Apple

Apple is usually described as a design company. That description explains its prices and none of its profits. The mechanism is simpler and less flattering: Apple sells a durable, expensive piece of hardware roughly every three to four years, and while you own it, it rents you the things that make the hardware usable — iCloud storage, App Store distribution, Apple Music, and the default search box that Google pays for. Hardware is the installment. Services is the annuity.

The numbers make the split explicit. In FY2025 Apple did approximately $416B of revenue, of which about $108B was Services. Services carried a gross margin near 75% against roughly 37% on products. That means Services, at about a quarter of revenue, contributed close to 40% of gross profit — and essentially all of the incremental gross profit over the past five years, because iPhone unit volumes stopped growing around 2015.

The strategic consequence is that Apple's job is no longer to sell more phones. It is to hold the 2.35 billion active devices already in circulation inside the account, and to raise revenue per account. Every product decision since 2015 — Watch, AirPods, iMessage reactions, Photos libraries, Family Sharing, Apple Pay — reads correctly only as a switching-cost decision.

The received explanation for Apple's moat is design, and design is not a moat — it is copyable within eighteen months and has been, repeatedly. The moat is switching costs, and they are unusually well engineered because Apple built them out of things a customer would never describe as lock-in.

Consider what a user actually abandons by leaving. A photo library that has been continuously ingested for a decade and does not export cleanly. An iMessage thread history that does not migrate, attached to a phone number that will start silently dropping messages from Apple users if deregistered incorrectly. Purchased apps and in-app subscriptions bound to the Apple ID. A Watch that becomes a paperweight. And, in the United States, a social cost: among American teenagers, iPhone ownership runs above 85%, and the group chat degrades visibly when one member leaves. Apple did not have to enforce that. Its users enforce it on each other.

The second layer is silicon. Since the M1 in 2020, Apple designs its own processors and buys manufacturing capacity from TSMC rather than buying finished chips from Intel or Qualcomm. That converts a supplier relationship into a cost advantage: Apple captures the margin an Intel would have charged, and it ships performance-per-watt that a competitor assembling merchant silicon cannot match on the same schedule.

The honest limit: this moat protects the installed base, not the next platform. Every Apple transition since 2001 was one Apple itself initiated. The current one is not.

Apple's capital allocation is the most consequential in the history of public markets by sheer size and the least discussed. Since 2012 the company has repurchased well over $700B of its own stock, retiring roughly 40% of shares outstanding. That means an investor who held Apple flat through the last decade watched their claim on earnings grow by two-thirds without buying a share. Roughly a third of Apple's per-share earnings growth is arithmetic, not operations — and the market has consistently rewarded it as though it were operations.

The corresponding criticism is that Apple has been an unusually timid acquirer. Its largest deal ever remains Beats at $3B in 2014. In an era when a language model with genuine capability could have been bought for a fraction of a single year's buyback, Apple bought back stock instead, and then had to license Google's Gemini to power Siri. That is the sharpest available indictment of buybacks as a default: the option value Apple declined to purchase was worth more than the shares it retired.