Apple

Every leg of the mechanism — installed base to premium hardware, a 75.6%-margin Services annuity, and a fortress FCF/buyback engine — verifies strongly against the primary Q3 FY2026 8-K: FCF $110.2B at a 30.2% margin in nine months, capex 1.87% of revenue and down 28.2% YoY, ~$62.2B net cash, installed base above 2.5B, and Greater China +22.4% YoY. What changed is the rate, not the machine: the cycle is extending at a decelerating reported rate rather than reaccelerating.

• The cycle extended a second quarter, at a decelerating reported rate — not a reacceleration: Q3 FY2026 revenue $109.42B (+16.4% YoY) after Q2's +17%, 9M FY2026 $364.4B (+16.1%), with June-quarter records for total revenue, EPS, operating cash flow, and for iPhone, Mac and Services, and double-digit growth in every geographic segment. iPhone $54.25B (+21.7%), Mac $10.35B (+28.7%)
• Q4 FY2026 is guided +9% to +11% reported against a disclosed 2.5pp FX headwind — ~+11.5% to ~+13.5% constant currency, midpoint ~+12.5%. On the only basis that speaks to the business, the guide is still low-double-digit, which is what the entry published. FX is an accounting translation, not a franchise variable
• The Services margin-mix engine verified intact and is the durable half of the annuity: Services gross margin 75.6% vs Products 40.1%, lifting blended gross margin to 50.1% in Q3 (~48.1% ex the ~2pp tariff-refund benefit) and 49.1% for 9M FY2026. The margin leg holds; the acceleration framing does not, and we drop it (see bear point 1)
• The cash engine is running at a materially higher rate: 9M FY2026 operating cash flow $117.0B (+43.1% YoY) less $6.8B capex = $110.2B FCF, a 30.2% margin, funding $62.1B of buybacks and $11.8B of dividends in nine months
• Balance sheet is strong: ~$62.2B net cash at 2026-06-27 ($146.5B cash and marketable securities vs $84.3B total debt — $2.0B commercial paper, $11.0B current term debt, $71.3B non-current)
• The asset-light differentiation strengthened rather than merely held: capex 1.87% of 9M FY2026 revenue and down 28.2% YoY ($6.8B vs $9.5B in 9M FY2025). This is the profile explicitly credited for Apple passing Nvidia as the most valuable company on 2026-07-17 and touching ~$5.036T intraday on 2026-07-28 — while the SOX sat >25% below its 2026-06-22 record close
• Greater China is a growth leg, not the erosion story we wrote: $18.82B in Q3 (+22.4% YoY) and $64.84B for 9M (+30.0% YoY). Our published signal — '-3.8% and eroding', lean Bearish — was directionally wrong, and wrong in the bear's favour. It is corrected to neutral-to-bullish here, with Siri AI regulatory approval in China still pending (only the original Apple Intelligence features 'just approved', management 2026-07-30) — so this growth was achieved with the AI leg unlanded
• Installed base is 'over two and a half billion active devices', an all-time high across all major product categories and geographic segments (company-stated, 2026-07-30) — above the ~2.35B we carried
• Guide-versus-result base rate is a beat, not a miss: the last completed pair was Q3 guided +14% to +17% revenue and delivered +16.4%, with gross margin of 48.1% ex-tariff landing at or above both candidate stored guide ranges (46.5-47.5% as we recorded it; 47.5-48.5% per a secondary source) — i.e. Apple met or beat its own memory-pressured margin guide under either reading

The annuity is genuinely softer than we published and the near-term print is capped on two sides. Services grew +12.1% in Q3 against +16% in Q2 and Q4 is guided 'largely similar' (~+14.5% constant currency after the 2.5pp FX drag) — flat, not compounding upward. Ex-tariff gross margin compresses from ~48.1% to a ~46.5% Q4 midpoint into what Cook calls 'a 100-year flood' in memory pricing, a newly named advanced-node supply constraint is expected to 'increase significantly sequentially' across iPhone, Mac and iPad, R&D stepped up 32.3% YoY, and the forward multiple may be wider than the ~31x we published.

• Correcting our own bull framing: Services is not 'compounding double-digit and accelerating'. It grew +12.1% ($30.74B) in Q3 FY2026 against +16% in Q2, and Q4 is guided 'largely similar' to +12% reported — ~+14.5% constant currency once the disclosed 2.5pp FX headwind is added back. The 75.6%-margin mix engine is intact; the acceleration claim is withdrawn. Note this is softer, not falsified — a flat mid-teens constant-currency annuity is still the highest-quality revenue in the business
• Ex-tariff gross margin compresses next quarter: Q3's reported 50.1% included ~2pp of tariff refunds (a $0.11 EPS benefit) and the Q4 guide of 47%-48% assumes only ~1pp — a ~46.5% ex-tariff midpoint against Q3's ~48.1%. Cook (2026-07-30) called memory 'a 100-year flood on the memory pricing, with exponential increases', said Apple 'reluctantly raised prices', and expects still-higher memory costs in the September quarter. This is an industry-wide input-cost shock, visible in the SOX closing 2026-07-31 at 11,311.1 (>25% below its 2026-06-22 record close of 14,634.72) and the KOSPI's -28.9% intramonth month
• A supply constraint caps the print: management expects advanced-node semiconductor availability constraints to 'increase significantly sequentially' across iPhone, Mac and iPad. Stated honestly, this is deferred revenue rather than lost revenue — a supply limit is not demand weakness — but it caps near-term reported growth, and the deferral is unproven until the units show up. It converts into a genuine bear fact only if the constraint 'resolves' without those units returning in Q1 FY2027
• Operating leverage narrows: R&D rose 32.3% YoY to $11.73B, a step-change, and Q4 opex is guided to $19.1-19.4B. A ~30%+ spend ramp against a guided low-double-digit constant-currency revenue rate compresses the incremental margin the bull case leans on. Balance-sheet corroboration exists (net intangibles rose from $11,093M at FY2025-end to $20,342M at 2026-06-27) but the cause is not disclosed and we do not attribute it
• The AI dependency is now architectural, not a lag: at WWDC26 (2026-06-08) Apple rebuilt Siri ('Siri AI') on a custom Google Gemini model running in Apple's own data centres, making Apple both payee and payer in the same relationship. This sharpens the partner-dependency risk rather than newly introducing it. Its economics (payment, model scale) are secondary-press only and are excluded here; we do not know what this costs
• The ~$20B/yr Google default-search payment remains unadjudicated: Google's opening appeal brief was filed 2026-05-22 defending the deal as won 'fair and square', oral arguments are unscheduled, and resolution is expected late 2026 or early 2027 — a live risk to near-pure-profit Services dollars, with no near-term catalyst either way
• Category and concentration weak spots: iPhone is roughly half of revenue ($54.25B of $109.42B in Q3) in a mature category, and iPad was the one declining line (-5.9%, $6.19B)
• Governance transition into a cost-inflating, supply-constrained quarter: Ternus becomes CEO 2026-09-01, Cook becomes Executive Chairman, Levinson lead independent director (announced 2026-04-20). Framed honestly, it is a watch item on capital-allocation continuity, not evidence of stress — the announcement was public for fourteen weeks before AAPL printed a ~$5.036T record on 2026-07-28
• The multiple is stretched: at the $308.91 close on 2026-07-31, stockanalysis.com shows 33.77x forward and 35.44x trailing. The forward year and the consensus behind the forward figure are not disclosed, so treat the forward multiple as indicative rather than precise — the premium is wide on either reading