Amazon Blocked ChatGPT. Now It’s Investing $50 Billion in OpenAI. Amazon recently announced it will invest $50 billion in OpenAI, alongside a broad partnership that includes OpenAI running massive workloads on AWS and using Amazon’s custom AI chips. On the surface, this reads like a straightforward infrastructure deal. OpenAI needs enormous amounts of compute to run and train its models, Amazon has one of the largest cloud platforms in the world, and both companies benefit from the arrangement. But the partnership becomes more interesting when you remember something that happened not long ago: Amazon blocked ChatGPT from accessing its product catalog. If AI assistants could freely crawl Amazon listings, they could recommend products without sending shoppers to Amazon at all. The decision about what to buy would happen inside the AI, not inside Amazon’s marketplace. For a company whose entire business depends on owning the moment of purchase, that was an obvious threat.
So why invest billions in the same company building that AI interface? The answer becomes clearer if you think about how commerce might work when AI agents start acting on behalf of users. Imagine asking an assistant to “buy more dog food” or “find a good espresso machine under $400.” The AI compares products, reads reviews, evaluates options, and decides which item to purchase. But once the decision is made, the assistant still needs somewhere to execute the transaction. It needs a reliable catalog of products, a payment system, shipping, logistics, customer service, and returns. There are very few places on the internet that can handle that process at global scale. Amazon already does.
That is the quiet strategic angle behind the deal. Amazon does not necessarily need to control the AI that recommends products. It just needs to make sure that when AI decides what to buy, the purchase still happens through Amazon’s commerce infrastructure. The press release hints at this future in a few places: OpenAI will run large-scale AI systems on AWS, and the two companies will build something called a stateful runtime environment, which essentially enables AI agents to remember context and perform tasks across systems. One of the most natural tasks those agents will eventually perform is buying things. If those agents are built inside Amazon’s infrastructure ecosystem, the path from “decide” to “purchase” naturally leads back to Amazon.
Today shoppers scroll through pages of listings and compare dozens of options before deciding what to buy. In an AI-driven interface, that browsing stage may shrink dramatically. The system might evaluate hundreds of products behind the scenes but present the shopper with only one or two recommendations. In that world the challenge shifts slightly. It is no longer just about appearing somewhere on page one of search results. It is about becoming the product the system feels confident recommending. The signals that matter are largely the same ones Amazon has always rewarded: strong reviews, reliable fulfillment, low return rates, clear product data, and consistent customer satisfaction. AI simply compresses the decision process that already happens inside it.
Seen this way, the Amazon–OpenAI deal looks less like a bet on a particular AI company and more like a familiar Amazon strategy. The interface for discovery may change, moving from search bars to AI assistants. But Amazon intends to remain exactly where it has always been: the system that actually completes the purchase.
Lumian is your extended team of dedicated human experts + AI employees that help you scale much faster on Amazon using AI.
In this week’s issue:
Marketplace: Amazon Canvas, Walmart Growth and Amazon x Meta/Snap
Seller Central Update of the Week: BuyerEmail deprecated
Tweet Spotlight: Amazon + OpenAI
Marketplace Madness
Amazon has introduced a new AI-powered “canvas” inside Seller Central that automatically builds interactive dashboards using a seller’s own data. Instead of manually pulling reports, sellers can ask questions like “How are my products performing?” or “What should I restock?” and the system generates a visual workspace showing sales trends, inventory risks, marketing performance, and recommended actions. The tool runs on Amazon’s Seller Assistant AI and updates in real time, allowing sellers to explore scenarios such as changing ad spend, adjusting prices, or restocking inventory while instantly seeing the projected impact.
Why it matters:
Sellers who learn to use these AI prompts will likely move faster on inventory, ads, and product expansion, while others may fall behind simply because they’re still pulling spreadsheets. It also signals where Amazon is heading as the platform guiding operational decisions directly inside Seller Central.
Walmart’s ecommerce business crossed $150 billion in annual sales for the first time, with online revenue growing more than 20% in Q4. Online now represents about 23% of Walmart’s sales, driven by faster delivery, heavy use of stores as fulfillment hubs, and expansion of higher-margin businesses like advertising and memberships. Walmart is also leaning into AI tools like its shopping assistant “Sparky,” which helps customers discover products and is already linked to higher order values.
Why it matters:
Walmart is quietly becoming a serious ecommerce platform. Faster delivery, rising ad spend, and AI-driven shopping suggest the marketplace will look increasingly similar to Amazon in how traffic and discovery work. For brands, this means Walmart is moving from a secondary channel to one that could meaningfully drive growth and deserves real attention.
Amazon is pushing deeper into social commerce through partnerships with Meta and Snap that allow shoppers to buy Amazon products directly from ads inside social apps. Instead of clicking an ad and leaving the platform, users can now complete a purchase from Amazon while staying inside apps like Instagram and Facebook. The goal is to blend social discovery with Amazon’s checkout and product catalog, essentially letting people discover products through social feeds and convert instantly without the usual multi-step funnel. Behind the scenes, the partnership also gives Amazon and the social platforms richer first-party data to improve ad targeting and measurement.
Why it matters:
The top of the funnel is moving outside Amazon, and this integration makes it easier for ads on social platforms to convert directly into Amazon sales. Brands that combine strong content, paid social, and Amazon retail media will have a clear advantage as discovery increasingly happens before shoppers ever open the Amazon app.
Seller Central Update of the Week
Here’s the link. Amazon quietly removed the final layer of buyer identity available to sellers this week. Obfuscated buyer emails have now been eliminated, meaning sellers no longer see any identifiable customer information, only aggregate performance data. The change applies across all Amazon marketplaces globally.
The implication is clear. Amazon fully owns the customer relationship. Sellers operate more like suppliers inside a closed retail ecosystem, able to transact at scale but without direct access to the buyer. If you want control over demand and customer insight, it increasingly has to be built outside the Amazon marketplace.
Tweet Spotlight
Meme Therapy
How did you like this week's newsletter?
Best,