Coverage · AI infrastructure · Alphabet Q3 2026 · Hold, probability-weighted value about $280
The Backlog Cleared. The Cash Did Not.
Alphabet’s third quarter passed two of the three tests we set in August. Google Cloud revenue grew 103% to $30.7 billion, and the Cloud backlog reached $560.8 billion, above our $560 billion line by eight hundred million.[1][2] The third test was the one the thesis rested on, and it failed: free cash flow was negative $9.1 billion for a second straight quarter, buybacks were zero for a third, and in nine months the company raised $138 billion from debt and equity markets while returning $8 billion.[1][2]
Management held 2026 capital spending at $195–205 billion and said 2027 would “increase significantly,” with the figure to come in February.[3] Our August framework treated that path as the bear case, weighted 25%, with a floor at $290. The shares went through the floor in one session. The funding turn is now our base case. We take the Street’s 2027 earnings whole and pay 18–20 times for them while the spending is financed externally, against 26–28 times in August.
Our view is Hold with a probability-weighted value of about $280, against the desk’s August figure of about $405. At $278.99, November 2 close, the value sits 0.4% above the price, inside our hold band. The whole cut is the multiple and the weights. What would change it is in section 5, and what the resting order in our own account did is in section 6.
One-page summary
| Item | Status |
|---|---|
| Rating | Hold. Our first published rating on the name; the August view it scores was built and never shipped (section 3) |
| Probability-weighted value | About $280, 12 to 18 months. Bull 20% $341–384, base 50% $273–303, bear 30% $202–224. The desk’s August figure was about $405 on a 30/45/25 tree |
| Price | $278.99, November 2 close, -14.1% from the $324.89 pre-report close[4]. Market value about $3.43 trillion on 12,294 million shares[2] |
| The quarter | Revenue $127.6B, +25%; operating margin 31.8%; Search & other +12.0%; Cloud $30.7B, +103%, operating margin 36.2%; EPS $2.12, or $2.67 before an $8.7B mark-down of equity holdings[1] |
| Proof tests | Cloud backlog $560.8B against our $560B line: passed. Cloud growth 103% against 75%: passed. Free cash flow positive with a buyback path: failed[2][8] |
| Cash | Operating cash flow $42.3B; capital expenditure $51.4B, 121% of it; free cash flow −$9.1B (Q2 −$5.9B). Nine-month capex $132.0B against $127.1B of operating cash flow[1] |
| Funding | Nine months: $76.3B of net new debt, $42.9B of common stock, $19.1B of preferred; buybacks $0 against $39.8B a year earlier. Long-term debt $125.1B from $46.5B at year-end[1][2] |
| Commitments | Purchase commitments $924.4B ($213.4B due within a year); leases signed and not yet started $130.5B; inventory $13.3B from $2.4B[2] |
| Valuation at the price | 18.4× FY2027 consensus EPS of $15.17; 24.8× this year’s earnings with the investment marks stripped out[5]. Microsoft trades at 22.1×, Meta at 17.9× |
| Next read | The fourth-quarter report, about February 3 (date not announced): the 2027 capex figure, fourth-quarter cash flow, any buyback, Search growth |
1. Two tests passed. The cash test failed.
The operating quarter was strong by any ordinary measure. Revenue grew 25% to $127.6 billion, the thirteenth straight quarter of double-digit growth, and operating income grew 30% to $40.6 billion.[1] Google Cloud more than doubled: $30.7 billion of revenue, $11.1 billion of operating income, a 36.2% margin against 23.7% a year ago, with TPU system sales “a meaningful contributor” and the finance chief guiding to “a meaningful revenue ramp in 2027.”[1][3] The Cloud backlog, which we had set as the first of three proof tests before the report, came in at $560.8 billion in the 10-Q, against the $560 billion we required and $513.9 billion in June.[2][6][8] It cleared by less than one percent, and it cleared.
Search grew 12.0% to $63.4 billion, down from 17% in the second quarter. The chief business officer attributed most of the slowdown to currency, retail and media, and acknowledged that “the strong growth in AI Mode queries had an impact because of the lower monetization” on those queries, while calling most of them incremental.[3] Reported earnings of $2.12 a share were cut by $0.55 from an $8.7 billion mark-down of equity holdings, chiefly the SpaceX stake; before the mark, the figure was $2.67.[1] We underwrite the operating lines and treat the marks as noise in both directions, as we said in August when they ran the other way.
The third test was free cash flow turning positive with a path back to buybacks. Operating cash flow was $42.3 billion and capital expenditure was $51.4 billion, so free cash flow was negative $9.1 billion, after negative $5.9 billion in the second quarter.[1] Repurchases were zero for the third quarter running; $69.5 billion of the April 2025 authorization sits unused.[2] The finance chief’s explanation was the TPU business: “we build the inventory ahead of getting the revenue stream in.”[3] Inventory did rise, to $13.3 billion from $2.4 billion at year-end.[2] But the inventory build is $11 billion over nine months; the capital expenditure is $132 billion. The cash test failed on spending, and the spending is going up.
Chart 1 — Capex overtook operating cash flow
Two quarters ago spending passed what the business generates. The guide implies a third.
Source: Alphabet 8-K exhibit 99.1 of October 28, 2026, statements of cash flows and the free-cash-flow reconciliation.[1] Free cash flow is operating cash flow less purchases of property and equipment, the company’s definition.
2. The funding turn, in the company’s own numbers
The August underwrite called the financing file the spine of the thesis: a company that returned about $13 billion a quarter had begun raising capital. The third quarter extended it. In the nine months to September, Alphabet issued $85.2 billion of debt net of costs and repaid $8.9 billion, sold $42.9 billion of common stock ($12.4 billion of it through the at-the-market program in the third quarter, with $27.6 billion of that program still available) and $19.1 billion of mandatory convertible preferred.[1][2] It bought back no stock. A year earlier the same nine months saw $39.8 billion of buybacks and $11.9 billion of net new debt.[1] Long-term debt stands at $125.1 billion, from $46.5 billion at year-end; the diluted share count is up 1.4% in a year, before the preferred converts in 2029.[1][2]
Chart 2 — From returning cash to raising it
Nine months ago the company returned $47 billion. This year it raised $138 billion.
Source: Alphabet 8-K exhibit 99.1, statements of cash flows for the nine months ended September 30, 2025 and 2026.[1] Net debt is issuance net of costs less repayments.
What the money is committed to is also on the record. Material purchase commitments reached $924.4 billion at September 30, from $811.0 billion in June; $213.4 billion is due within twelve months, and $802.1 billion is multi-year infrastructure, inventory, energy and content, mostly through 2031. Leases signed and not yet started add $130.5 billion. Property and equipment, net, is $372.8 billion against $246.6 billion at year-end, and depreciation grew 49% in the quarter.[1][2] The finance chief said the investment “will continue to weigh on the P&L in the form of higher depreciation expenses” and that third-party capacity coming online in the fourth quarter brings “margin headwinds in the near term.”[3]
The fourth quarter is already arithmetic. Nine-month capital expenditure was $132.0 billion; an unchanged $195–205 billion guide implies $63–73 billion in the fourth quarter, against operating cash flow that has run between $39 billion and $52 billion a quarter for five quarters.[1] Free cash flow will be negative again in the fourth quarter on the company’s own plan, and 2027 is guided higher from there. We stop treating that as a scenario. It is the path.
3. The August view, scored
First the record. On August 4 we underwrote Alphabet at Hold with a probability-weighted value of about $405: bull 30% at $450–480, base 45% at $390–420, bear 25% at $290–320, with buy gates at $335–340 and $318–325 and three proof tests for this report.[8] That piece was written for publication as the second note of the deployment series and never shipped. What readers saw was the series opener on July 31 with a wait-for-price alert at $318, two weekly notes in August that said the alert stood through the October report, and the August 14 essay recording that it never triggered.[7] Our coverage table has never carried the name. We score the August view here because we acted on it, and because an unpublished number scored honestly is worth more than a published one scored kindly. This is our first public rating on Alphabet.
Chart 3 — The August framework drawn on the price
Both gates traded. The bear range held for eleven weeks and gave way in one session.
Source: Yahoo Finance daily closes, August 4 to November 2, 2026; the November 2 figure is the close.[4] Ranges, weights, gates and the $265 thesis-break level as the desk set them on August 4 and August 18.[8]
From the August 4 anchor of $377.65 the shares are down 26.1% against the S&P 500 down 2.5%.[4] What the August view got right: it named the financing file as the thing the market grades, it refused to chase at 28 times earnings, it set the Microsoft-parity rule for the multiple, and two of its three proof tests read correctly. What it got wrong is the part that mattered. It priced the funding turn as a one-in-four bear case with a $290 floor, and its base case assumed funding would “normalize slowly.” The company instead spent 121% of its operating cash flow in a quarter and guided 2027 higher. The bear case was not a tail. It was the path, and the floor lasted one session.
4. The tree and the price
We take the Street’s 2027 earnings whole. Consensus is $15.17 a share on revenue of $628 billion, up 25%; the estimate has moved from $14.76 ninety days ago to $15.17 and has not yet absorbed this report.[5] On our arithmetic this year’s earnings are about $11 a share once the first-half investment gains and the third-quarter loss are stripped out, so the consensus asks for a 35% step in 2027. The bull and base cases accept it; the bear case cuts it to $13.50–14.00 as depreciation and third-party capacity land.
The multiple is the change. In August the base leg paid 26–28 times forward earnings. The rule we set then was that Alphabet should trade at or below Microsoft’s multiple until it funds itself again; Microsoft is at 22.1× its next fiscal year, which would be $335 for Alphabet, and Meta, the other buyer the market fines for spending, is at 17.9×.[5] Our base leg pays 18–20×, a 10–18% discount to Microsoft while the funding is external; the bull leg pays parity, 22–24×, if February brings a dated path back to positive free cash flow or a buyback; the bear leg pays 15–16×, where the shares bottomed in 2022 and in April 2025.
Chart 4 — The tree, before and after
The October bull case is the August base case. The August bear case is the October base case.
Ranges are twelve to eighteen months out. August weights 30/45/25 and ranges as set on August 4; October weights 20/50/30 on FY2027 consensus EPS at the multiples in the text. The price line is November 2.
| Case | Range | What has to happen |
|---|---|---|
| Bull, 20% $15.50–16.00 at 22–24× | $341–384 | The February call gives a 2027 figure with a dated path back to positive free cash flow in the second half, or a buyback; Cloud backlog keeps compounding and TPU revenue ramps as guided; Search holds 10% or better |
| Base, 50% $15.17 (consensus) at 18–20× | $273–303 | 2027 capex of $260–300 billion, free cash flow negative through 2027, financed with investment-grade debt and the at-the-market program, buybacks at zero; Cloud grows 60–70% with the guided margin headwind; Search 10–12% |
| Bear, 30% $13.50–14.00 at 15–16× | $202–224 | One of: a 2027 guide at or above $300 billion paired with another common-stock raise; Search under 10% as AI Mode monetizes below classic search; margins compressed by depreciation and third-party capacity so the 2027 estimate is cut |
| Weighted, 100% | about $280 | 0.20 × $362.5 + 0.50 × $288 + 0.30 × $213 = $280.4 |
Two cross-checks. A discounted cash flow on the same path, with stock compensation treated as a cost, revenue growth fading from 25% to 4% by 2035, capital spending of $280 billion in each of 2027 and 2028 and then fading toward a tenth of revenue, and a 10.8% discount rate, gives a base value of $203 a share, $263 with stock compensation added back; the price equals the base path at a 9.2% discount rate. A reverse check on earnings says the price needs 8.7% annual earnings growth from the 2027 estimate at a 20× exit in 2031, or 11.6% at 18×. The two disagree the way they should on a business spending more than it generates: the earnings say the price is undemanding, the cash says it is at the top of what it supports. The tree sits between them.
Chart 5 — The multiple against the peers
Alphabet trades at a 17% discount to Microsoft. The August rule said at or below parity until it funds itself.
Source: Yahoo Finance forward P/E on next-fiscal-year consensus EPS and October 29 peer prices; Alphabet at the November 2 close.[5] Alphabet at FY2027 consensus; Microsoft at its June 2028 fiscal year; Apple at September 2027.
Relative to the Street we are a long way below. The pre-report mean target was $430 across 54 analysts; the first post-report moves we have seen are to $300 from $350 at DA Davidson, which holds a Hold, to $400 from $425 at RBC and to $405 from $425 at Barclays.[9] We are consensus on the business and contrarian on the price.
5. What would change the rating
Our hold band is a twelve-to-eighteen-month probability-weighted value within roughly 10% below and 5% above the market price. At about $280, the view is Hold while the shares sit between about $267 and $311; a publication-day close above $311 would make the rating Reduce by the band, with no re-litigation, and we would say so. The conditions below were fixed and time-stamped before this note was written; any one of the first four breaks the thesis and triggers a re-underwrite.[8]
| Condition | The observable number | What it means |
|---|---|---|
| Price | A daily close below $200, under the entire bear range | Thesis broken; the view is withdrawn or lowered and re-underwritten |
| 2027 funding | A 2027 capital-spending guide at or above $320 billion at the February call, or any 2027 guide paired with a new common-stock raise beyond the $27.6 billion left on the at-the-market program | The bear case governs; an update within one trading day |
| Search | Search & other revenue growth below 8% in any quarter (this quarter: 12.0%) | The franchise leg is broken; re-underwrite |
| Estimates | The FY2027 consensus cut below $13.50, an 11% cut from $15.17, within thirty days of a report | The tree is re-run on the new number |
| Cash return (upside) | Free cash flow positive in any quarter, or a dated buyback resumption from the $69.5 billion authorization | Not a break: the bull weight rises and the base multiple moves toward Microsoft parity, a value near $310–335 |
| Margins | Cloud operating margin below 30% in a quarter (this quarter: 36.2%) | Review: the base earnings move toward the bear case |
The next test is the fourth-quarter report, when management has said it will give the 2027 figure. Alphabet reported the December quarter on February 4 this year; the company has not announced a date, and we will take it from the company.
6. The order in our own account
An order to buy a small Alphabet position at $319.78, entered August 18 at the lower of the two August gates, had rested since then. It filled on Monday, October 26, when the shares traded as low as $316.60, three sessions before the report.[4] We sold the position before this note was published. We hold no shares in Alphabet at publication, and the disclosure below says so.
The lesson is the one in section 3: the order was placed at a price level, and the price level was reached because the thing we were worried about got worse. A resting order in a name whose thesis is about funding should have carried a funding condition. The thesis-break table above now does.
Housekeeping
Scope. This note reflects the results release of October 28 and the Form 10-Q filed October 29, and the conference call of October 28. Company financials are taken from the filings; call quotations are from two published transcripts, a secondary source, checked against each other and attributed to the named speaker.[3] Consensus and comparables are from Yahoo Finance on October 29 and are labelled.
A correction to our own record. Our internal scorecard, our ledger and our trading-window register have carried the August view as “published.” It was not. The desk built it for publication and the series shipped a different piece. This note is the first rated Alphabet publication on this site, and the internal records have been corrected to say so.
Trading window. Our policy restricts trading in a covered name from ten days before a publication to five days after it. A note published November 3 carries a window of October 24 through November 8. The August order predates every window and was logged under the resting-order exception. The sale of the position fell inside the October 24 to November 8 window. It was a trade inside the window, logged as such in our trade log; it was not a protective order at a pre-approved level. No new order will be placed in the name before November 9. Protective exits at previously approved levels remain available at any time under our investment policy and are logged if used.
Sources
- Alphabet Inc. Form 8-K of October 28, 2026, Exhibit 99.1: third-quarter 2026 results, income statement, segment results, balance sheet, statements of cash flows, the free-cash-flow reconciliation, the at-the-market and notes disclosures. SEC EDGAR
- Alphabet Inc. Form 10-Q for the quarter ended September 30, 2026, filed October 29, 2026: revenue backlog ($567.7 billion, of which $560.8 billion Google Cloud), purchase commitments, leases not yet commenced, share repurchases, debt, legal matters, shares outstanding. SEC EDGAR
- Alphabet third-quarter 2026 earnings call, October 28, 2026, 1:30pm Pacific. Quotations from the published transcripts at Investing.com and TradingKey, secondary sources, checked against each other and attributed to Anat Ashkenazi (CFO), Philipp Schindler (Chief Business Officer) and Thomas Kurian (Google Cloud). Investing.com · TradingKey
- Yahoo Finance daily prices for GOOGL and SPY, August 4 to November 2, 2026; the November 2 figure as marked in the text. Price history
- Consensus estimates (FY2027 EPS $15.17, 52 analysts; revenue $627.8 billion), price targets (mean $429.68, 54 analysts, pre-report) and forward price-to-earnings multiples for Alphabet, Microsoft, Amazon, Meta, Apple and Nvidia from Yahoo Finance, October 29, 2026, a secondary source. Analysis page
- Alphabet Form 10-Q for the quarter ended June 30, 2026: revenue backlog $519.5 billion, of which $513.9 billion Google Cloud; purchase commitments $811.0 billion. SEC EDGAR
- Our published record on the name: the deployment-series opener (July 31, the $318 alert); the August 10 weekly and the August 17 weekly (the alert stands through the October report); the August 14 essay (the alert that never triggered).
- Our pre-registrations: the August 4 framework and its three proof tests, frozen in the desk’s ledger on October 20 before the report; the backlog print gate registered October 22 at 7:42pm Eastern (above $560 billion, 55%); the thesis-break conditions in section 5, registered October 29 before this note was written. Internal records with no public URL; the thresholds are quoted exactly as written.
- Post-report target changes: DA Davidson to $300 from $350 (Hold) via TipRanks, October 29; RBC to $400 from $425 and Barclays to $405 from $425 via a web search of the same day; secondary sources. TipRanks
Disclosures. The author holds no position in Alphabet at publication. A resting order entered August 18, 2026 filled on October 26, 2026 and the position was sold before this note was published, confirmed against brokerage records on the publication date; no order rests in the name. The author also holds other AI-infrastructure securities not discussed here.
Not investment advice. Brandon Leon is a publisher, not a registered investment adviser, and nothing here is a recommendation to buy or sell any security or a solicitation to do so. This is independent research prepared for general readership without regard to any individual's circumstances, objectives or risk tolerance. Forecasts are estimates and will be wrong in detail. Past performance of any call described here does not indicate future results. Do your own work or consult a licensed adviser before acting.
Methodology. The proof tests in section 1 and the thesis-break conditions in section 5 were set and time-stamped before the report and before this note respectively. Company figures come from Alphabet's own 8-K exhibit and 10-Q; call quotations are from named secondary transcripts; consensus figures are from a named secondary source and are labelled. Prices are official closing prices except where marked intraday. Revenue backlog and purchase commitments are as disclosed by the company under its own policies; this note does not characterise either as separately attested.