Last week at Davos, Andy Jassy diplomatically dodged questions about partnering with ChatGPT, while simultaneously critiquing "third-party agents" calling out their lack of buying history, personalization and misinformation around product and pricing details. Meanwhile, at NRF Google announced their Unified Commerce Protocol for end-to-end AI shopping alongside Shopify, Walmart, Target, and practically everyone except Amazon.

A structural shift is underway in e-commerce. AI-powered product discovery is creating new distribution channels, new monetization models, and imposing strategic decisions from every major player. Core to these decisions, is Amazon’s choice (or lack thereof) to partner with third-party agents. Clear incentives, though, are now pointing to a forthcoming ChatGPT-Amazon partnership, but there is a misconception on the owner of asymmetric leverage in this potential marriage.

The Agentic Commerce Myth

Let's start by addressing a fundamental misunderstanding: AI chatbots aren't going to autonomously purchase materially priced goods for you.

Eric Seufert captured this in his piece "Agentic Commerce is a Mirage." The vision of rogue AI agents running around with your credit card, restocking your fridge while you sleep? That's not happening. Nobody wants a chatbot making meaningful purchasing decisions on their behalf. What consumers actually want is straightforward: the easiest possible way to buy exactly what they need, with the least amount of friction.

Agentic commerce is not autonomous purchasing, and the distinction matters because it fundamentally changes how this technology gets built, who partners, and who wins.

Why Amazon Products Dominate AI Recommendations (Whether Amazon Likes It or Not)

Amazon's dominance in AI-powered shopping isn't about partnerships or ad spend, it's providing the underlying infrastructure (a natural strength of Amazon’s) for LLM curated product recommendations.

The key insight comes from understanding the signals LLMs prioritize when recommending products. Michael Morton, a senior analyst at MoffettNathanson, asked the AI models directly. The answer? Price, trustworthiness, and speed of delivery. In that order.

This is where the power of Amazon shines. Fast delivery through Prime, trusted reviews at scale, competitive pricing—Amazon has optimized their product catalog and infrastructure around the exact criteria that AI models prioritize.

Morton's team validated this with data. Using Profound, a tool for Answer Engine Optimization, they ran 500 e-commerce queries through multiple AI platforms. Amazon dominated product recommendations across the board, even without paying for placement. Their infrastructure investment has positioned them perfectly for the AI shopping era, whether they embrace it or not.

Google Just Forced Everyone's Hand with an Anti-Amazon Consortium

Google's UCP announcement at NRF shifts the competitive dynamics, but the pressure it creates falls primarily on ChatGPT, not Amazon.

The coalition backing UCP reads like a strategic alliance of Amazon competitors: Walmart, Target, Shopify, Etsy, Wayfair, Best Buy, Home Depot, and Macy's. Over 20 major players, all aligned around a single objective: make AI shopping work without Amazon. These partnerships signal that Google has every intention to meet and potentially leapfrog ChatGPT’s ambitions in agentic commerce. Quickly matching Google's breadth of retail partnerships and depth of catalog would be challenging. ChatGPT lacks the decade-long relationships with retailers and Product Listing Ads infrastructure Google spent years building. For ChatGPT, this creates an urgent problem.

Gemini’s share of global GenAI traffic increased from 5.7% to 21.5% while ChatGPT declined from 86.7% to 64.5%. ChatGPT is losing mindshare to Anthropic in B2B. Gemini has distribution through billions of Google Search users. Google can monetize AI shopping immediately through their existing retail media infrastructure, demand-side relationships, and measurement capabilities that ChatGPT lacks.

Given Google’s distribution advantages in consumer and Anthropic’s lead in the enterprise, ChatGPT needs to find a way to continue to maintain marketshare in consumer. And commerce is the category they simply need to win. Yes, commerce queries currently represent a small % of ChatGPT’s usage (1%-2%), but as agentic commerce becomes increasingly ubiqutious, this will undoubtedly only increase. Most important, commerce is a category that is incredibly easy to monetize (whether via affiliates fee, or ads, or both) and ChatGPT needs monetization channels to fuel their lofty ambitious and compute costs.

Why ChatGPT Needs Amazon More Than Amazon Needs ChatGPT

Amazon’s dominance in ecommerce looms. By definition, the winning agentic commerce chatbot must incorporate Amazon’s +40% marketshare of e-commerce transactions, global scale, and unrivaled catalog rich with review, price, and shipping metadata. Amazon won't partner with Google. They compete directly on product search, the foundation of e-commerce discovery. Partnering would hand their primary competitor access to their catalog, fulfillment infrastructure, and customer data. It's a non-starter—which points to an ChatGPT-Amazon partnership as the logical next move. Here’s why it is a strategic imperative for ChatGPT to partner with Amazon.

1. Current Retail Partnerships Aren’t Enough

ChatGPT’s retail partnerships with Shopify, Walmart and Etsy through Instant Checkout demonstrate initial traction. However, these partnerships collectively represent only a fraction of total e-commerce volume compared to Amazon.

The gap isn't just market share, it's also the infrastructure. When ChatGPT recommends products from Shopify merchants or Walmart, users still face uncertainty around delivery times, return policies, and seller reliability. Amazon solves this through Prime's two-day shipping guarantee, standardized return processes, and A-to-Z purchase protection. For AI shopping to scale, users need to trust that the purchase experience will be seamless. Only Amazon provides that infrastructure at a global scale.

2. Amazon is Succeeding Independently

Amazon already has Rufus, their AI shopping assistant. According to data from Sensor Tower analyzing 100,000+ Amazon shopping sessions over Black Friday, Rufus drove 3.5x higher conversion than non-Rufus sessions and accounted for 40% of sessions and 66% of purchases.

Amazon's strong position with Rufus and dominant market share means they can afford to wait and see which AI platform emerges as the winner. ChatGPT doesn't have that luxury—every month of delay is another month where Gemini gains share and ChatGPT's commerce ambitions remain constrained.

3. ChatGPT Needs to Monetize Beyond Premium Subscriptions, and Fast

ChatGPT just announced a 4% commission on sales. But with $1.4 trillion in infrastructure commitments and $8 billion in losses in just the first half of 2025, ChatGPT will need more than affiliate fees on 1-2% of query volume (even assuming a healthy conversion rate) to generate meaningful revenue.

Advertising is the only path to sustainable scale. We’re already starting to see contextual ads launch within free and Go tiers, but building an advertising business from scratch takes years. Partnering with the Amazon DSP shortcuts that timeline while providing immediate access to demand-side relationships and proven ad technology. The urgency is compounded by competitive pressure. As Gemini integrates shopping with Google's existing ad infrastructure, ChatGPT risks being outpaced not just on product selection but on the business model itself.

Amazon's Real Dilemma: The $68 Billion Question

So why is Amazon hesitating, despite having leverage? Consumer behaviour on ChatGPT presents a material downside risk to its advertising business.

Amazon generated approximately $68 billion in advertising revenue in 2025. The overwhelming majority comes from sponsored product ads sold on a cost-per-click basis. Amazon reports that consumers make an average of 14 clicks before completing a purchase. Approximately half of that appears on product pages consists of advertisements.

With agentic commerce, the entire funnel compresses dramatically. Even now, on ChatGPT, a user asks a question, receives 3-5 curated options, copies the product name, pastes it into Amazon search, and purchases. No extended browsing. No clicking through multiple product pages. Dramatically fewer ad impressions and touch points on the customer journey.

Opening up to ChatGPT creates two primary risks for Amazon:

1. Losing their direct relationship with the customer, but most importantly,

2. Watching billions in advertising revenue decline

Given ChatGPT’s need for Amazon and the downside risk to advertising revenue that ChatGPT presents, any partnership would require Amazon to secure a commensurately impactful revenue stream to its advertising business.

The Only Solution That Works: An Advertising Partnership on Amazon's Terms

Eric Seufert outlined what a partnership could look like: An advertising partnership structured on Amazon's terms means Amazon provides exclusive data access within ChatGPT (comprehensive product catalog, real-time inventory, dynamic pricing, delivery windows), and ChatGPT users gain seamless checkout capability through Amazon.

And Amazon already has the infrastructure to make this work:

1. The Amazon DSP as the first and only exclusive third-party demand partner for ChatGPT’s initial advertising rollout. While ChatGPT slowly rolls out ads they can afford to methodically curate an initial batch of advertisers. Assuming these ads work, supply will quickly become less constrained and ChatGPT will need demand at scale. The Amazon DSP brings immediate advertiser demand, allowing ChatGPT to monetize without building a self serve campaign manager and ad sales team. And for Amazon, exclusivity ensures they can replace declining on-site ad revenue with ChatGPT as a supply source bringing incremental revenue to its DSP.

2. Off-site Sponsored Products ads extended into ChatGPT. Amazon has invested significantly in offsite Sponsored Products ads, demonstrated through partnerships like Pinterest where Amazon product ads appear directly in search results. This proven technology can be extended into ChatGPT, allowing endemic Amazon brands to purchase inventory using existing tools and workflows.

3. Amazon DSP access to ChatGPT query signals for targeting. Signal sharing from ChatGPT queries enables enhanced targeting capabilities across Amazon's DSP. The Amazon DSP has been winning share from other enterprise DSPs due in part to its exclusive access to proprietary shopping signals. What better way to strengthen their hold on consumer shopping signals than having exclusive access to the world’s leading chatbot’s commerce queries and intent? ChatGPT can benefit here by owning a revenue share from the CPM markup that Amazon would charge for the use of ChatGPT audiences

ChatGPT takes a transaction cut (Amazon's traditional affiliate rates range from 3.5-4%); ChatGPT gains access to the share of e-commerce and catalog they desperately need, while Amazon replaces declining on-site advertising revenue with new ChatGPT inventory (on their terms), and strengthens the Amazon DSPs lead as the de facto enterprise DSP for commerce.

The question isn't whether they'll partner, it's whether ChatGPT can offer terms attractive enough to make Amazon move on something they don't urgently need. Some advantages, like controlling 40% of e-commerce and having the logistics infrastructure that AI models inherently prefer, are nearly impossible to overcome.

Note: this blog first appeared in our Cherry Picked newsletter. Subscribe here 🍒