Apple’s Siri AI: Spending $1B to Rent Google Gemini Instead of Building

The implications of this rent-versus-build divide are profound. For Apple, the near-term cost savings are dramatic; its $6.8 billion in capex is less than 1% of the combined peer spend. This translates directly into massive free cash flow that supports dividends, buybacks, and the new iPhone Upgrade leasing program with Klarna—starting at $17.99 per month—which could smooth demand cycles and lock in recurring revenue. Reporting suggests the leasing rollout, covering iPhones, Macs, and iPads, is now underway, deepening Apple’s transformation into a subscription-like hardware business. At the same time, rumors of a foldable iPhone have intensified, hinting at the next growth catalyst. For the AI industry, Apple’s strategy validates an alternative path: companies can offer competitive AI experiences by paying for access to foundational models rather than training and hosting them. That puts pressure on Google, OpenAI, and others to become the infrastructure layer for device-integrated AI, potentially shifting the balance of power from hardware owners to model providers. However, the arrangement also concentrates risk. Apple’s Siri becomes deeply dependent on Google’s model quality and pricing, and any strategic shift at Google could disrupt the experience. Meanwhile, the leasing model could provide Apple with a steady stream of used iPhones, enhancing its refurbished business and recycling margins.

Looking ahead, two transitions loom large. First, the September 1 handover from Tim Cook to John Ternus as CEO—Cook moves to executive chairman—introduces leadership uncertainty just as the company navigates supply constraints and a potential foldable iPhone launch. Ternus, a hardware engineering veteran, may double down on product-led growth, but his stance on AI investment will be closely watched. Second, the AI capital-spending gap is unlikely to persist indefinitely. As on-device AI demands rise, Apple may eventually need to invest in its own silicon or more advanced inference infrastructure, which could compress those eye-catching free-cash-flow margins. For now, Apple is betting that a $1 billion annual rental is a smarter wager than the $200-billion-plus annual outlay its competitors are committing to. If successful, it will show that AI leadership can be achieved without a data-center land grab—rewarding shareholders with cash returns while rivals grapple with monetization far down the road.