Built by Grok 4.6 High

Alphabet reports third-quarter 2026 results on Wednesday, October 28, 2026. Prepared October 26, 2026. The call is company-confirmed for 4:30 p.m. Eastern. The quarter ending September 30 has not been reported. Last reported quarter is Q2 2026, released after the close on July 22, 2026. The fiscal year is the calendar year. This page is a preview, not a review of a finished print.

GOOGL | Earnings Preview

Q3 2026 has not printed. The baseline is Q2 2026 versus Q2 2025, not versus Q1. Revenue was $119,796 million, +24.2% versus $96,428 million. That step was about +240 bps from Q1's +21.8% ($109,896 million versus $90,234 million). The four-quarter path is +15.9%, +18.0%, +21.8%, +24.2%. Consolidated growth is still accelerating. The engine inside it is not.

Google Cloud was $24,768 million, +81.8% versus $13,624 million, after +63.4% in Q1. That is about +1,840 bps of further acceleration. Search and other slowed from +19.1% to +16.8% ($63,271 million versus $54,190 million), about −230 bps. Services growth cooled from +16.0% to +14.5% ($94,540 million versus $82,543 million; Q1 was $89,637 million versus $77,264 million). YouTube ads were the ads line still speeding up, from +10.7% to +12.9% ($11,055 million versus $9,796 million; Q1 was $9,883 million versus $8,927 million). Subscriptions decelerated about 410 bps, from +19.3% to +15.2% ($12,911 million versus $11,203 million; Q1 was $12,384 million versus $10,379 million). Operating income was $40,770 million, +30.4%, and the margin expanded about +160 bps to 34.0%, after +220 bps in Q1. Profit is still growing faster than revenue. The gap is narrowing, and the expansion is getting smaller. GAAP diluted EPS of $9.11 is not the trajectory. The equity-securities gain was $99,031 million.

What the street is underwriting. Alphabet does not guide quarterly revenue or EPS. FMP, pulled October 26, has Q3 revenue at $127.908 billion, +25.0% versus $102,346 million, about +80 bps faster than the last print. FMP EPS is $3.03, +5.5% versus $2.87. Do not read that EPS rate as an operating slowdown. The operating bar is FMP EBIT of $45.002 billion, about +44% versus reported operating income of $31,228 million, and a 35.2% margin versus the reported 30.5%. That 30.5% included the EC fine. Management said the ex-fine margin was 33.9%. Versus that cleaner base, 35.2% is only about +130 bps, near the +160 bps just printed. The dollar is still tight: a repeat of the recent +30% operating-income pace is about $40.6 billion, under the $45 billion average. The same file's Q2 EBIT average sat about $2.6 billion above the operating income that was reported.

Cloud is what has to carry the +25%. Bank of America's October previews describe Street Cloud growth at about +89% and its own model at +93%, with Search at about +14% to +15%. That is one broker's description of consensus, not Visible Alpha and not Bloomberg. On the Q3 2025 Cloud base of $15,157 million, 89% is about $28.6 billion, another roughly 720 bps of acceleration on top of +81.8%. Search at 15% is about $65 billion, another roughly 180 bps of deceleration, which is what management already previewed. FMP does not publish segment estimates.

Search is still the franchise with the leading share. Cloud is the acceleration, and on the last cited industry split it is still third (AWS 28%, Azure 20%, Google Cloud 15%, Synergy via CIO Dive, October 22). Owning the ads leader is not the same trade as underwriting a third-place cloud business at a high-80s growth rate the street has already penciled in.

Watch items

• Guidance is capex, not a revenue bogey. The only range still in force is FY2026 capex of $195 billion to $205 billion, raised twice, midpoint $200 billion by arithmetic. Nothing since July 22 has moved it. First-half purchases were $35,674 million plus $44,924 million, $80.6 billion. The floor requires a second half of about $114 billion, about +121% versus second-half 2025 purchases of $23,953 million plus $27,851 million. A straight line of the Q2 quarter annualizes to about $180 billion, under the $195 billion floor. The range is aggressive versus that exit rate. It is not a sandbag. 2027 capex is "increase significantly," with no number. BofA is already near $300 billion. That is one broker, not a guide.
• The Q3 brakes are FX, the Search lap, and rented capacity. July's outlook flipped FX from a 1-point tailwind to a slight headwind, concentrated in Search and YouTube, and named the lap of the Search acceleration that started in Q3 2025. Third-party capacity is a bridge and "will create modest margin pressure," on top of a Wiz headwind management had already put at a low-single-digit Cloud-margin point for the rest of 2026. Depreciation of $7,104 million was still +42.1%. Tariffs are not the swing factor in the July outlook or in the news since.
• The revenue habit and the EPS habit are different. Revenue beat in 11 of 12 quarters, and the last four surprises averaged about +2.5%. Repeating that average would put the print about 2.5% above the $127.9 billion bar. That is arithmetic on the trailing surprise, not a forecast and not an internal model. The GAAP EPS streak is 12 for 12, and the widening beat is the equity-securities mark. $3.03 looks through that mark. Do not carry +217% forward.
• Meta is the only ad peer already out. Meta advertising revenue was $63,061 million, +25.9% versus $50,082 million, about −100 bps versus its own Q2 rate. Ad demand did not break. It does not say Google's Search deceleration has reversed. Microsoft reports the same evening. Amazon, Pinterest, and Reddit report the next day.

The management line the street has not fully priced is not another adjective on demand. It is that some near-term Cloud revenue is being left on the table so frontier training keeps the compute, while the capex ceiling has been wrong twice this year and 2027 is still unnumbered. Demand commentary got more bullish after July. The range did not.

Alphabet guides full-year capex. It does not guide the quarter. The standing range was set on the July 22 call and has not been revised. Daloopa's 2026Q3 document index was empty. Capex is Aggressive versus the Q2 exit rate, and it is not a sandbag. There is no revenue range to classify. FMP's +25% is mildly ahead of a cautious FX and Search-lap tone.

The capex path only moves up. Q4 2025 introduced 2026 at $175 billion–$185 billion. Q1 raised it to $180 billion–$190 billion for Intersect. Q2 raised it again, for faster delivery, not a restatement of costs. Q2 annualized is about $180 billion, under the new low. Hitting the $200 billion midpoint needs a second half of about $119 billion, about +130% versus second-half 2025 capex of $23,953 million plus $27,851 million. In 2025 every pre-final capex number was too low, and the year finished at $91,447 million, on the low end of the last range of $91 billion–$93 billion. The October call is when that last raise showed up. A hold of $195–205 billion would be new. Another walk-up would match last year.

July 2025's ad-comp caution did not stop Q3 2025 from accelerating. Do not treat the new Search-lap sentence as a forecast of a break. It is already underway: Search gave back about 230 bps last quarter, and Q3 2025 was the quarter the acceleration started, from +11.7% to +14.5%.

FMP, not a company guide:

FY2026 revenue growth is versus the sum of the four FY2025 quarters, $402.836 billion. FY2026 EPS of $20.60 embeds the equity-securities gains already reported. The FY2027 EPS step-down is that gain rolling off, not a sourced margin cut. Annual FMP EBIT of $305 billion is not the sum of the quarterly EBIT estimates and is not reported operating income. It is not in the model.

Guide columns are what management said. "Not guided" means the July 22 call did not give a number. Internal is n/a. Where no company midpoint exists, the percent is FMP versus the year-ago reported period, not a guide gap.

Current quarter — Q3 2026e

The +25% is a mix statement. Services, on the Q2 run-rate, does not produce another consolidated acceleration. Cloud does, if it stays near the stair-step. A quarter in which Search slows further and Cloud merely holds +82% can still land near $128 billion. A quarter in which the capacity bridge also slows reported Cloud growth is how that number gets missed. They said the bridge pressures margin. They did not say it slows growth.

FQ+1 — Q4 2026e

No quarterly dollar guide reaches Q4 either. The second-half comments are the same ones: TPU system sales stay a small piece of 2026 and ramp exiting the year, with the vast majority in 2027; Cloud margin carries the bridge and Wiz through the rest of 2026; depreciation and energy keep pressing. Capex is a full-year range. It does not split Q3 from Q4.

The street's shape from Q3 to Q4 is flat revenue growth and a higher EPS growth rate. The EPS step-up is not a company guide. Q4 2025 diluted EPS is a cleaner base than 2026's gain-inflated prints. FMP's Q4 EBIT margin is about 35.0%, versus a Q4 2025 operating margin of 31.6%, about +350 bps. That is still a wider expansion than the +160 bps just reported, in a half where the words are pressure.

FY guide still in force — FY2026 capex. FY2027 is not a number.

Eight reported quarters — trajectory into this print

Columns are calendar quarters. YoY is versus the year-ago quarter. Q3 2026 is not on this table. Margins and growth rates are computed from the cited dollars. They are not a separate Daloopa series.

Revenue growth re-accelerated for four compares, about +830 bps from Q3 2025 to Q2 2026. Search peaked at +19.1% and has already given back about 230 bps. Cloud has not. Capex dollars are still stepping up. The YoY rate decelerated for the first time in this window, from +107% to +100%. Free cash flow is the line that already broke: Q2 was −$5,855 million versus $5,301 million a year earlier. The quarter being previewed is compared with Q3 2025 free cash flow of $24,461 million, the cash quarter in this window. Cloud operating margin reached 35.6% in Q2 ($8,814 million on $24,768 million) versus 20.7% a year earlier ($2,826 million on $13,624 million), about +1,484 bps. The expansion barely narrowed from Q1. That is the line they just warned on. Cloud backlog was $514 billion versus $106 billion, +385%. Remaining performance obligations were $513.9 billion versus $108.2 billion, +375%, after +400% in Q1 ($462.3 billion versus $92.4 billion). 50% is expected over 24 months. Half of the book is not a Q3 revenue guide. Paid clicks held at +13% for a second quarter (Q1 was also +13%). Cost-per-click slowed to +3% from +5%. Volume is not the Search problem yet. Price is.

Five-year annual trajectory

Search was a high-single-digit grower in FY2022 and FY2023 (+9.1%, +7.7%), then a low-teens grower for two years (+13.2%, +13.4%). The mid-teens to high-teens quarterly rates of the last three prints are the break from that two-year regime, which is what Q3 is now lapping. Cloud grew +36.8% in FY2022, decelerated to +25.9%, then re-accelerated to +30.6% and +35.8%. Annual growth did not clear the high-30s. A full year near the recent quarterly rate would be a different company than FY2025's +35.8%. Free cash flow has been stuck near $70 billion for three years (+15.8%, then +4.7%, then +0.7%) while capex went from $32 billion to $91 billion. The $195–205 billion guide is the next step, which is why Q2 free cash flow went negative.

Confidence is high on the decision to spend and low on Q3's incremental margin. The bullishness moved from "the print" to "the opportunity is bigger than we said." The new sentences are about what Q3 will not have: FX help, an easy Search compare, and a clean Cloud margin.

They have raised the 2026 capex range twice and have not withdrawn the demand language. Pichai, on the July call, said 2027 returns look healthier than a year ago, which is why they will not cap 2027 spend to close the constraint. Ashkenazi's budget test is still "invest as long as the return is attractive." That is the line the street either believes or does not. The 2026 range itself has been absorbed: BofA near $200 billion versus a Street near $202 billion, inside $195–205 billion. The open item is 2027, which they have refused to number since February.

The caveats are specific, and they got tighter at each outlook. Q1's FX tailwind was about 3 points. Q2's was 1 point, and that is what they delivered (reported +24%, constant currency +23%). Q3 is a slight headwind. The Search lap is the first time in this four-call stretch that the outlook names a compare as a growth headwind rather than only FX. Supply has been tight "for multiple quarters." What is new is that they are paying up for someone else's capacity for about six months rather than only slipping revenue. Pichai called those bridge deals expensive near term and highly ROI-positive over a multi-year contract. "Modest" is not a basis-point guide.

Assumptions that the tone should not bury. Demand is not the scarce input. Allocation order is frontier training, then Search, YouTube, and Cloud serving. External TPU demand is pushed into the customer's data center. TPU hardware revenue is a 2027 item. Q2 growth accelerated even after excluding those sales, so +81.8% was not a one-quarter hardware dump. Headcount of 198,933 versus 187,103 was +6.3% against revenue +24.2%. They are not guiding a hiring freeze, and they are not guiding headcount to grow with revenue. The balance sheet funds the gap. Q2 operating cash flow was still $39,069 million. She described debt going from about $16 billion a year ago to about $100 billion, and said they are not going back to the equity market except the at-the-market program for stock-based-compensation taxes. The common dividend was held at $0.22. Waymo stays inside Other Bets. The Q2 operating loss was $1,799 million. It is not the Q3 algorithm.

No C-suite change was announced on the call. The defensive moments were coding, the six-month bridge, and a refusal to discuss a Waymo separation. That is narrower than a demand walk-back.

After July 22, the range did not move. Thomas Kurian at Goldman Sachs Communacopia on September 8 did not touch $195–205 billion. He added duration: customers exceeding commitments by more than 50%, a two-year payback on AI servers in aggregate, and a majority of AI-infrastructure contract value on five-year commits. The host's "nearly $100 billion" Cloud run-rate is four times the Q2 print. It is not a forecast. On October 20, in a Big Technology interview, Pichai said Cloud is forgoing short-term revenue so frontier work keeps compute, and that demand runs into 2027. That matches the July allocation stack. It is an interview, not a filing, and it is not a new guide. It does make "strong growth" less clean: growth can stay high and still be lower than the backlog would allow. Gemini 4 Argon was still "in training" on the July call, with coding named as the gap. It opened first to cyber defenders on September 30 and more broadly on October 13. Access inside the quarter was too late, and too limited, to explain Q3 Cloud revenue. It is a Q4 product and a call topic.

The stock is not a guidepost and is not used as a multiple here. The de-rating that followed the July capex raise and the negative free-cash-flow quarter is already public. The open question is whether Q3 Cloud margin and Search growth validate the new caveats, not whether demand commentary gets more bullish. September and October commentary got more bullish on demand and more explicit that some Cloud revenue is being deferred.

Ranked by what can change the model on October 28. "What the street believes" is this quarter. Segment rates are Bank of America's description, not FMP.

A print the street can live with is Cloud about +85% to +93%, Search +14% to +16%, gross revenue near $128 billion, operating income above the recent +30% pace but not necessarily the full $45 billion, capex inside $195–205 billion, and no 2027 number well above the path one broker has already published. A print that says the stair-step paused is Cloud at or below about +82% after they paid for third-party capacity, Search below 14%, operating income near $41 billion, and another capex raise. The contradiction is the one they have already set up. They say demand is still there, and that the Q3 headwinds are FX, the Search compare, and a temporary margin cost. The street believes the first half of that and has put a faster operating-income number under it than the last two quarters support.

Notable non-earnings news since the July 22 report, through October 26. There is no pre-announcement, no guidance cut, and no CEO or CFO departure. The Q3 revenue bar is unchanged. The tape is compute allocation, capex funding, a limited Gemini 4 launch, and remedies that hit Search and Play without an AdX breakup.

Also on the tape, and not given a full row: Waymo closed a $5 billion loan and named Madrid for 2027. That is Other Bets scaling on its own balance sheet, next to a Cloud business that did $24,768 million last quarter. Marvell granted a warrant tied to TPU-attached purchases through fiscal 2033. It adds a supplier around the TPU. It is not a Google revenue or capex disclosure. A London consumer claim over Play fees is open, with no ruling and no reserve found here.

Consistent beater on revenue. The 12-for-12 GAAP EPS record is no longer an operating beat. Over Q3 2023 through Q2 2026, diluted EPS beat FMP in 12 of 12 quarters, and revenue beat in 11 of 12. The last four quarters are 4 for 4 on both lines. The only clean revenue miss is Q4 2024, about −0.2%. Revenue-beat magnitude widened: the last four surprises averaged about +2.5%, versus about +1.4% over the prior eight, and the last two (+2.7%, +2.8%) are the widest of the window. GAAP EPS surprise widened from about +13% to about +86%, and that widening is the mark. Q2 EPS of $9.11 versus an FMP estimate of $2.87 is a +217% "beat" in the same quarter operating income was $40,770 million and the equity-securities gain was $99,031 million.

One FMP cell is overridden. The stored Q2 2025 revenue estimate of $99.826 billion is not the print-day figure. Scoring it would manufacture a miss on a quarter the street reported as a beat. Q2 2025 revenue is scored against $94.0 billion and marked with a dagger. Q3 2026 is not on the grid.

Green is a wider beat (revenue at least +2%, or EPS at least +15%). Blue is a mid-single to low-double-digit EPS beat, or a revenue beat of about 1% to 2%. Yellow is a thin beat. Red is the only miss. The two darkest EPS cells, Q1 2026 and Q2 2026, are the equity-securities marks. They are not a better business.

Alphabet does not discuss Street consensus on these calls. The prepared remarks are year over year. Ashkenazi's account of the Q2 EPS surprise was other income of about $98 billion, primarily unrealized gains. The operating account in the same remarks was revenue up 24% and operating income up 30%. Q3 2025 was the same split in the other direction on the margin: reported margin 30.5% included the EC fine, ex-fine margin was 33.9% on her words, and EPS still beat because other income was large. The quiet mark quarters beat EPS by roughly +4% on average. There is no quiet quarter inside the last four, so there is no recent organic EPS-surprise trend to copy.

Operating margin tells the same split. Expansion ran several hundred basis points through 2024, narrowed, compressed in Q3 and Q4 2025, then re-expanded by +220 bps and +160 bps. The compression quarters were still EPS beats. Do not read 12 for 12 EPS as 12 quarters of better operations.

Read against October 28, two habits are in the bar. The revenue habit is a beat of roughly 2% to 3%, a bit wider lately, on growth that has been accelerating. The EPS habit of the last two quarters must not be repeated mechanically. If Q3 other income is another outsized unrealized gain, GAAP EPS will "beat" $3.03 by a meaningless amount. The number that extends the actual streak is revenue versus $127.9 billion, and whether operating margin is still expanding against the Q3 2025 comparison of 30.5% reported.