Alphabet (GOOGL / GOOG) — Deep Dive

Verdict

Read in the desk’s order: (1) business quality → (2) moat durability → (3) price vs what’s priced in. Each section below maps to one step.

Catalyst calendar

Key Debates

What actually decides this stock

Settled by the record. (1) The Services profit pool — $153.8B TTM operating income (ex the EC fine), still growing double digits; the quality question is answered. (2) Cloud’s near-term growth — a $513.9B backlog against $24.8B of quarterly Cloud revenue is contracted cover, not hope. (3) The share-count direction — the count is rising (buybacks $0 in H1 2026, a $40B ATM armed and unsold, SBC compounding +13%/yr): the dilution regime is a fact, not a scenario.

Contested — the turbo trigger. Owner-earnings margin recovery. TTM owner-earnings margin is 5.6% of revenue; the demonstrated FY22–24 level is 14.65%. Everything else is second-order at the price: at the price-justifying corner of the grid below, ±1 point of revenue growth or ±1 point of mature margin is worth roughly ±$250B of EV (~6%). By contrast, restarting buybacks at the recent demonstrated pace moves the required bar by at most 2.2 points, and halving the deducted SpaceX mark moves it by 0.2 points (stress test below). The debate is over roughly eight points of growth-equivalent, and it is decided by one driver: whether the capex wave converts.

The kill attempt

1. Duration, stated plainly. The price needs owner earnings to compound +37.4%/yr for a decade off the TTM base — a base depressed by the very capex wave that is still accelerating. The friendlier FY25 base still needs +27.9%/yr. Delivered: −1.6%/yr across FY21→FY25. No large-cap in the desk’s coverage carries a wider required-vs-delivered gap.
2. The trough deepens before it shallows. Capex guidance has ratcheted $175–185B → $195–205B for FY26 across two quarters, with 2027 guided to “significantly increase” and FCF guided to “remain under pressure.” Depreciation follows capex with a lag; Q2 2026 was Alphabet’s first negative-FCF quarter (−$5,855M) — before the guide was raised.
3. The dilution regime. Buybacks were $70.7B in FY23 and $0 in H1 2026. SBC is $28.1B TTM and compounding +13%/yr; a $40B ATM is armed explicitly in part for employee tax withholding; the weighted count is rising (TTM diluted WASO 12,244M and climbing through 2026). The bar is quoted in total owner earnings — per share, it is higher.
4. Remedies can remove a layer, not just tax it. The ad-tech case seeks divestiture of the sell-side stack that monetizes the open web; the Search case’s conduct remedies (choice screens, data sharing) are live while both sides appeal. Distribution economics written over twenty years can be re-written by two courts.
5. Two single-name marks carry the deductions. The EV bridge deducts $373.9B of investment assets at carrying value. $80.0B of that is one restricted name (SpaceX); the $131.5B non-marketable line “primarily consists” of one unnamed private company per the filing — the same line that produced a $77.4B net remeasurement gain in Q2 alone. Reported earnings quality now runs through marks no outsider can verify.
6. Governance is a one-way ratchet. Page and Brin hold 52.7% of the vote on ~5.7% of the economics. Insider behavior over 24 months, full population: every sale (1.39M shares) ran under a pre-set 10b5-1 plan — zero discretionary sales, zero open-market purchases, zero pledges. The $29.5B Wiz acquisition closed with $22.7B of non-deductible goodwill and no disclosed return framework.
7. The central product question is unfalsifiable. Alphabet has never broken out AI-surface revenue or ads revenue on AI Overviews / AI Mode. Twelve quarters of parity assertions are the whole of the evidence on cannibalization — from the party with the most riding on the answer.

The bull case, answering in order

1. The bar is a margin statement, not a growth fantasy. Restated in revenue terms: at the demonstrated 14.65% owner-earnings margin, the price needs revenue compounding +24.7%/yr — against +11.8%/yr delivered FY21→FY25 before Cloud inflected to +81.8%. The gap is real (the bear point stands) but it is one variable deep: margin normalization does most of the work.
2. Capex is substantially pre-sold. Consolidated RPO of $519.5B — roughly 4.3 quarters of total revenue — is contracted demand for the capacity being built. Twelve quarters of calls say the same thing in sequence: sold out (2023) → supply-constrained (2024) → doubling the guide (2025) → backlog compounding (2026). Supply, not demand, has been the binding constraint at every step.
3. Conceded on the count — with the funding fact attached. The June 2026 raise ($30.5B of common + $19.1B of mandatory converts) pre-funded the wave in equity rather than debt markets; the ATM is armed but unsold ($0 at Jun 30). Whether buybacks restart is checkpoint 6 — a binary the Q3 10-Q answers.
4. Search remedies closed non-structural. The Dec 2025 final judgment imposed conduct remedies, not a Chrome divestiture; the ad-tech structural outcome is a 20–35% desk band, not a base case. TAC’s +42% rise is the real, already-visible cost — and Search & other still accelerated to +16.6% while paying it.
5. The marks are not the thesis — here is the test. Halving the deducted SpaceX mark raises EV by $40.0B and moves the governing bar by +0.2 points (table in the valuation section). The bar is a duration bet on the operating business; the marks are a disclosure-quality issue, sized and stress-tested, not a hidden fulcrum.
6. Conceded. Alignment is the grade’s weak leg (C+) and it is priced into the B, not explained away. What the record does show: the founders’ economic exposure is enormous in dollars, and the operating scorecard they control — execution A− — is the one that has compounded the asset.
7. The delivered check on the assertions. Parity assertions began Oct 29, 2024. Since then Search & other growth reached +16.6% by Q2’26, with paid clicks and CPC both positive in the latest print. Assertions followed by eight quarters of delivery are evidence — labeled as assertions-plus-delivery, which is exactly what they are.

The business: an ads engine, a cloud inflection, and a capital-markets pivot

Alphabet is three economics under one roof. Google Services (Search, YouTube, subscriptions, Android/Play) produced ~$153.8B of TTM operating income (ex the EC fine) — the profit pool that funds everything else. Google Cloud ($24,768M of Q2 revenue, +81.8%, at a 35.6% operating margin, with a $513.9B backlog — management splits growth about evenly between consumption of existing commitments and genuinely new workloads) has become the second engine. Other Bets burned $5.1B TTM on $1.7B of revenue, with Waymo the only asset of scale inside it (per the Q1 2026 call: 500K+ fully autonomous paid rides/week across 11 major US cities — operating scale disclosed; revenue and unit economics never disclosed).

The capital structure changed regime in June 2026. Alphabet raised ~$30.5B of common equity (including a Berkshire Hathaway affiliate private placement) and $19.1B of 6.25% mandatory convertible preferred (converts ~May 2029), armed a $40B at-the-market program (unsold at Jun 30), and had already issued ~$51.8B of notes in 2026 — total debt went from $16.2B at mid-2025 to $100.2B carrying value a year later, against $242.5B of cash and marketable securities. The AI build-out is now financed like infrastructure, in the open, at investment-grade coupons — the 100-year sterling tranche (6.125%, due 2126) is the tell. Wiz closed Mar 11, 2026 at $29.5B ($22.7B goodwill, not tax-deductible) into Cloud security. Control is not in question: Class B super-voting stock gives the founders 52.7% of votes.

The quarter: Q2 2026 — growth up, cash flow negative, guide raised

Three things decide how this quarter reads. First, the top line is real: Search accelerated, Cloud nearly doubled, and Services OI grew +20.2%. Second, the cash line is the regime: Q2 capex of $44.9B produced Alphabet’s first negative FCF quarter, and management raised the FY26 guide to $195–205B the same day while guiding FCF to “remain under pressure.” Third, the bottom line is not the business: net income of $112.2B contains a $99,031M gain on equity securities (SpaceX and an unnamed private company remeasuring) inside other income — strip it, and underlying net income is roughly the prior year’s level. The July payment of the $5.2B EC Android fine lands in Q3’s cash flow.

The series: five years plus TTM

*FY21 TAC as disclosed at the time (combined properties basis). SBC per the 10-K cash-flow statements (FY23–FY24 as comparatives in the FY25 10-K). †TTM operating income excludes the $5,391M EC Android fine charge carried in reported results; as reported, TTM operating income is $147,628M.

Read the table as one story: revenue compounded +11.8%/yr while owner earnings fell −1.6%/yr — the entire gap is capex (5.4× its FY21 level) and SBC (1.8×). Margins tell the same story in one line: FY22–24 owner-earnings margin averaged 14.65% in aggregate; the TTM print is 5.6%. Whether that 5.6% is a trough or a level is the valuation question, and it is answered only by delivery.

Twelve quarters of calls: the arcs that matter

• Capex ratchet (the master arc). “Sold out” (Q3’23) → supply-constrained with capex doubling promised (2024) → FY26 guide set at $175–185B (Q4’25) → raised to $180–195B (Q1’26) → $195–205B (Q2’26), with 2027 guided to “significantly increase” and FCF to “remain under pressure” (Ashkenazi, Q2’26). The same calls disclose the conversion lag in management’s own words: “$111B of capex in the last three years… just over $40B of incremental revenue” (Pichai, Q2’26, citing a ~2.5× modeled multiple on full useful life) — the desk prints that as a claim, not a result.
• AI monetization parity. Schindler, Oct 29, 2024: AI Overviews monetize “at approximately the same rate” as conventional results — the assertion was made on five calls (Q3 2024 on Oct 29, 2024; Q4 2024 on Feb 4, 2025; and Q1, Q2, Q3 2025) and was not repeated in Q4 2025, Q1 2026, or Q2 2026, where the Q2 2026 language weakened to “encouraged” plus new-format testing. The AI Max “billions of net new searches that weren’t really monetizable before” line is Q2 2026 (Schindler, Q&A) — mis-dated to Q1 2026 in this page’s prior edition; corrected in the transcript library and in the corrections section below. AI Mode passed 1B monthly users (Q2’26); management has never used the word “cannibalization” on a call, and has never broken out AI-surface revenue.
• Cloud. Backlog $513.9B; new-customer acquisition velocity more than doubled YoY; existing customers expanding >50% over commitments (Q2’26). First TPU system sales to customer data centers recognized in Q2’26 (small), with the “vast majority” of TPU-agreement revenue landing in 2027.
• Waymo. 500K+ fully autonomous paid rides/week, 11 major US cities (Pichai, Q1 2026 prepared remarks; the weekly figure had doubled in under a year). No Waymo revenue, per-ride, or unit-economics figure has ever been disclosed on a call or in a filing.
• AI operating metrics (Q2 2026 call, Pichai, prepared remarks — management-reported). Gemini app 950M monthly users; model APIs processing ~22B tokens/minute (up from 16B a quarter earlier); Gemini 4 in its largest pre-training run yet; ~500 Cloud customers each processed >1T tokens in the last year; ~90% of the Fortune 100 on Gemini Enterprise. Operating claims from the call record, labeled as such — not filing figures.
• Investor day / capital-markets day — none in 24 months (verified negative, Oct 16, 2026). Alphabet held no investor day or CMD in the 24 months to this edition (IR events record; 8-K census). Strategy claims on this page are confined to what calls and filings support; I/O and Cloud Next material enters only as carried in that record (Rule 4).

Valuation: the reverse DCF — what $349.22 has to believe

Method: solve on enterprise value at a 10% required return, 10 years, 2% terminal growth, flat share count as the governing convention. Base = owner earnings (FCF − SBC, its own line above). Market cap $4,247,139M (GOOGL $349.22 × 5,868M + Class B 835M + GOOG $344.91 × 5,527M; counts per the Q2 10-Q cover, Jul 15, 2026); EV $3,999,569M per the bridge in the next section.

Buyback-credited bars (only the demonstrated-pace scenarios; the buyback is currently paused, so flat governs): at a 1.7%/yr net shrink, +35.1% (TTM base) / +25.7% (FY25); at 2.5%/yr, +34.0% / +24.7%. Resumption would have to be demonstrated again before any credit applies.

The same price in belief language — implied EV ($B) at a constant revenue CAGR and a mature owner-earnings margin (today’s EV: $3,999.6B; cells at or above it justify the price):

The price is justified if revenue compounds ≥ +24%/yr at a ≥17.5% margin, or +18–20%/yr at a 25% margin, or +28%/yr at any shown margin. At the demonstrated 14.65% aggregate margin (FY22–24), the required revenue growth is +24.7%/yr; at today’s 5.6% TTM margin the required growth is higher still. Hurdle sensitivity: the TTM-base bar runs 31.8% at an 8% required return to 42.4% at 12%; the FY25-base bar runs 22.8% to 32.5%.

Concentration stress test (scenario, not a forecast). The bridge deducts investment assets at carrying value; as filed, $80.0B of that deduction is SpaceX shares inside marketable securities, subject to short-term sale restrictions (a further $14.1B of SpaceX sits in other non-current assets, restricted through Q3 2027, and is not deducted). Write the deducted mark down:

A 25% haircut to the entire $131.5B non-marketable line prints the same bars to 0.1pt. Conclusion: the marks move the bar by at most ~0.2 points — material to disclosure quality, immaterial to the price question. No sum-of-the-parts is computed in this edition: segment profit pools are shown in the series table, and the value question is answered by the consolidated reverse DCF alone.

Balance sheet & capital structure: the bridge, to the dollar

Enterprise-value bridge at Jun 30, 2026 ($M — printed in millions so it foots exactly):

Two components are carried on issuance/face bases rather than filing carrying values — mandatory converts at $19,100M vs $19,063M of filed net proceeds, and the current debt portion at its $2,000M face vs $1,999M carrying — a combined $38M (0.001% of EV); no printed bar moves at 0.1pt precision. Every other component ties to the Q2 10-Q to the dollar (cash + marketable securities subtotal as filed: $242,474M).

Debt. Note 6 of the 10-Q discloses grouped lines, not a per-series ladder: face value $101,085M across USD notes ($53,500M), Euro notes ($25,313M), and the 2026 sterling ($7,263M), Swiss franc ($3,772M), Canadian dollar ($5,985M), and yen ($3,566M) issues, plus $1,686M of other long-term debt; carrying value $100,164M, of which $1,999M is current. 2026 issuance was $20.0B USD + ~$31.8B foreign-currency; estimated fair value of the notes is $94.9B (Level 2) — the debt trades at a discount to carrying value. Commercial paper: $0 outstanding. Credit facilities: $11.7B, $1.3B drawn. The full grouped schedule is filed in the dossier (debt-tranches-2026-06-30.md). Net of cash and marketable securities alone, Alphabet is +$142.3B net cash before counting the non-marketable line — leverage is not the risk on this page.

The converts. 19M shares of 6.25% mandatory convertible preferred ($1,000 liquidation preference; $19,000M aggregate), issued Jun 5, 2026 as depositary shares GOOGM (Class A-indexed) / GOOGN (Class C-indexed); mandatory conversion on or about May 15, 2029 at 2.2520–2.8160 (A) / 2.2740–2.8420 (C) shares per preferred share; $1.0B of capped calls purchased alongside (caps ~$532.67 / ~$527.80). First dividend declared July 2026, paid Aug 15, 2026.

Equity machine, current state. June 2026: ~$30.5B common issued ($20.5B public + $10.0B Berkshire-affiliate private placement). ATM: $40B program entered Jun 1, 2026, $0 sold at Jun 30. Buybacks: $0 in H1 2026 (authorization is not activity; no credit is taken for resumption anywhere on this page).

Antitrust: one case closed non-structural, one judgment pending

• Search (D.D.C.). Liability Aug 2024; remedies judgment Sep 2025 — no Chrome divestiture, no default-payment ban; exclusive dealing curtailed, choice screens, data-sharing and syndication remedies, six-year term. Final Dec 2025; both DOJ and Alphabet have appealed. Trial-court remedies are the operative state of the world while appeals run.
• Ad tech (E.D. Va.). Liability Apr 2025 (publisher ad server + exchange monopolization; advertiser-tools claim dismissed). Remedies closing arguments Nov 2025; judgment pending at publication. DOJ seeks divestiture of the exchange (and possibly the publisher ad server, phased). Desk-judgment probability band: ~20–35% a structural divestiture is ordered; conduct remedies short of divestiture ~50%+; the remainder is appeal-stayed or minimal outcomes. This band is the desk’s judgment, labeled as such — not a market price or a filing fact.
• Europe. Android: the €4.342B fine was paid in cash in July 2026 ($5.2B) after the CJEU declined the appeal — a Q3 2026 cash outflow, not a contingency. Android Auto: €4.13B provisioned. Spain shopping fine €352.7M paid Q1 2026 (appeal continues). None of these threatens the model; the fines cadence is a recurring tax, disclosed item by item in the filings.

Management: graded B, now on a full-population record

The insider record, enumerated this time. All 349 Alphabet-issuer Form 4/4-A filings from Oct 1, 2024 to Oct 16, 2026 were pulled from EDGAR and parsed — every transaction code, footnote, and 10b5-1 checkbox. The population says: zero open-market purchases by any insider in 24 months; zero discretionary sales — every one of the 1,389,369 shares sold (117 filings) moved under a pre-set Rule 10b5-1 plan; zero pledges, prepaid forwards, or hedging disclosures anywhere in the population. Pichai’s biweekly plan (32,500 Class C shares a sale, adopted Dec 2, 2024) completed on Mar 18, 2026 after 24 sales totaling 780,000 shares — no sale since, and no successor plan disclosed; a finished plan is neutral evidence, and the desk reads it that way. Brin filed only conversions paired with gifts (zero sales); Page filed nothing at all. The honest summary: no insider has expressed a discretionary view on this stock, in either direction, for 24 months. One officer transition in the window — the chief accounting officer (O’Toole → Saraci, Jun 2026) — handled in the ordinary course. The prior edition’s insider note rested on a three-day window plus pattern assertions; this edition’s rests on the population, and the grade B stands on it, unchanged in every leg.

Charts

Qualtrim, dark exports; shown as-is in both themes.

Moat: the Fisher skeleton — what holds the moat up, and what we actually know

• Distribution defaults — answered by the record. Chrome, Android, and the Apple default remain in place; the Search remedies judgment (final Dec 2025) curtailed exclusivity without removing the defaults. The price of the default is visible and rising: TAC +42% FY21→FY25.
• Full-stack integration — answered by delivery. TPUs → Gemini → Search/Cloud/Workspace is one stack, and it is shipping: Gemini across the product line, first TPU system sales recognized Q2’26, Cloud compounding behind it.
• Data and scale feedback in Search and ads — thin. Directionally obvious, never quantified by the company; no filed figure isolates it.
• Cloud entrenchment — thin. The evidence is the backlog ($513.9B) and customers expanding >50% over commitments; churn and net retention are never disclosed.
• Habit and brand (Search, YouTube) — thin. Usage metrics are management-reported (950M Gemini app MAU; AI Mode 1B MAU); share data is third-party, not filed.
• Regulatory durability — unanswered. The E.D. Va. judgment is pending and the Search judgment is under appeal by both sides. The moat’s legal perimeter is a live question for years, which is why the verdict box carries the flag.
• The ads marketplace itself — answered historically, targeted directly. The network effects are the business; the ad-tech case aims at exactly this layer, which is what makes Debate 3 a moat debate and not just a legal one.

Downside Checklist (Pabrai) — the veto gate

No single flag vetoes the page’s conclusion; the valuation and moat flags together are why the verdict is framed as a duration question rather than a quality endorsement at the price.

Checkpoints — scored at the Oct 28 print (set Oct 16, 2026)

1. Cloud growth: pass ≥ +60% YoY; fail < +40% → Debate 1 resolves to “demand regime,” and the revenue leg of the bar is re-based down.
2. Cloud backlog: pass ≥ $540B; fail < $513.9B (any decline) → contracted cover is thinning; the pre-sold-capex answer to the bear case weakens materially.
3. Cloud margin: pass ≥ 33%; fail < 30% → mix or pricing is deteriorating as capacity lands; the margin-recovery driver is marked down.
4. FY26 capex guide: held at $195–205B = pass; raised above $205B = ratchet flag → the trough extends and the TTM-base bar’s premise is re-examined.
5. Q3 FCF: pass ≥ −$8B; fail < −$12B → the cash trough is deeper than the $5.2B EC payment explains; the conversion timeline slips a year.
6. Buybacks / ATM (binary): any repurchase > $0 restarts the shrink clock and re-opens the credited bars; $0 buybacks plus first ATM sales confirms the dilution regime for another quarter.
7. Search & other growth: pass ≥ +14%; fail < +11% → the first hard print consistent with cannibalization or the agentic shift; Debate 2 re-opens against the parity series.
8. E.D. Va. judgment (binary, when issued): structural divestiture ordered (desk band 20–35%) → the moat flag escalates to a thesis-level event; conduct-only remedies → the flag clears to monitoring.

Corrections in this edition (Oct 16, 2026)

This page’s prior edition (v2, same date) contained errors the desk’s own audit caught. They are listed, not buried:

1. Holdings line removed. Prior editions printed the author’s share count and portfolio weight in the header. Public pages carry no holdings data (Evidence Standard v3 holdings ruling).
2. Disclaimer added. The prior edition carried none; the standard footer now appears below.
3. EV bridge footing. The v2 bridge’s printed components summed to $3,999.2B against a stated $3,999.6B EV (market cap was rounded to $4,246.8B against the computed $4,247,139M). The bridge is now printed in $M and foots exactly to $3,999,569M, with the two off-basis components disclosed ($38M combined, 0.001% of EV).
4. AI Max quote re-dated. “Billions of net new searches” is Q2 2026 (Schindler, Q&A), not Q1 2026 as v2 printed. Dated corrections are filed in the transcript library itself (the 2024Q3, 2024Q4, 2026Q1 and 2026Q2 notes).
5. Cannibalization claim reworded. V2 printed “no evidence of cannibalization” as a Q2 2026 management statement. It traces to a desk annotation carrying a stale +12% Search figure (that was Q2 2025’s growth; Q2 2026 was +16.6%). No such statement exists in any call — management has never used the word. The page now carries the dated parity-assertion series plus the delivered record, labeled as what they are. A correction is filed in the library (2024Q1 note).
6. AI metrics sourced per figure. V2 asserted an I/O-era confinement rule it could not evidence. Each AI operating metric is now tied to the Q2 2026 call (Pichai, prepared remarks), verified against the primary record on Oct 16, 2026; the blanket attestation is withdrawn.
7. Waymo gap narrowed. V2 stated no Waymo ride-count series exists in Alphabet’s own record. The Q1 2026 call discloses 500K+ paid rides/week across 11 cities. The remaining gap is revenue and unit economics only.
8. SpaceX location corrected. V2’s bull case described the SpaceX mark as newly liquid while the valuation narrative grouped it with the non-marketable deduction. As filed: $80.0B sits in marketable securities (short-term sale restrictions) and $14.1B in other non-current assets (restricted through Q3 2027; not deducted in the bridge). No disclosed SpaceX sits in the non-marketable line — that line’s Q2 remeasurement gain is an unnamed private company.
9. Margin precision. V2’s “~15% demonstrated margin” is 14.65% in aggregate (FY22–24); required revenue growth at that margin is +24.7%/yr, not the +24.4% printed at 15%.
10. Share-count cell. The series table’s TTM WASO is now the weighted 12,244M; v2 printed the 12,230M spot count in a weighted-average row.
11. Verdict language. V2’s headline softened the bar (“duration, not magic”). The verdict box now carries the required-vs-delivered pair — +37.4%/yr required vs −1.7%/yr delivered — in the price line itself.
12. Model code hygiene. The filed model’s supplementary sum-of-the-parts block — judgment exit multiples, never published — has been deleted from the code and its outputs. The cleaned model was re-run and diffed field-by-field against the v2 output: identical, except a corrected FY22 revenue actual ($282,840M → $282,836M per EDGAR; the FY22–25 revenue CAGR moves +0.0005pt) and the removal of the unpublished SOTP keys. Nothing computed-but-unpublished remains in the filed model.
13. Management verification redone. V2’s insider record rested on a three-day window plus pattern assertions — including that Pichai’s selling cadence was “unchanged.” The full-population sweep (349 filings, Oct 1, 2024–Oct 16, 2026) shows his Dec 2024 plan completed on Mar 18, 2026, seven months before v2 described it as ongoing. The same sweep establishes what v2 only asserted: zero open-market purchases, zero discretionary sales, zero pledges in 24 months. Grade B affirmed on the full record.
14. Presented series corrected to the filings (Audit v3 fix pass, Oct 16, 2026). The series table’s SBC row did not tie to the 10-K cash-flow statements: FY21 printed $7,179M (filed: $15,376M), FY22 $19,359M (filed: $19,362M), FY23 $16,101M (filed: $22,460M — the printed figure appears nowhere in the filing), FY24 $24,906M (filed: $22,785M), FY25 $24,906M (filed: $24,953M). Dependent cells are corrected with it: owner earnings FY21 $59,833M → $51,636M, FY22 $40,665M → $40,648M, FY23 $53,394M → $47,035M, FY24 $60,324M → $49,979M, FY25 $48,314M → $48,313M; FCF FY22 $60,024M → $60,010M, FY24 $72,705M → $72,764M, FY25 $73,220M → $73,266M (all per the filed OCF and capex lines); capex FY22 $31,480M → $31,485M and TTM $134,968M → $132,402M (the model’s filed base input); FY25 operating income $128,962M → $129,039M; TTM revenue $445,837M → $445,866M. The printed FY24 row also failed its own arithmetic (printed FCF − printed SBC ≠ printed OE by $12,525M); every row now foots. The model’s governing FY25 base remains $48,314M ($1M of input rounding against the series’ exact-components $48,313M) and the governing TTM base is unchanged at $25,129M — no bar moves: every printed bar, credited scenario, and checkpoint threshold is as set. The delivered FY21→FY25 owner-earnings rate restates from −1.7%/yr to −1.6%/yr on the corrected FY21 base (FY22→FY25 stays +5.9%/yr); the figure quoted in item 11 is superseded accordingly. TTM operating income is now labeled where it appears: the $153,019M figure excludes the $5,391M EC Android fine charge; as reported it is $147,628M.
15. Diluted WASO row corrected (Oct 16, 2026). The series table’s “Diluted WASO (M)” row was mislabeled: its printed FY22–FY25 values (12,987 / 12,794 / 12,588 / 12,339) were period-end shares outstanding, not diluted weighted-average shares. They are replaced with the diluted WASO figures from EDGAR frames as used in the page’s governing model — FY22 13,159, FY23 12,722, FY24 12,447, FY25 12,230 — and the label is now accurate. The row’s other cells are conformed to the same diluted-WASO basis: FY21 remains “—” because no filed diluted WASO figure for FY21 is carried in the governing model/EDGAR frames, so no period-end count is substituted; TTM Jun-26 remains 12,244, which was already a diluted weighted-average figure on that basis.

Sources & method

Q2 2026 Form 10-Q (filed Jul 22, 2026): Note 3 (SpaceX $80.0B / $14.1B footnotes; non-marketable detail), Note 6 (debt groups), Note 7 (NCI; other income), Note 10 (legal proceedings), Note 11 (mandatory converts, ATM, June offering); balance sheet and cash-flow statement for the bridge. FY2021–FY2025 10-Ks and quarterly filings via EDGAR for the series (FY22 revenue as filed: $282,836M). Earnings calls: 12 quarters (Q3 2023–Q2 2026) located and read from publisher transcripts (Alphabet IR; Motley Fool) — texts at source; the desk library holds contemporaneous notes, not the full texts — cross-checked against Alphabet IR’s posted Q2 2026 transcript on Oct 16, 2026. DEF 14A (filed Apr 24, 2026) for votes and pay; Form 4 population, Oct 1, 2024–Oct 16, 2026 (349 Alphabet-issuer filings parsed in full from EDGAR) for the insider record. Q3 2026 date: Alphabet IR announcement, Oct 6, 2026. Prices: Yahoo chart API, Oct 16, 2026 (Oct 15 close). Charts: Figure 1 — revenue; Figure 2 — free cash flow; Figure 3 — net income (Qualtrim, dark exports). Reverse DCF: house method — 10% required return, 10-year horizon, 2% terminal growth, solve on EV, owner earnings = FCF − SBC; model and outputs filed in the desk dossier (reverse-dcf-2026-10-16-v3.json; v3-model-note-2026-10-16.md; debt-tranches-2026-06-30.md). This page is a dated research note, not a live feed: it is updated after each quarterly print, and the next update scores Q3 2026 against the eight checkpoints above.