American Express: A Fine Business Standing Behind a Tall Wall of Price
American Express presents a familiar puzzle for the value temperament: a genuinely excellent company wearing a genuinely inconvenient price tag. The judgment, as of mid-July 2026, is watch — the business passes the first three gates comfortably and fails the fourth decisively. Let us walk the gates in order.
Gate 1: Circle of Competence
Amex operates a closed-loop model: it is both the card network and the issuer. Revenue arrives from three streams — discount revenue from merchants (a higher merchant fee than Visa or Mastercard commands, thanks to premium customers’ high spending), net card fees (roughly $10 billion in 2025, up 18%, marking 30 consecutive quarters of double-digit growth), and net interest income from card lending. Because it sits on both sides of every transaction — cardmember and merchant — it holds data neither pure network can see, which feeds fraud prevention, targeted offers, and risk management.
The essential character is not “technology company.” It is a premium consumer brand plus a membership business.
Knowability passes on both levels. The information layer is straightforward: SEC filings, earnings materials, the CEO’s shareholder letters. The predictability layer holds too. “Premium consumers and businesses pay an annual fee and transact on the Amex card” is a demand pattern with over sixty years of demonstrated inertia. Newer evidence suggests the brand is aging downward rather than out: 73% of new accounts arrive on fee-paying products, and Gen-Z spending in the US consumer business grew 38%. The payments industry itself carries heavy regulatory and scale barriers. This sits squarely within the competence circle: a consumer franchise with brand-driven monopoly traits, an established network, and a regulated, capital-intensive industry.
One historical note: the thesis that carried this name in 1963 and 1991 — brand plus closed-loop network — remains alive today. But its center of gravity has shifted. The pure acceptance-network advantage has narrowed as Visa and Mastercard achieved near-universal coverage; the moat’s main body is now the premium brand, the annual-fee membership lock-in, and two-sided data. The argument has become more Coca-Cola than Western Union.
Gate 2: Moat
Three sources, all present. A two-sided network effect: concentrated high-income cardmembers induce merchants to accept premium fees, and wider acceptance increases card utility. A brand that has signified premium status since “Membership has its privileges.” And switching costs: accumulated membership rewards points, an ecosystem of credits and lounges, corporate expense-system integration.
The pricing power evidence is textbook. In September 2025, Amex raised the Platinum annual fee from $695 to $895 — a 28% increase — and new accounts still arrived 73% on fee-paying products, with card fee revenue up 18%. A company that raises prices annually without a prayer meeting is not common.
Direction: widening, with rising competitive pressure. Widening: 127.6 million cards in force (+8%), younger cohorts secured, and a growing share of recurring, less cycle-sensitive fee revenue. Narrowing: the Capital One–Discover merger in May 2025 created a rival with a similar integrated network; the premium-card war with Chase Sapphire and Robinhood (whose card costs $695) has intensified; Capital One’s acquisition of Brex pressures the SME side. The honest summary is that the moat is holding and expanding, but the cost of maintaining it — rewards and benefits — is structurally rising.
The AI era cuts in the moat’s favor. The closed loop produces proprietary two-sided transaction data that improves fraud detection, underwriting, and offers — a flywheel no competitor can replicate even as models commoditize.
Gate 3: Management
Stephen Squeri has been CEO since 2018, after 40 years at the company. Capital return ran $7.9 billion in 2024 ($5.9 billion in buybacks) and about $7.6 billion in 2025; the quarterly dividend rose from $0.82 to $0.95. Growth has been organic — no empire-building mega-mergers. The demerit: buybacks run as a standing program rather than a price-disciplined one, continuing above 20x earnings.
The retained-earnings test passes clearly: five-year average ROE of 32.5% with a standard deviation of only 2.6 percentage points — high and remarkably stable. EPS moved from $11.21 (2024) to $14.01 (2025), with 2026 guidance of $17.30–17.90.
Candor and alignment are solid: direct annual shareholder letters, consistent strategy, and Berkshire Hathaway’s 21–22% stake as a standing disciplinary presence. Share count falls steadily. Guidance reliability is good.
Gate 4: Price
For a financial company, owner earnings approximate net income: $10.185 billion for 2025, against reported net income of $10.1 billion. Attributing a “certain growth” multiple of 15x — supported by the 2026 EPS guidance and 30 quarters of double-digit fee growth — with a conservative 12x floor gives an intrinsic value range of $122 billion to $153 billion.
The market cap stands at $244.6 billion — a 60% premium to the top of that range. The P/E is 22.2, and the owner-earnings yield of 4.2% sits below the 10-year Treasury’s ~4.3%. No margin of safety. The practical re-evaluation trigger begins around a $150–160 billion market cap (an owner-earnings yield above 6.5%), a level reached historically only in crises of the 2020-COVID or 2023-banking-crisis variety.
Verdict
Watch. Gates 1 through 3 pass with distinction — the highest quality of business in the coverage set. Gate 4 fails: the market is already paying more for “Buffett’s Amex” than Buffett’s own discipline would pay. A fine company, at not a fine price.
What would change it: entry into the $150–160 billion market-cap range, which would reopen the valuation question — or, on the risk side, a sustained premium-card war of attrition led by Capital One–Discover that pushes rewards costs above fee revenue growth.
This analysis is AI-generated, educational, and not investment advice. Figures may contain errors or be delayed. Disclaimer