Apple is no longer the default “largest company on earth.” In September 2026 that title sits with NVIDIA. Apple is still number two, around $4.7–$4.9 trillion depending on the trading day — a consumer-hardware-and-services machine rather than an AI-picks vendor. The interesting case study is how it stays that rich when it is not the AI pure-play.
The flywheel (unchanged, still brutal)
- Sell a phone people replace on a cycle.
- Lock a high-margin accessory and wearables ring around it.
- Convert the installed base into Services (store, cloud, media, payments).
- Use retail and developer rules to keep the garden valuable.
NVIDIA sells to five giant clouds. Apple sells to hundreds of millions of individuals. That diversification is why a “lost AI narrative” has not collapsed the company the way it would crush a single-product GPU name.
AI is the awkward chapter
Apple Intelligence needed partners. Reporting in 2026 has pointed to on-device features and cloud models licensed from others (including Google Gemini for some Siri-class tasks). That is not failure — it is make-vs-buy. The risk is narrative: if buyers believe the iPhone is “late to AI,” mix shifts to Android in the upgrade cycle. The defence is still camera, battery, and the App Store bill, not a research paper.
Pricing power in a tired consumer
A reported ~$100 lift on a new iPhone generation is a test of whether the brand still taxes the installed base. Memory and component inflation is real. If households stretch, they keep last year’s phone — Apple’s Services revenue can still grow while unit growth stalls. Watch Services mix, not only launch-day queues in Mumbai.
India: factory + store, not just a TAM slide
India is assembly, stores, and a young upgrade ladder. For local retailers, Apple is a traffic magnet with thin authorised-reseller margins. For component MSMEs, it is a quality-system shock: you either make the list or you do not. For D2C brands on iOS, Apple’s tax is the store commission and privacy rules that make ads harder — budget CAC accordingly.
What a smaller brand can copy
- Attach, don’t discount the hero. Apple protects iPhone ASP and sells the watch and cloud around it. Kirana version: keep the hero SKU priced; bundle the cable and cover.
- Installed base > launch hype. Recurring services beat one festival sale.
- Buy AI if building it is slower. Apple licensed. Your SaaS can too — own the workflow, not the model.
- Retail as theatre. A clean store (or a clean WhatsApp catalogue) is part of price.
Mini case: an authorised reseller in Bengaluru
Launch week looks like a revenue miracle. Contribution margin may not. EMI plans, old-phone trade-ins, and staff overtime eat cash. Compare launch-week deposits to the following six weeks of accessory and AMC attach. If the spike is only hardware at thin margin, you ran a crowd, not a business.
FAQ
Did NVIDIA “beat” Apple forever?
Market-cap crowns rotate. Apple’s cash and buybacks are a different species of company from a GPU cycle. Do not treat the ranking as a morality play.
Is this investment advice?
No. It is a business-model read for operators who sell to the same humans Apple sells to.
Pair with the NVIDIA case study · All posts.