Apple Posts Its Best Year Yet, While Memory Costs Loom Over the Holiday Quarter
Apple reported its best fiscal year ever for both revenue and earnings per share. iPhone revenue rose 17.5% from a year earlier to a September-quarter record of $57.6 billion, according to The Verge. Mac revenue climbed 24% to $10.8 billion, its highest level in four years.
Apple also warned that rising memory costs would squeeze profit margins in the holiday quarter. Operating expenses rose 21% to $19.2 billion as the company increased research and development spending, including spending on AI.
On capital expenditure, which is money spent on data centers, servers and other long-lived infrastructure, Apple spent $2.2 billion last quarter. Amazon spent $77.3 billion over the same period and Alphabet spent $51.4 billion, figures reported by The Verge.
John Ternus took part in his first earnings call as Apple CEO. AI and Apple Intelligence were mentioned a record 45 times on the call, according to AlphaSense.
Apple describes its AI approach as a mix of on-device processing (models that run on the iPhone or Mac itself), its own servers, and outside AI technology. The server component builds on Private Cloud Compute, which Apple introduced in 2024. It handles requests too heavy for the device to process alone.
The revamped Siri, which Apple says is powered by Apple Intelligence, launched in beta for US English users in September. In October, Apple expanded it to French, Japanese, Korean, Portuguese, and Spanish.
The Siri models are developed with Google's Gemini technology. Reuters reported in January that Apple had entered a multi-year deal to use Gemini models for the revamped Siri, and the two companies announced at the time that Gemini would power the next-generation Siri. In November 2025, CNET reported that Apple would pay Google roughly $1 billion a year for a custom Gemini model to run Siri.
Apple shares rose 15% in July, according to Bloomberg, which reported ahead of the results that the stock had risen 22% in 2026 as of late July.
The broader context here is a divergence in how the largest platform companies spend on AI. Apple's capital outlay is a small fraction of its hyperscale peers', and its roadmap leans on on-device silicon, a private server tier and a licensed third-party model rather than on building data-center capacity at Amazon or Alphabet scale. That is a choice about where compute lives and who owns the model, and so far the earnings show it working in hardware sales.
In my view, the more open question is the holiday quarter. Memory costs are a supply-side input that touches every device Apple sells, and the company has already said they will compress margins. Whether Siri's expanding language support and the on-device model translate into upgrade demand will determine whether the AI push pays back through the hardware cycle. For users, the practical promise is a more capable assistant that does much of its work on the handset, with heavier requests routed to Apple's own servers. Whether that promise holds at scale is what the next few quarters will test.
