Apple

Apple makes the iPhone, Mac, and iPad. To place this company correctly in an AI story, start with how it makes money. OpenAI or Anthropic sell the model itself, but Apple sells the device. For Apple, AI isn't the product — it's a feature that makes people buy more devices. Most of why Apple behaves differently from everyone else traces back to that one fact.

The umbrella name for Apple's AI features is Apple Intelligence. It was announced in 2024, covering things scattered across the iPhone like notification summaries, writing tools, and image generation. In June 2026, Apple layered Siri AI on top of that — rebuilding Siri so it can see what's on your screen and carry out tasks across apps on your behalf. Articles calling this "the Siri that took two years to arrive" refer to the fact that a feature promised in 2024 didn't ship until 2026.

The most common misreading is thinking "Apple Intelligence is entirely AI built by Apple." That's only half right. The small models that run on the device itself are trained by Apple. But for tasks that need heavier reasoning, Apple brings in Google's Gemini. Even so, it doesn't send your question to Google — Apple runs that model on its own servers. Apple calls this server system Private Cloud Compute, and says it's encrypted so that even Apple can't see what's inside.

What sets Apple apart from other companies is its push to keep as much processing as possible on the device itself. Not sending data over the internet makes things slower and limits what's possible, but it also means less personal information leaves the device. That's both why Apple looks a step behind in the performance race, and why it chose to be. This is also why Apple stories rarely mention how much data center capacity or how many GPUs it bought — instead they talk about chips and memory, because the model has to run in your hand.

One exception is China. Since Google's Gemini isn't available there, Apple co-trains its China-specific AI model with Alibaba (as of August 2026). So the same iPhone can run different AI depending on the country.

"Gemini is going into Siri" doesn't mean Apple gave up on AI and handed it to Google. It means Apple builds the small models itself and only borrows Google's model for heavy tasks — and even then, runs it on Apple's own servers.