Apple Inc. (NASDAQ: AAPL)

What the company does

The problem Apple solves is not a shortage of computing power. It is the cost of assembling computing power into something that works. A buyer who wants a phone, a laptop, a watch and headphones that share files, calls, passwords, payments and photographs without configuration has, in practice, one supplier who will sell all of it as a single designed system. Apple charges a very large premium for removing that assembly problem, and the premium is the business.

Trace a single iPhone. Apple designs the processor and has TSMC fabricate it on a leading-edge node — TSMC earned a 67.7% gross margin in the June 2026 quarter with 77% of wafer revenue from 7nm and below, so this is expensive but contractually stable supply. Memory, displays, cameras and radios come from outside suppliers, many single-sourced: Apple’s FY2025 10-K notes that new products “often utilize custom components available from only one source.” Final assembly is done by partners located primarily in China mainland, India, Japan, South Korea, Taiwan and Vietnam. Sixty per cent of finished product moves through carriers, resellers and retailers; 40% through Apple’s own channels.

Apple collects cash fast. At 27 June 2026 receivables were $31.4bn against $466.8bn of trailing revenue, while accounts payable of $69.9bn exceeded receivables outright. Suppliers finance the working capital, not Apple and not the customer.

The second sale is the more important one

The phone becomes an active device and stays one for about three and a half years. Apple says its installed base exceeds 2.5 billion active devices and that paid subscriptions passed 1.5 billion. Counterpoint independently estimates more than a billion active iPhones — nearly one in four active smartphones worldwide. Against that base Apple earned $120.5bn of Services revenue in the twelve months to June 2026 at a 75.6% gross margin.

That is the whole architecture in one ratio: Services is 25.8% of revenue and 40.3% of gross profit. Services contains the App Store commission, advertising, iCloud, subscriptions, AppleCare, payments and licensing — chiefly the agreement under which Google pays for default search placement in Safari. Apple discloses none of them separately. Worth noting: the FY2025 10-K named “advertising, the App Store and cloud services” as Services growth drivers; the June 2026 10-Q names “advertising and cloud services” and drops the App Store. That omission is the only signal in the filings that the App Store may have stopped contributing.

Industry structure and what actually protects the position

The smartphone industry inverted in 2026, and no assessment written on pre-2026 assumptions survives it. IDC forecasts global unit shipments down 16.7% this year to just over a billion — “the steepest annual contraction the industry has ever recorded” — while market value rises 6.3% to $613bn on a 27.6% increase in average selling price. Units are collapsing and revenue is growing. Any framing of Apple built on unit share now describes the wrong variable.

Apple sold roughly 21% of the world’s smartphones in the June quarter and captured 49% of the industry’s revenue, its highest-ever second-quarter share. In the premium tier — $600 wholesale and above, now 29% of the whole market — Apple holds 65% and Samsung 19%.

The profit picture is starker, and it rests on competitors’ own filings rather than a tracker estimate. Samsung Electronics posted record group operating profit of KRW 89.5tn in Q2 2026 — and its Mobile eXperience division, the world’s largest smartphone business by units, lost KRW 0.7tn on KRW 33.2tn of revenue. Effectively all of Samsung’s record profit came from selling memory to the industry, not from selling phones. Xiaomi’s smartphone gross margin fell from 12.4% to 10.1% to 8.5% across three quarters. Against these, Apple’s 48.7% company gross margin is not an incremental advantage; it is a different order of business.

Which barriers actually bind

Several commonly cited barriers do not bind here. Manufacturing scale does not — Apple owns no fabrication and no assembly, and its long-lived assets outside the United States totalled just $9.6bn at FY2025 year-end. Brand alone does not — Samsung has one and lost money on handsets last quarter. Operating-system share does not — Android runs on roughly three-quarters of the world’s phones.

What binds is holding silicon, operating system and distribution simultaneously, and 2026 showed the mechanism plainly. As memory prices rose, every Android vendor faced a choice between margin and price. Most raised prices: Xiaomi’s average selling price rose 25.9% year on year while its gross margin still fell to 8.5%. Apple and Huawei held price. IDC’s account of China in Q2 2026 is explicit — the two “both held prices steady while the rest of the Android field raised them.” Apple grew 24.4% in a market that shrank 4.3%.

Apple could hold price because it had 40 points of product gross margin to spend and a Services annuity underneath it. Xiaomi, at 8.5%, had nothing to spend. A well-funded entrant could reproduce Apple’s hardware. It could not reproduce the 40-point buffer, because the buffer comes from an installed base that takes twenty years to accumulate rather than capital to buy.

The kind of company that wins here is the one that captures the value it creates rather than the volume it ships — volume in consumer hardware is a commodity, and the profit pool sits with whoever owns the customer relationship after the sale. Apple is that kind of company to an unusual degree: 21% of units, 49% of revenue, and on the evidence of Samsung’s and Xiaomi’s own filings, very close to all of the industry’s positive handset profit.

The growth engine

Apple grew revenue 16.1% in the nine months to June 2026 — roughly three times its FY2021–FY2025 average, which requires explanation rather than acceptance. Unusually, reported growth is organic growth: across FY2021–FY2024 the 10-Ks disclose no business combination, no goodwill line and no named acquisition. Maximum disclosed acquisition spend in any year was $306m, under 0.08% of revenue, and the acquisitions line was removed from the cash flow statement entirely from FY2023.

$ millions, FY2026 Q3 10-Q. Growth shares computed from the disclosed figures. Acquisitions contributed nothing — none is disclosed. iPhone and Services together produced 93% of all growth.

Margin, cash and capital allocation

Apple’s gross margin has risen for seven consecutive years, from 37.8% in FY2019 to 48.7% trailing. Almost none of it came from making hardware more cheaply.

Two mechanisms account for it. Mix: Services went from 17.8% of revenue to 25.8%, and carries a 75.6% margin against 40.1% for Products — shifting eight points of revenue from a 40-point margin to a 76-point margin lifts the blended rate by roughly three points before anything else happens. Services margin itself: up from 63.7% to 76.3%, as the mix within Services moved toward advertising and licensing, lines with almost no incremental cost of delivery.

The Products line is currently distorted, and this matters more than anything else on the page. It fell to 36.8% in FY2025 “primarily due to a different mix of products and tariff costs,” then rose to 40.1% in Q3 FY2026 “primarily due to a different mix of products and tariff refunds, partially offset by higher costs, including memory.” The refunds follow the US Supreme Court’s 20 February 2026 ruling striking down certain IEEPA tariffs; Apple applied for a refund and booked it as a reduction of products cost of sales. Apple does not disclose the amount. The record 50.1% consolidated gross margin in the June quarter therefore includes a recovery of prior-period costs that cannot recur and cannot be sized from outside.

Comparability: FY2023 spanned 53 weeks against 52 for FY2022 and FY2024, so any growth rate crossing it is distorted. FY2019 figures are the comparative columns of the FY2021 10-K, not the original filing. The FY2023 and FY2024 10-Ks each state that “certain prior period amounts have been reclassified” without specifying which. Segment expense detail was disclosed for the first time in FY2025 under ASU 2023-07. The trailing gross margin includes an undisclosed tariff-refund benefit. Trailing EPS is a sum of period EPS, not a recomputation on a weighted share count.

Cash conversion, and where the AI money is hiding

Cash converts better than profit: trailing operating cash flow of $146.7bn against net income of $128.9bn, less $10.0bn of capital expenditure, leaves $136.7bn of free cash flow — 29% of revenue. Depreciation and share-based compensation exceed a capital base that barely grows, and working capital is negative.

One line deserves attention. Capital expenditure in the nine months to June 2026 was $6.8bn, down 28% from $9.5bn — falling, while Alphabet guides to $195–205bn of 2026 capital spending, Amazon to approximately $220bn and Meta to $130–145bn. At the same time Apple’s gross intangible assets rose $12.1bn in nine months with no acquisition note explaining it, other purchase obligations roughly doubled to $29.3bn, R&D rose 32% in the June quarter on “infrastructure-related costs, including investments in artificial intelligence,” and inventories nearly doubled to $11.1bn. Apple has said it uses “some third-party cloud and we do our own data centers,” and announced in January 2026 that its next-generation foundation models will be based on Google’s Gemini models and cloud, with no commercial terms disclosed. The conclusion is that Apple’s AI spending is real and rising but lands in operating expense, cost of sales, intangibles and purchase commitments rather than capex (inferred). Its capex line understates it and is not comparable to a hyperscaler’s.

Where the cash goes

Ranked over FY2019–FY2025: buybacks $577.8bn, dividends $103.2bn, capital expenditure $75.7bn, debt reduction, and acquisitions last at effectively nothing. Diluted share count fell 20.9%. Buybacks have run at roughly 5.6× capital expenditure and around 40× the entire disclosed acquisition programme.

That ranking says something specific about how management thinks: this is a company that believes it already owns the assets it needs and treats its own equity as the best available use of surplus cash. It has never made a large acquisition, carries no goodwill, and has told shareholders only that it “intends to increase its dividend on an annual basis” — which it has, by roughly four cents a year, alongside a fresh $100bn repurchase authorisation each May. The balance sheet has quietly repaired: total debt down from $124.7bn at FY2021 year-end to $84.3bn, cash and securities at $146.5bn, a net cash position of $62.2bn against $33.8bn nine months earlier, and retained earnings back to a positive $11.3bn from an accumulated deficit. Apple is now buying back less than it earns.

No transaction post-dates the reported figures. One material event does: John Ternus became Chief Executive Officer on 1 September 2026, with Tim Cook moving to Executive Chair — the first change in fifteen years. Ternus’s FY2027 equity award is set at $55m of target value, 75% in performance shares vesting on total shareholder return relative to the S&P 500, the same structure and 55th-percentile target that governed Cook’s awards.

Cyclicality: where this business sits right now

On its own history, Apple is at a peak on every metric that matters. Revenue, gross margin, operating margin and Greater China segment profit are all at or near records; the June quarter’s 50.1% consolidated gross margin is the highest ever recorded, against 37.8% at the FY2019 trough. Greater China revenue of $64.8bn in nine months already exceeds the whole of FY2025.

The industry is at the opposite point of a different cycle, and that is the central tension. TrendForce reported conventional DRAM contract prices up 90–95% quarter-on-quarter in Q1 2026 — “the steepest increases in their history” — then 58–63% in Q2 and 13–18% in Q3; mobile LPDDR5X rose 78–83% in Q2 alone. IDC states NAND and DRAM costs are up more than 300% year on year and expects prices to keep rising until at least 2028. The supplier economics confirm where the profit went: Micron reported quarterly revenue of $41.5bn against $9.3bn a year earlier at an 84.6% gross margin; SK hynix grew operating profit 557% to a 76% operating margin. Apple sits on the paying side of that transfer. Samsung manufactures memory; Apple buys it.

The downside case is therefore not a demand recession but a margin compression Apple has already described. Memory contracted at 2026 prices flows into cost of sales with a lag of several quarters as inventory turns; the tariff refunds currently offsetting it are non-recurring; the pre-purchased inventory cushion is finite at $11.1bn; and the options are to absorb the cost in Products margin or raise prices — which the 10-Q says “may also reduce demand,” and which would forfeit the share gains won by holding price. IDC expects Apple’s 2026 units down just 1.3% against 24.3% for Android, so the exposure is to margin, not volume.

End-market exposure

$ millions, FY2026 Q3 10-Q segment note. Segment operating margins computed from disclosed segment operating income and net sales; they exclude $40.7bn of corporate R&D and G&A and are therefore not comparable to the 33.6% company operating margin.

What to monitor, and where it is published

Risks beyond the cycle

The Google search payment is a large, undisclosed, near-100%-margin revenue line under active appeal. The trial record in United States v. Google put Google’s 2022 payments to Apple at $20bn, and total distribution payments across all partners at $26.3bn in 2021. Judge Mehta declined to ban the payments in his 2 September 2025 remedies opinion but limited agreements to one-year terms and prohibited exclusivity. The DOJ and co-plaintiff states are now cross-appealing precisely to have a payment ban reinstated. Because this revenue carries essentially no cost of delivery, it falls to Services gross profit roughly one-for-one — and Apple has never disclosed the amount, so an outside investor cannot size what is at stake. Apple’s own 10-K says the remedies the DOJ originally proposed “could materially adversely affect the Company’s ability to earn revenue from such licensing arrangements.”

The App Store’s commission architecture is being reset in three jurisdictions on three timetables, and it is already visible in the data. In the US, Apple has collected no commission on external-link purchases since April 2025 and has asked the District Court to approve 15%, 10% and 5% rates. In the EU, the Core Technology Fee and Initial Acquisition Fee are eliminated on 1 October 2026 and replaced by a flat 5% Core Technology Commission. Japan’s Mobile Software Competition Act came fully into force in December 2025. Independent trackers report Apple’s US App Store commission revenue down 18% in 2026 to date and US App Store consumer spending down 6% year on year in the June quarter — the first decline in a decade. The compounding risk is that each jurisdiction’s remedy becomes a template for the next.

The Digital Markets Act carries a penalty that is large relative to the business. The Commission fined Apple €500m in April 2025 for breaching the anti-steering obligation, under appeal; the separate Article 6(4) proceeding remains open, and Apple’s 10-K notes the Commission “may impose fines up to 10% of the Company’s annual worldwide net sales” — roughly $41.6bn on FY2025 revenue. The General Court dismissed Apple’s gatekeeper-designation challenge entirely on 8 July 2026. Separately, Apple announced in June 2026 that its new Siri will not ship in the EU on iPhone or iPad, with “no timeline” for availability — regulation removing a product from a market rather than changing its price.

Manufacturing geography concentrates the same counterparties the component shock is already stressing. Substantially all final assembly sits with partners primarily in Asia, many components are single-sourced, and a US Section 232 semiconductor investigation is open. A trade measure and a component shortage would arrive through the same supply chain at the same time — and Apple’s FY2025 10-K declines to quantify tariff exposure at all, saying only that “the ultimate impact remains uncertain.”

Every senior executive role turned over within twenty months. CEO on 1 September 2026, CFO in January 2025, COO in July 2025, General Counsel in March 2026, Principal Accounting Officer in January 2026. Each transition was planned and internal, and Apple’s succession record is good. The risk is timing rather than capability: a new leadership team is taking a margin decision on component costs, an AI platform transition and three concurrent regulatory resets, none of which it has faced before in these seats.

The AI position depends on a partner and has a two-year delivery record behind it. Apple announced a personalised Siri in June 2024, delayed it publicly in March 2025 to “the coming year,” and at WWDC in June 2026 was still describing it as a beta shipping “later this year.” Counterpoint’s assessment of whether AI is yet driving hardware upgrades is negative: the capability “has yet to give consumers a compelling reason to upgrade.”

What the sources could not answer

App Store revenue and gross billings, which Apple has never disclosed. The Google licensing payment, for which the only sourced figure is $20bn for 2022 from the trial record. The FY2026 tariff refund and residual tariff cost, both narrative-only — so the sustainable Products gross margin cannot be computed. The $12.1bn nine-month increase in gross intangible assets, for which the 10-Q offers no note. iPhone unit volumes, unreported since FY2019. Memory contract terms — how much of Apple’s 2027 requirement is contracted, at what price, for how long, which is the single most valuable disclosure that does not exist. And management’s own framing: the research folder holds no transcripts, investor-day materials or shareholder letters, and the Item 2.02 Forms 8-K contain only cover pages rather than the press-release exhibits, so analyst pushback and management candour were not assessable.

Takeaways