What exactly did you think you were buying when you paid flagship prices for a phone?
Most people would answer: the hardware, the software, and the vague promise that Apple’s incentives were aligned with theirs. You pay up front, so you don’t get sold to later. That was the deal, or at least the story everyone told themselves about the deal. In 2026, Apple stopped pretending. Persistent ads are now baked into iOS, pushing iCloud+ and AppleCare+ at users who already handed over a four-figure sum for the device doing the pushing.
The reaction has been about what you’d expect. One user summed it up with “I wish Apple would just stop that crap.” A MacRumors forum poster put it more bluntly, saying Apple “seems to have completely lost the plot in 2026.” These aren’t Android partisans scoring points. These are people who bought into the ecosystem on purpose and are now watching it monetize their attention a second time.
The services trap is a math problem
None of this is mysterious. Apple’s ads are part of a strategy to grow revenue from services, and that strategy has a structural problem: hardware sales plateau, but shareholders don’t accept plateaus. Services revenue has to keep climbing. Once you’ve sold storage, insurance, music, TV, and news to everyone willing to buy them voluntarily, the only lever left is putting the pitch somewhere people can’t avoid it.
And Apple is not being subtle about the direction. Ads have been expanding into Apple Maps, where a search for something like “sporting goods” can surface a promoted local shop. More App Store ad inventory arrived in March. A business management platform for advertisers launched in April, with ad products rolling out in the US and Canada over the summer. That’s not an experiment. That’s infrastructure.
Infrastructure doesn’t get uninstalled. It gets expanded.
Why I care about this on an AI site
I review AI tools and agents for a living, which means I spend most of my week watching companies figure out how to charge for software that costs them real money to run. Inference is expensive. Subscriptions cover some of it. Enterprise contracts cover more. But the consumer tier, the free or cheap tier that everyone actually uses, has never covered its own costs and probably never will.
So when the most profitable consumer tech company on earth, one with no cost pressure and a sterling privacy reputation to protect, decides that persistent in-OS promotion is acceptable, it sets a ceiling. Or rather, it removes one. Every AI company with a burn rate and a board is watching a permission slip get handed out.
Here’s what makes agents different from a banner in Settings. A billboard is annoying but legible. You see it, you identify it as an ad, you ignore it. An agent that recommends, ranks, books, buys, and summarizes on your behalf doesn’t give you that clarity. When your assistant suggests a vendor, a product, a travel option, or a tool, you have no reliable way to tell whether that came from your stated preferences or from a commercial arrangement you never saw.
The thing that makes agents useful is exactly the thing that makes them dangerous as an ad surface: you delegate judgment to them.
What to watch for in the tools you use
I’m not telling anyone to throw their phone in a lake. I’m saying the Apple situation is a preview, and you should calibrate accordingly. When I evaluate an AI product now, these questions carry more weight than they did two years ago:
- Does the company make money any way other than your subscription? If yes, find out how.
- Are recommendations, rankings, or suggested actions ever influenced by paid placement, and is that disclosed in the interface rather than buried in terms?
- Can you turn off promotional content entirely, or only reduce it?
- Does the pricing page mention an ad-free tier? That phrasing tells you the free tier’s future.
- Is your usage data used to target anything, including the company’s own upsells?
The part that actually stings
One iOS user, talking about leaving the ecosystem, said the only thing keeping them there is iMessage and the group chats they’d lose. Everything else has a solid Linux alternative. That’s the whole dynamic in one sentence. The lock-in isn’t quality anymore, it’s social cost. Once a company knows leaving is painful, the quality bar it has to clear drops.
AI agents are building that same kind of lock-in right now, and faster than Apple ever did. Your context, your memory, your connected accounts, your workflows. Switching gets harder every month you stay. Which means the window where your displeasure still functions as use is open now and closing steadily.
Use it. Complain loudly, cancel things that annoy you, and pay attention to how the tools you rely on make money. Apple just demonstrated that a trillion-dollar balance sheet is no protection against this. Your favorite AI startup has considerably less to lose.
🕒 Published: