The next great battle in e-commerce may not be about who has the biggest warehouse, the cheapest shipping or the largest product catalogue.
It may be about something much more fundamental:
Who controls the moment when a customer decides what to buy?
This week, that question became unusually visible.
Amazon blocked Meta's Muse AI agent from shopping on Amazon.com, saying the agent was unauthorized and violated its Conditions of Use. At almost exactly the same moment, Shopify announced an integration that allows Muse users to shop across Shopify-powered stores and complete purchases through Shop Pay.
On the surface, these look like two companies making different technology decisions.
Underneath, they reveal two very different economic models.
Amazon has built an enormous business around controlling the shopping journey. Shopify has built a business around enabling merchants to complete transactions.
AI shopping agents threaten that distinction.
Amazon Doesn't Just Sell Products. It Sells Attention.
Amazon is obviously an enormous retailer and marketplace.
But increasingly, another business sits on top of that marketplace:
advertising.
Amazon reported $68.635 billion in advertising-services revenue in 2025, up from $56.214 billion in 2024. Its advertising services include sponsored ads, display and video advertising.
That makes the Amazon shopping page more than a catalogue.
It is also an advertising surface.
When you search for “running shoes,” “coffee maker” or “laptop,” you're not simply receiving a neutral database query.
You're entering an environment where sellers can pay to put products in front of you.
The search page itself becomes part of the monetization machine.
And that creates a problem for AI agents.
Because an AI agent doesn't necessarily need to show you the page.
It can simply ask:
Which products meet the user's requirements?
Then it can compare them.
Then it can make a decision.
Then it can buy.
The customer may never see the sponsored listing.
The AI Agent Is Trying to Remove the Page
This is the deeper significance of Muse.
Traditional e-commerce works roughly like this:
Customer → search → results page → sponsored products → product pages → comparison → checkout.
Agentic commerce potentially changes it to:
Customer → AI agent → product data → recommendation → checkout.
The interface disappears.
The search results disappear.
The advertising layer can disappear.
The customer doesn't necessarily browse ten products.
The agent does the browsing.
And that changes who controls the economics of discovery.
This is why Amazon's objection isn't merely technical.
Amazon says Muse wasn't authorized to access its platform and raised concerns about transparency, security and how the agent interacts with customer accounts. The company says it asked Meta to remove Amazon from Muse, but Meta declined.
There are legitimate security and platform-governance questions here.
But there is also an obvious commercial question:
What happens to Amazon's advertising machine if consumers stop looking at Amazon's shopping interface?
The Agent Doesn't Need Amazon's Recommendation Engine
This could become even more disruptive.
Amazon has spent years building sophisticated recommendation systems.
It knows what you've bought.
It knows what you searched for.
It knows what products you clicked.
It knows what products other consumers bought.
It has enormous amounts of behavioural data.
But a personal AI agent introduces a competing layer of intelligence.
Instead of asking:
“What does Amazon recommend?”
the customer can ask:
“What is the best option for me?”
That is a fundamentally different relationship.
The AI becomes the intermediary.
And whoever controls that intermediary increasingly controls the customer's path to purchase.
Shopify Is Playing a Different Game
Shopify's incentives are different.
Shopify isn't primarily trying to convince consumers that they should buy from Shopify.
Its business is built around providing infrastructure for merchants.
In 2025, Shopify facilitated $378.4 billion in gross merchandise volume, while generating $11.6 billion in revenue. Merchant solutions—including payments—accounted for 76% of total revenue.
That distinction is critical.
When a Shopify merchant sells another product, Shopify can make money from the transaction infrastructure.
More merchants.
More orders.
More payment volume.
More commerce.
That's why AI agents can potentially look less like a threat and more like another distribution channel.
If an AI agent brings a customer to a Shopify merchant and completes a purchase, Shopify still has an opportunity to participate in the transaction.
Shopify Just Opened the Door
The Meta partnership makes that strategy explicit.
Shopify and Meta announced that Muse will be able to shop across Shopify-powered stores and use Shop Pay's agentic checkout to complete purchases.
And Shopify has been preparing for this shift.
Its 2025 annual report explicitly describes investments in agentic commerce, including Universal Commerce Protocol, Catalog and Agentic Storefronts.
That tells you something important.
Shopify isn't treating AI agents as an external threat that needs to be kept outside the walls.
It is trying to make its infrastructure compatible with them.
The philosophy is essentially:
If the interface changes, make the infrastructure work with the new interface.
Amazon and Shopify Are Making Opposite Bets
This is why the contrast is so interesting.
Amazon's model has historically benefited from owning the entire shopping environment.
Search.
Discovery.
Recommendations.
Reviews.
Advertising.
Checkout.
Fulfillment.
Prime.
The customer comes to Amazon.
Shopify's model is almost the reverse.
Shopify provides the infrastructure that allows merchants to own their storefronts and customer relationships.
The customer doesn't necessarily need to know Shopify exists.
The merchant does.
That means an AI agent could actually strengthen Shopify's model.
The consumer asks Muse for a product.
Muse searches across merchants.
It finds a Shopify-powered store.
It completes the purchase through Shop Pay.
The customer gets the product.
The merchant gets the sale.
Shopify processes the commerce.
Everyone potentially gets paid.
This Is Why AI Agents Are Different From Search Engines
Google's search business doesn't completely disappear when a consumer searches for a product.
Google can still show advertisements.
It can show shopping results.
It can monetize the click.
But an autonomous shopping agent changes the unit of interaction.
The user doesn't necessarily click.
The agent acts.
That means the future could look less like:
Search → click → website → purchase
and more like:
Intent → agent → transaction
That tiny change could have enormous consequences.
Because the economic value of the internet has historically been built around the click.
AI agents threaten to make the click optional.
The Search Result Could Become Obsolete
Imagine saying:
“Find me a 55-inch TV under $600 with good reviews, low energy consumption and free delivery.”
Today, you might receive pages of results.
You open five websites.
You compare specifications.
You read reviews.
You look at prices.
You search again.
An agent could potentially perform the entire process.
It doesn't need to show you 50 options.
It might return:
“I found three that meet your requirements. This one is $479, has the lowest energy consumption and can arrive Thursday. Want me to buy it?”
Once the consumer approves, the transaction happens.
That is agentic commerce.
And the implications for advertising are enormous.
The End of “Pay to Be Seen”?
This doesn't mean advertising disappears.
But it could change.
Today, a retailer can pay for visibility.
The consumer sees the sponsored product.
The consumer clicks.
The consumer evaluates it.
An AI agent might not care whether a company paid for visibility.
It cares about the parameters it was instructed to optimize:
Price.
Quality.
Availability.
Reviews.
Delivery time.
Compatibility.
Brand preference.
Maybe sustainability.
Maybe something else.
That means the old advertising question—
“How do I get in front of the customer?”
—could become:
“How do I become the product the agent selects?”
That's a very different marketing problem.
The New SEO Might Be Agent Optimization
This could eventually create an entirely new digital marketing discipline.
For two decades, businesses optimized websites for search engines.
They learned:
SEO.
Then came social algorithms.
Businesses learned:
social media optimization.
Now AI agents are becoming another gatekeeper.
Businesses may increasingly need to optimize:
product feeds, structured data, catalogues, pricing, availability, reviews and machine-readable product information.
The goal isn't necessarily to rank #1 on a Google results page.
It is to become the product an AI agent considers when a consumer expresses an intention.
That could fundamentally change e-commerce marketing.
And This Is Where Shopify Has an Interesting Advantage
Shopify sits underneath millions of independent merchants.
Its platform already manages product catalogues, inventory, orders, payments and storefront infrastructure.
That makes it potentially useful infrastructure for an agentic internet.
Instead of every AI company having to negotiate separately with millions of merchants, an integration with commerce infrastructure can provide access to a large merchant ecosystem.
That's essentially what makes the Meta-Shopify relationship interesting.
The agent doesn't need to negotiate with every individual store.
The infrastructure can handle much of the complexity.
But Amazon Has Something Shopify Doesn't
Amazon has something incredibly valuable too:
intent.
Millions of people deliberately visit Amazon because they want to buy something.
Amazon also has enormous fulfilment infrastructure, seller relationships, customer data and logistics capabilities.
And it isn't sitting still.
Amazon has its own AI shopping initiatives, including Buy for Me, which allows its shopping assistant to find products on external brand websites. Amazon has also been trying to develop agentic shopping experiences while restricting unauthorized third-party agents from interacting with its platform.
So this isn't simply:
Amazon versus AI.
It's more accurately:
Amazon versus external AI control of the shopping journey.
The Real Battle Is Over the Customer Relationship
This is the part investors, merchants and marketers should watch closely.
Who owns the customer relationship when AI becomes the buyer?
Today:
Amazon owns much of the shopping experience.
Tomorrow:
The AI agent may own the interface.
The merchant owns the product.
Shopify may own the commerce infrastructure.
Amazon may own its marketplace.
Meta wants Muse to become the personal interface.
Google has Gemini.
OpenAI has its own commerce ambitions.
Perplexity has been developing agentic shopping.
Everyone is converging on the same question:
Who gets to stand between the consumer and the transaction?
The AI Agent Could Become the New Browser
The browser liberated users from individual software applications.
You didn't need to install a separate program for every website.
You opened Chrome or Safari and went everywhere.
AI agents could do something similar for commerce.
Instead of manually visiting Amazon, Walmart, Target, Shopify stores and thousands of individual websites, the agent could potentially navigate the commercial internet on your behalf.
That makes the agent itself a new layer of infrastructure.
And whoever controls that layer could have extraordinary influence over commerce.
This Is Why Amazon's Decision Matters
Amazon could have decided:
Let Muse in.
Instead, it chose to block it.
Amazon says the decision is about authorization, security and transparency.
Those are legitimate issues, especially when an outside agent can potentially access accounts, order histories and perform transactions. Amazon says Muse does not identify itself as an agent while browsing, which it views as a significant concern.
But the commercial implications are impossible to ignore.
Amazon makes tens of billions of dollars from advertising.
If consumers increasingly bypass the Amazon interface and allow AI agents to make purchasing decisions, the value of sponsored placement could eventually be challenged.
That's not necessarily happening at scale today.
Agentic shopping is still young.
But Amazon has every reason to think about what happens if it does.
And Shopify Is Betting on the Opposite Future
Shopify's bet is effectively:
If agents are going to become the new shopping interface, make Shopify the infrastructure they use.
That's a fascinating strategy.
Because the company doesn't necessarily have to own the consumer's attention.
It can make money when the transaction happens.
Its 2025 filings show just how closely its merchant-solutions business is tied to commerce volume: merchant solutions generated $8.8 billion in revenue in 2025, with Shopify Payments as a major component.
So if AI agents produce more transactions, Shopify has a potential economic path to benefit.
Muse Is the Early Warning
Meta launched Muse on September 8.
Within days, it climbed to the top of the U.S. iPhone free-app rankings, surpassing ChatGPT, according to multiple reports.
That doesn't prove consumers are already abandoning traditional shopping.
They aren't.
But it does demonstrate that consumers are interested in AI systems that do things, rather than merely answer questions.
And that distinction is enormous.
Chatbots give you information.
Agents can potentially execute the task.
The next stage of e-commerce is therefore not necessarily:
“AI helps me shop.”
It is:
“AI shops for me.”
The Internet Is Moving From Pages to Intent
This may ultimately be the biggest change.
The first internet was organized around pages.
Search engines organized those pages.
Social media organized attention around feeds.
Mobile apps organized services around applications.
The agentic internet could organize digital activity around intent.
You don't need to know where to go.
You tell the agent what you want.
The agent figures out where to go.
That's why Amazon blocking Muse and Shopify embracing it aren't merely two corporate announcements.
They represent two different visions of the next internet.
One side has built enormous value by controlling the destination.
The other has built enormous value by powering the transaction.
Amazon Charges Brands to Be Seen. Shopify Charges for the Commerce Layer.
That is the simplest way to understand the split.
Amazon's advertising business monetizes visibility.
Shopify's merchant-solutions business is heavily connected to commerce volume.
An AI shopping agent threatens to reduce the importance of visibility while potentially increasing the importance of transaction infrastructure.
The agent doesn't care about your homepage.
It doesn't care about your banner.
It doesn't necessarily care about your sponsored placement.
It cares about satisfying the user's request.
And that could make the next decade of e-commerce very different from the last.
The Shopping Mall Is Becoming an Algorithm
For decades, e-commerce tried to recreate the shopping mall online.
Thousands of products.
Search boxes.
Categories.
Recommendations.
Advertisements.
Product pages.
Checkout.
AI agents could remove much of that architecture.
The consumer simply states an intention.
The algorithm becomes the shopping mall.
And suddenly the most important question isn't:
“Where does the consumer shop?”
It becomes:
“Which AI does the consumer trust to shop for them?”
That is the battle Amazon, Meta, Shopify, Google, OpenAI and others are beginning to fight.
And yesterday's Amazon block and today's Shopify partnership may be an early glimpse of how that battle will unfold.
Amazon built one of the world's largest digital shopping destinations.
Shopify built infrastructure for millions of merchants.
Meta is trying to put an AI agent between consumers and both.
The winner of the next e-commerce era may not be the company with the biggest catalogue.
It may be the company that controls the agent standing between human intent and the transaction.
And that is why the future of shopping may look less like:
“Google it. Click it. Buy it.”
and increasingly like:
“Tell the AI what you want. Let it handle the rest.”
Comments
Post a Comment