ChatGPT Just Became the World's Biggest Storefront — And OpenAI Isn't Running the Register

Instant Checkout launched on September 29, 2025, with U.S. Etsy sellers live and over a million Shopify merchants — including Glossier, SKIMS, Spanx, and Vuori — promised as 'coming soon.' Six months later, Shopify president Harley Finkelstein confirmed that the bottleneck was on the AI side: only 12 Shopify merchants had actually integrated. On March 24, 2026, OpenAI retired the product and pivoted to 'merchant-owned checkout.'

The September 29 keynote turned ChatGPT into a 4,000-app surface — but the commerce layer OpenAI actually built is upstream of checkout. The headline launch was Dots: always-on agents powered by GPT-6 Astra, each running on a dedicated cloud computer, each connecting to the plugin ecosystem that already spans Shopify, HubSpot, Salesforce, and 3,997 other apps.

Amazon captured the shopping agent war before ChatGPT joined it. Rufus, launched in 2024 and renamed Alexa for Shopping on May 13, 2026, already reaches more than 300 million active customers and is delivering an estimated $12B in annualized incremental sales. OpenAI's September announcements barely changed that — Amazon's ad-supported, retail-owned agent was already in the search bar where ChatGPT is now trying to insert itself.

The mechanism that lets Amazon win is one OpenAI cannot copy: every Alexa for Shopping interaction ends inside Amazon's own storefront, where Amazon captures both the 1P margin and the high-margin ad click. Customers who use the assistant are 60% more likely to complete a purchase in the same session, and the assistant has driven monthly active users up 149% over the past year.

The Instant Checkout retreat ran Shopify's catalog while OpenAI ran discovery — and Shopify's merchants almost universally voted with their feet. Yet the parent company's Q2 2026 numbers, reported August 5, 2026, show the rest of the business is stronger than the agentic-commerce narrative suggests.

Shopify Catalog — the integration that puts every Shopify merchant's product feed into ChatGPT without per-merchant work — is now the practical answer to the Instant Checkout retreat. The 12-merchant ceiling on native checkout does not constrain catalog exposure, and management is investing behind Agentic Commerce Protocol tools that let merchants compete for attention inside ChatGPT without paying the 4% take rate.

Walmart built the right defensive moat before OpenAI arrived — and the Q2 FY2027 numbers, reported August 20, 2026, show it is paying off. Global e-commerce grew 23%, the 10th consecutive quarter of 20%+ growth, while the company's Sparky AI shopping agent is up 70% year over year inside the Walmart app. Crucially, Walmart built its own ChatGPT app — Sparky-powered, with account linking, loyalty, and Walmart Payments — that turns ChatGPT into a discovery front end for the Walmart storefront rather than a checkout OpenAI controls.

The strategic lesson: Walmart does not pay OpenAI a take rate, owns the customer identity, and uses ChatGPT as a top-of-funnel for its own checkout. That is the architecture OpenAI's other big partners — Sephora, Target, Best Buy, Home Depot, Lowe's, Wayfair — are now cloning via the Agentic Commerce Protocol.

The brief's claim that ChatGPT 'quietly added a virtual try-on feature' after DevDay does not match the verified record. The only live try-on inside ChatGPT is makeup — a L'Oréal/Maybelline partnership announced June 17, 2026 at VivaTech and rolling out across the U.S. over the summer using L'Oréal's ModiFace technology. The widely reported clothing virtual try-on remains in development as of late August 2026, per leaked UI references and no OpenAI confirmation.

The brief's framing — that ChatGPT collapses the value chain for Amazon, Shopify, Inditex, and DTC peers — overstates what actually changed. OpenAI built the rails for every future shopping agent but is no longer trying to be the storefront itself. The new commerce stack has four layers, and each is being captured by a different winner.

OpenAI's 4% take rate was the original sin of the first agentic-commerce stack: it asked merchants to fund a parallel checkout on top of their own. The pivot to merchant-owned checkout fixes that — but it also means OpenAI is upstream of merchants, not competing with them. That is good for Amazon and Walmart, manageable for Sephora and Target, and a margin leak for the apparel DTCs that have to feed ChatGPT a catalog and watch the basket close somewhere else.

The agentic-commerce trade is two horizons stacked on top of each other. Short-term catalysts will print on retailer earnings calls through Q4 2026; the structural call is whether merchants accept the new economics by 2028.