Apple has finally shipped the Siri overhaul it first teased two years ago. The new version, called Siri AI, arrives with iOS 27 and can dig through your texts, emails, photos, and calendar to answer personal questions, read what is on your screen without you asking, write your messages for you, and carry out actions inside your other apps.
On paper, this closes a gap that has embarrassed Apple since ChatGPT and Google Assistant made Siri look outdated. But the more interesting story is what is actually running underneath the friendly voice.
Apple is not powering this new Siri with its own AI. It signed a multi year deal to license a massive AI model built by Google, called Gemini, reportedly paying around one billion dollars a year for the privilege. That model, not anything Apple built in house, is doing the heavy thinking behind personal search, screen reading, and natural conversation.
This is a real reversal for a company that has spent decades selling total control over its own hardware and software as a selling point. Apple's earlier fix for a smarter Siri was routing certain requests to a plug in from OpenAI's ChatGPT starting in late 2024. That did not go far enough, so Apple went shopping again, this time for something bigger and more deeply wired into the assistant itself.
The delay behind all this was not free. Apple first showed off a "more personal" Siri at its 2024 developer conference and then advertised it heavily when the iPhone 16 launched that same year. The features never arrived on schedule, and Apple eventually agreed to pay two hundred and fifty million dollars to settle a lawsuit from buyers who said they were sold a phone based on capabilities that did not exist yet.
For business owners, there are two practical takeaways here. First, Siri AI only runs on the iPhone 15 Pro and newer, because the older chips cannot handle the processing load. If your company has staff on older company phones, budget for that gap before promising anyone a smarter assistant.
Second, and more important, pay attention to who actually owns the AI inside the products you use. Apple's pitch has always been that your data stays private because processing happens in a locked down system it built, but the model doing the actual thinking was built by a company whose entire business is knowing everything about you. Brand names can hide who is really doing the work inside a product, and that will matter more as AI features spread into every tool your business depends on.
Apple's bigger AI ambitions have clearly slipped. Buying rather than building is not a failure by itself, plenty of smart companies do it. But it is a signal worth watching: if even Apple, with its enormous cash pile and famous engineering culture, needed two years and a billion dollar annual check to Google just to catch up, that tells you how far ahead the leading AI labs really are, and how hard it will be for any single company to go it alone.