AI Agents: What’s the risk for SMID eCommerce? By Investing.com

Morgan Stanley pivots to valuing e-commerce marketplaces against logistics peers (~13x P/E on 2027 estimates) and calls logistics an "underappreciated moat" amid agentic AI disruption. The bank names Allegro a Top Pick with a PLN 38.00 target (≈38% implied upside), trims Kaspi.kz net income estimates by up to 5% for 2026-28, and raises Zalando adjusted EPS for 2026/27 by 4-5% driven mainly by a buyback. Analysts warn agentic AI could raise paid-traffic costs vs. organic reach and squeeze platform margins, leaving sector profitability and long-term structural margins uncertain.

Analysis

Agentic AI changes discovery economics: as assistants internalize intent, merchants will need to convert intent off-platform, pushing paid-traffic spend up to reclaim demand. We model a plausible 15–40% rise in effective CPMs and CPA over 12–24 months as merchants compete for the same finite clicks, which mechanically pressures marketplace advertising margins by a few hundred basis points unless LTV improves.

Physical logistics becomes a differentiator not because it stops disintermediation but because it expands optionality — platforms that own density and control fulfillment can monetize through higher take rates, bundling, and cost capture. Expect logistics capex-to-sales to rise for winners and for investors to re-rate those assets from software multiples toward industrial/asset-backed multiples over a 2–4 year window, compressing pure-play marketplace valuations in the process.

Immediate infra and adtech beneficiaries are the hidden suppliers of compute and programmatic spend: server vendors and mobile ad stacks will see secular demand even if platform economics get choppy. That creates a bifurcated market where capital-light marketplaces decelerate earnings growth while capital-heavy logistics operators earn steadier, lower-volatility cash flows.

The contrarian risk is that consensus overstresses physical moats and underweights brand/data moats: AI agents will prefer high-quality branded feeds and trusted merchant relationships, which could preserve advertising yields for top marketplaces. The clean test will be the next 6–18 months of AI agent rollouts and merchant CAC trends; a rapid rise in on-platform conversion or direct-checkout adoption would blunt the worst-case re-rating scenario.

Key Decisions for Investors

• Long SMCI (6–18 months): buy shares or 2027 LEAPS sized 2–3% NAV. Thesis: incremental AI infra orders drive revenue/take-rate expansion; target +40–80% if hyperscalers accelerate refresh cycles. Risk: order pull-ins and inventory destocking could remove 30%+ of upside — set 20% stop or hedge with 20% notional put protection.
• Long APP (AppLovin) (6–12 months): accumulate stock or buy 9–12 month calls (delta ~0.6) reflecting programmatic reallocation into app ecosystems. Reward ~30–50% if CPMs reprice upward and mobile monetization holds; downside ~25% on privacy/regulatory shocks — cap position sizing to 1–2% NAV.
• Pair trade (12 months): long select last-mile/logistics operators (e.g., UPS/FDX) vs short high-valuation pure-play marketplaces (e.g., ZALG or similar). Rationale: capture re-rating into asset-backed multiples; aim for asymmetric 1.5:1 reward:risk targeting +25% gross on long and -25% on short. Monitor same-store parcel volumes and merchant CAC as exit signals.
• Tactical hedge: long UBS (3–6 months) as a macro/flow hedge (2% NAV) into potential market re-rating and buyback-driven EPS upside. Expected payoff is modest (+15–25%) if markets re-risk; downside limited to ~15% in a flash sell-off — use tight stops or convert to covered calls after initial move.