American Express Stock Is Cheap, But Does That Make It a Buy Now?

American Express (AXP) is trading down ~21% YTD versus the S&P 500's ~4% decline, but the article argues the stock is undervalued given durable moats. 2025 revenue rose ~10% to nearly $19B and net income jumped ~13% to ~$2.5B (13% net margin); analysts forecast ~9% revenue growth and ~14% EPS growth this year. The primary risk discussed is AI agent-driven price hunting and use of low-cost payment rails (e.g., stablecoins) that could erode fee revenue, but the author cites rewards programs, debt/credit convenience, and brand prestige as resilient defenses and concludes AXP is a buy.

Analysis

AI-driven price agents create a credible threat to interchange revenue only if three things happen in sequence: (1) mass consumer adoption of agentic shopping, (2) merchant acceptance of alternative low-fee rails at scale, and (3) seamless fiat on/off ramps that are cheaper than card rails. Each link has different incumbency advantages — banks and card networks own the easiest-to-integrate rails and the richest identity/rewards datasets, which materially raises the switching cost for agents that want to deliver frictionless checkout.

Second-order winners are tokenization and API orchestration providers inside the card ecosystem: firms that can monetize routing decisions by offering agent-friendly, issuer-endorsed discounts will capture newly created margin without destroying interchange. Conversely, direct-to-merchant crypto rails and standalone stablecoin routing face a capital-conversion and trust problem — merchants must absorb FX/settlement friction or pay intermediaries, creating a persistent niche rather than a universal substitute.

Regulatory and timing dynamics matter: expect pilot-driven fee pressure localized to travel, digital subscriptions, and high-ticket e-commerce within 6–18 months, but broad retail defection will take multiple years unless a dominant agent standard emerges. The most credible path to material revenue loss for issuers is regulatory intervention (interchange caps or mandated routing) triggered by visible consumer harm — a tail risk that could crystallize inside 2–4 years rather than overnight.

Key Decisions for Investors

• Long AXP (6–12 months): establish a 2–3% portfolio position via a covered-call or long-call-spread to capture premium-card pricing power and rewards stickiness; hedge with a 10–15% OTM put or pair against regional banks. Rationale: issuer data and rewards create durable switching costs; target asymmetric upside if market re-rates issuer moats.
• Long V or MA (9–18 months): buy 12–18 month call spreads (moderate delta) sized to 1–2% portfolio each. Reason: networks can monetize agent integration and tokenization services; expected returns skew positive as they convert routing control into SaaS-like fees.
• Relative-value: long AXP / short JPM (6–12 months): 0.5–1% net exposure (long AXP outsized). JPM is more exposed to consumer banking cyclical risk and co-branded card fee pressure. Use this to isolate issuer premium vs. bank balance-sheet cyclicality; cut if macro liquidity spreads widen >50bp.
• Tactical long NVDA (12 months) as a macro hedge: 1% position in long-dated calls to capture upside if agent adoption accelerates AI compute demand. This is an insurance trade—high optionality vs modest capital outlay—benefits if agent proliferation becomes a rapid compute consumption story.