Apple spent $14 billion building what everyone else needed $670 billion to chase, and the gap tells you which companies actually believe their own hype.
The Summary
- Apple finally shipped Siri AI at WWDC 2026, two years after first promising it, built on Google's Gemini models but using none of Google's assistant code, search infrastructure, or client deployment systems.
- The new Siri becomes a systemwide AI interface layer that lets users act on app content across iPhone, iPad, Mac, Apple Watch, and Vision Pro without developers building separate chatbot interfaces.
- Apple's EU launch is blocked indefinitely as regulators refuse to engage on the company's privacy architecture.
- Apple will spend $14 billion on AI capex in 2026 while Amazon, Google, Meta, and Microsoft committed $670 billion combined, a 48x difference that signals fundamentally different beliefs about what AI requires.
The Signal
The Siri that shipped this week is not trying to beat ChatGPT or Claude. It's designed to replace the free tier of ChatGPT for the billion-plus people already holding iPhones. That's the tell. Apple isn't racing toward AGI. It's embedding useful AI into the workflow layer where people already live.
Craig Federighi's post-keynote tech talk made the Google partnership clearer than Monday's stage presentation. Apple uses Gemini models but nothing else from Google's stack. No Gemini Assistant code. No Google Search. No shared infrastructure. The models run on-device and through Apple's Private Cloud Compute. Federighi's exact words: "The amount of the Google Assistant we use is none."
"Apple is turning Siri into a systemwide AI interface for apps, data and workplace actions across iPhone, iPad, Mac, Apple Watch and Vision Pro."
This is the real shift. VentureBeat caught what most coverage missed: Siri AI is now an app action layer. Enterprise developers can expose app content through App Entities, make it searchable via Spotlight's semantic index, and define actions through App Intents and App Schemas. Users can ask Siri to find, summarize, update, or act on app content without opening the app or navigating menus.
The demos showed this in action. Mike Rockwell demonstrated finding specific photos and sharing them with a group without opening Photos. He showed Siri pulling restaurant recommendations from old messages, surfacing hotel confirmation numbers from buried emails, and using onscreen awareness to answer questions about visible content. Example: if you get a text about a potluck, you can brainstorm with Siri on what to bring and add a recipe to Notes, all without switching apps.
The business model is elegant. Apple doesn't need to own the frontier models. It needs to own the interface layer between users and their devices. By making Siri the orchestration layer for apps, Apple turns every third-party developer into an AI feature contributor. The more apps adopt App Intents, the more useful Siri becomes. The more useful Siri becomes, the stickier the iPhone gets.
Key technical specs:
- New Apple Foundation Models built in collaboration with Google's Gemini
- On-device processing plus Private Cloud Compute for larger queries
- Personal context understanding across messages, emails, photos, calendar
- Systemwide app actions through App Intents framework
- Onscreen awareness for contextual queries
The capital expenditure story is the quiet earthquake here. Apple projects $14 billion in AI capex for 2026. Amazon, Google, Meta, and Microsoft are spending $670 billion combined. That 48x delta isn't about Apple being cheap or behind. It's about Apple believing AI doesn't require hyperscale infrastructure to be useful. They're not training foundation models from scratch. They're not building massive GPU clusters. They're building the integration layer and letting Google (and previously OpenAI) handle the compute-heavy lifting.
"When tech companies spend hundreds of billions on capital expenditures to add sparkly AI features to Office, Gmail, and Instagram, I only see fearful atheists."
The EU regulatory block adds geopolitical texture. Apple says it can't launch Siri AI on iPhones or iPads in the European Union and that regulators are refusing to engage on the privacy architecture. This isn't a technical problem. It's a jurisdictional one. Apple built a system where user data doesn't leave Apple's control, even when using Google's models. European regulators either don't understand the architecture or don't believe it, and Apple isn't willing to compromise to ship.
The Verge noted Apple recently settled a $250 million class action lawsuit for "misleading consumers" about Apple Intelligence's availability and performance after the 2024 WWDC promises didn't materialize. That delay cost real money and credibility. Shipping Siri AI now, two years late, is both a technical milestone and a reputational repair job.
The Implication
If you're building software for Apple platforms, you now have to decide how much of your app to expose through Siri's AI layer. The companies that adopt App Intents early will become more discoverable and useful. The ones that resist will feel increasingly buried. This is the same dynamic that played out with Spotlight search and Siri Shortcuts, but the stakes are higher because Siri AI can now orchestrate multi-step workflows across apps.
For the agent economy, Apple just validated the interface-layer strategy over the foundation-model-ownership strategy. You don't need to own the LLM. You need to own the context layer where users live. The $656 billion capex gap between Apple and its peers will be studied in business schools as either the greatest arbitrage of the 2020s or the moment Apple got left behind. Early evidence suggests the former. Watch which developers rush to integrate and which hesitate. That will tell you whether Apple's bet lands.
Sources
Business Insider Tech | Daring Fireball | VentureBeat | Mashable Tech | TechCrunch AI | Bloomberg Tech | Fortune Tech | The Verge AI