Apple's AI Strategy: Let Everyone Else Pay for the Servers

Apple just had its best fiscal year on record for revenue and earnings per share, and AI is all over the story — even though the company is spending a sliver of what its biggest rivals put into the technology.

iPhone revenue hit a September-quarter record of $57.6 billion, up 17.5% from a year earlier. Mac revenue climbed 24% to $10.8 billion, its best showing in four years. Shares still dipped in premarket trading Tuesday after Apple warned that rising memory costs would eat into margins this holiday quarter. But the underlying numbers point to a company finding its footing in AI without betting the balance sheet on it.

Apple hasn't said how much of that growth comes from AI specifically versus other factors like pricing. But listening to executives talk about it, the enthusiasm is hard to miss. AI and "Apple Intelligence" came up 45 times on the earnings call, according to financial research platform AlphaSense — a new record, topping even the call after Apple first unveiled the technology and a revamped Siri back in June 2024.

"We see tremendous opportunity in AI," said John Ternus, on his first earnings call as CEO.

Three ways to get paid

CFO Kevan Parekh laid out what he called a "wide range of ways that AI can contribute to revenue growth over time." Strip away the corporate phrasing and it comes down to three moves, none of which require Apple to out-build OpenAI or Google on data centers.

First, AI gives people a reason to buy new Apple hardware. "We feel AI offers a compelling reason for customers to upgrade and switch in order to get the latest capabilities," Parekh said. Second, Apple collects a cut when customers subscribe to AI apps through its App Store. Third, it can eventually charge its own users for deeper access to Apple's AI features, the same way iCloud storage tiers work today.

That last piece is just getting started. Siri AI, the overhauled assistant, launched in beta for US English users in September and expanded to French, Japanese, Korean, Portuguese, and Spanish in October. Ternus said features that lean on server-side processing — Siri AI and intelligent photo editing among them — will carry usage limits, and customers who want more "will be able to purchase expanded access in the future." He compared it to how iCloud evolved from a simple storage product into something with layered capabilities and pricing.

Spending less than everyone else

What makes Apple's position unusual is how little it's spending to get there. The company laid out $2.2 billion in capital expenditures last quarter. Amazon spent $77.3 billion. Alphabet spent $51.4 billion. Those figures cover each company's total spending, not just AI, and don't capture every AI-related cost, but the gap is stark regardless.

Apple's approach blends on-device processing, its own servers, and other companies' models. Siri AI runs on technology built with Google's Gemini, and Apple is reportedly planning to let users pick among outside AI providers for tasks like writing tools and image generation. Operating expenses did rise 21% to $19.2 billion, partly from increased R&D spending on AI, and Siri AI's cloud costs will show up in services cost of sales going forward. Parekh said it's too early to know how much traffic will need server processing versus staying on-device — meaning those costs could grow. Even so, Apple's capital spending remains a fraction of what its AI-infrastructure-heavy competitors are laying out.

The Mac's unexpected AI moment

One of the more interesting threads from the call: Apple's computers are becoming a destination for running AI locally, without paying a cloud provider per request. Ternus said the Mac is "having an AI moment all its own," and the company pointed to real customers to make the case. Gap is running AI entirely on Mac Mini computers for loss prevention across hundreds of stores, skipping the token costs that come with cloud-based AI. Christie's is running local models on Mac Studio machines to improve workflows, including pricing estimates.

Ternus described the Mac Mini and Mac Studio as "pushing the boundaries of local, private AI, giving users the ability to do things once considered impossible." It's a notable turn for the Mac Mini in particular — a small, screenless desktop that's found a new audience among developers and hobbyists looking for an affordable way to run AI models without a subscription. Earlier this year, then-CEO Tim Cook said recognition of the Mac Mini and Mac Studio's AI chops was growing faster than Apple expected, pushing demand higher than forecast.

Collecting a toll on other people's AI

Apple also profits when customers pay for AI products it didn't build. Earlier this year, The Wall Street Journal reported that Apple was on track to pull in more than $1 billion in App Store fees from generative AI apps in 2026, citing analytics firm AppMagic.

That revenue stream has a vulnerability, though: it only works if customers pay through Apple's system. The same Wall Street Journal reporting noted that OpenAI has started steering potential subscribers toward paying on its own website instead of through the App Store, sidestepping Apple's commission entirely. It's a preview of the tension built into Apple's strategy — the company is positioning itself as the toll booth for AI traffic across more than 2.5 billion active devices, but the companies paying that toll have every incentive to find a way around it.

For now, Apple doesn't need to win the AI model race to win financially. It can sell the hardware people use to run AI, take a cut when they subscribe to someone else's AI app, and eventually charge for more of its own. That's three revenue streams built on a fraction of the capital spending its rivals are committing — a bet that owning the interface matters more than owning the infrastructure.