Apple 把 Siri 交給 Google：認輸，還是這十年最聰明的一步？
蘋果史上第一次，把自家最核心的語音助理，交給死對頭 Google 的模型來跑。這到底是技不如人的妥協，還是一步高招？
If you had told any tech editor three years ago that "Siri will be powered by Google's AI in the future," you would have been laughed at. Apple and Google, one selling hardware and the other relying on advertising, are the most classic arch-enemies in the mobile world. But at WWDC 2026, this scenario became a reality - the new Siri, renamed "Siri AI," is powered by Google's Gemini. The internet is filled with voices saying "Apple has admitted defeat." I think, however, that this is the most sober move Apple has made in the past decade.
Admitting defeat depends on what you're competing in
If it's about "who has the strongest model," then Apple has indeed lost, and quite miserably. But the question is - the battle for smartphones has never been about comparing models.
You buy an iPhone because its AI is the smartest? No. It's because it's user-friendly, secure, seamlessly integrated with your other Apple devices, and you're used to it after a decade. Apple's true strength lies in this entire "can't-live-without" experience. The strength of the model has never been the reason it sells smartphones.
Once you understand this, you'll see why Apple dared to outsource Siri - it knows where its strengths lie and is clear that Siri's "brain" is not the core. Since it's not, using the most powerful existing one is perfectly fine.
What Apple is really betting on
Handing over the core assistant to a competitor, of course, comes with risks. But Apple has apparently calculated a trend: large language models are becoming like electricity.
Think about it, no company insists on generating its own electricity or building its own network - it's too expensive, too stupid, and not core to their business. Models are heading in the same direction. When GPT, Gemini, and Claude are all strong enough to be "good enough for ordinary people," spending billions to train a similar model is simply a waste.
Apple is betting that in the future, models will become so cheap that they're readily available, and the things that others can't take away from it - hundreds of millions of devices, operating systems, and user trust - will become more valuable. Outsourcing what will depreciate (models) and holding tight to what will appreciate (ecosystem) is a calculation that Apple has made very precisely. To understand the differences between various models, see 2026 Mainstream AI Model Comparison.
Is this move vulnerable
Yes, and significantly so. Tying the most core assistant to a competitor means that if Google raises prices, withdraws its support, or manipulates the data, Apple will be at a disadvantage. Therefore, it's almost certain that Apple is secretly developing its own model, and this move is just "using the strongest one as a temporary solution."
In other words, the current Siri AI is more like Apple's "rental phase" - first, move into the best house, and then slowly build its own. When its own house is ready, it can move in at any time. To see how Apple plays this calculation to the finest detail, read Apple's 7 Masterstrokes to Save Siri.
TheAI Academy Summary and Commentary
So, back to the question: is this admitting defeat or a clever move? My answer is - admitting defeat in areas where it's not good at, in exchange for continuing to win in areas where it excels. This isn't surrender; it's about seeing things clearly.
The hardest thing for tech companies is not creating something but acknowledging that "someone else does this better than me" and letting go. Many giants have died clinging to things they're not good at. Apple didn't make this mistake this time, and that's worth noting. Extended reading: Why Even Apple Doesn't Make Its Own Models.
A one-sentence commentary: Apple isn't admitting defeat; it's seeing things clearly - admitting defeat in areas where it's not good at, in exchange for continuing to win in areas where it excels.
(This article is TheAI Academy's commentary based on publicly available information from WWDC 2026, with details subject to Apple's official announcements.)
Sources
Frequently Asked Questions
Apple 把 Siri 交給 Google 是認輸嗎？
與其說認輸，更像務實——Apple 承認自家模型不是強項，把面子讓給工程、把好用的體驗留給使用者，同時守住裝置與生態主導權。
Siri AI 用 Google Gemini 有什麼風險？
長期有議價權、資料流動與依賴對手的風險，這次更像過渡期的務實選擇，Apple 自研模型應仍在發展。
這對 AI 產業代表什麼趨勢？
當連 Apple 都不自己做模型，代表競爭重心從『誰模型最強』轉向『誰整合得最好』，模型趨向基礎建設化。