While competitors pour hundreds of billions into AI infrastructure, Apple chose a different path: a $1B/year licensing deal with Google. Here's what the numbers reveal.
Apple has restructured AI under Software Engineering (SVP Federighi), hiring ex-Google Gemini lead Amar Subramanya as VP. This organizational move signals that Apple treats AI as a software feature layer to be shipped, rather than a standalone research division[9], following internal turbulence and researcher exits[10].
Instead of massive CapEx, Apple signed a $1B/year deal to use Google's Gemini models (Spring 2026 rollout)[5]. This capital-efficient approach leverages Apple's $20B search revenue leverage but creates a strategic dependency—potentially reducing differentiation from Android devices running the same models[11].
Big Tech is spending at unprecedented levels to build AI infrastructure. These figures represent 2025 capital expenditure commitments.
The magnitude of difference between Apple's licensing approach and competitors' infrastructure investments.
Key considerations for evaluating Apple's AI positioning.
Apple avoids tens of billions in uncertain AI infrastructure investments while gaining access to Google's 1.2T parameter Gemini models. The $1B annual cost is roughly 0.25% of Apple's annual revenue.
If AI capabilities are licensed from Google, Apple's software experience becomes harder to distinguish from Android devices running the same models. The moat narrows to UI and hardware integration.
Apple's new AI VP (from Google Gemini) reports to Software Engineering, not the CEO — AI is positioned as a feature layer, not a strategic function. This aligns with the licensing approach.
Whether Apple can create meaningful differentiation through fine-tuning, UI design, and hardware ecosystem integration — or whether the model layer becomes the primary driver of user experience and value capture. The answer likely depends on whether AI model capabilities continue to diverge rapidly or begin to plateau and commoditize.