What: AI shopping assistants threaten marketplaces' advertising and commission profits by taking over purchase decisions, without needing to replicate their fulfilment networks.

Why it is important: Department stores building retail media revenue face the same exposure as Amazon: if agents intercept the purchase decision, sponsored placement loses its value before sales do.

Marketplaces earned their customer relationships through operational reliability, and those habits now underpin advertising businesses. In 2025, Amazon generated $69 billion in advertising revenue against $34 billion in operating income excluding AWS; at an assumed 70% contribution margin, advertising contributed more than reported operating income. The same holds for DoorDash and Instacart, while Alphabet and Meta took nearly $500 billion in advertising revenue.

Assistants cannot rebuild those networks, but they can compare providers and route orders among them. Whether a platform welcomes that depends on two variables: how much incremental demand an agent brings, and how much operating income rests on commissions and advertising. Instacart, with room to win new grocery baskets, integrated Muse; Amazon, whose breadth makes any assistant that cannot buy from it less useful, blocked it. Shopify, Toast and Square lean in because they earn software and payments revenue without an ad business to defend.

The unresolved question is whether agents create enough new purchases to offset lower margin on orders that would have happened anyway. Airbnb's differentiated inventory gives it leverage, whereas Expedia may swap one acquisition cost for another, and a booking won through an assistant must be won again next time.

IADS Notes: The contest the article describes, over discovery rather than fulfilment, is already measurable at department-store level: at John Lewis, searches originating from AI agents grew from 0.3% to 2.5% of the total in a year, prompting investment in the influencer and video content that AI systems draw on (Reuters, September 2026). The profit pool at stake is the one retailers have been building most aggressively, with retail media now a major earnings engine for thin-margin players and, as a result, exposed to the warning that it risks becoming a money grab if ad revenue outweighs shopper relevance (The Robin Report, August 2026). Sponsored placement also loses its grip when the shopper is an algorithm: research shows only ratings and competitive pricing consistently influence AI-driven choices, while scarcity cues, countdown timers and strike-through pricing are ineffective or counterproductive (Harvard Business Review, May 2026). The bargaining dynamics the article sketches have precedent on both sides: OpenAI's withdrawal of Instant Checkout showed that consumers still prefer retailer-controlled environments for transactions, making data ownership and direct customer relationships the decisive lever as AI platforms mediate the path to purchase (Journal du Net, March 2026), while the 4% fee on OpenAI-powered checkouts for Shopify merchants was judged steep by some yet cheaper than Amazon's 8–15% referral fees by others, provided AI checkout lifts conversion and lowers acquisition costs (Internet Retailing, January 2026).

Who gets paid when AI does the shopping?