Meta's newest AI agent rocketed to No. 1 in the App Store. Amazon slammed the door within two weeks. Ben Thompson calls it the start of the "Aggregator Wars" — aggregator versus aggregator, with a trillion dollars on the table.
If you have been following the tech news this September, you know the shape of the story: Meta launched Muse, its new personal AI agent, on September 8, 2026. Less than two weeks later, Amazon blocked it from shopping on Amazon.com. On the surface it looks like a standard corporate squabble. Under the surface, it is the first real skirmish in a war over who owns the customer relationship in the age of AI agents — and, as Stratechery's Ben Thompson put it this week, it is genuinely "aggregator v. aggregator."
Here is the fun part: the two companies are also business partners. Amazon items have been purchasable inside Facebook and Instagram since 2023, and in April 2026 Meta signed a multibillion-dollar deal to run agentic AI workloads on Amazon's Graviton chips. King Kong and Godzilla, still splitting the lunch check while planning to destroy each other's cities.
The timeline is short and brutal.
Muse is free, with paid subscription tiers, available on iOS, Android, muse.ai, and WhatsApp. It is Meta's biggest swing at turning its assistant into the front door for your whole digital life — including your credit card.
Amazon's public case, per GeekWire and company statements:
Amazon's spokesperson put it pointedly: "We think it's fairly straightforward that third-party applications that offer to make purchases on behalf of customers from other businesses should operate openly and respect service provider decisions about whether or not to participate." Translation: food-delivery apps have deals with restaurants, travel agencies have deals with airlines — agents that buy things should have deals with merchants too.
There is also a more human reason Amazon might not rush into agentic commerce: if Muse makes a bad order, Amazon is the one cleaning it up — the angry customer and the angry vendor both. As TechCrunch noted, Muse has one of the lower hallucination rates among AI models, but it is still far from zero.
Here is where Ben Thompson's analysis gets interesting. Amazon is under no legal obligation to let Muse in — and unlike nearly every other business on the internet, it can survive being left out of the agent revolution entirely.
Why? Because Amazon's moat is not informational. It is physical. Warehouses, logistics networks, last-mile delivery, Prime's two-day promise. An AI agent cannot replicate a fulfillment center. If agents want to buy physical goods, they still need the trucks. That is why Amazon can block every agent it dislikes and its customers will barely notice.
Compare that with Expedia or Google, whose core value is being the digital middleman between you and information. An AI agent that shops for you is an existential threat to them, because the agent replaces the middleman. For Amazon, being skipped by an agent is a nuisance. For Expedia, it is an obituary.
And there is the ad business. Amazon generated more than $68 billion in ad revenue last year, built on people browsing its pages and seeing sponsored products. An agent that shops invisibly — one that "doesn't see ads," as Thompson notes — is not just a missing customer. It is a missing audience.
Aggregators got rich by commoditizing suppliers and owning the user relationship — Google with pages, Facebook with friends, Amazon with products. The new AI agents threaten to do the same to them: commoditize the aggregators by owning the user relationship one layer up. When your agent does the shopping, Amazon's storefront becomes a supplier; Amazon's homepage, with its sponsored carousels, becomes skippable overhead.
That is why Thompson dubs this the Aggregator Wars. It is King Kong vs. Godzilla precisely because both monsters are aggregators — Meta aggregating attention and social data, Amazon aggregating demand and logistics — and neither can afford to be the one that gets commoditized.
This is not Amazon's first fight with an AI shopper. The track record:
So the scoreboard on the legal front reads: agents 1, Amazon 0. And the user-side experience — a popup addressed directly to shoppers, reminding them of terms they agreed to — shows Amazon pivoting to the battlefield where it still has weapons: contracts, and the customer's own loyalty.
If Amazon's moat is physical, Meta's counter-move is to lease a different set of trucks. The retailer lineup Meta announced at Meta Connect — Walmart, Best Buy, Gap, Sephora, Wayfair, Dick's, Fanatics, Ulta, plus Shopify and Stripe rails — is not random. It is a coalition of everyone who would like a shot at Amazon's share of American wallets.
Thompson's argument: Meta needs Walmart to wait out Amazon. Walmart is the only retailer with a comparable physical logistics footprint in the U.S., and it has every incentive to make agent-driven shopping work. Groceries, essentials, same-day pickup — the mundane stuff Amazon delivers so reliably is exactly what Walmart can offer Muse users instead. If agents increasingly route around Amazon, Amazon loses the ad impressions and the data exhaust that feed its next decade of growth.
Zuckerberg's promise that Muse stays free for most users is part of the same play: maximum distribution first, monetization later — the classic land-grab, with your shopping cart as the territory.
Google is hedging in public. It is reportedly testing purchases from Walmart-owned Flipkart through Gemini and AI Mode in India — while Amazon has previously blocked Google's AI shopping agents from its marketplace. Expedia is fighting to keep its middleware position before agents eat it. And where is Google in the endgame? Thompson's own headline asks the question: whither Google? — because an agent that searches, decides, and buys makes the search box optional.
Amazon blocked Muse because it could, and because it had to: its $68-billion ad business depends on being the storefront, not the supplier behind someone else's butler. Meta kept Muse out of the App Store doghouse and into the coalition-building phase because it has no retail moat of its own — only 3 billion humans and an agent that checks with you before spending your money.
The Aggregator Wars have begun. King Kong has the warehouses; Godzilla has the eyeballs and the agent. And the shopping cart — that most mundane, necessary thing — is now the most contested real estate in tech.
Sources & further reading: Stratechery, "Amazon Blocks Muse, Amazon's Moat, Aggregator v Aggregator" and "More on Muse, Amazon, and Walmart" (Sept 22, 2026); GeekWire (Todd Bishop, Sept 20, 2026); TechCrunch (Sept 21, 2026); Business of Fashion; The Information; WSJ; Meta's Muse launch announcement (Sept 8, 2026); Ninth Circuit opinion in Amazon v. Perplexity (Aug 4, 2026).
