Apple (AAPL) — The Control Group Gets the Bill

Apple closed its best fiscal year with iPhone revenue up 21%, again spending seven cents of capex per dollar of net income. The AI build reached it anyway, through memory: gross margin without tariff refunds fell to about 46.9%, and the December guide is 45% to 46% with memory pricing "unprecedented" and still rising. We value Apple at $171.34, from $208 in July, against $338.32 at the close before the release.

TON618 Capital Research · As of 2 November 2026 · Download PDF · Download model

Our July note called Apple the control group of the AI build: the one large technology income statement with no depreciation wave coming, because it was not building. That finding held. What changed is where the build reaches Apple. Memory makers are moving capacity to servers, and the company that spends seven cents on plant now pays for the shortage in its gross margin. This note asks what a record year is worth once that bill is counted.

Terms used throughout, defined once. Fiscal years: Apple's year ends in late September, so FY27 runs from 27 September 2026 and the December quarter is its first. Ex-refund: a figure without the tariff refunds Apple disclosed, about 2 points of gross margin and $0.11 of EPS in the June quarter and about 1 point and $0.05 in the September quarter. Capex: payments for property, plant and equipment. Net additions: the rise in net property, plant and equipment plus non-current intangibles over the year, plus depreciation and amortization; it counts assets acquired whether or not cash has been paid. Free cash flow to equity (FCFE): net income plus depreciation and amortization, less additions; stock compensation stays an expense. Carry-in inventory: components bought earlier at lower prices and used now. ROIC: after-tax operating income over operating capital, against the weighted average cost of capital (WACC). Terminal growth: the rate the business grows at forever after the forecast ends, set here from long-run GDP projections.

The quarter beat, and the guide is the news: at the midpoints, December revenue grows 8.5% and EPS about 1.4% to 2.1%, a house calculation. That gap is the note's subject. Net sales were $114.5 billion, up 11.7%, above July's guide of 9% to 11%; the release rounds to 12%. iPhone rose 17.5% to $57.6 billion, Mac 23.7% to $10.8 billion, and Services +9.8% to $31.6 billion, while iPad fell 10.4% and Wearables, Home and Accessories fell 7.9%. Greater China rose 21.0%. Chief financial officer Kevan Parekh said supply limited iPhone, Mac and iPad, and that currency cost nearly two points against the June quarter.

Gross margin was $54.8 billion, 47.9%, including about 1 point from tariff refunds, so about 46.9% without them. Parekh said Apple has now received "the vast majority of the refunds," so we do not expect them to recur at this scale, and we strip them from every run rate. Products margin was 37.2% as reported, against 36.2% a year ago; Services margin was 76.0%. Research and development rose 32.3% to $11.7 billion, and operating income rose 9.9% to $35.6 billion. Net income was $30.7 billion.

One earnings number, on one basis. Diluted EPS was $2.10, up 14%, against a $1.98 estimate in our brokerage feed. The $2.10 includes $0.05 of refunds, so $2.05 without them. The quarter's tax rate was 15.0% against 17.0% for the year; at the full-year rate EPS would have been about 5 cents lower, a house calculation. For the year, EPS of $8.98 compares with our July estimate of $8.83, both including refunds; without them, $8.82 against $8.66, so the beat is 15 cents as reported and 16 without refunds.

The December guide, from the call. Revenue up 7% to 10%, which Parekh said "reflects our best view of constrained supply": about $156.0 billion at the midpoint. Services up 8% to 9%. Gross margin 45% to 46%. Operating expenses of $20.4 billion to $20.8 billion, up 11% to 13%. At the midpoints and 14.5 billion shares, EPS is $2.90 ($2.88 on the 14,592 million count the model uses), a house calculation, in line with the $2.90 in the feed and against $2.84 a year ago.

Cash and capital. Operating cash flow for the year was $145.1 billion and capex $9.04 billion, down 28.9%, leaving free cash flow of $136.1 billion. Apple bought back $86.4 billion of stock and paid $15.7 billion of dividends, and it ended with $56.8 billion of net cash.

The record says both: each jump set a higher level, and growth paused after it. On our model the difference moves fair value between $169.30 (base case only) and $180.68 (all three scenarios). Before FY26 the iPhone had jumped three times in seven years (Exhibit 1). After FY15's 52% jump, revenue fell 11.8%; after FY18's 16.7%, it fell 13.6%; after FY21's 39.3%, it rose 7.0%, then stalled for three years. The average year after a jump was −6.1%. The level mostly holds: iPhone revenue has stayed above $190 billion since FY21, and from FY15 to FY26 it grew 4.6% a year. But growth pauses: from FY21 to FY25 it was 2.2% a year.

What differs this time. FY27 has a second reason to grow that the earlier cycles lacked: a new form factor at $1,999 against $1,199 for the 18 Pro. Most of FY27's growth is price and mix, an inference from those list prices, since Apple reports no units and Parekh said "it's early to really read price elasticity." Khan said it is "really hard to predict when supply and demand will be in balance" on Duo, so some FY27 demand may move into FY28, which argues for a shallower give-back. Against that, FY26 was already a record for upgraders before Duo shipped, and those upgraders are less likely to buy again in FY27 or FY28.

What we carry. Our base case takes iPhone up 9% in FY27 and down 2% in FY28, a third of the average give-back, then 3% and 4% a year; it reaches $327 billion in FY33, 3.7% a year from FY26. The bull takes 13% and 4%, with no give-back at all; the bear 4% and −10%, a little shallower than the two sharper give-backs. On July's method the path is worth −$5.46 a share, because July's base carried one rate fading from 7.5% and this one pauses. Cycle or level matters less than it sounds. These change the base case only, at half the weight. If FY28 is flat, fair value is $172.33, up $0.99. With no give-back, 9% in FY27 and about 5% a year after, $176.35, up $5.01; the same 5% a year after each scenario's own FY27 growth, in all three scenarios, gives $180.68. If FY28 takes the full average give-back, $169.30, down $2.05. All of them start from a level the whole model already accepts.

Exhibit 2 is the quarter's most important figure, and memory is the reason. Without refunds, gross margin ran 49.3% in March, 48.1% in June and about 46.9% in September, and the guide midpoint is 45.5%. Parekh said the one-point fall from June to September was "driven entirely by memory impacts," after offsets from carry-in inventory, other components and price. Chief executive John Ternus was direct about what comes next: "What we're seeing with memory pricing right now is unprecedented," December brings "even higher memory costs," and "looking beyond the December quarter, we see the pricing for memory purchases continuing to increase." TrendForce projects fourth-quarter contract prices up 10% to 15% for conventional DRAM and 15% to 20% for NAND, with suppliers moving capacity to server products and PC makers buying ahead of a tight 2027.

The inventory build delays the cost; it does not avoid it. Inventories went from $5.7 billion a year ago to $6.7 billion in March, $11.1 billion in June and $14.5 billion in September. A launch does not explain that: September 2025 was also a launch quarter and held $5.7 billion, and the build started in June, before any launch. The 10-K shows that $8.3 billion of the $8.8 billion rise is components, which went from $2.1 billion to $10.4 billion. It does not say how much is memory, and Duo parts are in the figure too; our call is that most of those components are memory, bought ahead. On rising costs and short supply, the filing says Apple "expects these trends to intensify for memory and storage components," and manufacturing purchase obligations rose from $57.0 billion in June to $89.7 billion, with $82.7 billion payable within 12 months. Management said the carry-in benefit continues in December, and Ternus expects "a decreasing benefit from the carry-in inventory on memory over time." If all $8.3 billion were memory and prices rise 15% a quarter, the holding saves about $1.2 billion for each quarter it defers a purchase, about 0.8 point of December's gross margin, a rough ceiling and a house estimate. So the December guide already has this help, and the guide is 45.5%.

What we carry. The midpoint implies a products margin of about 37.7% in December if Services margin falls to 75% (37.5% if it holds 76%), normally Apple's strongest quarter for products. Our base takes 35.5% for FY27 and FY28, which implies about 34.4% in the three quarters after December, then 36.5% in FY29 and 37% from FY30, the level of FY24 and FY25. Prices stay up and supply catches up; we do not assume the 39% of FY26, which had the refunds and pre-shortage memory. Services falls from 76% to 74% because Parekh said Siri AI's inference costs "will be recognized through services cost of sales," and that the guide includes "some costs related to the ramp of Siri AI." Company margin comes out at 45.8% in FY27 and 48.4% in FY33, against 48.1% in FY26 without refunds. On July's method these gross margins cost $8.88, the largest data line in the bridge.

Operating expenses are the other half, and they outlast memory. They were 15.8% of revenue in FY26 and have risen as a share every year since FY21's 12.0%; the December guide is up 11% to 13% on revenue up 7% to 10%. We grow them 11% in FY27, the low end of that guide, then 7% and 5%, and with revenue after that, so they reach 17.0% of revenue in FY28, when the iPhone pauses, and stay near 17.1%. We stop the rise, which the trend does not do, but we do not let the share fall back: six years have shown no operating leverage, and while Parekh said "this growth rate is coming down," he added "we continue to significantly increase our overall investments in AI"; we carry the slowdown in the dollars and not in the share. On July's method that path costs $5.86. Back at FY26's 15.8% from FY29, fair value would be $174.74 in the base case alone and $178.41 in all three. This is where the long-run gap sits: FY33 gross margin, 48.4%, ends 0.3 point above FY26's without refunds (48.1%), so operating expense, which rises 1.3 points of revenue, accounts for more than all of the 1.0-point gap between FY26's 32.3% operating margin without refunds and the 31.3% we carry. Base-case FY27 EPS is $8.88, against $8.98 in FY26 and $8.82 without refunds.

Length matters less than level. In the base case alone, one more year of squeeze, with the recovery a year later, changes fair value by 13 cents. One point of products margin in every year and every scenario moves it about $3.88, because the terminal value capitalizes the level (about $2.00 in the base case alone). No recovery at all in the base case gives $168.69; a return to 39% by FY29 in the base case gives $175.13.

Less than it looks. Services grew 13.9%, 16.3%, 12.1% and +9.8% over the four quarters of FY26 (Exhibit 3). The fall from March to September is 6.4 points, and arithmetic more than covers it. Parekh attributed about four points to currency, which he had told investors to expect on the prior call: "This ended up being a significant headwind." The comparison also got harder: a year earlier Services grew 11.6% in the March quarter and 15.1% in the September quarter. Adding each quarter's growth to the year-earlier growth, the two-year figure was 27.9% in March and 24.9% in September, a fall of 2.9 points, smaller than the currency effect alone.

These items are real, but they are not what moved the number. Parekh said "the performance on the App Store was similar to the prior year," with continuing weakness in mobile gaming, changes to the App Store's business model in certain countries and a US court ruling on link-out transactions. December adds another half point of currency and a new item, sequential headwinds "primarily from third-party licensing." The call did not say what that licensing is. The largest licensing line we know of in Services is search payments, whose remedy is still in the D.C. Circuit (§7). The 10-K lists third-party licensing arrangements within Advertising and describes search licensing arrangements with Google, but does not say what Parekh's headwind is, so we do not read it as search until a later call or filing does. Video, cloud, payments and advertising each set all-time records.

What we carry. Base Services growth is 9% in FY27, from the 8% to 9% guide and with currency lapping in the second half, 9.5% in FY28, then fading to 4% by FY33, when Services is 31% of revenue, from 25.8%. The bull takes 11% to 12.5% with a paid Siri AI tier, for which Ternus said "anyone who needs greater capacity will be able to purchase expanded access in the future." The bear loses search payments in FY28: growth of 6%, then 1%, and a margin falling to 71%. July carried no separate line; on its method this path adds +$3.21. In the base case alone, Services at about 6% a year (7% in FY27) gives fair value of $165.14, and at about 12% (11% in FY27), $178.43. The slower path costs more than the iPhone can give back: Services at about 6% takes $6.21 from fair value, more than the $5.01 that removing the iPhone's give-back adds, because each dollar of Services carries about twice the gross margin of a dollar of products. The bear's search path in the base case gives $157.90.

Yes, for what the series tests, and no longer in the sense that it has no AI bill. The series asks whether long depreciation clocks on rising capex flatter earnings. Capex fell to $9.04 billion, $0.07 per dollar of net income, the same as the nine-month $0.07 in July (Exhibit 4). Re-running July's method on FY26 actual capex, a three-year clock against a five-year clock moves FY26 depreciation by −$58 million, or 0.3 cents of EPS; the sign flipped from July because capex has now fallen, and the size still rounds to zero. Whatever else is true, the EPS is clean.

But the AI bill exists, and it runs through four other lines. Research and development rose $11.2 billion to $45.8 billion, 8.5 cents per dollar of net income, and the 10-K attributes the rise primarily to "higher infrastructure-related costs, including investments in artificial intelligence, and headcount-related expenses." Siri AI's server costs will enter Services cost of sales as the rollout widens, and from the December quarter part of hardware revenue is deferred for it. Non-current intangibles rose 83% to $20.3 billion, all of it by March. And memory is the fourth line, the one Apple did not choose. Capital stays small even on the wider counts: net additions were $24.9 billion, $0.19 per dollar of net income, 2.7 times the cash capex, or $28.1 billion and 21 cents with the current portion of licensed rights the 10-K breaks out.

The intangibles step is licensed rights with a term. The 10-K shows gross licensed intangibles up from $16.1 billion to $29.6 billion, $13.5 billion of the $14.5 billion total rise. It names no licensor and gives no useful life, but its schedule for all amortizable intangibles, $24.2 billion including $3.2 billion owned, runs off 91% by FY30: $6.6 billion of amortization in FY27, against $4.9 billion in FY26, with $0.8 billion in FY31 and $1.4 billion in total after. Off the balance sheet, unconditional purchase obligations with a remaining term over one year rose from $13.3 billion to $24.8 billion, at $5.5 billion to $6.7 billion a year through FY30, then $0.2 billion in FY31. The filing ties neither to AI: the intangibles note does not say what the rights cover, and the obligations note lists supplier arrangements, distribution rights and licensed intellectual property and content. Reading them as part of the AI bill is our inference, from timing alone: the rights came by March and the unconditional obligations by June, as Apple built Siri AI and R&D rose on infrastructure including AI. The label moves no number: the rights sit in the margins and the 15 cents whatever they license. We model no cliff, which assumes Apple renews. Intangible amortization ran at $3.6 billion and $3.7 billion in FY24 and FY25, before this step, so we read renewal as the book's record, untested for the step itself.

The liability is the size of the rights, and we leave it out of net cash. Other non-current liabilities rose $12.8 billion to $54.3 billion. Leases and non-current deferred revenue explain about $1.0 billion, and unrecognized tax benefits fell. The remaining $11.8 billion is the size of the $11.8 billion rise in net licensed rights, though that figure includes a current portion and the filing neither links the two nor names the liability. It also does not say where payments for licensed rights are classified; the investing "Other" line is only $2.4 billion. We read the liability as rights bought on deferred terms. It is not a deduction: our cash flows treat the cost of rights as paid in the year, through additions at least equal to amortization, and hold working capital flat, so taking the payable off net cash as well would largely charge the same rights twice if the payable is rolled over. If it is paid down with nothing replacing it, the cost is about $0.81 a share.

So Apple is buying its AI mainly through operating expense, on a mix of its own data centers and rented compute, with what we read as a one-time step in intangibles. That is the opposite of the accounting the series questions: expensed now, deferring almost nothing. The cost of that choice is that the rent can rise. "In addition to its own data center infrastructure, the Company relies on third-party cloud service providers to meet these compute and storage needs," the 10-K says, and it gives no split. Parekh left the owned side open at "any further requirements we might have on the compute side." If Apple needed 30 cents of capital per added revenue dollar instead of our 15, fair value would be $163.83.

The print's largest effect is on margins, the cash-flow method and the rate add only $2.12 and $1.43, and the largest single move is a decision on a weight, shown last so it can be judged alone.

The model. Free cash flow to equity in three scenarios weighted 25/50/25, stock compensation expensed, the diluted count held flat at 14,592 million because buybacks are the distribution of the cash flows we discount, and net cash added once. Every input and method below was set as the most defensible choice on today's evidence. To show what moved, we first re-ran July's configuration, which reproduces its fair value exactly, $207.93 (published as $208): free cash flow to equity $181.58, price-to-earnings value $247.46, blended 60/40. Each factor in Exhibit 5 and the attribution table re-runs that blend with one more change, so the factors sum to the move.

The quarter's data, on July's method: $192.34. FY26 actuals and rolling the date add +$2.57. The share count, 14,532 million outstanding at 26 September plus the quarter's 60 million of dilution against July's 14,715 million, adds +$0.95. Tax at 17.0%, the FY26 rate and the December guide, against 17.5%, adds +$1.27. The iPhone takes $5.46, Services adds +$3.21, and a slower line for Mac, iPad and Wearables takes $3.39; those three grew 4.1% together in FY26, with supply short in December. Gross margins take $8.88 and operating expenses $5.86 (§3).

Four changes to how the cash flows work: +$2.12 net. The forecast now runs to FY33, with every line landing on terminal growth, where July stopped at 4.5% against a 3.5% terminal: +$0.17. Terminal growth is set from long-run nominal GDP: the Federal Reserve's projections of 16 September 2026 give 2.0% real growth and 2.0% inflation in the longer run, so 4.0% in the base and bull cases and 3.0% in the bear, against July's 3.5%, 4.0% and 3.0%: +$3.29. The bear stays a point below because it loses search payments and iPhone share; at 4.0% fair value would be $174.72. Capital now grows with revenue: additions equal depreciation plus 15 cents per added revenue dollar, FY26's ratio of plant and non-current intangibles to revenue, where July held capex below depreciation forever: −$3.84. Counting the current portion of licensed rights the ratio is 16.2 cents; at 16 cents fair value is $170.84. Without non-current licensed rights the ratio is about 12 cents; at 12 cents fair value is $172.85. We model no working-capital drag: FY26's $14.3 billion is more than explained by $8.9 billion of inventory and the final $8.8 billion of repatriation tax, and FY25's $25.1 billion includes the $15.8 billion State Aid payment. Net cash of $56.8 billion is added at face value and its interest taken out of the cash flows: +$2.50.

The cost of equity: 9.7% for FY27 and FY28 (−$2.83), 9.4% for the terminal (+$4.26). We argue the rate for the company and show the standard reads beside it. The 10-year Treasury was 4.91% on 2 November, against 4.67% on 29 July. Apple's own 60-month beta against the S&P 500 is 1.06, with a standard error of 0.15; blended toward the 1.04 median of Microsoft, Alphabet and Sony it barely moves (1.06), and with a 4.5% equity risk premium that is 9.7%, against July's 9.53%. The 36-month beta is 0.74, which would mean 8.2%. The 60-month window reads about 1.3 over its first two years and 0.74 over its last three, and the gap rejects neither. We keep the longer window for its smaller error (0.15 against 0.26) and higher correlation with the market (0.68 against 0.44), not because the recent reading is wrong, and we show it: $218.68 at that rate throughout. The terminal value describes a company growing with the economy with debt at 1.6% of capital, so we discount it at a beta of 1.0, 9.4%, reached in a straight line over FY29 to FY33. The peer median of 1.04 would give 9.6% ($166.53); we prefer 1.0 because two of those three peers are mid-build and the third is a conglomerate. At July's 9.53% throughout, fair value would be $168.96. At a 4.0% equity risk premium with the same betas, the rates are 9.1% and 8.9% and fair value $187.66.

How long each part of the business grows faster than normal. Each stage's length comes from what drives it, checked against base rates of growth persistence, which find long runs of above-median sales growth uncommon (6.3% of firms against 3.1% by chance).

Return on capital. Apple's capital is a judgment before it is a number. Equity plus debt less all cash and securities averaged $48 billion over the last year, because suppliers and customers fund much of the business: accounts payable are $82.3 billion against $39.9 billion of receivables from the same vendors. On that base the return is 276%, too sensitive to a small denominator to read a trend from. Leaving $137 billion of cash in gives 72%, but that charges the operations for securities that earn nothing in operating income. Our headline counts net supplier float as capital, because Apple's purchase obligations show that float has a cost when supply is short: $87 billion on average, and a return of 151%. The WACC is 9.62%: equity at 9.7% and 98.4% of capital, debt at about 4.5% after tax and 1.6%, so it sits within a tenth of a point of the cost of equity; the workbook shows the build. No definition puts the return near the hurdle.

The mechanical label is fading, because the spread fell 37 points in four quarters. Read it from the business. After-tax operating income rose from $101 billion to $131 billion over the window, and capital rose from $35 billion to $87 billion as Apple paid its remaining long-dated tax in cash, built inventory and added intangibles; with cash left in the denominator the return rose. The return earned on new capital over the window is 58%, against the 173% our terminal value assumes on 15 cents of capital per added dollar. That gap is large and we explain it: the window's added capital includes $22.0 billion of repatriation tax paid down from September 2023, which supports no growth, and without it the earned figure is about 101%, a house estimate. The rest is the inventory build and the intangibles step. If new capital earned only that 101% forever, fair value would be $165.98; at the unadjusted 58%, $156.32.

What the price requires. Holding every other input in all three scenarios, fair value reaches $338.32 with costs of equity of 6.9% for the forecast years and 6.6% for the terminal, 2.8 points below ours, which at a 4.91% 10-year and a 4.5% premium means a beta near 0.45. Or with positive growth rates 2.9 times ours in every line through FY32, the give-back years left as they are: in the base that is 12.9% a year, reaching $1.12 trillion of revenue in FY33 against $661 billion. Or with terminal growth 3.2 points higher in every scenario, 7.2% in the base. No margin we can argue gets there: each point of operating margin in every year and scenario is worth $5.54, so it would take about 30 points. The stock is at 38.1 times our base FY27 EPS. Terminal value is 71% of operating value in the base case, so the long-run inputs carry most of the number (Exhibit 7). Combinations also count: the bull at 8.2% is worth $279.25, the closest of the cash-flow routes we show, or $291.12 at July's weights, and it needs both the higher path and a rate we did not choose.

What investors pay. The cross-check's 32 times is the top of Apple's post-2020 band in our July note, and the market pays more today. On FY26 EPS, trailing multiples in our feed for the six durable franchises we pulled run from Procter & Gamble's 24.4 ($219) through Microsoft's 27.3 and Coca-Cola's 27.7, Visa's 37.1 and Walmart's 44.8 to Costco's 48.8 ($438), a range that brackets the price. We leave out Alphabet's 14.6, whose trailing earnings carry marks on private stakes. That is what investors pay for durable franchises now; it is not a value we can derive from these cash flows at our rates.

Our July note armed seven tests, and three carried due dates of 30 October that predate this print, so we score them on it. The stock's rise since July is the first line on the record: we put fair value at $208 against a $308.91 close, 33% below the price (the price was 49% above value), and the stock is now 9.5% higher.

Each tripwire has its reading rule set now, so the next update scores it rather than reinterprets it. Effects are on fair value at today's other inputs.

1. December-quarter gross margin (FQ1-27 release, late January). At or below 44.5%, with a March guide midpoint at or below 44.5%: base products margin moves down one point in every year, about $169.35 in the base alone ($167.46 if every scenario moves). At or above 46.5%, with a March guide midpoint at or above 46.5%: up one point, about $173.34 ($175.23 if every scenario moves). Between those, the base holds.
2. Carry-in and the build (each 10-Q). Inventories below $10 billion by the June quarter with ex-refund margin under 45%: the carry-in is spent and the squeeze stage extends a year, $171.21. Inventories above $15 billion with margin at or above 46%: the build is still working, and the base holds with the recovery brought forward a year. Purchase obligations are recorded each quarter; a fall below $60 billion reads as supply normalizing.
3. Services growth and margin (FQ2 to FQ4-27). Read after currency lapped, from the June quarter. Two quarters at or above 12%: the 12% path, $178.43. Two quarters at or below 6%: the 6% path, $165.14. Services margin at or above 75.5% through FY27, with Siri AI rolled out: the 74% we carry for its cost comes off, which lifts fair value to about $172.59 in the base alone (+$1.25). Third-party licensing named as search-related, on any call or filing: read with tripwire 8.
4. The iPhone give-back and Siri AI (FY27 release, late October 2027; waypoints each quarter). FY27 iPhone growth at or above 13% with FY28 guided or tracking flat or up: bull's path. At or below 4%: bear's. Siri AI is scored as a driver only if management attributes upgrades to it with a number, or a paid tier reports revenue; otherwise it stays a cost.
5. Licensed rights and commitments (FY2027 10-K). We read FY26's rise as a single step with a term of about four years. Another rise of $10 billion or more in gross licensed intangibles, about three-quarters of FY26's, or unconditional purchase obligations with a remaining term over one year above $40 billion, about $22 billion of new commitments once FY27's $6.7 billion runs off, makes it a recurring bill and re-opens our 15 cents of capital. A trip only re-opens the rate: our base funds $20.0 billion of additions in FY27, so net additions at or below that leave the 15 cents standing. We have no mapping from a trip to a capital rate; for scale, 26 cents of capital per added revenue dollar, where the terminal return on new capital equals the 101% Apple earned over the window without the $22.0 billion of repatriation tax, gives $165.98. Neither rise, with FY27 amortization near the scheduled $6.6 billion, keeps the step reading and the 15 cents. If gross licensed rights fall and obligations fall with them, Apple is not renewing, which is the assumption behind the 15 cents: we would re-open it the other way, $172.85 at 12 cents, before any margin the lapsed amortization returns. If the $11.8 billion liability is paid down with no new rights behind it, flat working capital costs about $0.81 a share.
6. Owned AI infrastructure (each quarter). FY27 capex guided or tracking above $14 billion, or $0.10 per dollar of net income on cash capex: Apple is building, and the capital rate moves to 30 cents, $163.83, and the control-group claim is re-opened.
7. The P/E weight (our rates, each quarter). The weight stays at zero while 26 times needs terminal growth above the 5% ceiling on our path. That breaks with our rates about 0.5 point lower, a 10-year near 4.4% on today's beta, or the same fall in the rate from a lower beta. Below that for a full quarter, a 20% weight returns, about $183.62 at today's inputs and more with the lower rate.
8. DOJ v. Google (D.C. Circuit, any ruling). An affirmed remedy with payments intact: no change. A payment ban ordered: the bear's Services path becomes the base, $157.90. A ruling in Google's favor on liability removes this from the bear.
9. Price. A close at or below $171.34 resolves the gap. $128.51 is a 25% margin of safety to fair value. Replaces July's $156. Scored against this note if, at the FY28 print, AAPL's total return from $338.32 beats the S&P 500's with base inputs tracking.

We value Apple at $171.34 a share, from $208. This was a record year, the earnings are clean and the cash is real, and our July estimates were too low on revenue and EPS. Apple is still the control group: seven cents of capex per dollar of net income, and a re-mark that rounds to zero. But the bill for the AI build has reached it through memory, and the print's largest effect on value is the margin path. An iPhone cycle worth a few dollars either way does not offset that, and the operating-expense share outlasts the squeeze. Arithmetic more than covers the Services slowdown to September; the App Store and third-party licensing items are real but are not what moved the number. Two-thirds of the fall from $208 is our decision on the weight, and at July's weights the value is $195.89. Holding our other inputs, the price takes a cost of equity about 2.8 points below ours, or 12.9% revenue growth for seven years, and our bull case, $216.53, stops well short of it. The late-January call, with the December margin and the March guide, is the next test. The Fund holds no position in AAPL.