I can’t know, and I would rather say so than pretend. I cannot see the quality of the loan book beyond what the firm reports. The wholesale loan book grew 16% in a year, and I do not know how much of it is to other lenders, funds and nonbank financial firms, though Dimon’s letter warns that credit standards in leveraged lending have been loosening and that losses in a cycle will be higher than expected for the environment.³⁶ I cannot know what final shape the Basel III and surcharge rules will take, and a point of required capital is worth about $22B of equity C (calculated).⁵ I do not know whether trading will stay at a $34B nine-month pace. I do not know how AI will change costs; the letter says it will affect every function and cannot yet name winners.⁶ And I cannot know when and how Dimon hands over. The range between my bear and bull is wide because these are wide questions.

Multiply it out

The price needs several things together. A return on tangible equity averaging 19% or more for ten years: I would give that 30%. Capital rules no tougher than the re-proposals: 65%. Credit losses staying inside a normal cycle, without another 2008: 70%. And a market in 2036 still paying about three times tangible book: 30%. Multiplied, that is about 4% E (estimate). They overlap, since a good return supports the multiple, so the true figure is higher, perhaps a fifth. And missing one does not ruin the case. The point is that the price is not a simple bet. It needs a lot to go right at once.

Quality and price are two questions

How good is the business? Among the best I have studied: a first-ranked deposit base, a return on tangible equity higher than any big universal rival, a record of not losing money in a crisis year, a bench of successors, and a balance sheet that the regulator tests every year. How good is the price? Fair to full. It is 15% above my base value, below my bull value, and it pays for the recent run to continue. A great business can be a poor purchase at the wrong price. This looks closer to a good business at a full price than to a poor purchase, and I say so without calling it a bargain or a trap.

It helps to compare it with the next best use of the same dollar. A ten-year Treasury pays 4.94% with no risk to the dollars promised.¹⁷ JPMorgan returns about 8.7% a year on my base, three and a half points more, with all the risk of a bank share. On this site, American Express is priced level with its base value at the same 9.5% rate, on a return on tangible equity of 41% in 2025, and Berkshire Hathaway is 16% above its base. Among the other banks, Bank of America trades at 1.7 times tangible book on a 12.6% return, 14.9% before a settlement charge, and Wells Fargo at 1.9 on 16.3%.⁷⁸ I have not valued them, and the multiple tracks the return in each, so none looks cheap beside JPMorgan by that measure alone.

The margin of safety

My margin-of-safety price is $190, 35% below my base value. At $190 my bear case returns about 9.5% a year, close to the 9.5% I require, and the base returns about 12.7% C (calculated). I use a wider cushion than I would for a business whose sales I can see, because I cannot see a bank’s credit. The price is 77% above it. The stock last closed below that level in April 2024.¹

Where this could be in 5 and 10 years

These are my estimates of what a share is worth at later dates, from the model above. The value at a date is the present value, at that date, of the owner cash that comes after it, on today’s 2,655.6M shares and in that year’s dollars. It includes the value of the bank beyond 2036 and leaves out cash paid before the date, which the fourth column adds up from now to the end of 2036, including the fourth quarter of 2026.

For the base to stand at five years, revenue has to grow about 5% a year to 2031, the overhead ratio has to drift from 54% to 57.5%, and credit costs have to run at 0.85% of loans. That keeps the return on tangible equity near 18% in 2031 and pays out about $95 a share by then E (estimate). For it to stand at ten, the return has to settle near 16%, which is worth about 2.2 times tangible book. $425 is that multiple on $196 of tangible book per share E (estimate).

The bear and the bull differ mostly in the return on tangible equity. In the bear a 2028 recession cuts revenue 5% and the return settles near 14%. In the bull it stays at 21 to 23% throughout. The discount rate is 9.5% in all three.

The most likely way the base is wrong is the fade. If the return stayed at 20% for ten years and after, instead of easing to 16.5%, a share would be worth about $539 at the end of 2036 and $355 today, above the price E (estimate).

These are estimates of value, not forecasts of the share price. The market price can sit above or below value for years.

VIIThe case for the price

What most people miss

First, more than a quarter of the equity is not working hard. Corporate held $120.9B of equity in 2025 and earned $4.5B on it, 3.7% C (calculated), and Dimon puts about $40B of the firm’s capital in the excess class, earning about 4% after tax.²⁶ Second, even the firm’s worst case is not a loss. Its investor-day recession scenario, with stocks down 40% and credit losses doubled, still gave a return on tangible equity of about 10%, above my cost of equity.⁶ Third, tangible book per share grew 8.5% a year over nine years while the firm paid out 70% of its earnings C (calculated).¹⁶²¹⁴

The bull case

JPMorgan is the best large bank, it earns about 20% on tangible equity against its own 17% through-the-cycle target, and it has spare capital it has not used. The clearest version is Jamie Dimon’s, in the annual report letter dated 6 April 2026. He puts the excess at about $40B, says it earns about 4% after tax, believes it can be deployed at excellent returns over several years, and calls AI a genuine shift, not a speculative bubble.⁶ The one named outside holder I could confirm is Viking Global Investors, with 2,899,386 shares at 30 June 2026, worth $949M, up from 1,905,982 at 31 December 2025.²³ Giverny Capital held 47,876. None of Altimeter, Atreides, Duquesne, Coatue, Tiger Global, Pershing Square, Berkshire Hathaway, ARK, Akre or Fundsmith reported any in its latest 13F.²⁴ I found no published view on JPMorgan from Gerstner, Baker, Wood, Gurley, Palihapitiya, Sacks or Ackman, so I attribute none. A 13F shows a position, not an argument.

Against reality

Much of the bull case is true today. The return is 20%, or 21% without the securities losses, and the deposit franchise, the banking rank and the credit record back it.⁴ My bull case gives it all of that for a decade, an average return of 21.8%, and is worth $389 E (estimate). If JPMorgan does that and hits its targets, $337.04 is a fair price, about 13% below that value. The price needs less: about 19% held for ten years, against a target of 17% and a ten-year average of 17.9% C (calculated). Today’s return is already there. The distance is in the other indicators. The overhead ratio rose from 52% to 54% in a year, and buybacks fell from $6.7B in the second quarter to $1.0B in the third.³

What a 10x would take. Ten times $337.04 is $3,370 a share, about $8.95T of market value C (calculated). At 15 times earnings that needs about $597B of net income to common, 9.6 times the normalized 2026 figure of $62.4B. At today’s 3.0 times tangible book it needs about $3.0T of tangible equity, ten times the $299B the firm has, which at a 20% return with nothing paid out takes twelve and a half years. The stock rose 4.9 times in the last ten years, from a company worth roughly $250B, and 1.8 times of that rise was a higher multiple, which cannot repeat at these levels C (calculated).¹ From $895B, a tenfold move is close to impossible, and I would not build a case on it.

Where I come down

The firm’s own scenario work and its capital record moved me: they make my bear case a return of about 11% in a bad year, not a loss. The excess capital did not move me, since the firm says deploying it may take years and it has just cut buybacks. What would raise my values is that capital earning above 15% while the overhead ratio stays near 53%; what would lower them is a return at 17% or below without a recession.

VIIIRisks, and what would change my view

The three ways this goes badly wrong are at the start of Valuation. These are the rest.

• Credit. The card book is $252.7B with a net charge-off rate of 2.90% in the third quarter and about 3.2% guided for 2026. The firm has $1.25T of card lines customers can draw on. The Apple Card portfolio will add to it when the deal closes, about 24 months after December 2025, and the firm already set aside $2.2B for it.³⁵² Wholesale loans rose 16% in a year to $938.8B.³ Dimon wrote that losses on leveraged lending in the next credit cycle will be higher than expected for the environment, because underwriting has weakened.⁶
• Rates and securities. A 100 basis point fall in rates takes $2.2B from annual net interest income, and a rise adds $1.4B.³ But rising long rates cut tangible book through the securities book: the accumulated loss grew from $4.3B to $19.8B in nine months. Dimon’s third-quarter comment lists large fiscal deficits, sticky inflation, elevated asset prices and rising interest rates among the things shifting under the surface.³⁴
• Trading. Markets revenue was $34.0B in nine months, 21% of revenue, up 23%. It was a strong stretch for client activity, and Dimon noted activity moderated during the third quarter.³⁴
• Regulation and capital. The stress buffer is 2.5% until 30 September 2027 and gets recalculated after that on revised models. The GSIB surcharge reaches 5.5% in 2028 under the present rule and 5.2% if the re-proposal is adopted. The Basel III re-proposal would raise required capital about 6% by itself and about 4% with the surcharge change, in the firm’s estimate.⁵
• Law and politics. President Trump, in a personal capacity, and several affiliated companies sued JPMorgan Chase Bank and its CEO in Florida state court in January 2026, in a matter the 10-Q files under fair access to banking. The defendants have moved the case to federal court, and the plaintiffs are challenging that. The firm also faces a class action over the interest paid on its brokerage cash sweep program, in which the court let breach-of-contract claims go forward in February 2026 and the plaintiffs asked in June for class status, and a consent order with the OCC on trade surveillance data, entered in March 2024, remains in force.⁵² The 10-Q lists other matters, including an Indian enforcement fine of about $31.5M that the firm is appealing.⁵
• Succession. Dimon was 69 at the end of 2025. The retention awards to the four likely successors, $100M, cost shareholders money to reduce this risk, and they vest on a return of only 12%.²⁰ Dimon has been chief executive for almost 21 years, so no one inside the firm has yet done the job.
• Technology and competition. The letter says AI will affect every function and that agentic commerce raises the risk of data misuse. Stablecoins and AI-run payments could take fees from card and payments businesses, and the letter does not claim to know who will win.⁶
• Geopolitics. The 2025 letter, dated 6 April 2026, names the war in Iran, the war in Ukraine and tensions with China, and warns that oil and supply-chain shocks could lead to stickier inflation and higher rates than markets expect. The next subsection puts numbers on it.⁶

Geopolitical risk, and what it means for the value

JPMorgan is mostly a US bank with a global trading and payments arm. In 2025 the Commercial & Investment Bank booked $30.5B of its $78.5B revenue outside North America, 39% of its revenue and about 17% of the firm’s $182.4B C (calculated).² At 30 June 2026 the largest foreign exposures were Germany at $107.7B and the United Kingdom at $99.1B, and Mainland China was $16.7B, up from $13.2B at the end of 2025, which is 5.6% of tangible equity C (calculated).⁵³ The direct exposure is small; the damage would come through markets, credit and the rules.

Russia is the history. The firm’s exposure was about $500M at the end of 2022 and it has since called what remains not material.¹⁵² Russian courts have entered judgments, one for $439M, that exceed the firm’s assets there, and the firm says those assets could be seized in full.⁵ That costs under $1 a share.

The base is $293. The price, $337.04, needs about 19% on tangible equity for ten years, so it leaves no room for the first row.

Who decides, and the odds. Tariffs and export controls are decided by Washington and Beijing. The letter says to expect “bumps in the road,” maybe large ones.⁶ Tariffs on China have gone up and down since 2018, so I put the slowdown in row 1 at about one in four over ten years, and a sanctions cut-off of China, which has no precedent at that size, under one in ten. These are guesses, not measurements. The agencies decide the capital rules; the March 2026 re-proposals are their second try, and I put something close to them at about two in three.⁵ Treasury yields are set by markets. The 10-year was at 4.94% on 30 October.¹⁷ The letter cites a Congressional Budget Office path from 100% debt to GDP to 120% in 2036.⁶ The accumulated loss already grew from $7.7B to $19.8B in the third quarter.³ Another move like it is close to even over a few years. It costs little as bonds pull back to par; the harm is a recession, which is row 1.

What is in the price. Russia and the $19.8B loss are in my starting numbers. The re-proposals are public, but I cannot see how much the market has taken. A slowdown is not in the price. I predict none of these.

What would raise my base value: the fourth-quarter results in January showing a return of 20% or more on normal credit costs, an overhead ratio at or under 54%, and buybacks resuming. What would lower it: an overhead ratio rising faster than a point a year, card charge-offs over 4%, or required capital moving up in the final rules while the excess sits idle.³⁵