The Aggregator Wars Are Here. What That Means for Your AI Strategy.

I watched something shift this week that I realized I needed to name clearly: the companies that own distribution channels are moving to block the companies that want to sell through them. Not by price, not by feature. By outright refusal.

Amazon announced it would block Meta's new agent from accessing Alexa. GM dropped Apple CarPlay and Google Android Auto to build its own system. These aren't product disagreements. They're distribution wars. And they matter to you because they're about to reshape where, and how, your customers encounter AI.

Let me walk through what I'm seeing and what it means for how you think about AI adoption.

The Pattern: Control the Last Mile

An aggregator, in the way I use the word, is anything that sits between you and your customer. Amazon has Alexa. GM has the car dashboard. Apple has Siri. Meta wants to build an agent that works across the web. These platforms are fighting because they understand something fundamental: the company that controls how the customer accesses your service often matters more than the company that builds the service.

For twenty years in sales and operations, I saw this with mobile first. Companies that owned the app store, Apple, Google, didn't have to be the best at any one thing. They just had to own the door. Everyone else had to ask permission.

AI is going through the same fracture right now. And the fracture matters to you because it changes the path from your business to your customer's hands.

Amazon blocking Meta's agent isn't about technology. It's Amazon saying: if you want to reach customers through Alexa, you use Amazon's AI, not Meta's. You build for Amazon's rules. GM dropping Apple's CarPlay isn't about maps or voice assistants. It's GM saying: the dashboard is our aggregator, and we control what runs there.

This is the beginning of a world where the distribution platform is also the AI platform. And you have to make a choice about how you build around it.

What This Means: You're Now Choosing an Aggregator

Five years ago, if you built a service, say, a scheduling tool for appointment businesses, you could sell it directly to customers, charge a subscription, and live. The distribution was yours.

Today, you still can. But your customers are increasingly accessing AI through platforms they already use: their phone's assistant, their car's interface, their web browser. If the AI they want is locked to one aggregator, they either accept it or don't use it at all.

For a small business owner, this creates a new constraint: which aggregator do I build for?

Let me be specific. A property-management client I worked with earlier this year was deciding between building a chatbot for their own website versus building an agent that runs inside Alexa. On paper, it sounds like the same product. It isn't.

If you build for Alexa, Amazon controls the interface, the payment, the customer data, and the ranking in results. You pay them a cut. Your customers access you through Amazon's rules. You cannot change the voice, the interaction pattern, or the business model without Amazon's approval.

If you build for your own website, you control everything. You also own all the distribution risk.

That client spent weeks deciding. The question wasn't "which is better?" It was "what am I willing to give up, and what can I not afford to lose?"

She chose her own website. But she had to choose. And that choice is now a line item in every AI project plan.

The Conflict Is Real, and It's Escalating

Amazon and Meta aren't fighting because one of them is evil. They're fighting because the value of controlling distribution is enormous, and if one aggregator wins, the others lose leverage with customers.

GM didn't drop CarPlay and Android Auto out of spite. GM realized that every interaction that happens inside Apple's or Google's ecosystem teaches those companies something about GM customers that GM cannot see. Apple and Google can use that data to build competing services, to negotiate better terms, to lock GM out entirely. So GM is saying: we will own that data flow, and if you want to reach GM drivers, you build through us.

This creates a new world for small operators. The path from your service to the customer now goes through a gatekeeper that has reasons to block you, to favor competitors, to extract terms.

It also creates opportunity. If you pick the right aggregator early, you get favored-partner status, better economics, and faster growth. If you pick wrong, or if you're split across multiple gatekeepers with competing interests, you get stuck explaining yourself to multiple systems instead of serving customers.

What You Actually Need to Do

I'm going to give you three concrete steps, not principles.

First: Map where your customers already are. Not where you think they are. Where they actually spend time. If your customers are already in Alexa, competing in Google Search, or checking their car's screen, those aggregators matter to your strategy. If they're not, don't build for them yet.

A training-company client I advised was thinking about building an agent because competitors were. But their customers, corporate L&D teams, lived in Outlook and Slack, not in consumer platforms. The real aggregator for them wasn't Amazon or Meta. It was Microsoft, because Office is their operating system. That changed their entire build plan.

Second: Understand the terms before you commit. When Amazon blocks Meta, it's exercising the terms in Amazon's agreement. You need to read those terms for any aggregator you build into. Can Amazon change the economics without notice? Can they deprioritize your service if it competes with theirs? Can they ban you for any reason?

Most small operators skip this. They build a great agent and then discover the aggregator changed the payout structure, and suddenly the unit economics don't work. Read the terms first.

Third: Keep your own customer data. This is the one thing you can control that aggregators will always want to take from you. If you build on Alexa, Google, or Meta, they will offer you insights, dashboards, and analytics about your customers. But they own the primary data. If they change terms, ban you, or go out of business, you lose access.

The best small operators I've worked with build on aggregators but keep a separate channel, usually email or a private SMS list, where they own the relationship directly. It costs more. It's also insurance.

The Larger Game: This Isn't About These Three Companies

Amazon, Meta, Apple, Google, and GM are just the first movers. Over the next eighteen months, you'll see every major platform clarify whether it owns AI distribution or rents it. Insurance companies will control agents for claims. Banks will control agents for account access. Retailers will control agents for shopping. Hospitals will control agents for patient intake.

Each one will be making the same choice GM and Amazon made: do I let competitors access my customer interface, or do I build my own and keep the data?

Most will choose to build their own.

This means the distribution landscape is fragmenting. In the mobile era, you could live on the App Store and that was enough. In the AI era, you're going to have to build for multiple aggregators, each with different rules and economics. That's work. It's also reality.

What I Got Wrong

Two years ago, I thought the main AI story was about which model was best: GPT, Claude, Gemini. Model quality matters. But it turned out the real story was simpler and harder: who gets to sit between the model and the customer.

That's the aggregator fight. And it started this week. You need to see it.

Your Next Step

If you're building or planning any AI product, ask yourself this one question: which aggregator do my customers already use? Don't guess. Look at your actual customer data. Then make a decision about whether to build on that aggregator or alongside it.

The companies that get this right early will have a significant edge. The ones that build for an aggregator that doesn't matter to their customers will waste time. And the ones that try to build for every aggregator equally will exhaust themselves.

Pick one. Get it right. Then expand from there.

The war for distribution has begun. The question is which side you're on, and whether you chose it on purpose or by accident.