AXP $AXP Energy Ltd(AXP.AU)$ reported a solid 3Q26. Revenue rose 9% to $20.06B. Adjusted EPS rose 13% to $4.68. Card fees rose 16%. Credit improved. But GAAP EPS rose 31%, and about $1.11 of that came from the GBTG equity sale. The headline overstates underlying momentum.
Our concern is not demand. It is flywheel economics. CMS/card fees rose to 68% from 58%. Benefit costs are absorbing more of each card-fee increase. NIM expansion is deposit-driven and near its limit. At 16-17x forward P/E, the stock looks fairly valued. We stay Neutral.
Earnings quality:
The 31% GAAP EPS growth isn't the right operating benchmark. Adjusted EPS +13% is. Three one-offs boosted GAAP: the GBTG sale, provision release, and a lower tax rate. Provisions fell to $985M from $1.287B, adding about $300M pre-tax. About $215M came from a CECL model update. The tax rate was 18.1% vs. 24.1%, mainly due to discrete benefits. That will not repeat. Normalized growth is solid, but it is not 31%.
Revenue:
Card fees are the cleanest signal. They rose 16% to $2.97B. The seven-year CAGR is about 17%. The premium-card flywheel is still working. Billed business rose 8% to $454B. USCS rose 9%, with Gen-Z +35% and Millennials +11%. ICS rose 11% FX-adjusted. Commercial Services rose 3%, or 5% excluding the exited small-business portfolio. Small-business spending is still cautious. Service fees looked flat, but that is accounting noise. Ex-GBTG, service fees rose about 8%.
The key debate:
Can card-fee pricing outrun the costs of benefits? CMS rose 36% to $2.01B. The CMS/card fee ratio rose to 68% from 58%. For every extra $1 of card fees, AXP now spends about $0.68 on benefits. Net retention is about $0.32. This is partly deliberate investment behind the Platinum refresh. It is not proof the flywheel is broken.
But it does mean the payback period is longer. We would turn more positive on flywheel margins only when CMS growth falls below card-fee growth. On current trends, that looks more like 1H27 than 2H26.
Credit:
Credit quality improved across the board. Net write-offs were 1.8% vs. 1.9%. 30+ day delinquency was 1.2% vs 1.3%. Provisions fell 23%. But reserve coverage fell to 2.6% from 2.9%. That means less room for future provision releases. Credit is not deteriorating.
But part of the profit beat came from releasing reserves, not from core earnings. CET1 was 10.9%, inside the 10-11% target. Q3 capital return was $2.2B, about 59% of net income.
NIM:
Net interest income rose 10%. Volume added 7 points; price added 3. The price contribution came almost entirely from lower deposit costs. NIM was 8.5%, +30bp. The average deposit rate fell to 3.2% from 3.7%.
Loan yields did not improve much. Savings rates are already at 3.0%. Further deposit-cost relief is limited. If the Fed slows, NIM may peak in 2027. We see NII growth slowing to 5-7% in 2027.
Moat:
AXP’s moat is not network scale. Visa and Mastercard are larger. The moat is premium customers, a closed-loop network, and member benefits. T&E is 29% of billed business. Premium customers are fee-tolerant and benefit-sensitive. The closed loop lets AXP earn discount revenue, card fees, and interest. That is different from pure networks.
But the moat is narrowing. Chase Sapphire Reserve and Capital One Venture X are investing aggressively. Agentic commerce could shift the payment entry point. Merchant surcharging is legal in some states. These are real medium-term risks.
What would change our view:
Upside requires Q4 billed business above 8%, 2027 card-fee growth guidance of at least 12%, and a stable CMS/card fee ratio. Downside risks are a consumer slowdown, provision rebuilding, and tighter regulation. The market is pricing a neutral scenario. We need evidence that flywheel economics are stabilizing before getting more constructive.
Comments