Apple Collected Nearly $1 Billion From AI Apps It Didn’t Build — And the Strategy Is Working
Apple spent the last two years watching OpenAI, Google, Anthropic and every other AI lab race to build the most capable model. Then it handed them all an invoice. In 2025, Apple collected nearly $900 million in App Store commissions from generative AI applications alone — without building a competitive AI product of its own. It is, by any measure, one of the most elegant monetization strategies in tech history.
The figure comes from mobile intelligence firm AppMagic, whose analysis was first reported by The Wall Street Journal. Under Apple’s standard terms, AI companies pay a 30% commission on subscriptions in the first year, dropping to 15% annually thereafter. Every ChatGPT subscription sold through an iPhone, every Claude or Gemini upgrade purchased on iOS — Apple takes its cut. The company is now on track to cross $1 billion in AI app revenue in 2026.
ChatGPT Carries the Load
The revenue is heavily concentrated. ChatGPT, which recently reported 900 million weekly active users, accounts for roughly 75% of Apple’s AI commission income. Grok, Elon Musk’s xAI chatbot, comes in a distant second at around 5%. Apple’s monthly AI revenue climbed from approximately $35 million in January 2025 to a peak of $101 million in August, before slipping somewhat as ChatGPT download growth slowed.
What makes this position extraordinary is the asymmetry of effort. Amazon, Alphabet, Meta and Microsoft collectively spent close to $400 billion on AI infrastructure in 2025. Apple spent roughly $12.7 billion — a fraction of its rivals — and generated near-billion-dollar AI revenue anyway. Its 2.5 billion active devices and 850 million average weekly App Store users give it a structural toll-road position that no amount of model training can replicate.
Apple doesn’t need to win the AI race. It just needs AI to remain primarily delivered through smartphones. As long as the iPhone is the world’s dominant consumer device, every AI subscription sold to an iPhone user flows through Apple’s payment infrastructure — regardless of which model powers it.
Siri Remains the Weak Link
The irony is hard to miss. Apple’s own AI offering — Siri — is widely regarded as the weakest product in the field by modern standards. The company has been working to overhaul Siri with large language model capabilities, but the rollout has encountered significant delays. Testing revealed the upgraded assistant can be slow to handle requests and sometimes fails to process queries entirely. The full AI-powered Siri, which would bring contextual awareness, improved follow-up handling, and the ability to control apps by voice, has been pushed to later in 2026.
Rather than race to catch up internally, Apple announced in January that Google’s Gemini will power a revamped version of Siri. The partnership effectively outsources the hardest AI problem to a competitor, with Bloomberg reporting the deal is valued at approximately $1 billion annually. The arrangement adds a layer of financial complexity: Google already pays Apple around $20 billion per year to remain the default search engine on iPhones. Now Apple is paying Google for AI capability — though at a dramatically smaller scale.
Apple is prioritizing on-device AI — smaller, privacy-preserving models that run locally — over the large data center infrastructure its rivals are racing to build. Gemini fills the gap for tasks requiring more powerful cloud-based reasoning, keeping capital expenditures flat while delivering a credible Siri upgrade.
The Threat That Could Undermine the Whole Setup
Apple’s toll-road advantage depends on one assumption holding: that consumers keep using iPhones as their primary interface with AI. OpenAI is working to disrupt exactly that. Following its $6.5 billion acquisition of AI device startup io — founded by former Apple designer Jony Ive, who created the iPhone and iPad — OpenAI is developing a new AI-native device intended to compete directly with smartphones. If that device, or any successor hardware category, gains meaningful traction, it would route AI subscriptions outside the App Store entirely, cutting Apple off from its passive AI revenue stream.
For now, that threat remains speculative. OpenAI’s device hasn’t shipped, and smartphones aren’t going anywhere in the near term. But the longer Apple delays a competitive AI product of its own, the more it depends on the hardware status quo remaining intact.
Apple’s AI strategy is essentially a bet that distribution beats capability — that controlling the device in people’s pockets is more durable than controlling the best model. So far, the financials support that thesis. Whether it holds as AI moves toward new form factors is the defining question for Apple’s next decade.
Sources
Entrepreneur, “Apple Made Nearly $1 Billion Last Year From a Technology It Doesn’t Even Lead” — Sherin Shibu (March 2026)
MacRumors, “Apple Made Nearly $900 Million From Generative AI Apps Last Year” — Tim Hardwick (March 2026)
