American Express Raises the Revenue Guide, Holds EPS, and Falls 4.3% Again: The Reinvestment Trade Now Has a Track Record

Results vs. Consensus

Year-over-year comparison

Sequential comparison (vs. Q1 2026)

Quality of Beat/Miss

• Revenue: Growth of 10% is identical on a reported and FX-adjusted basis, so none of it is currency. The mix is the point: net card fees grew 15% and now compound off a repriced Platinum book, net interest income grew 11% while balances grew 9%, and discount revenue grew 9% on 9% billings. This is a revenue base where the fastest-growing line is the most recurring one. The 0.3% shortfall against consensus is not a demand signal; billed business accelerated to 9.4% FX-adjusted from roughly 8.5% in Q1.
• Margins: Expenses grew 12% against 10% revenue, and the driver is identifiable rather than diffuse. Card Member services rose 50% to $1,949 million, which is the U.S. Platinum refresh landing in the cost line before the matching fee revenue finishes amortizing in. Marketing grew only 6% and operating expenses grew 6%, so the reported negative jaws are almost entirely the deliberate value-proposition investment plus benefit usage on higher spend. Provisions fell 23%, which offsets a meaningful part of it at the pretax line.
• EPS: The $4.53 is honestly earned. Pretax income supplied $521 million of year-over-year growth, the buyback removed 3% of the share count, and the tax rate worked against the result by roughly five points. Strip the tax normalization and the underlying earnings growth is materially better than the 11% printed. The $2.2 billion of Q2 repurchases is the same lever management declined to lean on harder, which is precisely what the market objected to.

Segment Performance

Segment revenue sums to $19,742 million and segment pretax income to $4,640 million; the differences against the consolidated figures are Corporate & Other. Consolidated values are the income-statement amounts and govern. GMNS is shown on total network volumes rather than billed business.

U.S. Consumer Services

This is the segment that justifies the entire reinvestment argument. Revenue grew 11% to $9,524 million and pretax income grew 23% to $2,065 million, a level that exceeds the prior four quarters by a wide margin. Billings grew 11.4%, which management identified as the fastest rate since the first quarter of 2018 once the pandemic discontinuity is excluded. Provisions in the segment fell 40% to $498 million from $829 million, so credit improvement is concentrated exactly where the spend acceleration is. Expenses grew 15%, with rewards, business development and Card Member services together up 20% and marketing up just 2%.

"One of the biggest contributor to the acceleration is the platinum refresh. It is, by far, the biggest product we have. It has the fastest growth rate. There's truly a lot of momentum." — Christophe Le Caillec, CFO

Management attributed the acceleration to a combination of new acquisition, tenured Card Members consolidating spend onto the refreshed product, and upgrades from other cards in the wallet, referencing a 600-basis-point acceleration across the U.S. consumer Platinum portfolio that it had quantified at Q1.

Assessment: A 23% pretax gain on 11% revenue in the largest segment, with provisions down 40%, is the cleanest possible evidence that the September 2025 refresh is earning its cost. The market is discounting the consolidated expense line while the segment doing the spending is producing the operating leverage.

International Card Services

Billings grew 13% reported and 12% FX-adjusted, revenue grew 12% to $3,619 million, and net interest income grew 26%, the fastest of any segment. Management said four of its top five countries grew at a double-digit rate and that growth was broad-based across consumer and business customers. Pretax income, however, grew only 3% to $477 million and fell sharply from $781 million in Q1, with segment expenses up 14% against 12% revenue growth.

"International also delivered another strong quarter. We spent up 12% FX adjusted. Growth remains broad-based across consumer and business customers and across geographies with 4 of our tough 5 countries growing at a double-digit rate." — Christophe Le Caillec, CFO

The international premium playbook is being run deliberately. Management said it has refreshed the Platinum Card in roughly 80% of the countries where the card is issued, which has driven 20% FX-adjusted growth in international Platinum spending this year, and that around 70% of new consumer Platinum accounts outside the U.S. come from Millennials and Gen Z.

Assessment: The top line is doing what the thesis requires and the bottom line is not, because international is earlier in the same invest-then-harvest cycle the U.S. is now exiting. The $781 million to $477 million sequential drop deserves a direct explanation that the call did not supply, and we would treat it as the single most under-examined number in the quarter rather than as a warning.

Commercial Services

Revenue grew 7% to $4,503 million and pretax income grew 7% to $970 million, with billings up 5% and net interest income up 15%. Management said U.S. small and medium enterprise and large global customers grew at the same 5% pace, and characterized commercial as early in its product roadmap. The softness is specifically located: management said middle market is where it has seen weakness, while small business has been very strong and large, global and corporate have been moving along nicely. In the quarter the company launched a pilot of a new expense-management platform to an initial group of middle-market customers and introduced a $300 ChatGPT business annual statement credit for U.S. Business Platinum and Gold Card Members.

"It's middle market where we have seen the softness. Small business has been very strong and large in global and corporate been moving along, I believe, very nicely." — Stephen Squeri, Chairman and CEO

Assessment: Commercial has now put up two quarters of sequential billings improvement off a low base, and 5% growth with matched 7% revenue and pretax is a respectable holding pattern rather than a fix. The expense-management pilot is the right competitive answer to fintech encroachment in middle market, but it is a pilot, and the segment remains the part of the franchise where the premium moat is thinnest.

Global Merchant & Network Services

Revenue grew 8% to $2,096 million and pretax income grew 7% to $1,128 million on total network volumes of $516.8 billion, up 9%. Non-interest revenues grew 9% and expenses grew 10%, with marketing up 4%. Segment assets grew 6% to $19,401 million.

Assessment: GMNS is the quiet compounder that tracks billings with a small lag and does not require a narrative. Its strategic value in this quarter is optionality rather than growth: the closed loop is the asset management is positioning for agentic commerce, and that argument is made on data rather than on discount rate.

Key KPIs

Q1 2026 billed business is derived as the six-month total of $883.8 billion less the Q2 figure of $455.8 billion. Q1 FX-adjusted spend growth is approximate, taken from management's statement that Q2's 9.4% was "almost 1 point higher than Q1."

Key Topics & Management Commentary

Overall Management Tone: Pre-emptive and unusually direct. Management opened by naming the objection it expected rather than waiting for it in Q&A, then defended the reinvestment choice as a repeat of a strategy it has run for years instead of treating it as a new decision requiring apology. Tone was more assertive than at Q1, where the same choice was explained defensively; the one place the posture outran the disclosure was international profitability, which was described in top-line terms without addressing the sequential pretax decline.

1. The Guidance Question, Answered Before It Was Asked

The most striking feature of the call is that the CEO articulated the bear case in his prepared remarks and then rejected it. Rather than let the raise-revenue-hold-EPS decision emerge under questioning, management framed it as an explicit capital-allocation choice between buying back more stock and funding growth.

"I'm sure the question on your minds right now is if you're outperforming your expectations, and you are raising your revenue guidance, why aren't you also raising EPS guidance? I'll answer it. We have a choice. We can either drop the overperformance to the bottom line and buy back more shares or we can invest to grow the business further through the wide range of attractive growth opportunities we have across our business, both in the U.S. and international. We've chosen the latter." — Stephen Squeri, Chairman and CEO

The justification offered was returns rather than sentiment: management pointed to the company's high ROE as evidence that reinvested dollars have historically compounded better than repurchased shares, and noted this is what it has consistently done over the past several years.

Assessment: Pre-empting the question is a confidence signal, and being right about which question was coming is not the same as satisfying it. A 36.4% ROE is a strong empirical defense of the policy. What management still has not supplied is any framework for how an investor should size the payoff or its timing, which is why the same decision has now been sold twice.

2. The Second-Half Arithmetic Nobody Discussed

First-half EPS was $8.81 against a maintained full-year range of $17.30 to $17.90. That leaves $8.49 to $9.09 for the second half, and the midpoint requires $8.79, which is fractionally below what the company already earned in the first six months. Holding Q2's $4.53 flat for two more quarters produces $17.87, three cents below the top of the guide.

Assessment: This is the most important number in the quarter and it appeared in neither the release nor the call. The maintained EPS range is not a statement about earnings capacity; it is a spending authorization. Whether that is bullish depends on whether the spending compounds, and USCS pretax up 23% is the first hard evidence that it does.

3. Where the Reinvestment Is Actually Going

Pressed on the specific destinations, management named four: the proposed TheFork acquisition with its deal and integration costs, technology platform work pulled forward into the second half, continued card acquisition, and agentic commerce capability. Management was explicit that TheFork was not in the original plan for the year and will require second-half investment.

"And so across a wide range of technology platforms, we're able to pull some of those investments into the second half of the year. And so to be able to get to things quicker, is a huge advantage for us because we're going to have to make these investments over time. And why not make those investments sooner rather than later to update the platforms and so forth." — Stephen Squeri, Chairman and CEO

The framing throughout was that scale creates an obligation to keep spending. Management characterized the business as one that requires continuous fuel rather than one that can coast on installed advantage.

Assessment: Pulling technology spend forward is the most defensible item on the list, because the work is committed either way and doing it during a period of revenue outperformance is the cheapest time to absorb it. The weakest item is the vagueness: "a wide range of technology platforms" is not a disclosure an investor can underwrite, and it is the reason the reinvestment reads as open-ended.

4. Net Card Fees and the Amortization Mechanic

Net card fees reached a record $2,862 million and grew 15.4%, the fastest-growing revenue line and the 32nd consecutive quarter of double-digit growth. Management reiterated that growth accelerates in Q3 and exits the year in the high teens, and supplied the mechanism for why the refresh has not yet produced its usual post-refresh spike.

"The reason why it takes a while to hit the P&L is because, as you know, we started repricing car members from January only. And when we move car members to the new price point, we amortize it over 12 months. So it's kind of like a slow process." — Christophe Le Caillec, CFO

Management added that the full cycle takes something like two years to find its way into the P&L, which is why the growth rate shows the pattern it does.

Assessment: This is the most useful disclosure on the call and the strongest support for the rating. The card-fee acceleration is not a forecast dependent on demand; it is the mechanical amortization of price increases already taken on a book already repriced. That converts the single most important second-half revenue driver from an assumption into a schedule.

5. Credit Improved Rather Than Merely Held

Provisions fell 23% to $1,084 million and included a reserve release. The reserve balance on Card balances fell to $5,866 million, or 2.7% of balances, from 3.0% a year ago, while the write-off rate was flat at 2.0% and delinquencies improved to 1.2%.

"The strengthening we have seen in our credit performance is a deliberate outcome of our strategy to invest in value propositions that attract customers with high credit quality." — Christophe Le Caillec, CFO

Management said provision expense included a reserve release of $191 million, mostly reflecting further strengthening of portfolio credit performance, and that both delinquency and write-off rates remain below 2019 levels with delinquency rates between 1.2% and 1.3% for over three years.

Assessment: Credit is doing more than not breaking. A reserve release against an improving delinquency rate on an 8% larger balance sheet is the signature of a book whose mix is getting better, which is exactly what the premium-acquisition strategy is supposed to produce. This pillar is now the most durable part of the thesis.

6. The Stress-Test Proof Point

Management used the Federal Reserve's CCAR results as third-party validation of the credit argument, and the specific claim is unusually strong.

"The strength of our model also holds in a stressed environment as demonstrated by the Fed's recently released CCAR results, which show that under a severely adverse scenario, we have the lowest projected credit card loss rate across all banks and a pretax ROE of 3.8% over 9 quarters." — Christophe Le Caillec, CFO

The CFO returned to the same point later, noting that the credit reserve rate has fallen from 2.9% at the end of 2019 to 2.7% now as an outcome of focusing on premium Card Members, and that the company remains profitable through the cycle.

Assessment: Remaining profitable through a severely adverse scenario while carrying the lowest projected card loss rate of any bank is the answer to the consumer-credit bear case, and it comes from the regulator rather than from management's own modeling. For a lender trading on a premium-customer thesis, this is the most credible evidence available that the premium mix is real.

7. Spend Acceleration Was Broad, Not Narrow

Overall spend grew 9.4% FX-adjusted, almost a point higher than Q1, with goods and services up 9% and travel and entertainment up 10%. Retail spending grew 13% FX-adjusted, restaurant spending, the largest T&E category, grew 10%, airline spending grew 10%, and global travel bookings grew 22%. U.S. consumer spending grew 11%.

Management also addressed the geopolitical question directly, noting gas spend increasing significantly but representing only around 2% of total billings, and some visible impact on travel to and through the Middle East.

"So there is like some impacts, but there is no evidence of a general slowdown and people are offsetting these transactions with all the transactions in other categories. So it's not really visible at a macro level." — Christophe Le Caillec, CFO

Assessment: Airline spending at 10% is the number that matters most here, because late-quarter airline softness was a live watch item from Q1 and management now describes it as the highest growth in six quarters. That specific bear point has resolved favorably rather than lingering.

8. Cohort Quality and the Fee-Paying Mix

The company acquired 3 million new cards in the quarter, with 75% of new accounts acquired on fee-paying products, which management described as the highest level since it increased its focus on premium products. Across the year, 65% of new consumer accounts came from Millennials and Gen Z, and those cohorts were the fastest-growing in U.S. consumer spending.

On the durability question, the CFO argued the younger-customer focus is an embedded growth asset rather than a mix risk, on the basis that these customers arrive with lower income and grow with the franchise.

Assessment: A 75% fee-paying acquisition mix is the leading indicator for the card-fee line two years out, and it is running at a record while the fee line is still only mid-teens. The cohort argument is sound but unfalsifiable on a quarterly cadence; the fee-paying mix is the version of it that shows up in the P&L.

9. Retention Under Repricing

The most direct test of premium pricing power is whether Card Members accept a higher fee. Management said roughly a quarter of the U.S. Platinum portfolio had been repriced by the end of Q1 at a $200 higher fee, and that retention did not deteriorate.

"And the retention rate were not only through the roof, they were flat year-over-year. And last year, there was like no price increase." — Christophe Le Caillec, CFO

Management characterized attrition as unchanged, describing downgrades as largely life-stage driven, and noted a meaningful volume of upgrades onto Platinum as a result of the new value proposition.

Assessment: Flat retention through a $200 price increase on a quarter of the flagship book, against a prior year with no increase, is the cleanest pricing-power datapoint the company has produced in this cycle. It also de-risks the high-teens card-fee exit rate, since the repricing schedule only works if the members stay.

10. Dining Platforms as Acquisition Infrastructure

The proposed TheFork acquisition would add 50,000 restaurants across 11 European countries to a dining network that already includes Resy and Tock. Management was explicit that these are not evaluated as standalone businesses.

"When you look at Resy, Tock and TheFork on a stand-alone P&L basis, we don't look at it that way. So because it's part of the value proposition. That's how we look at it. It's a benefit." — Stephen Squeri, Chairman and CEO

The strategic logic offered was that these are open platforms serving non-Card Members too, which makes them a cheaper acquisition channel, and that Card Member spend at Resy restaurants runs at twice the rate of overall restaurant spend. Management said Resy and Tock will be combined on the front end while TheFork stays standalone given its European focus, with travel representatives able to book across both.

Assessment: Declining to run a standalone P&L on an acquisition is a governance flag in most companies and a defensible position in this one, because the asset is a benefit rather than a business. The 2x spend multiple at Resy restaurants is the closest thing to a return metric on offer, and it is suggestive rather than conclusive given the obvious selection effect.

11. AI as Capacity, Not Savings

Asked when AI-driven operating leverage would show up, management drew a deliberate distinction between efficiency and cost reduction.

"And from a technology perspective, what we're seeing is anywhere from a 30% to 40% decrease in cycle time from a coding perspective. Now that's really not a savings because what that does is it allows us to do more." — Stephen Squeri, Chairman and CEO

On servicing, management said equipping customer service and travel agents with AI-powered tools has produced a lack of acceleration in hiring even as the business grows, and that headcount would be expected to decline over time through attrition. The CFO framed operating leverage as the ratio of operating expenses to revenue, said operating expenses run annualized between $16 billion and $17 billion, and reaffirmed mid-single-digit growth for the year against 6% in the quarter.

Assessment: Converting a 30% to 40% coding cycle-time gain into throughput rather than savings is the right long-run choice and the wrong short-run answer for anyone hoping AI offsets the reinvestment. Management explicitly declined to commit to P&L savings in 2026 or 2027, which removes a source of upside some investors were carrying.

Guidance & Outlook

The shape of the guide matters more than the direction. Revenue growth was raised to a point estimate of 10%, which is the top of the prior range rather than an extension above it, and every incremental expense line moved up with it. The VCE ratio was lifted to 44% to 45%, which management tied to higher-than-expected spend in reward-earning categories including airlines, and second-half marketing steps to roughly 10% growth on increased customer acquisition. Operating expenses stay mid-single digits for the year.

Implied second-half ramp: First-half EPS of $8.81 against the $17.30 to $17.90 range leaves $8.49 to $9.09 for the second half. The midpoint requires $8.79, marginally below the first half. Two more quarters at Q2's $4.53 would deliver $9.06 and a full-year $17.87.

Street at: Full-year consensus EPS of $17.67 sits near the midpoint of the maintained range, so holding the range is not a cut relative to where the Street already was. The raised revenue guide implies roughly $79 billion of full-year revenue, in line with where consensus sat.

Guidance style: Conservative and consistent. This is the same company that reaffirmed rather than raised after an 11% Q1, and it has now raised revenue only to the top of the existing band while pre-committing the upside to expense. The pattern is a management team that would rather bank optionality than harvest it.

Analyst Q&A Highlights

Sustainability of the 11% U.S. consumer acceleration

The opening question went straight to whether the fastest U.S. consumer billings growth in eight years is a refresh artifact or a durable rate, asking specifically how much came from new account acquisition versus mature customers spending more. Management attributed it to all three available sources and quantified the Platinum contribution by reference to the prior quarter's disclosure.

Q: "So the 11-plus percent growth in USCS is truly impressive. I'm curious if there's a way to parse apart how much is coming as a result of the strong account acquisitions you've seen of late versus like your core mature customers spending more on the card. I'm just trying to think about the sustainability of that outperformance?"
— Sanjay Sakhrani, KBW

A: "We see that momentum coming from new card member acquisition as well as the tenured card members that are increasing their spend as well as either people who have either another card in their wallet and grading to platinum card, right? And so platinum is definitely pulling a lot here in the acceleration. If you go back to Q1, we quantified that acceleration, and I talked about it was in a Q1 number, but we talked about a 600 basis point acceleration across the entire platinum portfolio in U.S. consumer. So it's truly coming from a lot of tenured card members that are just consolidating their spend because they enjoy the new value proposition."
— Christophe Le Caillec, CFO

Assessment: The most useful part of the answer is the tenured-member consolidation, because spend that migrates onto a refreshed card from a competitor's card in the same wallet is stickier than newly acquired spend and does not require continued acquisition dollars. That said, an acceleration whose largest single driver is a product refresh laps by construction, and management has not separated the durable share from the refresh share.

What the reinvestment buys and when

The dominant theme of the call was the reinvestment decision, and the most direct challenge asked what the money buys and what it does for the ability to sustain top-line growth into 2027. The answer was a list of destinations rather than a return framework, closing on an argument about the nature of the business.

Q: "So Steve, you noted the decision to reinvest the upside in revenue growth. Maybe just talk a little bit more about the areas that you're investing in. I know you referenced customer acquisition and technology and maybe just talk a little bit about what you think this will do in terms of your ability to sustain these types of top line growth levels into 2027?"
— Ryan Nash, Goldman Sachs

A: "And to my point about sustaining the growth, you need to continue to invest in acquiring high revenue-generating cardholders and high spending cardholders to continue the really good revenue growth that we've seen over the last few years... So it's not a stagnant business, and it's not a business that you don't need to put gas in a tank. And so that's what we're doing. And I think it's a strategy we've employed for a number of years. I think it's really served us well."
— Stephen Squeri, Chairman and CEO

Assessment: The exchange is the quarter in miniature. The question asked for a return on the incremental dollar and the answer defended the policy of spending it. Four consecutive years of double-digit revenue and mid-teens EPS growth is a real track record and a legitimate defense, but it is a backward-looking one, and until the payoff is sized the market will keep pricing the spend as certain and the return as speculative.

The card-fee acceleration mechanic

A recurring line of questioning pushed on why the Platinum refresh had not produced the sharp card-fee acceleration prior refreshes did, and what underpins the high-teens exit rate. The response supplied the amortization schedule, which is the most concrete forward disclosure of the call.

Q: "Can you maybe just give a little bit more color on your net card fee growth guidance in the back half of this year, kind of what's contemplated in that, whether it's just from a step-up the annual fee from the back book? Or are you contemplating more acquisitions? And maybe just taking a step back, I think historically, you've seen more of a notable acceleration after a refresh, and you just haven't seen this thus far at Platinum?"
— Terry Ma, Barclays

A: "So there is inertia in that number, in that growth rate. And it takes it from beginning to end something like 2 years to find its way into the P&L. That's why you have this kind of like cycle. But to be clear, because that's the most important thing, we expect that either card fees are going to exit this year with a growth rate in the high teens."
— Christophe Le Caillec, CFO

Assessment: Management answered a skeptical question with a mechanism rather than a reassurance, which is the right way to handle it. The two-year amortization tail also cuts the other way: it means the fee benefit from this refresh will still be flowing in during 2027, extending the revenue visibility beyond the guide horizon.

Decomposing the net interest income deceleration

An analyst noticed net interest income growth slowed by 150 basis points sequentially and asked what to make of it against balance growth that had stepped up. The answer separated the accounting discontinuity from the underlying trend.

Q: "A question on NII growth. First, in the quarter, growth there slowed 150 bps from 1Q to 2Q. So just wondering if there's anything to call out there."
— Robert Wildhack, Autonomous Research

A: "But in Q2, specifically in the month of April, we transferred one of this portfolio, the lowest portfolio to be specific. It's not a large portfolio, but it was large enough to impact the NII growth rate by about 1 percentage point. So the decline you see in the growth rate from 12% to 11% is largely attributable to the transfer of that portfolio to the new issuer."
— Christophe Le Caillec, CFO

Assessment: A clean, checkable answer: the deceleration is a portfolio transfer, not margin compression or slowing balance growth. The forward disclosure attached to it is the more important part, since the effect builds to roughly 2.5 points of net interest income headwind by Q4 and persists until it laps. Investors modeling the line off reported growth rates will understate the underlying trend for three quarters.

Retention under a $200 repricing

A question on attrition asked for both the quantitative retention picture on new customers and the qualitative reasons members leave. The answer was the most emphatic of the call and pointed back to a prior-quarter disclosure.

Q: "Look, you guys have talked a lot about customer acquisition. Can we talk a little bit about attrition, both quantitatively and qualitatively, can we sort of think through what the 1- to 2-year retention rates are on new customers versus what you've seen historically? And also, to the extent you get feedback when customers don't renew the card. What are the reasons that a customer might walk away?"
— Richard Shane, JPMorgan

A: "The executive summary to you -- the answer to your question is we're not seeing anything on attrition... by the end of Q1, we had repriced, if you want, about 1/4 of the U.S. Platinum portfolio. And the retention rate were not only through the roof, they were flat year-over-year. And last year, there was like no price increase."
— Christophe Le Caillec, CFO

Assessment: Flat retention against a year with no price increase is the strongest form this datapoint can take, and it is the load-bearing assumption under the entire fee-acceleration story. The answer stayed qualitative on the one-to-two-year new-customer cohort curve that was actually asked about, which remains undisclosed.

Middle market softness and the fintech question

A question probed whether commercial weakness is a growth-rate issue or a competitive one, specifically whether middle-market customers are defecting to fintech competitors. Management located the softness precisely and pointed to the newly launched expense-management platform as the response.

Q: "Steve, it sounds like you're feeling a little bit better about SME, launched a pilot on the expense management software. I guess I'm trying to understand, is this more of a discussion around like the pace of billed business growth? Or could we see some type of middle market customers moving away to the fintechs where there could be some lumpiness?"
— Donald Fandetti, Wells Fargo

A: "It's middle market where we have seen the softness. Small business has been very strong and large in global and corporate been moving along, I believe, very nicely. So my perspective is we've seen an uptick in SME billings over the last 2 quarters. And I think the road map that we have and this financial -- the launch of the middle market expense management software will certainly help us."
— Stephen Squeri, Chairman and CEO

Assessment: Naming middle market specifically, and conceding that named competitors are present there, is more candid than the segment's aggregate 5% growth would require. The answer implicitly concedes that the competitive gap is a product gap rather than a pricing or brand gap, which is fixable but not fixed.

The rewards-liability assumption change

One question went after a quieter item: an adjustment to the ultimate redemption rate assumption underpinning the Membership Rewards liability, and how much of the quarter's cost performance it explains. This matters because a rewards-accrual change is exactly the kind of item that can flatter a print.

Q: "Wanted to ask about the adjustment in the URR and how material the benefit was in the quarter? And that helps us gauge the rest of the year and when you grow over it next year."
— Craig Maurer, FT Partners

A: "And so when we did that, we came up with a slightly lower ultimate rate of redemption, but it really hasn't changed it much. I think we report externally the ultimate rate of redemption is like 96% rate, and it really hasn't changed that much. So it has a small benefit in the quarter in terms of the MR cost. But at the full year level, it's just going to be like a -- it's going to be de minimis. So it's not -- it's more hygiene and making sure that we have the right accounting process more than anything else."
— Christophe Le Caillec, CFO

Assessment: The right question to ask and a satisfactory answer, though "small benefit in the quarter" was never sized. Card Member rewards expense grew 9% against 9% billings growth, which is consistent with the change being immaterial, so we take the answer at face value while noting an unquantified tailwind sits inside a quarter that already beat on EPS.

What They're NOT Saying

1. The international pretax decline. International Card Services pretax income fell from $781 million in Q1 to $477 million in Q2 while revenue rose sequentially. Nobody asked and management did not volunteer an explanation. Segment expenses grew 14% against 12% revenue, but that does not obviously account for a 39% sequential profit decline.
2. The size of the reinvestment. Management named the categories of incremental spend and never quantified the total. The only way to infer it is the gap between first-half EPS and the maintained full-year guide, which the company did not draw attention to.
3. Any 2027 EPS framing. The reinvestment case rests entirely on 2027 and beyond, and management declined to characterize what 2027 looks like beyond referencing a long-term growth aspiration. An investor being asked to fund a payoff has been given no dating for it.
4. The GBT proceeds and their use. The sale of the Global Business Travel Group equity interest is expected to close in the second half and is excluded from guidance, with no indication of size or whether proceeds fund buyback, the reinvestment program, or TheFork.
5. New-customer cohort retention curves. Asked directly for one-to-two-year retention on new customers versus history, management answered on aggregate attrition and the repriced Platinum book instead. Given that 75% of new accounts are now fee-paying, the cohort curve is the metric that would validate the acquisition spend.
6. Billings guidance. Management explicitly declined to guide billings for the balance of the year, offering only that it is a momentum business and flagging the roughly one-point co-brand headwind. The billings assumption embedded in the raised 10% revenue guide was asked for and not given.
7. Quantified AI savings. Management volunteered that it is not committing to P&L savings from AI in 2026 or likely 2027. That is candid, and it removes an offset investors might otherwise apply against the expense step-up.

Market Reaction

• Pre-print setup: AXP closed at $340.84 on July 23, down 7.9% year to date against the S&P 500's 8.2% gain, up 10.6% over trailing twelve months and down 0.5% over the trailing thirty days. The 52-week closing range entering the print was $292.27 to $384.89, so the stock sat in the middle of its own range and well behind the index.
• Reaction session (July 24, before-open reporter): Opened at $326.12, a 4.3% gap down, traded a range of $318.20 to $329.05, and closed at $326.17, down 4.3% or $14.67. At the intraday low the stock was down 6.6%.
• Volume: 6.4 million shares against a 3.1 million 30-day average, 2.1 times normal.
• Market context: The S&P 500 was unchanged on the session, so the entire move was company-specific.

The decline is a composite of three things, and only one of them is about the business. The revenue miss was 0.3%, roughly $60 million, and it broke a run of top-line beats, which gave a crowded pre-print setup a reason to unwind. The guidance construction did the rest: raising revenue while holding EPS makes the reinvestment explicit and unavoidable, and management compounded the effect by naming the alternative it rejected, a larger buyback. Finally the optics of the expense line, 12% growth against 10% revenue, plus a tax rate stepping from 18.7% to 23.6%, produced a headline in which costs rose faster than sales and net income grew only 8%.

What is notable is the symmetry with the prior quarter. In April the stock fell 4.3% on a bigger EPS beat because management reaffirmed rather than raised. In July it fell 4.3% on a smaller EPS beat because management raised revenue but not EPS. The market has now priced the same objection twice at the same magnitude, which suggests it is not reacting to the news in either quarter so much as repeatedly re-rating the capital-allocation policy itself. The recovery from a 6.6% intraday decline to a 4.3% close is the one sign that some buyers were willing to separate the operating quarter from the guide.

Street Perspective

Debate: Is reinvestment compounding or is it a treadmill?

Bull view: The spending is producing measurable acceleration rather than defending a flat line. U.S. consumer billings hit an eight-year high and USCS pretax income grew 23%, so the incremental dollar is visibly earning a return in the segment where it is being spent. Four consecutive years of double-digit revenue and mid-teens EPS growth argue the policy works.

Bear view: A growth algorithm that requires ever-larger reinvestment to sustain the same growth rate is a treadmill, not a flywheel. Each refresh cycle raises the cost base permanently while the revenue benefit amortizes and then laps, and management has now twice declined to let outperformance reach the bottom line.

Our take: The bulls have the better of it this quarter, on evidence rather than argument. A 23% pretax gain in the largest segment alongside a 40% decline in that segment's provisions is not what a treadmill looks like. The bear case is really a disclosure complaint dressed as an economic one, and it would be answered by management sizing the payoff rather than by spending less.

Debate: Does the card-fee acceleration arrive on schedule?

Bull view: The high-teens exit rate is mechanical, not forecast. Repricing began in January and amortizes over twelve months on a book already a quarter repriced by the end of Q1, with retention flat through a $200 increase. The revenue driver is a schedule, and it keeps flowing into 2027.

Bear view: Card fees grew 15.4% and the promised inflection keeps living one quarter out. Prior refreshes produced visible acceleration sooner, and a company guiding to a high-teens exit while printing mid-teens is asking to be taken on faith in the same quarter it asked to be taken on faith about the reinvestment.

Our take: The bulls, with a caveat. The amortization mechanic is checkable and the retention evidence is strong, so the exit rate is more likely than not. The caveat is that it converts Q3 into a must-deliver quarter: card fees need to visibly accelerate from 15.4%, and if they do not, the credibility cost lands on the reinvestment argument too.

Debate: Is the valuation a de-rated compounder or a fair price for a slowing financial?

Bull view: At $326.17 the stock trades near 18.5 times the midpoint of a maintained EPS guide whose own arithmetic implies no second-half growth is needed to reach the top end. That is a modest multiple for a 36% ROE franchise with best-in-class credit, a record fee-paying acquisition mix, and a raised revenue outlook.

Bear view: Published targets clustering in the high-$300s were struck on an assumption of EPS upside that management has now twice refused to deliver. If EPS compounds at the guide rather than above it, the multiple is fair rather than cheap, and the co-brand headwinds plus the normalized tax rate make 2027 estimates look optimistic.

Our take: The bulls, but the catalyst is calendar-driven rather than available on demand. The gap between an 18.5x multiple and a Street target set in the high-$300s is the market declining to underwrite a payoff it cannot date. That gap most plausibly closes in Q4, when the Platinum refresh laps, variable engagement cost growth decelerates by management's own guidance, and the card-fee line exits in the high teens. Investors unwilling to hold through two more quarters of this expense optic should not own it.

Model Update Needed

Valuation impact: The raised revenue guide and the second-half EPS arithmetic together argue for modelling the top of the company's range rather than the midpoint, which mechanically improves the multiple at any given price. At $326.17 against $17.87, the stock trades near 18.3 times, and against a 2027 estimate built on the company's own mid-teens aspiration it trades in the mid-teens. We see no reason to change the shape of the model; we do see reason to stop treating the guide midpoint as the expected value.

Thesis Scorecard Post-Earnings

Overall: Thesis strengthened on fundamentals and weakened on the path to recognition. Five of seven pillars moved favorably or held, credit improved beyond "stable," and the one commitment we most wanted graded, a back-half guide raise, was half delivered: revenue yes, EPS no. Against that, the catalyst-vacuum bear point has moved from a risk to an observed fact, having now cost roughly 4.3% twice.

Action: Hold the position and treat weakness as accumulation rather than warning. The operating business is compounding faster than the guide admits, and the arithmetic that makes the guide conservative is available to anyone who subtracts $8.81 from $17.90. The recognition event is most likely Q4, when the refresh laps, variable engagement cost growth decelerates by management's own guidance, and card fees exit in the high teens. Investors who need the catalyst sooner than the calendar allows should size accordingly.