34 days ago Amazon announced the Amazon DSP will soon support ChatGPT Ads. Amazon stated a select group of customers were “currently testing ChatGPT ads” as part of a pilot. Since then, zero public updates. At unBoxed, Amazon Ads’ annual conference, ChatGPT Ads was barely mentioned.

What should we make of the lack of update? Will this partnership be merely another demand source for ChatGPT? Or will this alliance forge a new architecture of consumer ecommerce?

I think it will be the latter. And I think this partnership will illuminate Amazon’s strategy and define an ongoing playbook for how they partner with and embrace third-party consumer AI agents.

In those same 34 days, Amazon did make news with another consumer AI agent: it famously blocked Meta’s Muse. Their incentive to do so is simple. Ceding control of the customer experience to products like ChatGPT and Muse, presents a material downside risk to their advertising business.

The Risk To Amazon Ads

Amazon’s advertising business is roughly $80 billion, growing 20%+ annually. Based on an extremely rough estimate (note: zero sources from Amazon on this figure), approximately 65%, or $50+ billion, is generated from Sponsored Ads: pay-per-click ads on search and product pages, where marketers bid on keywords and product SKUs (aka ASINs) that appear natively on Amazon. Sponsored ads monetize purchasing intent and if agents are able to make browsing and purchasing decisions on Amazon that collapse the touchpoints on the path to purchase, then Amazon’s unique ability to profitably monetize purchasing intent may cease to exist. Every purchase or decision an agent makes is a click on a keyword or SKU that a human doesn’t make, and a touchpoint that Amazon no longer monetizes.

The financial implications are enormous. According to research firm Stratably, Amazon’s advertising business can be estimated to operate at roughly 40% margins. Applying that assumption to Amazon’s reported 2025 financials suggests advertising generated approximately $27.5 billion in operating profit, representing 80% of Amazon’s $34.4 billion in total operating income outside AWS. Put differently, an estimated four out of every five dollars of Amazon’s non-AWS operating profit comes from advertising. Preserving that business is central to the company’s enterprise value.

Despite Amazon’s perfectly economically rationale decision to block consumer agents in the near term, there is a clear generational shift in consumer behaviour unfolding before our eyes and Amazons knows it would be overly myopic to pursue long term incentive alignment with consumer agents.

Partnerships Are Inevitable

Over the past year, Amazon CEO Andy Jassy has repeatedly acknowledged that the company expects to partner with third-party AI shopping agents, while emphasizing the need for a mutually beneficial “value exchange.” Examples:

October 30, 2025 | Q3 2025 Earnings Call

“But we’re also having conversations with and expect over time to partner with third-party agents.”

Jassy also stated:

“We’ve got to find a way to make the customer experience better and have the right exchange value.”

Source: Amazon Q3 2025 Earnings Call Transcript, The Motley Fool

February 5, 2026 | Q4 2025 Earnings Call

“We will have relationships with third-party horizontal agents that can enable shopping as well.”

He qualified that expectation by emphasizing:

“A value exchange that makes sense for both parties.”

Source: Amazon Q4 2025 Earnings Call Transcript, StockAnalysis.com

April 29, 2026 | Q1 2026 Earnings Call

“We’re having conversations with all those folks to try and make that better and find something that works for customers and all the companies.”

Source: Amazon Q1 2026 Earnings Call Transcript

Jassy’s language has evolved from expecting partnerships in October 2025 to explicitly confirming ongoing conversations with third-party agents in April 2026. Yet the recurring qualification around a mutually beneficial “value exchange” raises a fundamental question: How does Amazon embrace agentic commerce without sacrificing the advertising economics that make its retail business so profitable?

The ChatGPT Partnership

Based on Jassy’s public remarks, a partnership between ChatGPT and Amazon was inevitable. What’s most striking is the timing, and the sequence of events that led to it.

On January 16, 2026, OpenAI announced plans to introduce advertising in ChatGPT, officially launching its advertising business on February 9. Just 203 days later, on August 31, OpenAI announced that its advertising business had reached a $1 billion annualized revenue run rate.

In that remarkably short period, OpenAI began building the foundations of a sophisticated advertising platform: creating entirely new, highly valuable advertising inventory, integrating third-party demand, improving targeting capabilities, and partnering with independent measurement providers. The speed with which OpenAI assembled this infrastructure underscores both the seriousness of its ambitions and the enormous commercial opportunity ahead.

That Amazon Ads announced a forthcoming partnership with ChatGPT Ads just 10 days after OpenAI disclosed its $1 billion run rate is unlikely to be a coincidence. The timing suggests advertising is precisely the “value exchange” Jassy has repeatedly named as a prerequisite for partnering with third-party AI agents. And it’s an exchange that could only exist once ChatGPT built a scaled advertising business.

The Value Exchange in Practice

The exchange works like this. OpenAI gets Amazon’s advertiser demand and its retail signals. Amazon gets a toll. When a brand buys ChatGPT inventory through the Amazon DSP, Amazon collects platform and audience fees on that spend. I’d bet it also negotiated a share of the media revenue itself. Amazon now has a mechanism to offset the erosion of its own advertising economics as discovery and purchasing decisions migrate toward AI agents. Rather than surrendering highly profitable advertising revenue to the platforms disrupting traditional commerce, Amazon is positioning itself to monetize purchase intent regardless of whether that intent originates on Amazon.com or inside ChatGPT.

Amazon is heavily incentivized to be more than another third-party demand source for ChatGPT Ads. Bringing its unparalleled retail signals, including ratings, reviews, pricing and inventory, into ChatGPT would enrich the commerce experience and strengthen both companies’ positions in consumer AI and ecommerce. The opportunity is to connect discovery and checkout seamlessly between the two products, with Amazon supplying the retail intelligence that makes those interactions more valuable.

Amazon is also uniquely positioned to capitalize on the new supply ChatGPT creates. Amazon Ads has a breadth of ad products that could extend naturally into the ChatGPT consumer experience. Some immediate examples:

• REC ads (Responsive eCommerce Creative): These package the information that helps sell a product on Amazon into a lower-funnel display ad: product imagery, live pricing, star ratings, Prime eligibility and promotional offers. The creative updates dynamically and automatically excludes out-of-stock products, bringing Amazon’s retail signals directly into the ad unit. Amazon’s newly announced carousel format makes the fit even more obvious: multiple products in a single, swipeable unit that could sit naturally alongside a ChatGPT shopping conversation. For endemic brands, this would extend a familiar advertising format into an entirely new source of purchase intent.

• Sponsored Prompts: These let brands answer the questions standing between a consumer and a purchase. Powered by Amazon’s first-party signals, prompts surface relevant product information directly on the page or open a conversation with Amazon’s shopping assistant. The reported performance is hard to ignore. On Amazon’s Q2 2026 earnings call, Andy Jassy stated:

“Shoppers who click a Sponsored Prompt convert to a sale 48% more often and spent 21% more on average than those who don’t.”

• Branded Conversations: Amazon is taking this a step further with Branded Conversations, announced at unBoxed and scheduled to launch in January 2027. These extend Sponsored Brands prompts into multi-turn conversations informed by a brand’s catalog, product guides, compatibility information and other supplied content. Advertisers pay for the initial conversation-starting click. Follow-up exchanges are included. The fit with ChatGPT is obvious. A consumer evaluating an espresso machine could ask about counter space, milk drinks and maintenance, with the brand participating throughout that decision. Amazon would bring the product knowledge and advertising infrastructure. ChatGPT would bring the consumer and the context of what they’re trying to buy.

Jassy calls this a “value exchange.” I’d call it an unprecedented marriage of agentic commerce and retail media, with the potential to not only protect Amazon’s downside risk in ads but transform consumer ecommerce behaviour for the next decade. My bet: within a year, 10% of every endemic brand’s advertising budget is spent on ChatGPT inventory. And, with improved conversion rates and Amazon’s ability to monetize DSP spend through platform and audience fees, it’s possible this partnership creates incremental advertising revenue for Amazon.

A Blueprint for Consumer Agents

What does this mean for personal AI agents like Muse and Instinct? If they have any aspirations of engendering e-commerce behaviour they need to sell ads first.

Look at what the ChatGPT deal is. It is an ads deal. Nothing announced lets ChatGPT’s agent browse and buy on Amazon. Ads come first. Shopping access comes second, if it comes at all.

The logic follows from the risk. An agent that shops on a user’s behalf takes clicks away from Sponsored Ads. Amazon will only accept that loss if the agent gives it somewhere else to monetize the same intent. ChatGPT now can. Muse can’t yet. But, given Meta’s world-class ability to monetize engagement, it will, very very fast.

OpenAI’s last eight months show what “can” requires:

• Scale. Advertisers don’t move budget to inventory that can’t absorb it. OpenAI got the call at a $1 billion run rate.

• Inventory open to outside buyers. Amazon collects its fees only if brands can buy the inventory through the Amazon DSP.

• Formats that carry retail signals. Price, ratings, Prime eligibility and availability have to render inside the ad. This is what makes Amazon more than another demand source.

• Closed-loop measurement. Endemic brands need to see that spend on the agent drives sales on Amazon.

If the assumptions behind this sequence of events hold, Amazon is telegraphing a blueprint for every consumer agent. Build ad inventory Amazon can sell and it will embrace you. Fail to and it will block you.