Shopping Inside AI: Where It Stands, How to Win, and Who Takes the Market

Somewhere between 33% and 83% of shoppers used AI to help with their holiday shopping in 2025, depending on which survey you trust. The range tells you something useful: the behavior is real and growing, but the measurement is still messy. What is not messy is the direction. AI-assisted product discovery is becoming a primary channel, and the question of whether to transact inside the AI interface — rather than clicking through to a merchant site — is the defining commerce question of 2026.

The answer, as of today, is more complicated than the headlines suggest. ChatGPT launched Instant Checkout with enormous fanfare in September 2025. By March 2026, it had scaled back. Perplexity charges merchants nothing and has built a working hybrid checkout. Google has quietly assembled the most complete commerce infrastructure of any AI platform, backed by a coalition of every major retailer and payment network on the planet. And Amazon, which invested $15 billion in OpenAI, has blocked ChatGPT crawlers from accessing its product pages.

This is not a story about one winner. It is a story about four distinct platforms, each with different economics, different audiences, and different bets on how AI commerce actually works.

The Platform Landscape

The four platforms merchants need to understand are ChatGPT, Perplexity, Google AI Mode and Gemini, and Amazon. Each operates differently enough that treating them as a single channel is a strategic mistake.

**ChatGPT** has the largest user base of any AI platform, with 810 million daily active users as of November 2025. Its Agentic Commerce Protocol (ACP), co-developed with Stripe, is an open standard that lets AI agents complete purchases on behalf of users. The vision was compelling: a shopper asks a question, gets a recommendation, and buys without leaving the chat. The reality was that only around 12 of Shopify's millions of merchants ever went live with Instant Checkout before OpenAI pivoted in March 2026. The bottleneck was not merchant willingness — Shopify president Harley Finkelstein confirmed the hold-up was on OpenAI's side. The company had not even set up systems to collect and remit state sales taxes, which tells you volumes never reached meaningful scale. ChatGPT now routes purchases through retailer apps (Instacart, Target, Expedia) rather than native product listings. The 4% transaction fee remains. The merchant dashboard does not exist yet.

**Perplexity** took the opposite approach. Zero listing fees. Zero commissions. Zero transaction fees. Merchants keep 100% of revenue from every sale. Perplexity eliminated all advertising in 2026 and is betting its business model entirely on user trust and its Pro subscription. The feed spec requires a Google Shopping CSV delivered via SFTP, with GTINs mandatory — products without UPCs simply do not appear. Real-time pricing accuracy matters heavily because Perplexity cross-references your feed against live site data and suppresses listings with price mismatches. Onboarding takes four to six weeks. For merchants who qualify, the economics are straightforward: it is the only AI shopping channel with no downside cost.

**Google AI Mode and Gemini** are where the most complete infrastructure lives. Google's Universal Commerce Protocol (UCP), launched in January 2026, is an open standard backed by Shopify, Walmart, Target, Etsy, Wayfair, Visa, Mastercard, Stripe, Adyen, Best Buy, Macy's, and The Home Depot. That coalition is not a coincidence. Google has spent 20 years building Merchant Center relationships with every major retailer on earth, and UCP is the protocol that converts those relationships into AI-native commerce. For merchants already running Google Shopping, AI Mode is an additional placement surface that costs nothing incremental. Gemini goes furthest of any platform: it can browse products, compare options, check real-time inventory, and complete purchases via Google Pay without the shopper leaving the interface. It even supports conditional purchases — buy this if the price drops below a threshold. The analytics are the most mature of any platform, reported through Google Ads and Search Console.

**Amazon** is the wildcard. It has the largest product catalog of any retailer on earth, has invested $15 billion in OpenAI, and has explicitly blocked ChatGPT crawlers from accessing its product pages, prices, and reviews. Amazon is building its own AI shopping experience through Rufus, its conversational shopping assistant. The company that controls the most product data in the world has decided not to share it with the AI platforms trying to displace traditional search. That decision will either look prescient or catastrophic in three years.

Why ChatGPT Checkout Failed

The three structural problems that blocked ChatGPT Instant Checkout are worth understanding because they apply to every protocol-based approach to AI commerce.

The first is merchant adoption lag. ACP requires merchants to explicitly opt in, integrate their catalogs, and maintain their product data for AI surfaces. Even with Shopify handling the heavy lifting for its merchants, adoption stalled at a handful of large brands. Small and mid-market merchants lack the engineering resources to justify the investment when agent-driven transaction volumes are still negligible.

The second is product data standardization. Merchant product information — pricing, availability, shipping options — needs to be standardized and constantly updated for chatbots to access accurate data. Static catalog feeds go stale. Prices change. Items go out of stock. Any system that relies on pre-ingested catalog data will always be working with information that is at least partially wrong.

The third is fraud and trust safeguards. Merchants and payment firms need safeguards against AI initiating fraudulent or erroneous transactions. Protocol-based systems address this through explicit merchant consent, but that leaves the vast majority of merchants — those who have not opted in — completely out of reach.

Google UCP faces the same structural challenges, but it has two advantages ChatGPT does not. First, it can leverage Merchant Center existing data infrastructure, which already handles real-time inventory and pricing for billions of product listings. Second, its coalition of payment networks and retailers means the protocol has institutional backing that ACP lacks.

Who Will Win

The honest answer is that Google is best positioned to win AI commerce at scale, for reasons that have nothing to do with the quality of its AI.

Google has existing data relationships with every major retailer. It has the payment infrastructure via Google Pay. It has the consumer trust built over two decades of Shopping. It has the most mature analytics of any AI platform. And it has UCP, backed by a coalition that includes every major payment network and most of the largest retailers in the world. The question is not whether Google can win — it is whether antitrust regulators will let it.

ChatGPT has the largest user base but has demonstrated that user base alone does not translate to commerce volume. The checkout pivot is a significant signal: 810 million daily active users and fewer than a dozen merchants live on Instant Checkout is not a commerce business. It is a discovery business with commerce ambitions.

Perplexity has the cleanest merchant economics of any platform and a user base that skews toward high-intent research queries. The zero-fee model is a genuine differentiator. But Perplexity lacks the payment infrastructure, the retailer relationships, and the consumer habit formation that Google has. It is the best early bet for merchants who want to experiment at zero cost.

Amazon will not lose the commerce war. It will fight it on its own terms, with its own AI, on its own platform. The question for brands that sell on Amazon is whether they need to be present in AI platforms that compete with Amazon, or whether Amazon own AI discovery is sufficient. For most consumer brands, the answer is both.

What Brands Should Do Right Now

The practical priority stack for any brand that sells products is straightforward.

Google Merchant Center is table stakes. If you are not already optimized for Google Shopping, start there. AI Mode is an additional placement surface on top of existing Shopping infrastructure, so the work compounds. Add conversational attributes to your product descriptions — answers to common questions, compatible accessories, intended use cases — because these are the fields AI Mode rewards that standard Shopping feeds do not require.

Submit to Perplexity merchant program. It is free, the onboarding is manageable, and the zero-fee economics mean there is no downside. GTINs are mandatory, so ensure your catalog has clean UPC data before applying. Real-time pricing accuracy is critical — Perplexity suppresses listings with price mismatches, so automated feed updates are worth the investment.

Apply to ChatGPT merchant program at chatgpt.com/merchants. The checkout pivot means native transactions are less likely in the near term, but product discovery through ChatGPT is real and growing. Being in the merchant program positions you for whatever commerce infrastructure OpenAI builds next.

Add Schema.org Product markup to every product page. All four platforms use structured data as a signal for product discovery. This is the single highest-leverage technical action for AI commerce visibility across all platforms simultaneously.

For Shopify merchants specifically: enable Shop Pay and ensure your Google channel integration is active. Shopify Agentic Storefronts auto-syndicate your catalog to ChatGPT. The Google channel feeds Merchant Center, which feeds AI Mode and Gemini. The integration work you do once propagates across multiple AI surfaces.

The brands that win AI commerce will not be the ones who waited to see which platform survived. They will be the ones who were already present on all of them when the volume arrived. The volume is arriving now.