Should You Forget Upstart and Buy American Express Instead?

Upstart, an AI-driven lending platform that has originated roughly $50 billion in loans and automated 91% of loans in Q3, saw revenue fall 39% in 2023 and a $240 million net loss that year but reported 71% top-line growth in the most recent quarter and two consecutive GAAP profitable quarters. American Express reported Q3 2025 revenue (net of interest expense) up 11% year-over-year to $18.4 billion, driven by 9% payment-volume growth and 5.7 million net new active cards, with management guiding to $15.35 diluted EPS (midpoint), ~10% above 2024; the stock trades at a P/E of ~25.3 and benefits from a large network (151 million cardholders, 160 million merchant locations). The piece positions Upstart as higher-risk/higher-reward exposure to AI-driven credit expansion while framing AmEx as a stable, premium franchise for risk-averse investors.

Analysis

Market structure: AI-driven lenders (UPST and its bank partners) are the primary beneficiaries as improved underwriting expands credit to thin-file borrowers, potentially raising originations by 20–50% in addressable niches over 2–3 years. Incumbent players that rely on FICO-only models will lose pricing power in those segments while payment networks (AXP) benefit from higher spend and premium cardholder cohorts, supporting stable APRs and merchant fees.

Risk assessment: Principal tail risks are regulatory/model risk (CFPB or EU-style AI audits), a macro credit shock that spikes delinquencies >300–400 bps, or model drift leading to higher loss severity; each could erase >50% of UPST equity value in 6–12 months. Near term (days–weeks) expect volatility around Fed moves and quarterly prints; medium term (3–12 months) performance hinges on funding costs and charge-off trends; long term (2–5 years) on sustained automation adoption and bank partner economics.

Trade implications: Tactical allocation favors defensive payments exposure (AXP) for income and low single-digit EPS growth risk, and a small, defined‑risk speculative allocation to UPST for optionality. Use pair and option structures to express views: buy AXP and buy capped UPST upside via call spreads to limit downside while monetizing AXP with OTM covered calls or cash‑secured puts.

Contrarian angles: Consensus underestimates how quickly AI can reprice marginal credit; if UPST sustains >50% y/y originations and two straight GAAP‑profitable quarters, equity could re-rate materially. Conversely, the market may be underpricing regulatory risk — implied vol is high, so prefer defined‑risk structures (spreads) over naked directional exposure to UPST.

Key Decisions for Investors

• Establish a 2–3% portfolio long position in American Express (AXP) over the next 2–6 weeks; if P/E compresses to <22 (price ≲ $338 assuming EPS $15.35) scale to 4–6%. Hedge by selling 3‑month covered calls at +8–12% OTM to generate ~2–4% incremental yield per quarter.
• Allocate a 0.5–1.0% position-size speculative trade in Upstart (UPST) using a 12‑month defined‑risk bull call spread (buy 2026 LEAP ~ATM, sell 150–200% OTM depending on premium) and cap max loss at the premium paid. Liquidate or re-evaluate if UPST fails to: a) post 2 consecutive GAAP‑profitable quarters, or b) shows >20% q/q decline in originations within 6 months.
• Implement a pair trade: long AXP (notional X) and short UPST (notional ~0.25X–0.5X to reflect higher beta) for 6–12 months to capture spread compression between stable payments earnings and high‑beta fintech risk. Close if relative performance reverses by >15% in 30 days or on major regulatory developments.
• Reduce generic high‑beta fintech exposure by ~25% and reallocate to payments/large‑cap banks (AXP, BK, JPM) within 1 month to lower portfolio volatility; monitor CFPB/SEC AI rule announcements for 90 days and raise cash if a supervisory action targets model‑based underwriting.