If you browse Reddit or HN, you’ll see the same pattern: people are buying Mac Minis specifically to run AI agents with the use of computers. They are installing headless machines whose sole job is to automate their workflow. OpenGL—the open-source framework that lets you run cloud, GPT-5, or whatever model you actually want to control your computer on—has become a killer app for Mac hardware. Not final cut. No logic. An AI agent that clicks buttons.
This is exactly what Apple Intelligence was supposed to be.
Apple had everything: the hardware, the ecosystem, the “it just works” reputation. They could send an agentic AI that would automate your computer instead of actually summarizing your information. Imagine if Siri could actually file your taxes, respond to emails, or manage your calendar by actually using your apps, rather than through some brittle API layer that breaks every update.
They could have charged $500 more per device and people would have had to pay it. Margins would have been obscene. And they won’t win the AI race not by building the best model, but by being the only company that can ship an AI you can actually trust with root access to your computer. That trust built over decades was their ditch.
So why didn’t they do it?
Maybe he didn’t even see it. It sounds simple, but it’s probably the most common reason companies miss opportunities. When you’re Apple, you’re thinking about chip design, manufacturing scale, and retail strategy. An open-source project that lets AI agents take control of computers can’t ping your radar unless it’s already happening.
Or maybe they’ve seen it and decided it wasn’t worth the risk. If you’re Apple, you wouldn’t want your AI agent to automatically buy things, post on social media, or make irreversible decisions. The risk of liability would be huge. It’s better to send something safe and limited than something powerful and unpredictable.
But there is another dynamic at play. Look who’s going to be angry about OpenGL-style automation: LinkedIn, Facebook, walled gardens, and anyone with a careful API strategy. These services depend on friction. They want you to use their apps, see their ads, be in their ecosystem. An AI that can automatically remove friction that threatens survival.
If Apple had made it, they would have been fighting Instagram over ToS violations by Tuesday. He will testify before Congress about fraud by AI agents. Each technology platform will be updating its terms to explicitly ban Apple Intelligence.
By letting a third party do this, Apple gains plausible deniability. They are just selling hardware. It’s not their fault what people drive on it. This is the same strategy that made them billions in the App Store, while maintaining that they are “not responsible for what the developers do.”
But I think this is short-term thinking.
Here’s what people forget about trenches: They’re mixed. The reason for Microsoft’s dominance on PCs wasn’t just that they had the best OS. It’s like everyone built for Windows, which made Windows more valuable, which made more people build for Windows. Network effect.
If Apple had an agent layer, they could create the most defensible moat in technology. Because the more an AI agent knows about you, the better it gets. And Apple already has all your data, all your apps, all your devices. They could create an agent that works seamlessly across your iPhone, Mac, iPad, and Watch—something no one else can do.
More importantly, they could own the API. Do you want your service to work with Apple Agent? You play by Apple’s rules. Suddenly Apple isn’t fighting with the platforms – these are the platforms the platform needs to integrate with. It’s the App Store playbook again, but for the AI age.
Mac Mini Rush is a preview of this future. People want agents. They want automation. They want to pay for it. They’re literally buying extra computers to run someone else’s AI on Apple’s hardware.
Apple is getting hardware revenue but platform revenue is decreasing. This may seem smart in this quarter. But platform revenue is what has made Apple a $3 trillion company. And the platforms themselves create a trillion-dollar moat.
I suspect that ten years from now, people will look back on 2024-2025 as the moment when Apple had a clear chance to own the agent layer and decided not to take it. Not because they couldn’t build it – they clearly could – but because they were optimizing for this year’s legal risk rather than the next decade’s platform power.
People buying a Mac Mini to run AI agents aren’t just early adopters. They are showing Apple exactly the product they should have made. Whether Apple is paying attention is another question entirely.
<a href