Amazon Is Now Selling Its AI Shopping Brain to Its Competitors

TL;DR

• Amazon is licensing its Alexa for Shopping AI technology to other retailers via AWS, allowing them to launch branded AI shopping tools "in as little as 60 days."
• Kate Spade (Tapestry) is the first named customer, with additional retailers "currently in testing."
• This is the AWS playbook, applied to AI commerce. Amazon takes infrastructure built for itself, productises it, and sells it to competitors — who become dependent on Amazon's stack.
• Google launched its competing Universal Cart at I/O on 19 May, creating a two-horse race for agentic commerce infrastructure. Both companies want to own the layer between shoppers and merchants.
• The strategic question for every retailer: build your own AI shopping experience, partner with OpenAI/Google, or buy from Amazon — the company that also competes with you for every transaction.

What Happened

On Wednesday 27 May, Amazon announced it is packaging the "architecture, starter code and learnings" from Alexa for Shopping — its recently rebranded e-commerce AI agent — and offering it to other retailers through Amazon Web Services. The service lets retailers deploy AI shopping tools tailored to their own storefront, catalogue, and branding, with Amazon claiming a deployment timeline of "as little as 60 days." ¹

The first named customer is Kate Spade, the luxury fashion brand owned by Tapestry, which has used the service to launch a gifting assistant. Amazon says additional retailers are "currently in testing." ¹

This is the same playbook Amazon used with AWS roughly two decades ago: take internal infrastructure built to solve Amazon's own problems, productise it, and sell it to other companies — including direct competitors. Amazon later repeated the pattern with cashier-less checkout technology (Just Walk Out), warehousing, and supply chain services. ¹

The announcement comes two weeks after Amazon rebranded its e-commerce chatbot from Rufus to Alexa for Shopping and enabled it by default in search queries on its own store. ¹

What It Actually Means

The AWS Playbook, Applied to AI Commerce

This is not a product launch. It's a market-structure move.

When Amazon launched AWS in 2006, the pitch to startups was simple: don't build your own servers, rent ours. The pitch to enterprises took longer to land, but the economics were inexorable. By 2025, AWS was a $100-billion-plus revenue business built on the proposition that even Amazon's competitors should run on Amazon's infrastructure.

The AI shopping licensing play is structurally identical. Amazon is saying to retailers: you can build your own AI shopping agent, or you can license ours — the one we built for ourselves, battle-tested on hundreds of millions of shoppers, and now available with your branding on top.

The 60-day deployment timeline is the hook. Building a competent AI shopping agent from scratch — one that handles natural-language product discovery, comparison, personalisation, and transaction — is a multi-year engineering project. Amazon is offering to compress that to two months.

The cost is not just the licensing fee. The cost is that your AI shopping stack — the layer that mediates between your customers and your products — is owned by Amazon.

Why AWS, Not Amazon Retail

Amazon is routing this through AWS rather than its retail division. That's deliberate.

Retailers who would never share customer data or sales data with Amazon-the-marketplace might be willing to buy infrastructure from AWS, which has a two-decade track record of not stealing customer data from its cloud clients. The AWS brand provides a firewall — real or perceived — between the infrastructure provider and the competing retailer.

Whether that firewall holds under pressure is an open question. But it's the same question Netflix asked before moving to AWS, and Netflix moved anyway.

The Competitive Landscape Just Shifted

The AI shopping space now has three lanes:

1. Build your own — Walmart, Target, and eBay have taken this route, building proprietary AI shopping tools.
2. Partner with an AI platform — OpenAI, Google, and Perplexity have all rolled out shopping agents. Retailers can integrate with these platforms, but they cede the customer relationship to the AI intermediary.
3. License from Amazon — the new option. You get Amazon's technology with your branding, but you're buying from the company that competes with you for every transaction.

Amazon's pitch explicitly targets option 2. In Wednesday's blog post, Amazon warned retailers against "relinquishing control of the shopping experience to an intermediary," arguing that "retailers already possess deep vertical knowledge about their products, customers, and categories that no general-purpose AI can match." ¹

This is Amazon positioning itself as the retailer-friendly AI partner — the one that lets you keep your brand, your customer relationship, and your data. It's a clever frame, and it's not entirely wrong.

The Google Countermove

This story cannot be read in isolation. On 19 May, at Google I/O, Google announced three commerce components that together constitute a competing vision for agentic commerce: ²

• Universal Cart — a persistent, AI-powered shopping cart that follows shoppers across Google properties (Search, Gemini, YouTube, Gmail), tracking products, offers, and prices, and ultimately completing transactions.
• Universal Commerce Protocol (UCP) — the infrastructure layer that tells Google how to interact with merchants for shopping, checkout, and fulfilment.
• Agent Payments Protocol (AP2) — a payment layer that lets AI agents complete purchases on a shopper's behalf, subject to user-defined rules and limits.

Google's Universal Cart is scheduled for US release in summer 2026. ²

The two visions are mirror images:

• Google's model: the cart lives above the merchant. Shoppers add products from anywhere — Search, YouTube, Gmail — into a Google-owned cart. Google monitors prices, recommends alternatives, and completes the purchase. Merchants still fulfil the order, but they don't own the purchase intent or the product discovery. ²

• Amazon's model: the AI lives on the merchant's site. The retailer owns the cart, the branding, and the customer relationship. Amazon provides the AI infrastructure underneath.

In Google's world, retailers become fulfilment endpoints. In Amazon's world, retailers keep the storefront but run it on Amazon's AI rails.

Both models remove the merchant from the centre of the shopping experience. They just do it from different directions.

Stakeholder Landscape

Cross-Layer Implications

Security & data. Every AI shopping agent needs access to product catalogues, pricing, inventory, and — eventually — customer behaviour data. If Amazon's AI shopping stack becomes the default for mid-market retail, Amazon gains visibility into a significant portion of non-Amazon e-commerce data. The AWS firewall is real, but the data flows through Amazon infrastructure.

Regulatory. The EU's Digital Markets Act and the UK's Digital Markets, Competition and Consumers Act are both designed to prevent gatekeeper platforms from using infrastructure dominance to advantage their own retail operations. Amazon routing this through AWS is partly a regulatory strategy — it's harder to argue that AWS infrastructure creates a retail conflict of interest. But regulators are getting more sophisticated about structural separation.

Platform dynamics. This is a platform play in the classic sense: Amazon is building a two-sided market where retailers are the customers and Amazon's AI is the platform. If it works, Amazon captures value from every AI-assisted shopping interaction — even on competitors' sites.

Talent. The AI shopping race is creating demand for engineers who understand both large language models and e-commerce infrastructure. That's a narrow pool, and Amazon, Google, Shopify, and OpenAI are all fishing in it.

What This Means for You

If you run an e-commerce operation (mid-market, $10M–$500M revenue):
You now have a genuine build-vs-buy-vs-partner decision on AI shopping. Amazon's 60-day deployment claim is aggressive but credible given the underlying technology. The questions to ask: (1) Do you trust AWS with your catalogue and customer interaction data? (2) Does your AI shopping experience need to be differentiated, or is "good enough and fast" the right answer? (3) What happens to your negotiating position if Amazon's AI becomes the default and you want to switch later?

If you're on Shopify or a similar platform:
Watch Shopify's AI roadmap closely. Shopify cannot afford to let Amazon become the default AI shopping layer for its merchants. Expect accelerated AI feature releases from Shopify in the next two quarters. If you're evaluating platforms, AI shopping capability is now a first-order selection criterion.

If you're a consumer:
Nothing changes today. But the infrastructure being laid down now will determine whether your shopping experience in 2028 is a choice between merchant websites or a choice between AI agents that shop on your behalf. The convenience is real. The concentration risk is also real.

If you're an investor:
Amazon's move is a long-term platform play that won't move revenue materially for several quarters. The signal is strategic: Amazon intends to own the AI commerce infrastructure layer the way it owns cloud infrastructure. Google's Universal Cart is the counter-bet. Watch retailer adoption rates and — critically — whether any major retailer (top-50 e-commerce) signs on.

Uncertainty Ledger

• Adoption beyond Kate Spade. One luxury brand is a proof of concept, not a trend. The "currently in testing" retailers are unnamed. Until a major mid-market or enterprise retailer signs publicly, the scale of adoption is speculative.
• Data governance. How Amazon handles retailer data through the AWS AI shopping service is not yet detailed. The terms of service will determine whether retailers' concerns are addressed or validated.
• Google's Universal Cart timeline. "Summer 2026" could mean June or September. The earlier it launches, the more pressure on Amazon's licensing model.
• Consumer behaviour. It is not yet clear whether shoppers want AI agents to complete purchases on their behalf. Early data from Shopify and BoF/McKinsey suggests AI-assisted product discovery is growing rapidly (4,700% increase in shopping-related generative AI searches between 2024 and 2025), but AI-completed transactions are still nascent. ³
• Regulatory risk. Neither the EU nor the UK has specifically addressed AI-commerce infrastructure in competition regulation. That will change if one or two providers become dominant.

Bottom Line

Amazon is doing to AI shopping what it did to cloud computing: taking the infrastructure it built for itself and selling it to competitors. The pitch is compelling — deploy in 60 days, keep your brand, don't cede control to Google or OpenAI. The cost is that your AI shopping brain lives inside Amazon. Google's Universal Cart is the counter-offer: let Google own the cart and the discovery, and you just fulfil. Both models remove the merchant from the centre. The difference is who owns the relationship — and whether retailers are willing to buy their AI future from the company they've spent two decades competing against.

Sources: