AXP — Q2 2026 Financial Report Analysis

Revenue up 10%, pretax up 15% — but most of the earnings gain came from the reserve line, not the business

American Express reported second-quarter 2026 total revenues net of interest expense of $19.64 billion, up 10% from $17.86 billion a year ago, and diluted earnings per share of $4.53, up 11%. Cardholder spending grew 9% — management's Stephen Squeri called it "the highest rate we've seen in three years on an FX-adjusted basis" — and the company raised its full-year revenue growth guidance to 10% from a prior 9–10% range while leaving EPS guidance of $17.30–$17.90 unchanged.

The revenue acceleration is real and broad. The earnings growth is a different story. Pretax income rose 15% to $4.07 billion, but $321 million of that $521 million increase came from a lower provision for credit losses — and that provision decline was an accounting reserve movement, not lower actual losses. Strip provisions out and the operating business (revenue minus expenses) grew just 4% against 10% revenue growth, because Amex is spending heavily on the benefits attached to its refreshed Platinum cards.

Headline metrics

Source: American Express Form 10-Q for the quarter ended June 30, 2026 (filed 2026-07-24, accession 0000004962-26-000322) and the Q2 2026 earnings release and statistical tables furnished on Form 8-K the same day.

Two terms worth defining before going further. Billed business is the total dollar value of purchases Amex cardholders put on their cards — it is the volume figure that drives Amex's biggest revenue line, discount revenue (the fee merchants pay Amex on each transaction). The provision for credit losses is the charge Amex books each quarter to cover expected future losses on money cardholders owe it; it consists of the losses actually written off in the quarter plus or minus a change in the reserve set aside against future losses. That distinction is the crux of this quarter.

The provision line did the heavy lifting

Amex released reserves this quarter instead of building them. On card balances, the $1,017 million provision was $190 million less than the $1,207 million actually written off, drawing the reserve down. A year ago the reverse happened: a $1,320 million provision against $1,122 million of write-offs, a $198 million build. That swing alone is worth roughly $388 million of pretax income — before counting a smaller favorable swing on other loans and receivables, which brings the total provision benefit to $321 million year over year.

Underneath the accounting, actual credit losses got slightly worse in dollar terms and stayed flat in rate terms:

• Net write-offs on card balances rose 8% to $1,207 million, roughly in line with the 8% growth in card balances to $218.1 billion.
• The net write-off rate on principal only — the share of loan balances Amex gave up as uncollectible, annualized, the measure banks compare across peers — was 2.0%, unchanged from a year ago.
• The 30+ day delinquency rate, which counts balances where a payment is at least a month late and is the best forward indicator of write-offs two to three quarters out, fell to 1.2% from 1.3%.

So the reserve release is defensible: with delinquencies falling, the loss rate Amex expects on tomorrow's balances is genuinely lower, and the reserve ratio dropped to 2.7% of card balances from 3.0%. Management attributes this to the credit quality of customers its premium products attract. But a release is a one-time pull-forward of income from a stock of reserves, and it cannot repeat indefinitely. At 2.7% of balances and 248% coverage of past-due amounts (down from 252%), the cushion is thinner than a year ago, and this tailwind should fade over the next several quarters.

Takeaway: Amex's spending and revenue momentum is genuine and accelerating, but the quarter's profit growth is not the same thing. Of the $521 million pretax increase, $321 million came from a credit reserve release that cannot repeat and $200 million from everything else — and the operating business's pre-provision profit (revenue minus expenses) grew only 4% against 10% revenue growth, because the Platinum refresh's benefit costs are currently growing faster than the fees it collects.

The Platinum refresh: fees up 15%, benefit costs up 50%

Amex refreshed its U.S. Consumer and Business Platinum cards in 2025, raising annual fees and adding benefits. Both sides are now visible in the P&L, and the cost side is running ahead.

Card Member services — the cost of delivering benefits like airline credits, lounge access and statement credits — jumped 50%, which the 10-Q attributes "primarily due to higher usage of Card Member benefits and the new U.S. Platinum benefits." That $648 million increase is $266 million more than the entire increase in net card fees. Together, the three variable customer engagement lines (rewards, business development and services) rose 17% to $8.76 billion, against 10% revenue growth, lifting them from 42.0% to 44.6% of revenue.

This is why total expenses grew 12% and pre-provision profit grew only 4%. It is not a sign the refresh failed — the refresh is also what is driving the spend and fee growth — but it does mean the economics are front-loaded on cost. Management called Platinum "the fastest growing" portfolio in U.S. Consumer, and U.S. Consumer Services billed business grew 11%, its fastest rate in the five quarters disclosed.

One nuance on rewards expense: the 9% increase would have been larger but for "a benefit from enhancements to the models that estimate future redemptions of Membership Rewards points by U.S. Card Members." Amex accrues rewards cost based on an assumed ultimate redemption rate (URR) — the share of points issued it expects to eventually be cashed in. That rate held at 96% year over year, but a model refinement reduced the accrual this quarter. The filing does not quantify it. Readers should treat the 9% rewards growth as flattered by a modelling change, not purely a reflection of underlying rewards economics.

Where the fee growth is actually coming from

Net card fees grew 15% while proprietary basic cards-in-force grew only 3%, to 67.5 million. The gap is price: the average annualized fee per proprietary card rose 12% to $131 from $117. Card acquisition volume was slightly down — 3.0 million new proprietary cards in the quarter versus 3.1 million a year ago, and 6.1 million year-to-date versus 6.4 million.

That is a materially different growth engine than "we are signing up more people." Amex is charging existing and new premium customers more and, so far, keeping them (management cited "strong Card Member retention"). It works as long as the benefit package justifies the fee — which is exactly what the 50% increase in services expense is buying. The risk is that a price-led fee model is more exposed to a competitive response or a consumer pullback than a volume-led one, and slightly lower new-card counts mean less of a growth cushion if attrition ticks up.

Segments: U.S. Consumer's profit growth is almost entirely the provision

U.S. Consumer Services looks like the standout at +23% pretax, but the arithmetic is unflattering: revenue rose $971 million while expenses rose $913 million — a $58 million operating gain — and the provision fell $331 million, from $829 million to $498 million. Roughly 85% of the segment's pretax increase came from the credit line.

International Card Services is the fastest-growing on volume (+13%, +12% excluding currency effects — meaning the growth is not an artifact of a weaker dollar) and revenue (+12%), but pretax income was essentially flat at $477 million because expenses rose 14%. Salaries and other operating expenses in the segment were $884 million, up from $783 million a year ago and sharply up from $630 million in Q1 2026. Consolidated "Other, net" expense also rose to $294 million from $147 million in Q1, which the 10-Q attributes partly to "an increase in legal reserves and higher technology costs." ICS is buying share and absorbing cost to do it; its profit contribution is not yet tracking its volume.

Commercial Services remains the laggard at +5% billed business. Management flagged a coming headwind: "We expect a moderation in spend growth as we exit small business cobrand held-for-sale portfolios." Note also that the exit of one such portfolio already dampened net interest income growth this quarter, and a gain on the sale of another portfolio sits inside the improved "Other, net" line — portfolio churn is quietly touching several lines.

Spending mix and who is spending

Total transaction count grew 10%, slightly ahead of the 9% dollar growth, meaning average ticket size was roughly flat to marginally down — consistent with volume-driven rather than inflation-driven growth. By category:

• Goods & services (71% of billed business) grew 9%, "driven by continued momentum in retail spending."
• Travel & entertainment (28%) grew 10%, "driven by sustained strength in restaurant spend and further acceleration in airline spend."

T&E outgrowing G&S matters for Amex specifically, because T&E carries higher merchant discount rates and is where its premium value propositions concentrate — but it is also the first category to contract in a consumer slowdown. The 10-Q identifies Millennial and Gen-Z cardholders as "our largest and fastest-growing cohort," which is the demographic argument behind the premium-fee strategy: younger high earners accepting a high annual fee today represent a longer fee annuity than the aging affluent base Amex historically served.

Why net income grew less than pretax income

Net income rose only 8% against 15% pretax growth because the effective tax rate jumped to 23.6% from 18.7%. The 10-Q is explicit: the prior year benefited from "discrete tax benefits... related to the resolution of certain prior-year tax items." Applying last year's 18.7% rate to this quarter's pretax income would have produced about $3.31 billion of net income rather than $3.11 billion — roughly $200 million, or about $0.29 per share. In other words, the prior-year quarter is an easy comparison on pretax and a hard one on net income, and neither the 15% nor the 8% figure alone describes the business.

EPS then grew 11%, faster than net income's 8%, because the average diluted share count fell 3% to 679 million. Amex repurchased 7.1 million shares at an average price of $315.77 (a $2,241 million outlay) and paid $645 million in dividends — $2,887 million returned, or 93% of net income available to common shareholders in the quarter, and 86% year-to-date. The quarterly dividend rose 16% to $0.95 per share. CET1 — the core regulatory capital ratio banks must hold against risk-weighted assets — was 10.4%, inside the company's 10–11% target and down from 10.6% a year ago.

The buyback pace is aggressive against 8% balance-sheet growth. Returning 93% of earnings while growing loans 8% is sustainable only while credit stays benign and CET1 stays above 10%; a genuine credit turn would force a choice between the buyback and the balance sheet.

Guidance and trajectory

Management raised full-year 2026 revenue growth guidance to 10%, up from the 9–10% range reaffirmed in April, and maintained EPS guidance of $17.30–$17.90 (FY2025 EPS was $15.38, so the range implies 12–16% growth). Squeri was direct about why the revenue raise did not carry an EPS raise: the company will "reinvest this outperformance in growth initiatives given the significant opportunities we see ahead."

First-half EPS of $8.81 leaves $8.49–$9.09 implied for the second half, against $7.67 in the second half of 2025. The midpoint implies second-half growth of roughly 15%, in line with the 14% delivered in the first half — so the maintained guidance is not conservatism so much as a decision to convert the revenue beat into spending rather than earnings.

Three things to watch:

1. The provision comparison gets harder. With reserves already at 2.7% of balances and delinquencies at 1.2%, there is limited room for further release. Absent one, second-half pretax growth has to come from the 4%-growing operating line — which is a very different profile than the 15% reported this quarter.
2. Whether services expense growth decelerates. The 50% jump reflects the first full-year lap of the refreshed Platinum benefits; it should moderate as the comparison normalizes into 2026's back half and 2027. If it does, operating leverage turns positive quickly, since fees keep compounding at a repricing-driven 12% per card. If usage of benefits keeps climbing instead, the refresh's payback period extends.
3. A sizable one-off gain is coming. Amex agreed to sell its roughly 30% stake in Global Business Travel Group as part of that company's pending acquisition, and expects "a sizable pre-tax gain" on closing, to be booked in Other expense. That gain is not in guidance-relevant operating performance and should be excluded when judging the underlying run rate — though it also gives management room to absorb further investment spend without missing the EPS range.

My read: the demand signal here is stronger than the earnings signal. Nine percent spend growth with delinquencies falling, and 12% per-card fee growth with retention holding, is a durable combination that few card issuers can currently show. But the market is being shown 15% pretax growth that is 62% reserve release and 11% EPS growth that leans on a 3% share-count reduction, while the part of the business that actually has to compound — revenue less expenses — grew 4%. That gap closes either through services expense decelerating or through the reserve tailwind reversing. Which one happens first is the question the next two quarters answer.