Alphabet GOOGL

Reference table

1. The business and the segments

Alphabet reports three segments — Google Services, Google Cloud and Other Bets — with certain costs held back as unallocated "Alphabet-level activities," which the 10-Q says "primarily reflect expenses related to our shared AI research and development." The company had 198,933 employees as at 2026-06-30 (the 10-Q figure; the vendor profile still carries 190,820, and the filing wins). Chief executive: Sundar Pichai.

Revenue by line, FY2023-FY2025 (vendor seg_prod, cross-checked to the reported total):

The six lines sum exactly to reported FY2025 revenue of $402.963B, which validates them. The vendor adds a seventh row, "Other Segments" at −$127M, which would break the reconciliation and is discarded (Section 8). The vendor also relabels the subscriptions line as "Google Inc." in the FY2025 record and as "Google Subscriptions, Platforms, And Devices" in every prior year — a label instability worth naming.

The June 2026 quarter, from the 10-Q's own narrative. Total revenue $119.796B, +24.2% year over year. Within it: Search and other +$9.1B; YouTube advertising +$1.3B; Network −$51M; subscriptions, platforms and devices +$1.7B; Google Services in aggregate +$12.0B, or +15%; Google Cloud +$11.1B, or +82%, which puts cloud at roughly $24.6B for the quarter — a ~$99B annual run-rate. The 10-Q attributes cloud growth to "Google Cloud Platform largely from infrastructure and platform services" and adds a genuinely new fact: "in the second quarter of 2026, we began recognizing revenue from the sale of TPU systems" — the company has started selling its custom accelerator hardware to third parties, converting an internal cost advantage into an external market.

Segment operating income — the strongest number in this dive. Absolute segment operating income is not recoverable from the extracted filing text (Section 8), but the year-over-year movements are stated explicitly and they are extraordinary:

Google Services delivered a 54% incremental operating margin while absorbing a $1.5B legal charge booked into its own general-and-administrative line. Google Cloud delivered the same incremental margin while scaling 82%. The FY2025 Other Bets loss of $7.5B included a $2.1B fourth-quarter compensation charge relating to the autonomous-driving unit, which then raised $16.0B in February 2026 with "the significant majority" funded by Alphabet.

Geography (vendor seg_geo, FY2025): United States $194.209B (48.2%), EMEA $117.045B (29.1%), Asia-Pacific $67.680B (16.8%), Americas excluding the US $23.902B (5.9%). Unlike many names in this feed, seg_geo is fully populated here and grew in every region (US +13.9%, EMEA +14.6%, APAC +19.1%).

2. The 94% net margin — the single most important thing to understand

Alphabet's reported earnings for the two most recent quarters are dominated by unrealised marks on private equity stakes, and are therefore close to useless as a measure of earning power.

The vendor's quarterly income statements are, unusually, correct — the filing confirms them. The problem is not the data, it is the interpretation. Here is what actually happened:

The 10-Q filed 2026-07-23, at Item 2: "OI&E of $98.0 billion for the three months ended June 30, 2026 included net gains on equity securities of $99.0 billion, primarily related to unrealized gains in our equity securities portfolio from SpaceX and a private company." The 10-Q filed 2026-04-30 says the same of the March quarter: "OI&E of $37.7 billion for the three months ended March 31, 2026 included net gains on equity securities of $36.9 billion, primarily related to unrealized gains on our non-marketable equity securities." And the 10-K: "OI&E of $29.8 billion for the year ended December 31, 2025 included net gains on equity securities of $24.1 billion."

The tax note is the clincher. The 10-Q explains the jump in the effective rate as "primarily attributed to higher pre-tax earnings from unrealized gains on equity securities that are deferred tax liabilities at the statutory tax rate." A deferred tax liability is what you book when income has been recognised for accounting but not for tax — i.e. no cash has arrived and none has been paid out.

Deriving core earnings, stated method first. Core pre-tax income = operating income + (other income and expense − net equity-security gains). Tax at 17%, which is Alphabet's FY2025 effective rate adjusted for the statutory rate applied to the marks. Divide by the diluted share count in the same quarter:

• Q2'26: $40.770B + ($97.983B − $99.0B) = $39.75B core pre-tax → $2.68 core EPS. Adding back the $2.1B legal charge disclosed in the same filing (a $1.5B damages award in general and administrative expenses plus $581M of interest in other income) gives $2.82 — against a $2.87 consensus. An in-line quarter.
• Q1'26: $39.696B + ($37.716B − $36.9B) = $40.51B → $2.75 core EPS, against a $2.64 consensus. A modest beat.
• FY2025: $129.166B + ($29.660B − $24.1B) = $134.73B → $9.14 core EPS, against reported diluted earnings of $10.81.

What this does to the multiple. Trailing reported diluted earnings, summing the four earn_cal actuals, are $19.91 — and $377.65 ÷ $19.91 = 18.97x, exactly the vendor's priceToEarningsDilutedRatioTTM. That figure is rejected. Trailing core earnings are at most $11.12 (Q3'25 $2.87 + Q4'25 $2.82, both upper bounds because the FY2025 equity gains are not split by quarter, plus core Q1'26 $2.75 and core Q2'26 $2.68), giving a trailing core multiple of at least 34.0x. A 34x trailing multiple is a completely different investment case from a 19x one, and every screen in the market shows the latter.

The corroborating evidence is in the cash flow statement. The vendor's incomeQualityTTM — operating cash flow divided by net income — has fallen to 0.76. For FY2025 the same ratio was 1.25 ($164.713B of operating cash flow against $132.170B of net income). Earnings that outrun cash by 32% at a company with Alphabet's collection profile are not earnings.

And it cuts both ways. These are marks, not sales. If the private valuations that produced +$99.0B in one quarter move down, the same mechanism produces a symmetric loss. The knowledge base's most specific structural warning on this point, from forward_guidance on 2026-06-12 at conviction 75, names it precisely: "marked-up private stakes coming to market."

Therefore: this dive values Alphabet on core operating earnings and forward consensus. Reported trailing earnings, trailing price-to-earnings, trailing return on equity (50.8%), trailing EBIT margin (67.4%) and trailing pre-tax margin (67.1%) are reported for completeness and used for nothing.

3. Growth — a mega-cap re-accelerating

The second derivative is what matters here, and it is positive:

Consensus continues the acceleration and then fades it: $495.195B FY2026E (+22.9%, 35 analysts), $605.867B FY2027E (+22.3%, 39), $721.625B FY2028E (+19.1%, 38), $815.730B FY2029E (+13.0%, 29), $936.258B FY2030E (+14.8%, 35). Analyst coverage is deep — 39-42 on the near forward earnings lines — which is the opposite of the thin-coverage problem this feed usually presents.

Three observations on the quality of that growth.

First, margins are expanding into the acceleration, not contracting. Gross margin rose from 59.5% in the June 2025 quarter to 61.6% a year later; operating margin from 32.4% to 34.0%. The 10-Q attributes the cost-of-revenue increase to "depreciation expense, TAC, inventory and other costs, content acquisition costs," and separately notes the traffic-acquisition-cost rate fell from 20.6% to 19.8% on mix shift away from the network properties. Operating expenses grew 27% against revenue growth of 24%, so the margin gain came from gross profit.

Second, the incremental margins are outstanding. Both reporting segments converted incremental revenue to incremental operating income at 54% in the June quarter (Section 1). At that conversion rate, consensus revenue growth of $110B in FY2027 alone would carry roughly $59B of incremental operating income — which is broadly what the consensus EBIT line implies ($196.791B FY2027E against $160.844B FY2026E, a $36B step, i.e. consensus is assuming the incremental margin degrades to about 32%).

Third, some of the step was bought. The Wiz acquisition closed 2026-03-11 for $29.5B after purchase-price adjustments and is consolidated within the Google Cloud segment. Intersect, a renewable-energy developer, closed 2026-03-10 for $5.9B and is allocated across segments. Neither was in the FY2025 base. The 82% cloud growth is therefore not entirely organic and the filing does not disaggregate it — a real gap.

4. Balance sheet — the capital raise, the corrected EV, and $811B of commitments

The vendor balance sheet is seven months stale and the seven months in question are the most important in the company's financing history. bal_a runs only to 2025-12-31. Everything below at 2026-06-30 comes from the 10-Q and supersedes it.

The vendor's enterprise value is wrong in the usual way. km_ttm.enterpriseValueTTM is $4,627.207B against a market capitalisation of $4,570.362B, i.e. it adds $56.845B of implied net DEBT. The company has net cash of roughly $140.1B. Corrected enterprise value is approximately $4,430.3B, a $196.9B (4.3%) overstatement by the vendor — small in percentage terms and larger in absolute terms than most companies in this universe. All EV-derived multiples in the payload are recomputed or rejected accordingly (Section 8). On the corrected figure, EV to FY2027 consensus revenue is 7.31x.

What happened in June 2026 — the capital raise, in full. This is the largest change to Alphabet's capital structure since it listed, and none of it is in the vendor file.

• 2026-06-04, public offering: 29 million Class A shares at $355.1982 and 29 million Class C shares at $351.8018. Net proceeds $20.5B.
• 2026-06-04, private placement: 14 million Class A and 14 million Class C shares to an affiliate of Berkshire Hathaway Inc. Net proceeds $10.0B.
• 2026-06-05, mandatory convertible preferred: 385 million Series A and Series B depositary shares representing 19 million shares of 6.25% mandatory convertible preferred stock, liquidation preference $1,000 per share ($50 per depositary share). Net proceeds $19.0B. Converts on or about 2029-05-15 at 2.2520-2.8160 Class A shares (Series A) and 2.2740-2.8420 Class C shares (Series B) per preferred share. Capped calls cost a $1.0B premium with cap prices of $532.6704 (Class A) and $527.7974 (Class C), a 50% premium to the offering prices.
• 2026-06-01, at-the-market programme: authority to sell up to $40.0B of Class A and Class C stock through sales agents. No shares sold as at 2026-06-30; the full $40.0B remains available.
• Debt, first half 2026: $20.0B of US-dollar notes (Q1, 4.80% weighted coupon, 15-year weighted maturity) plus $31.8B of foreign-currency notes across Sterling, Swiss Franc, Euro, Canadian dollar and yen — $51.8B in six months, of which $20.3B net was raised in the June quarter alone.

Aggregate net equity proceeds of $49.6B, explicitly earmarked in the filing "for general corporate purposes, including capital expenditures to scale AI infrastructure and global compute." The dilution itself is modest — 86 million new common shares is 0.70% of the diluted count, potential conversion adds another 43-54 million (0.35-0.44%), and the full at-the-market authority at today's price would be ~106 million shares (0.86%). The signal, not the arithmetic, is what matters: a company that repurchased $45.4B of its own stock in FY2025 sold $49.6B of it in June 2026.

The preferred also carries a real cash cost: 19 million shares at a $1,000 liquidation preference and a 6.25% rate is $1.19B a year of preferred dividends, which the 10-Q confirms is deducted in arriving at net income available to common. The Board declared $12.15 per preferred share in July, payable 2026-08-15.

The commitments, which are the part almost nobody reads. As at 2026-06-30 the 10-Q discloses:

• $811.0B of purchase commitments and other contractual obligations, of which $200.7B is short-term (Item 2), described in Note 10 as $707.0B of expected future fixed or guaranteed commitments, "the significant majority of which related to long-term supply agreements," generally fulfilled through 2030, plus energy take-or-pay contracts running to 2054.
• $85.2B of future lease payments on leases entered but not yet commenced, primarily data centres, commencing between 2026 and 2031 — up from $58.5B at 2025-12-31. Plus a June 2026 short-term lease with a $5.8B non-cancellable commitment commencing in the third quarter.
• Backstops to third parties: $7.6B of financial guarantees plus $43.8B of credit-derivative notional = $51.4B of maximum potential future payments, plus $24.1B of agreed future backstops to support data-centre and energy build-out. The trajectory across three consecutive balance-sheet dates is the story: $5.7B + $16.9B = $22.6B (2025-12-31) → $9.0B + $28.4B = $37.4B (2026-03-31) → $7.6B + $43.8B = $51.4B (2026-06-30). The credit-derivative notional — backstops on third-party data-centre leases — has grown 159% in six months.
• $20.0B of future capital funding commitments with a private company, contingent on operational and financial milestones through 2030, accounted for as an equity derivative.

The legal and regulatory ledger. This is the largest single unquantifiable on the name and the 10-Q gives it four pages.

• Android (Europe): the European Court of Justice denied the final appeal in July 2026. The decision is now final and $5.2B in fine plus accrued interest was paid in cash in July 2026 — previously accrued, so a cash event rather than an earnings event.
• Search (US): the District Court ruled against the company in August 2024; final judgment was entered in December 2025, imposing restrictions on distribution and requiring the sharing of certain search data with, and syndication services to, competitors. The company appealed in January 2026 and moved to pause certain remedies; the motion to stay was denied as premature. The DOJ and state Attorneys General also appealed, in February 2026.
• Advertising technology (US) — the big one. The Eastern District of Virginia issued a mixed decision in April 2025: advertiser tools and prior acquisitions were not anticompetitive, but publisher tools "unfairly excluded rivals." A remedies proceeding ran in September 2025 with closing arguments in November 2025, and judgment is awaited. The filing states: "The DOJ's remedy proposal includes structural remedies that could have a material adverse effect on our business." "Given the nature of these matters, we cannot estimate a possible loss." A separate state case in the Eastern District of Texas follows the remedies decision.
• Advertising technology (Europe): a €3.0B fine in September 2025 for "self-preferencing," against which a $3.5B charge was taken and bank guarantees placed in lieu of cash payment; under appeal. The UK issued a Statement of Objections in September 2024.
• Private follow-on actions: in July 2026 the Stockholm Patent and Market Court awarded a plaintiff approximately $2.1B in principal damages plus interest over shopping-search ranking; $1.5B was booked in Google Services general and administrative expenses and $581M in other income in the June quarter. Under appeal. The 10-Q warns explicitly that this is a template: "private claims stemming from regulatory proceedings in which Google's liability has been fully determined and the remaining dispute concerns potential damages."
• Digital Markets Act (EU): two investigations, preliminary findings of non-compliance issued March 2025. "We cannot reasonably estimate a probable loss."

Risk factors DID change. Part II Item 1A of the 10-Q lists material changes since the 10-K, and they map exactly onto everything above: a new paragraph on supplying accelerator systems ("may increase our costs and operational complexity"), a new paragraph on "financial guarantees, such as backstops to support the build-out of third-party data centers and power infrastructure," an entirely new risk factor headed "Disruptions in our ability to access the capital markets, obtain future financing, or manage our indebtedness…", and a new section headed "Risks Related to Ownership of Our Stock" covering dilution from the at-the-market programme and the convertible preferred, and warning that the company "may be required to suspend share repurchases" if it is issuing equity. A company writes those risk factors when it knows what it is doing to its own share count.

5. Cash flow — the free-cash-flow inversion

This is the inversion. Operating cash flow grew 31% in FY2025 and free cash flow grew 0.7%, because capital expenditure grew 74%. In the June 2026 quarter capital expenditure of $44.9B exceeded operating cash flow of $39.1B, and free cash flow was approximately −$5.8B. The knowledge base's independent reading of the same print, from eurodollar_university on 2026-07-30 at conviction 60, puts it at −$5.9B — an independent triangulation of our arithmetic.

Three supporting facts. Capital expenditure ran at 37.5% of revenue in the June quarter. The vendor's capexToDepreciationTTM is 5.25x — the asset base is being built more than five times faster than it is being written off, which means the depreciation drag is almost entirely in the future. First-half depreciation on property and equipment was $13.6B against $9.5B a year earlier, +43%, while capital expenditure rose +104%. The gap between those two growth rates is the deferred earnings headwind.

And the guidance direction is unambiguous and unquantified. The 10-K, the March 10-Q and the June 10-Q all say the same sentence: "In 2026, we expect to significantly increase, relative to 2025, our investment in our technical infrastructure." FY2025 was $91.4B. The filings never put a number on 2026. Earnings-call transcripts are not available on this research plan, and the 8-K of 2026-07-22 carries the earnings release, not the call — so if a specific capex figure was guided verbally, it is not in our sources and we will not invent one. The knowledge base carries two independent estimates, labelled as such and not as company figures: eurodollar_university on 2026-07-30 cites a "2026 budget rises to $195-205B", and the same source on 2026-07-23 cites a "2027 budget nearing ~$400B per JP Morgan." Treat both as third-party estimates.

Shareholder returns have gone to approximately zero and are turning negative. Repurchases were $45.709B in FY2025 and $0.000 in both quarters of 2026, with $69.5B of the April 2025 authorisation still available — unchanged between 2025-12-31 and 2026-06-30, which is the arithmetic proof that nothing was bought. The dividend was raised 5% to $0.22 per quarter in April 2026, a yield of 0.233%. Against that: $49.6B of equity issued, a $40.0B at-the-market facility live, and $1.19B a year of new preferred dividends ranking ahead of the common. On any honest accounting, the net shareholder yield in 2026 is negative.

The freeCashFlowYieldTTM of 1.17% and priceToFreeCashFlowRatioTTM of 85.8x are, for once, not vendor defects. They are the correct description of what has happened.

6. Valuation — priced in or room?

At $377.65 (market cap $4.570T; ~12.31B diluted shares in the June quarter):

Why FY2026 consensus EPS of $20.267 is rejected, with the proof. Actual first-half reported earnings were $5.11 + $9.11 = $14.22. The quarterly consensus for the September quarter is $3.02 (earn_cal). $20.267 − $14.22 = $6.047, which implies a December-quarter estimate of $3.03. In other words, the FY2026 annual consensus is a hybrid: two quarters of GAAP actuals containing roughly $8.7 a share of unrealised investment marks, bolted onto two quarters of core estimates. It is not a forecast of $20.27 of earning power and must never be used as a valuation denominator. Implied FY2026 CORE EPS = $2.75 + $2.68 + $3.02 + $3.03 = ~$11.48, which is 32.9x. The FY2027 line, at $15.006 on 42 analysts, is a clean forward number and is the primary anchor of this dive.

A note in the vendor's favour, tested rather than assumed. The house contract warns that est EBITDA/EBIT rows are corrupt in roughly 70% of files. For GOOGL they are not. FY2025 ebitAvg of $130.021B against actual operating income of $129.166B is a 0.7% match; FY2024's $113.838B against $112.390B is 1.3%. The EBIT row is therefore accepted here and used to build the operating-only EPS ladder above, which strips the ~$1.65 a share of assumed non-operating income embedded in the FY2027 consensus earnings figure.

Peer context. The vendor peer set — Apple ($4.544T), Nvidia ($5.133T), Microsoft ($3.659T), Broadcom ($1.989T), Taiwan Semiconductor ($2.164T), Meta ($1.498T), Cisco, Flex, Garmin and Magic Software — supplies prices and market capitalisations only, no multiples. No peer-multiple comparison is drawn. The vendor's own composite rating is B+ with an overall score of 3, with the sub-scores telling the story: returns on equity and assets both 5 (top marks, but computed on the inflated earnings), price-to-earnings 2 and price-to-book 1.

6a. What today's price assumes (the inversion)

At $377.65 — 25.2x FY2027 consensus earnings, ~28.3x on an operating-only basis, ~32.9x current-year core — the price embeds roughly the following falsifiable claims:

• Revenue compounds ~22% in both FY2026 and FY2027, reaching $605.9B. (Consensus-derived: $495.195B on 35 analysts, then $605.867B on 39.) That requires cloud to keep compounding near 50%+ off a ~$99B run-rate and search to keep growing double digits with no share loss. The June quarter's +24.2% supports it; nothing in the filings guarantees it.
• Earnings per share reach $15.01 by FY2027, up ~31% from our derived FY2026 core of ~$11.48, with operating margin held near 32.5%. (Consensus for the numerator; our derivation for the base.) This is the most fragile assumption in the price. It requires the depreciation from roughly $200B of 2025-2026 capital expenditure to land without breaking the margin, at a company whose first-half depreciation grew 43% while capital expenditure grew 104%. The catch-up has not happened yet.
• Capital expenditure inflects. (Our number, derived.) At the June quarter's 37.5% of revenue, FY2027 revenue of $605.9B implies capex of ~$227B against consensus operating cash flow that is nowhere near it. The price implicitly assumes the capital-intensity ratio falls — the filings say only "significantly increase," with no number and no peak.
• The advertising-technology structural remedy is worth zero. (Our reading of the filing.) Judgment is awaited after November 2025 closing arguments, the government's proposal is structural, the company says it "could have a material adverse effect" and "cannot estimate a possible loss." The market is carrying a nil provision for an outcome the company will not size.
• The multiple does not de-rate despite a 1.17% free-cash-flow yield, a zero buyback and a live $40.0B share-selling programme. (Vendor freeCashFlowYieldTTM, 10-Q Note 11.) For a decade the multiple was supported by a structural bid from repurchases. That bid is gone and, at the margin, reversed.

6b. The return bridge (why the multiple moves)

Expected return over 12-24 months decomposes as: core earnings growth (+31% from our derived FY2026 core of ~$11.48 to consensus FY2027 of $15.01) + multiple drift (COMPRESSION in our base) + shareholder yield (+0.23% dividend, ZERO buyback, and negative once the at-the-market programme is used).

Our base case explicitly assumes multiple compression, and we say so rather than hiding it in a rolldown. Today the market pays 25.2x on the next full year (FY2027). Our $390 base is 22.0x on FY2028 — i.e. one year forward from here the market pays roughly 22x on the then-next full year rather than 25x. That is a genuine ~13% de-rating, not the mechanical rolldown the addendum warns about.

Why compression rather than hold. Four reasons, all traceable. The shareholder yield that supported the multiple has gone to zero and is turning negative. The free-cash-flow yield is 1.17% and structurally so for at least two more years. The balance sheet has moved from a pure equity story to one with $98.2B of debt, $19.0B of preferred ranking ahead of the common, $51.4B of third-party credit backstops and $811.0B of purchase commitments. And an unresolved structural-remedy proceeding sits on the largest profit pool. Against that, revenue is accelerating, which argues for expansion — and that is precisely why the compression assumed is mild rather than severe.

If we are wrong and the multiple simply holds at 25.2x on the next full year, the price in a year is ~$446 (+18%) — that is effectively the path to our bull case. If the market de-rates a re-accelerating platform to 15x, the price is $266 — that is the bear. Almost none of the expected return in the base case comes from re-rating; almost all of it comes from earnings growth, and the multiple takes most of it back. That is the honest arithmetic of a Hold.

6c. Variant perception (where we differ, what would surprise)

• We differ from every screen in the market on the trailing multiple, and it is not a close call. The vendor, and by extension the retail-facing data that runs on it, shows Alphabet at 18.97x trailing earnings. The honest figure on core earnings is at least 34.0x. The gap is entirely a $99.0B non-cash mark on private stakes disclosed in one sentence of a 10-Q. This is the single largest perception gap on the name, and it runs in the direction of the stock being more expensive than it looks, not less. Watchable number: the September-quarter other-income line against the $3.02 core consensus.
• We differ on the shareholder yield. The market still carries Alphabet as a buyback compounder. Repurchases were $0.000 in both quarters of 2026, $69.5B of authorisation is untouched, $49.6B of equity was issued in June, and a $40.0B at-the-market facility is live and unused. The 10-Q's own new risk factor concedes the company "may be required to suspend share repurchases" while issuing equity. Watchable event: any share sold under the at-the-market programme, which would be the first confirmation that dilution is being used rather than merely authorised.
• We are LESS bullish than the street, which is unusual for us. Consensus is $423.44 with 71 buy-side ratings against 11 holds and one sell; our base is $390, 8% below. We think the street is capitalising an acceleration that is real while under-provisioning for a capital intensity that is also real and a remedy proceeding it cannot price.
• We do NOT differ on the growth. Revenue +24.2%, cloud +82%, 54% incremental operating margins in both segments and expanding gross margin are facts from the filing, not forecasts. Anyone bearish on this name on the basis of business deterioration is looking at the wrong quarter. Our disagreement is entirely about price and financing.
• Positive surprise that would force us higher: a capital-expenditure figure that flattens, free cash flow returning to positive, or a resumption of repurchases at scale — any of which would restore the multiple's structural support. A favourable advertising-technology remedy removing the structural tail. Or a September quarter with core earnings materially above $3.02 on continued 80%+ cloud growth. Watchable numbers: quarterly capex, quarterly free cash flow, buyback dollars.
• Negative surprise that would break the thesis: a downward revision of the private-stake marks. The mechanism that produced +$99.0B in one quarter is symmetric; a 30% write-down would be a ~$30B pre-tax hit and could put a quarter into loss with no change in the operating business. Also: a structural remedy in advertising technology; another capital-expenditure increase with no free-cash-flow path; or actual use of the at-the-market programme.

Synthos fair values

All three anchors are multiples of FY2027 consensus earnings of $15.006 (42 analysts), cross-checked against FY2028 consensus of $17.716 and against the operating-only ladder derived from the verified ebitAvg row.

• Bear ~$270 — 18.0x FY27E. Cross-checks: 15.2x FY28E, 17.1x the FY2028 operating-only figure. The scenario: the advertising-technology judgment imposes a structural remedy, or capital expenditure rises again with free cash flow still negative into 2027 and the depreciation wave compresses the operating margin below 30%. The market re-rates a capital-intensive platform toward a mature multiple. −28.5%.
• Base ~$390 — 26.0x FY27E. Cross-checks: 22.0x FY28E, 19.2x FY29E, ~29.2x the FY2027 operating-only figure of ~$13.36. The scenario: consensus is roughly right, revenue compounds 22% and margins hold, but the multiple gives back ~13% because the shareholder yield is gone and the free-cash-flow yield stays near 1%. +3.3%, plus a 0.23% dividend. This is a fully-valued outcome, and that is the finding.
• Bull ~$475 — 31.7x FY27E, 26.8x FY28E. Note this is exactly the street's high target, which is a useful independent check that we have not built a fantasy. The scenario: cloud sustains 50%+ growth, the accelerator-systems line becomes material, capital expenditure inflects in 2027 and free cash flow snaps back, the advertising-technology remedy is behavioural rather than structural, and the crowd keeps paying a growth multiple. +25.8%.

Base is 3.3% above spot; asymmetry is roughly 0.9:1 (28.5% down against 25.8% up) — very slightly negative, on a stock that has risen 99.7% in twelve months. That is a Hold: own it if you own it, do not add here, and note that the bear case is larger than the bull.

7. Knowledge base — 613 claims, 132 of them from the company's own executives

Raw hits: 613 across the entity aliases GOOGL, Google and Alphabet (case-insensitive, exact entity match) over all 51,928 records, spanning 2011-06-23 to 2026-08-01, from 58 distinct sources. Before any of it can be read as conviction, four deductions are mandatory:

The independent count is the number that matters, and it changes the reading completely. All 613 claims read 65% bullish. The 427 independent claims read 240 bullish / 102 bearish / 85 neutral. And in the most recent window — since 2026-06-01 the independent lane is 17 BEARISH against 16 bullish and 2 neutral, a dead heat. Mean conviction is 70.8 on the bullish side and 66.0 on the bearish: the bears are nearly as convinced as the bulls. A lane whose single largest voice is the company's own chief executive is not a conviction signal, and we score it Mixed.

A note on quotation. Many claims in this lane name specific model families and research organisations. House style prohibits reproducing those names, so where they appear they are replaced with a bracketed generic and the claim is marked accordingly. Everything else is verbatim from the stored claim text.

The bearish lane — the capital-structure and credit argument, which deserves the airtime because it is the one this dive's own arithmetic corroborates:

> 2026-07-30 · eurodollar_university · bearish · conviction 60

> "Alphabet's first-ever negative free cash flow (-$5.9B) as capex hits $45B and 2026 budget rises to $195-205B shows AI spending overwhelming even the strongest cash generators, with each dollar buying less capacity."

> 2026-07-30 · eurodollar_university · bearish · conviction 72

> "The AI bubble is a credit bubble, not an equity one — CDS spreads and weak bond demand show the debt cycle turning as capex is financed at the wrong price, risking terrible returns."

> 2026-08-01 · eurodollar_university · bearish · conviction 55

> "AI-buildout debt is drawing growing caution — CDS on Oracle (215bps), SpaceX, Nvidia, Meta, Amazon, Alphabet moved higher in July; Amazon's $25B deal saw its order book shrink from $62B to $41B."

> 2026-06-12 · forward_guidance · bearish · conviction 75

> "Bearish Mag7/hyperscalers — record equity issuance and reversed buybacks, drunken low-return AI capex, marked-up private stakes coming to market; a $6T company can't sustain a growth multiple. Underperformance continues."

> 2026-02-06 · forward_guidance · bearish · conviction 75

> "Soaring AI capex is killing buybacks and forcing debt/equity issuance, removing the structural bid; shorting Mag 7 makes sense for the first time."

> 2026-06-21 · jordi_visser · bearish · conviction 90 — the single highest-conviction claim in the entire lane

> "Core repeated trade: long his AI thematic portfolio, short the hyperscalers — 'your capex is my opportunity'; massive capex spend is a depreciation time bomb driving multiple compression."

> 2026-07-29 · we_study_billionaires · bearish · conviction 45

> "With ~90% search share Google has maxed its market and become overvalued, with no near-term growth runway since new bets like Google Classroom take a decade to pay off."

> 2026-07-21 · all_in · bearish · conviction 50

> "Market leaders are borrowing billions on top of cash flow for capex amid an existing private-credit problem; the ~$100B AI spend is priced to perfection and must return in earnings, not just revenue."

> 2026-02-06 · darius_dale · bearish · conviction 78

> "Hyperscaler capex (~$650B in 2026, near doubling) is an unprecedented AI CapEx bubble that will end in a secular bear market."

> 2026-06-18 · odd_lots · bearish · conviction 75

> "The magic seven were seven easy monopolies dominating separate niches; now they're all fighting in one AI market — capital-heavy this time, not capital-light — a bloodcurdling dogfight where only one can be first mover."

Why this lane is unusually credible: it made a specific, dated, checkable prediction and the filing confirmed it. The February 2026 forward_guidance claim — "soaring AI capex is killing buybacks and forcing debt/equity issuance, removing the structural bid" — was made four months before Alphabet stopped repurchasing shares, issued $49.6B of equity and established a $40.0B at-the-market programme. The June 2026 claim added "marked-up private stakes coming to market," nine days before the quarter in which a $99.0B mark on private stakes produced a 94% net margin. We do not usually get to grade a KB lane against a filing this cleanly, and this one graded well.

The bullish lane — the backlog and franchise argument, which is equally serious:

> 2026-07-26 · jordi_visser · bullish · conviction 75 — from the same source as the conviction-90 bear claim above, seven weeks later

> "Google's backlog is the whole ball game — grew 106B to 514B in a year, cloud +82%, spending chases visible demand; capex likely value-accretive over 3-5 years."

> 2026-07-21 · compound_and_friends · bullish · conviction 72

> "Alphabet is the last clean AI hyperscaler; its earnings are the make-or-break moment for tech — strong cloud/capex guidance would reverse the capex sell-offs hard to the upside"

> 2026-01-22 · we_study_billionaires · bullish · conviction 78

> "Google holds a ~90% search monopoly with no real competitor; a creative monopoly that captures far more value than it creates, though LLMs are a long-term wildcard."

> 2026-04-03 · empire · bullish · conviction 80

> "Comfortable owning Google multi-year — it's a monopoly unlikely to be unseated, and trading down ~20% YTD." (Note the entry context: the stock is 95% above its 52-week low today.)

> 2026-07-18 · jordi_visser · bullish · conviction 75

> "Not worried about hyperscaler debt: ~$2T capex is matched by ~$2T contracted compute demand (RPO); government will ensure the AI trade works vs China. Capex keeps accelerating."

> 2026-01-31 · lex_fridman · bullish · conviction 70

> "Google's TPU vertical integration avoids NVIDIA's insane chip margins; head start on datacenters gives structural cost advantage in high-lead-time infrastructure."

> 2026-02-19 · raoul_pal · bullish · conviction 85 — the highest-conviction bullish claim, and the one most damaged by subsequent events

> "Big-four AI CapEx isn't over-investment; strong ROI over 1-2 years lets them pay it back in few years given minimal debt." The "minimal debt" premise no longer holds: long-term debt went from $46.5B to $98.2B in six months, plus $19.0B of preferred and $51.4B of third-party backstops.

The most useful claim in the lane is the neutral one:

> 2026-07-07 · lyn_alden · neutral · conviction 55

> "Hyperscaler free cash flows have collapsed from aggressive AI capex and whether it pays off is a genuine question, so she wants valuation discounts before taking the risk — was less enthused on Alphabet after its run-up."

That is, almost word for word, the conclusion this dive reaches independently from the filings: the business is fine, the question is the price and the capital intensity, and the run-up has removed the discount.

Management's own lane, at half weight and stated as such. The 132 Alphabet-executive claims are 97% bullish and centre on integration of a single model stack across products, full-stack ownership from silicon to applications, and a universal assistant. They are useful as a statement of intent and worthless as evidence. One is worth recording precisely because it cuts against the rest: on 2026-06-01, demis_hassabis_mgmt — the head of Alphabet's research unit — filed a bearish claim at conviction 35 conceding the company "lacks a competitive product by usage in the two seminal moments" and is "losing the zeitgeist." When a company's own research chief is the bear, it is worth more than the 128 bullish management claims combined.

What we discarded and why. 11 quarantine_misattributed claims excluded under the four-lane attribution policy. 132 Alphabet-management and 35 peer-management claims down-weighted to half. 8 promoter-class claims down-weighted. Roughly a third of the independent lane names specific model families or research organisations and was read for substance but not quoted, under house style. No claim was excluded for being inconvenient, and the bearish lane is quoted at greater length than the bullish one deliberately, because it is the one this dive's own arithmetic supports.

8. Data integrity — what we rejected from the vendor file and why

Eleven items. The first is the one that matters.

1. Trailing price-to-earnings is computed on unrealised investment marks — REJECTED as a lens. ratios_ttm.priceToEarningsRatioTTM (18.751x) and priceToEarningsDilutedRatioTTM (18.968x) are arithmetically correct on trailing reported diluted earnings of $19.91. The vendor reports trailing earnings of $19.91 a share and a multiple of 18.97x; the 10-Q filed 2026-07-23 shows that $9.11 of the June quarter's $9.11 diluted result rests on $99.0B of unrealised gains on private equity stakes, and the March 10-Q shows $36.9B of the same in Q1; we use core earnings of at most $11.12 and a multiple of at least 34.0x. Also rejected on the same grounds: returnOnEquityTTM (50.84%), returnOnAssetsTTM (26.49%), ebitMarginTTM (67.39%), pretaxProfitMarginTTM (67.11%), netProfitMarginTTM (54.76%), earningsYieldTTM (5.32%), interestCoverageRatioTTM (296x), grahamNumberTTM ($154.39) and both PEG ratios. The tell is in the file itself: ebtPerEbitTTM of 2.026 — pre-tax income at twice operating income is not a business, it is a mark.

2. est FY2026 EPS of $20.267 is a hybrid of actuals and estimates — REJECTED as a valuation denominator. First-half reported earnings were $14.22; the September consensus is $3.02; the residual implies a December estimate of $3.03. The annual figure therefore contains roughly $8.7 a share of unrealised marks and is not a forecast of earning power. est FY2026 netIncomeAvg of $253.552B against ebitAvg of $160.844B — net income 58% above operating profit — is the same defect surfacing. We use FY2027 ($15.006, 42 analysts) as the primary forward anchor and a derived FY2026 core of ~$11.48.

3. The balance sheet is SEVEN MONTHS STALE and the interval contains the largest financing events in the company's history. bal_a ends at 2025-12-31. The vendor reports cash and short-term investments of $126.843B and long-term debt of $46.547B; the 10-Q shows $242.5B of cash, equivalents and short-term marketable securities and $98.2B of long-term debt at 2026-06-30; we use the 10-Q. Between the two dates the company issued $49.6B of equity, $51.8B of notes, established a $40.0B at-the-market programme and closed $35.4B of acquisitions. No quarterly balance sheet is supplied in this feed.

4. km_ttm.enterpriseValueTTM ($4,627.207B) treats a net-cash company as carrying $56.845B of net debt — CORRECTED. On the 10-Q figures the corrected net position is +$140.1B of net cash and corrected enterprise value is approximately $4,430.3B, a $196.9B overstatement. evToSalesTTM (10.376x), evToEBITDATTM (14.205x), netDebtToEBITDATTM (0.175x) and enterpriseValueMultipleTTM (14.205x) are all recomputed or rejected — the EBITDA-based ones additionally because the denominator inherits defect 1.

5. Segment revenue: a spurious seventh row and unstable labels — the row DISCARDED, the labels corrected against the filing. seg_prod FY2025 carries "Other Segments" at −$127M; the six substantive lines sum to exactly the reported $402.963B without it, so it is discarded. The subscriptions line is labelled "Google Inc." in the FY2025 record and "Google Subscriptions, Platforms, And Devices" in FY2018-FY2024; the YouTube line alternates between "YouTube Ads" and "YouTube Advertising Revenue." The 10-K's naming is used throughout Section 1. Separately, seg_prod reports revenue lines, not the three reportable segments — Google Services, Google Cloud and Other Bets — and supplies no segment operating income at all.

6. Financial-statement tables are absent from the extracted filing text — stated as a gap, not estimated around. Every "The following table presents…" in the 10-K and both 10-Qs is followed by a caption and no numbers, so the consolidated statements, the segment operating-income table, the other-income breakdown and the tax table are not machine-readable in our copy. Every filing figure in this dive comes from narrative prose, which the filings supply in unusual abundance. Absolute segment operating income is therefore not recoverable; the year-over-year movements are, and are used.

7. Two 52-week ranges disagree within the same payload. quote gives a high of $408.61 and a low of $193.67; tech gives $402.62 and $194.67. On the quote figures the stock is 7.6% below the high and 95.0% above the low; on tech's, 6.2% and 94.0%. Both are reported; nothing in this dive turns on the difference. The same applies to the moving averages (50-day $358.74 in quote, $358.23 in tech; 200-day $326.27 and $327.55).

8. Quarterly revenue does not reconcile to the annual total, three different ways. The four FY2025 quarters in inc_q sum to $402.904B against the annual $402.963B, a $59M gap. earn_cal reports December-quarter revenue of $113.828B where inc_q reports $113.896B and the residual implies $113.955B. Immaterial in size, but it means the quarterly series is not a clean decomposition of the annual one and should not be used for precise sequential arithmetic.

9. Employee count and dividend are stale in profile. The profile carries 190,820 employees; the 10-Q states 198,933 as at 2026-06-30 — the filing wins. lastDividend of $0.85 predates the April 2026 increase to $0.22 per quarter ($0.88 annualised), also per the 10-Q.

10. Diluted earnings per share are not populated anywhere in the payload. epsdiluted is null in every annual and quarterly record; the eps field is the basic figure ($9.23 for the June quarter) while earn_cal.epsActual is the diluted one ($9.11). We use the diluted series and say so. Related: implied shares from marketCap ÷ price is 12.102B, below the June quarter's 12.309B diluted count, so the reported market capitalisation may understate by up to ~1.7%; we report it as supplied.

11. The mandatory convertible preferred is invisible to the payload. Nineteen million shares with a $1,000 liquidation preference ($19.0B), a 6.25% coupon costing $1.19B a year ranking ahead of the common, converting in May 2029, appear nowhere in the vendor data. Non-equity tripwire check: the Series A and Series B depositary shares are separately listed instruments priced near a $50 par-like level and must not be valued as common equity. This dive covers GOOGL common only; the tripwire is noted, not triggered.

Not defects, correctly reported: freeCashFlowYieldTTM (1.17%) and priceToFreeCashFlowRatioTTM (85.8x) look absurd for Alphabet and are right — they are the correct description of the capital-expenditure surge. capexToDepreciationTTM (5.25x) likewise. incomeQualityTTM (0.76) is not a defect but a diagnosis. And, against the standing warning in the data contract, the est EBIT rows are verified accurate for this name (FY2025 within 0.7% of actual operating income, FY2024 within 1.3%) and are used.

9. Technicals — a double, then a stall

• Price $377.65, +1.11% on the day. −7.6% from the 52-week high of $408.61 and +95.0% above the 52-week low of $193.67. Maximum drawdown from peak over the trailing year is only −6.2% — this stock has essentially not corrected.
• Above BOTH moving averages: +5.4% over a 50-day of $358.23 and +15.3% over a 200-day of $327.55, both rising.
• RSI 52.6 — squarely neutral, neither overbought nor washed out. MACD −0.36 — fractionally negative, the one hint of loss of momentum.
• Relative performance, and the split matters: 12-month +99.7% against SPY +24.3% and QQQ +30.8% — a 69-to-75 point outperformance and very nearly a double. 6-month +9.9% against SPY +11.1% — already in line. 3-month −1.5% against SPY +7.6% and QQQ +7.7% — a nine-point deficit. The outperformance is entirely historical; the stock has been flat to down for a quarter while the market rose.
• Practical read on entry. The June-quarter print landed on 2026-07-22 and the stock has gone sideways since; today's +1.11% comes on a day when semiconductors moved violently (several names up 7-13%), so it is sector beta rather than a company event. There is no technical case for urgency: the stock is mid-range, mid-RSI, above its averages, with no catalyst for 85 days. For anyone not already positioned, the honest advice from the chart is the same as from the valuation — buying a name 95% above its low, 8% above our base fair value, into a three-month stretch of flat-to-down price action, is chasing. For anyone already positioned, nothing here argues for selling into strength either.

10. Insiders — routine, tiny, and plan-driven

Eight transactions are supplied, all in Class C stock, and not one is a discretionary open-market purchase or a discretionary sale of size:

The reading, stated carefully. The conversions and in-kind dispositions are restricted-unit vesting with shares withheld for tax. The two sales are 82 and 449 shares — trivial. The 10-Q independently confirms the mechanism: it discloses that Marsida Saraci, appointed Principal Accounting Officer on 2026-06-02, "previously adopted an employee trading plan on May 19, 2025" effective until 2026-08-01 to sell the net shares from vesting of approximately 1,804 gross Class C shares. These are pre-programmed, not decisions.

What the data does not contain is more informative than what it does. There is no insider activity from the chief executive, the chief financial officer, or the founders in this window — and the founders' Class B holdings are the controlling block. The insider file is a non-signal on this name and should be read as one. One incidental observation: the last transactions were struck at $333.20-$333.39, and the stock is 13% higher thirteen days later.

11. Verdict, kill-criteria and flip conditions

Hold.

What we are saying is not that the business is bad. It is that the business is good, the price already knows, and the way the growth is being financed has changed in a manner the multiple has not yet acknowledged.

The bull side, in facts from the filings: revenue +24.2% year over year, the fastest in five years and accelerating from 8.7% in FY2023; cloud +82% at a ~$99B run-rate; 54% incremental operating margins in both reporting segments; gross margin up 210 basis points and operating margin up 160 basis points year over year; a new hardware revenue line begun in the June quarter; roughly $140.1B of net cash and $242.5B of liquid resources; and a search franchise that independent voices in the knowledge base put at ~90% share.

The bear side, from the same filings: trailing earnings are 87% non-cash marks and the honest trailing multiple is ≥34x, not the 18.97x every screen shows; free cash flow went negative in the June quarter with capital expenditure at 37.5% of revenue and 5.25x depreciation; buybacks are zero with $69.5B of authorisation unused; $49.6B of equity issued and a $40.0B at-the-market programme live; $51.8B of notes issued in six months taking long-term debt to $98.2B; $51.4B of third-party credit backstops plus $24.1B agreed; $811.0B of purchase commitments; $17.4B of accrued legal fines; and a structural-remedy judgment awaited that the company itself says "could have a material adverse effect" and refuses to size.

Base fair value $390, 3.3% above spot, 8% below the street's $423.44, with a bear at $270 that is larger than the bull at $475. On a stock that has risen 99.7% in twelve months, that is a Hold.

Pre-registered KILL criteria — what would move this to Avoid:

• A structural remedy in the US advertising-technology case. Judgment is awaited after November 2025 closing arguments. A forced divestiture of the publisher-side stack would hit the highest-margin part of the advertising engine and re-rate the whole company.
• A material downward revision of the private-stake marks. The $99.0B that produced the June quarter's 94% net margin is unrealised and symmetric. A 30% write-down is roughly $30B pre-tax and could put a quarter into loss with the operating business unchanged.
• A third consecutive quarter of negative free cash flow accompanied by a further capital-expenditure increase and no stated peak. The filings currently say only "significantly increase, relative to 2025" ($91.4B) with no number.
• Actual sales under the $40.0B at-the-market programme while repurchases remain at zero — the confirmation that the shareholder yield is not merely absent but negative.
• Cloud incremental operating margin falling below 35%, which would mean the scaling economics that justify the capital intensity are not there.
• Credit-derivative backstop notionals continuing to compound at the current rate ($16.9B → $28.4B → $43.8B across three quarters). At that pace the off-balance-sheet exposure to third-party builders passes $70B by year end, and Alphabet becomes a credit provider to the buildout as well as a buyer within it.

Pre-registered FLIP TO BUY (upsize):

• A price below ~$310 — roughly 20.7x FY2027 consensus, at which the base case offers +26% and the asymmetry inverts. That is a 18% drawdown from here and entirely achievable in a capex-led sell-off.
• A capital-expenditure figure that flattens year over year, or free cash flow returning decisively positive — either restores the free-cash-flow yield and with it the multiple's support.
• Resumption of repurchases at scale against the $69.5B authorisation, which would signal management believes the equity is cheap at a moment it has just been selling it.
• A behavioural rather than structural outcome in the advertising-technology remedy, removing the largest unquantifiable on the name.
• Two consecutive quarters of cloud revenue growth above 60% with incremental operating margins above 50%, which would make the FY2028-FY2030 consensus ladder look conservative rather than heroic.

Where GOOGL fits in the Synthos Framework Portfolio. The mega-cap platform / AI infrastructure sleeve, as a 2-3% position if already held, with no addition at this price. The justification for holding rather than exiting: this is one of very few $4T-plus businesses whose revenue growth is accelerating rather than decaying, and exiting a compounding franchise on a valuation call has a poor historical record. The justification for not adding: the base case offers 3.3%, the bear case is larger than the bull, the free-cash-flow yield is 1.17%, and the single largest risk cannot be sized by anyone including the company. New money waits for either a lower price (~$310) or a resolved remedy.

Single biggest risk: the US advertising-technology structural remedy. In April 2025 the Eastern District of Virginia found that the publisher-side tools "unfairly excluded rivals." A remedies proceeding ran in September 2025, closing arguments were held in November 2025, and judgment has not arrived. The 10-Q's language is unusually direct for a filing: "The DOJ's remedy proposal includes structural remedies that could have a material adverse effect on our business," and "Given the nature of these matters, we cannot estimate a possible loss." The market is carrying this at zero. A separate state case in Texas follows, and private damages actions across Europe follow determined liability as a matter of course — the $2.1B Stockholm award booked in the June quarter is the first instalment of that pattern, not the last. This is why the risk score is 6 and not 4.

Most fragile assumption in the price: that FY2027 earnings of $15.01 arrive with the depreciation from roughly $200B of 2025-2026 capital expenditure fully loaded, without breaking the ~32.5% operating margin that consensus assumes. First-half depreciation grew 43% while first-half capital expenditure grew 104%; capital expenditure is running at 5.25x depreciation. The entire gap between those growth rates is a future charge against the income statement that has not yet appeared, and every valuation anchor in this dive — ours and the street's — assumes it lands gently.

Provenance & disclosures

• Traceability: 613 tagged knowledge-base claims name Alphabet, Google or GOOGL across 51,928 records, spanning 2011-06-23 to 2026-08-01 from 58 sources. After the four-lane attribution policy: 11 quarantine_misattributed claims EXCLUDED; 132 claims from Alphabet's own executives (sundar_pichai_mgmt 84, demis_hassabis_mgmt 41, GOOGL_mgmt 7 — 97% bullish) held at half weight and explicitly not treated as independent conviction; 35 peer-company management claims at half weight; 8 promoter-class claims at low weight. Independent claims: 427 across 46 sources (breadth 46), reading 240 bullish / 102 bearish / 85 neutral — and 17 bearish against 16 bullish since 2026-06-01. Net conviction mixed. Claims naming specific model families or research organisations were read for substance but not reproduced, under house style; all other quotes are verbatim from the stored claim text. Searches covered the exact entity aliases GOOGL, Google and Alphabet (case-insensitive); subsidiary-level aliases (Waymo 41, YouTube 30, Google Cloud 7) exist separately in the base and were not merged into the count.
• Data as-of: fundamentals — income statement 2026-06-30 (Q2'26, reported 2026-07-22, 10-Q filed 2026-07-23, thirteen days before this dive); balance sheet and cash flow figures for 2026-06-30 taken from the 10-Q narrative because the vendor balance sheet ends 2025-12-31; annual statements FY2025 (10-K filed 2026-02-05) · estimates 2026-08-04 · prices 2026-08-04 ($377.65 close, +1.11%; 50-day $358.23; 200-day $327.55; RSI 52.6) · knowledge-base claims 2026-08-04. Every figure comes from the Synthos vendor payload or from the SEC filings on disk; no figure comes from memory, recall or external retrieval.
• Filings used: 10-K filed 2026-02-05 (FY2025); 10-Q filed 2026-04-30 (Q1'26); 10-Q filed 2026-07-23 (Q2'26); 8-K 2026-06-05 (Items 1.01/3.03/5.03 — the mandatory convertible preferred offering, capped calls and certificates of designations); 8-K 2026-06-11 (Item 5.07 — 2026 annual meeting: the 2021 Stock Plan reserve increased by 200,000,000 Class C shares, auditor ratified, and ten shareholder proposals defeated, including on climate disclosure, water usage, equal shareholder voting, AI board oversight and AI data-usage oversight); 8-K 2026-07-22 (Item 2.02 — the June-quarter results release furnished as Exhibit 99.1).
• Data rejections (detailed in Section 8): trailing price-to-earnings (18.75x / 18.97x) rejected as a lens in favour of core earnings of ≤$11.12 and ≥34.0x, together with trailing return on equity (50.84%), return on assets (26.49%), EBIT margin (67.39%), pre-tax margin (67.11%), net margin (54.76%), earnings yield (5.32%), interest coverage (296x), the Graham number and both PEG ratios, all of which inherit the same defect; est FY2026 EPS of $20.267 rejected as a valuation denominator (a hybrid of first-half GAAP actuals containing ~$8.7 a share of unrealised marks and second-half core estimates), along with the FY2026 netIncomeAvg/ebitAvg pair; bal_a superseded by the 10-Q as seven months stale across the largest financing events in the company's history; km_ttm.enterpriseValueTTM corrected from $4,627.207B to approximately $4,430.3B (a $56.845B implied net-debt position against roughly $140.1B of actual net cash), with all EV-derived multiples recomputed or rejected; the seg_prod "Other Segments" row of −$127M discarded as breaking the reconciliation, and its unstable labels corrected against the 10-K; financial-statement tables absent from the extracted filing text and stated as a gap, with segment operating income available only as year-over-year movements; two conflicting 52-week ranges reported side by side; quarterly revenue not reconciling to the annual total by $59M with three different values for the December quarter; profile employee count (190,820) and lastDividend ($0.85) superseded by the 10-Q (198,933 employees, $0.22 quarterly); epsdiluted null throughout, with earn_cal.epsActual used as the diluted series; and the $19.0B of mandatory convertible preferred stock, its $1.19B annual dividend and the $40.0B at-the-market programme entirely absent from the payload.
• Basis note — the most important one in this dive: Alphabet's reported earnings are dominated by unrealised marks on non-marketable equity securities. The June-quarter 10-Q states that other income and expense of $98.0B "included net gains on equity securities of $99.0 billion, primarily related to unrealized gains in our equity securities portfolio from SpaceX and a private company"; the March 10-Q states $36.9B of the same within $37.7B; the 10-K states $24.1B within $29.8B for FY2025. Core earnings, derived as operating income plus other income and expense less net equity-security gains, taxed at 17%, are $2.68 for the June quarter ($2.82 before a disclosed $2.1B legal charge, against a $2.87 consensus), $2.75 for March (against $2.64) and $9.14 for FY2025 (against $10.81 reported). This dive values Alphabet on core and forward consensus earnings; reported trailing earnings are shown for completeness and used for nothing.
• Cash-flow note: free cash flow was approximately −$5.8B in the June quarter (operating cash flow $39.1B against capital expenditure $44.9B), the first negative quarter on record in this data. First-half capital expenditure was $80.6B against $39.6B a year earlier. Capital-expenditure guidance is directional only — "we expect to significantly increase, relative to 2025" ($91.4B) — with no figure in any filing; the $195-205B and ~$400B figures quoted in Section 5 are third-party knowledge-base estimates, labelled as such. Share repurchases were $0.000 in both quarters of 2026 with $69.5B of authorisation unused, against $45.709B in FY2025.
• Estimate coverage: deep — 39 analysts on FY2026 EPS, 42 on FY2027, 34 on FY2028, 30 on FY2029, 16 on FY2030; 35-39 on the revenue lines. The FY2027 row is the primary forward anchor. Contrary to the standing warning that est EBIT/EBITDA rows are unreliable, this file's EBIT row was tested against actuals and accepted (FY2025 within 0.7%, FY2024 within 1.3%), and is used to build the operating-only earnings ladder in Section 6. Note that the implied share counts inside the estimate rows are internally inconsistent (12.511B for FY2026, 12.225B for FY2027) and reflect neither the June equity issuance nor the potential preferred conversion.
• Peer note: the vendor peer set (Apple, Nvidia, Microsoft, Broadcom, Taiwan Semiconductor, Meta, Cisco, Flex, Garmin, Magic Software) supplies prices and market capitalisations only — no multiples — so no peer-multiple comparison is drawn. Several of the smaller entries are not comparables in any meaningful sense.
• Fair-value caveat: the $270 / $390 / $475 anchors are 18.0x / 26.0x / 31.7x FY2027 consensus earnings of $15.006, cross-checked against FY2028 consensus of $17.716 (15.2x / 22.0x / 26.8x) and against the operating-only ladder derived from the verified consensus EBIT row. Stated arithmetic, not a discounted cash flow. The base case explicitly assumes ~13% multiple compression over the next twelve months on the loss of the buyback bid and the 1.17% free-cash-flow yield; if the multiple merely holds, the price is ~$446. The bull case of $475 is exactly the street's published high target, which is stated deliberately as an independent check.
• Transcript gap: earnings-call transcripts are not available on this research plan. Management commentary and analyst question-and-answer from the 2026-07-22 call are therefore absent from every source used here, and the 8-K of that date furnishes the earnings release, not the call. Any capital-expenditure figure guided verbally on that call is not in our sources and no such figure has been invented. Management framing that does appear — the 132 knowledge-base claims from Alphabet's own executives — is held at half weight throughout and labelled at every point of use.
• Non-equity tripwire: checked and noted, not triggered. Alphabet's Series A and Series B depositary shares (each a 1/20th interest in a share of 6.25% mandatory convertible preferred, $50 per depositary share) are separately listed instruments that inherit parent fundamentals in feeds of this type and must not be valued as common equity. This dive covers GOOGL common stock only.
• Timing: the June quarter was reported 2026-07-22 and the 10-Q filed 2026-07-23, thirteen and twelve days before this dive, so the most recent quarter is fully reflected. The next print is 2026-10-28, 85 days away — well outside the five-day earnings-proximity rule, so no banner applies.
• Not investment advice. Independent research, educational and informational only, never personalised. No recommendation to buy, sell or hold any security is made to any person.
• Version: 2026-08-04-full.