What does ChatGPT’s Agentic pivot mean for internet stocks?

40–55% of ChatGPT users already use AI to research products and 20–40% say they have made purchases based on AI, Morgan Stanley notes as ChatGPT pivots to an agentic model that directs users to merchants rather than transacting in‑platform. The bank says this could advantage large search platforms, major e‑commerce retailers and online travel sites (via catalogs, logistics and consumer data) and materially lower customer acquisition costs if single‑digit commission rates scale versus paid search, while vertical retail agents (e.g., Walmart's Sparky, Amazon's Rufus) may gain traction faster.

Analysis

The economic lever that matters here is not the AI front-end itself but the re-pricing of customer acquisition. If a merchant can shave 20–40% off paid-search and affiliate spend by routing discovery through lower-fee AI interfaces, expect near-term EBITDA tailwinds of ~50–150bps for high-volume, low-margin retailers (WMT/AMZN style) within 6–12 months as fulfillment fixed-cost absorption improves.

Second-order supply-chain effects will amplify winners and losers. Greater direct-to-merchant routing favors firms with fast last-mile capacity and tight inventory turns: a 2–5% uplift in fulfillment utilization can translate to ~30–80bps gross-margin improvement for integrated networks, while marketplaces that rely on third-party logistics face higher return rates and marginal working capital stress as discovery becomes more opaque.

Key risks are regulatory and measurement frictions that can flip incentives quickly. Two clear reversal catalysts in the 3–12 month window: (1) a major OS/search provider embeds superior in-line checkout or measurement (reclaiming CAC), and (2) merchants refuse to accept a new commission structure, forcing AI agents back to lead-gen rather than transaction routing.

The consensus underestimates how vertically-focused agents concentrate economic rents. Players with proprietary purchase histories and closed-loop fulfillment will capture the lion’s share of upside; conversely, horizontal adtech that commoditizes traffic is exposed. That bifurcation creates simple asymmetric trades into incumbents with logistics + data moats versus pure-play measurement/ad-tech intermediaries.

Key Decisions for Investors

• Overweight WMT (6–12 months): initiate a 6–8% portfolio allocation to WMT equity or buy 1.5x notional via 9–12 month call spread. Rationale: direct margin uplift from lower CAC and higher fulfillment utilization. Target: 12–18% upside; stop-loss: 8% or if company guidance cuts gross margin by >50bps on same-store metrics.
• Long AMZN (9–12 months) with upside collar: buy AMZN shares or 0.75x notional via 9–12 month calls and sell shorter 3–6 month calls to finance. Rationale: marketplace & ads optionality plus logistics scale. Target: 15–25% upside; hedge: cap loss to 8–10% via sold calls or protective puts if regulatory headlines escalate.
• Pair trade (3–9 months): long SMCI (infrastructure exposure) + short APP (adtech exposure). Size: 2–4% gross each side. Rationale: compute infra benefits from enterprise vertical agents while adtech loses CAC share. Risk/reward: asymmetric — buy-side optionality (calls on SMCI) vs cheaper puts or outright short on APP; unwind if ad spend re-accelerates or SMCI reports softer order book.
• Tactical hedge (0–6 months): buy out-of-the-money puts on a consumer discretionary basket or allocate 1–2% to cash if evidence emerges that AI routing is limited to search intent (i.e., no transaction flow). This limits downside from a scenario where incumbents re-aggregate demand and push CAC back onto merchants.