Alphabet Cloud Revenue Doubles to $30.7 Billion; Search Growth Slows to 12% as Executives Point to FX, AI Mode Monetization

4:21:10 PM ETFlex it. When it's shaking in your face, go

4:21:12 PM ETand flex it, when they double in your face,

4:21:14 PM ETgo

4:21:14 PM ETflex it when they're shaking in your face go ahead flex it when they're

4:21:18 PM ETEat a bump, eat the sick, right?

4:21:22 PM ETUh-huh. That's

4:21:24 PM ETit, right?

4:21:26 PM ETYeah. That's it. When the shaking in your feet's

4:21:30 PM ETgoing

4:21:34 PM ETYo, yo, yo, yo, yo, rearrangement,

4:21:36 PM ETThat's unnegotable. Go, go,

4:21:38 PM ETexit.

4:21:40 PM ETPeople are shocked when they find

4:21:42 PM ETout that it's just the two of us.

4:21:44 PM ETWe're a small, scrappy business compared to the big

4:21:44 PM ETpeople are shocked when they find out

4:21:46 PM ETguys. In order to

4:21:46 PM ETit's just the two of us we're a small scrappy business compared to the big guys

4:21:48 PM ETgrow, we have to multiply our

4:21:50 PM ETcapability. Gemini and Google

4:21:50 PM ETin order to grow we have to multiply our capability sorry gemini

4:21:52 PM ETWorkspace absolutely levels the playing

4:21:54 PM ETfield for small businesses like

4:21:56 PM ETours. We're able to act like we're much

4:21:56 PM ETand google workspace absolutely levels the playing field for small

4:21:58 PM ETbigger. People think that we have a whole army

4:21:58 PM ETbusinesses like ours we're able to act like we're much bigger people

4:22:00 PM ETof folks. Payroll, visuals,

4:22:02 PM ETmarketing strategy, emails,

4:22:02 PM ETthink that we have a whole army of folks payroll visuals mock

4:22:04 PM ETplaybooks with Gemini it feels like we

4:22:06 PM EThave a multi-million dollar

4:22:06 PM ETemails playbooks with gemini it feels like we have a multi

4:22:08 PM ETteam. We show up as a big brand even though we're

4:22:10 PM ETtiny.

4:22:10 PM ETmultimillion dollar team we show up as a big brand even though we're tiny

4:22:12 PM ETIt's NFL Sunday. Who's coming

4:22:14 PM ETover? Everyone. For what

4:22:14 PM ETit's nfl sunday who's coming over everyone for

4:22:16 PM ETgame? All of them.

4:22:18 PM ETDave and Lizzie are from California. We'll need the new

4:22:20 PM ETNiners and Chargers. Check!

4:22:20 PM ETwhat game all of them dave and lizzie are from california we'll need the

4:22:22 PM ETKyle's mom's freaking out. She needs every

4:22:24 PM ETgame for fantasy. Check. We're gonna be

4:22:24 PM ETniners and chargers check kyle's mom's freaking out she needs every game

4:22:26 PM ETthe worst hosts ever. We're

4:22:28 PM ETgood. We got NFL Sunday ticket and YouTube

4:22:28 PM ETfor fantasy check we're gonna be the worst hosts ever we're good

4:22:30 PM ETTV. NFL

4:22:32 PM ETSunday Ticket and YouTube TV. Get

4:22:32 PM ETwe get nfl sunday ticket in youtube tv

4:22:34 PM ETevery game, every Sunday, all in one

4:22:36 PM ETplace. NFL with Multi-View.

4:22:36 PM ETnfl sunday ticket in youtube tv get every game every sunday all in one place nfl

4:22:38 PM ETCheck. New users get NFL Sunday ticket for

4:22:40 PM ETeight payments of $30 a

4:22:40 PM ETnfl with multi view check new users get nfl sunday ticket for eight payments

4:22:40 PM ETthirty dollars a month

4:22:42 PM ETmonth. Dawn in the

4:22:46 PM ETsky, you know how I feel?

4:22:48 PM ETIt's a new dark it's a

4:22:50 PM ETnew day, it's a new

4:22:52 PM ETlife, and I'm

4:22:52 PM ETit's a new dog it's a new day it's a new life

4:22:54 PM ETfeeling good.

4:22:56 PM ETand i'm feeling good

4:23:00 PM ETIt's a new dog, it's a new

4:23:02 PM ETday, it's a new

4:23:04 PM ETlife, and I'm

4:23:06 PM ETfeeling good.

4:23:14 PM ETThis public sculpture that

4:23:16 PM ETI sketched seven years ago

4:23:16 PM ETthis public sculpture that i sketched

4:23:18 PM ETwas dormant and the

4:23:20 PM ETquestion became: how are we

4:23:20 PM ETseven years ago was dormant and

4:23:22 PM ETactually going to fabricate

4:23:24 PM ETthis? My work is

4:23:24 PM ETthe question became how are we actually going to fabricate this

4:23:26 PM ETrooted in research, so I'm using Chrome.

4:23:28 PM ETto look at the

4:23:28 PM ETmy work is rooted in research so i'm using

4:23:30 PM ETprocess, and I've got 200 tabs

4:23:32 PM ETopen, and I don't want to

4:23:32 PM ETchrome to look at the process and i've got

4:23:34 PM ETclose anything because it's

4:23:34 PM ETtwo hundred tabs open and i don't wanna close anything because

4:23:36 PM ETone thing to dream

4:23:38 PM ETit, and then it's another thing to

4:23:40 PM ETactually bring it to life.

4:23:40 PM ETit's one thing to dream it and then

4:23:42 PM ETanother thing to actually bring it to life

4:23:48 PM ETYeah,

4:23:50 PM ETtransetter, whoa, league of my own, it don't

4:23:52 PM ETget better. No, read what I

4:23:54 PM ETwrote, I'm a bestseller. Yeah, reach

4:23:56 PM ETfor my goals, hit it dead

4:23:56 PM ETno read what i wrote i'm a best seller yeah reach for my goals hit

4:23:58 PM ETcenter, hit it dead center. Better, better,

4:24:00 PM ETbring it better, bad because I'm bigger,

4:24:02 PM ETbetter, solo, single, selling several shows at every

4:24:04 PM ETsingle second. Peter Piper picked a peck a

4:24:04 PM ETbetter bat because i'm bigger better solo single selling several shows

4:24:06 PM ETof pickle peppers Connor came and crushed the

4:24:08 PM ETcompetition. Hope they get the message.

4:24:08 PM ETat every single second beat a piper picked a peck of pickled peppers kinda came and crushed the competition hope they get

4:24:10 PM ETlook, always showing up

4:24:12 PM ETLast year's song still holding

4:24:12 PM ETget the message look always showing up right

4:24:14 PM ETup. See the name now and they

4:24:16 PM ETknow what's up. Price hit the top

4:24:16 PM ETyear's song's still holding up see the name now and they know what's

4:24:18 PM ETand it's going up by my

4:24:20 PM ETWhoa. Leak of my own. It

4:24:20 PM ETup price hit the top and it's going up my friends set a train

4:24:22 PM ETdon't get better. No, read what I

4:24:24 PM ETwrote, I'm a bestseller.

4:24:26 PM ETyeah. Reach for my goals, it is dead sending

4:24:28 PM ETHit it dead center.

4:24:28 PM ETyeah reach for my goals hit it dead center why hit it dead center

4:24:32 PM ETYou can be a fan your entire life and still ask

4:24:34 PM ETwhat counts as a catch these days

4:24:34 PM ETshit hot you can be a fan your entire life and still ask what counts

4:24:36 PM ETanyway? Two feet for the

4:24:38 PM ETpros and one for the rest.

4:24:38 PM ETas a catch these days anyway two feet for the

4:24:40 PM ETNow that's the catch. And what is

4:24:42 PM ETdoes this mean?

4:24:42 PM ETand one for the rest now that's the catch and what does this mean why

4:24:44 PM ETWhy we all scared on the turtle, ready? Look left,

4:24:46 PM ETthrow right. Gotta hit the tight

4:24:46 PM ETreal scat on the turtle ready look left throw

4:24:48 PM ETend.

4:24:50 PM ETAnd what was the first touchdown

4:24:50 PM ETright gotta hit the tight end

4:24:52 PM ETdance? The high

4:24:54 PM ETsteps. Old school. That's the thing about being

4:24:54 PM ETand what was the first touchdown dance the high

4:24:56 PM ETa fan man. You can never know

4:24:58 PM ETenough.

4:24:58 PM ETstep old school that's the thing about being a fan man you can

4:25:00 PM ETnever know enough

4:25:10 PM ETWhat is the

4:25:12 PM ETI

4:25:26 PM ETI don't see how I see.

4:25:50 PM ETIf I'm

4:25:52 PM ETclear, you

4:26:00 PM ETUh

4:26:18 PM ETAnd

4:26:40 PM ETAnd I'm going to go to the bathroom.

4:26:44 PM ETIt's going down

4:26:46 PM ETeverywhere today. Buckle up, Brad.

4:26:46 PM ETit's going down everywhere today buck

4:26:48 PM ETFloor it, Tom. Over to Jen, a

4:26:50 PM ETcrafty player, building her bag biz.

4:26:50 PM ETup brad floor it tom over to jen a crafty pledge

4:26:52 PM ETShe opens Gemini, takes a pick,

4:26:54 PM ETthen asks to see the bag on some

4:26:54 PM ETbuilding her bag biz she opens gemini takes a pick then asks

4:26:56 PM ETstudio backgrounds. That's how bullets

4:26:58 PM ETHobo goes pro,

4:26:58 PM ETto see the bag on some studio backgrounds that's how boho hova

4:27:00 PM ETTom. Next, we have Mr.

4:27:02 PM ETBeast. Uh what's the game plan here? Well

4:27:02 PM ETgoes pro tom next we have mister beast what's the game plan here

4:27:04 PM ETfirst, he takes a pic of the pool and asks

4:27:06 PM EThow much slime he'll need to fill it

4:27:08 PM ETup. Gemini crunches the numbers and

4:27:08 PM ETwell first he takes a pic of the pool and asks how much slime

4:27:10 PM ETwhoa! Take me to

4:27:10 PM ETneed to fill it up gemini crunches the numbers and woah

4:27:12 PM ETFlying Town! Brad, I'm

4:27:14 PM ETjust speechless. Well, it's both of

4:27:14 PM ETtake me to slime town brad i'm just speechless well

4:27:16 PM ETa third and a logistical

4:27:18 PM ETtriumph. Finally, we have

4:27:18 PM ETit's both a third and a logistical triumph

4:27:20 PM ETNick. Yesterday, he asked Gemini to

4:27:22 PM ETlet him know as soon as a good deal on a

4:27:24 PM ETsquat rack pops up. That's a pro

4:27:24 PM ETwe have nick yesterday he asked gemini to let him know as soon as a good deal

4:27:26 PM ETmove, Tom. It's scanning

4:27:28 PM ETeverywhere. Scanning?

4:27:28 PM ETon a squat rack pops up that's a pro move tom

4:27:30 PM ETScanning? Gemini doesn't quit!

4:27:30 PM ETit's scanning everywhere scanning and scanning gemini

4:27:32 PM ETOh, it found one! We got a deal

4:27:34 PM EThere, Tom. Nick makes the snag

4:27:34 PM ETdoesn't quit oh it found one we got a deal here tom

4:27:36 PM ETfor cheap. He's saving his loot

4:27:38 PM ETand blasting his glue. Working

4:27:38 PM ETnick makes the snag for cheap he's saving his loot and blasting it

4:27:40 PM ETquads and saving

4:27:42 PM ETwads. I don't think we needed to double down there,

4:27:42 PM ETgloop working quads and saving was i don't think we needed to double down there

4:27:44 PM ETTom.

4:27:44 PM ETtom

4:27:52 PM ETCan

4:28:10 PM ETMm-hmm.

4:28:16 PM ETOh my god.

4:28:16 PM EToh my

4:28:20 PM ETname is Maria.

4:30:28 PM ETWelcome, everyone.

4:30:28 PM ETwelcome everyone

4:30:30 PM ETThank you for standing by for the

4:30:32 PM ETAlphabet Third Quarter

4:30:34 PM ET2026 Earnings Conference Call.

4:30:34 PM ETthank you for standing by for the alphabet third quarter

4:30:34 PM ETtwenty twenty six earnings conference call

4:30:36 PM ETAt this time, all participants

4:30:38 PM ETare in a listen-only mode.

4:30:38 PM ETat this time all participants are in a listen only mode

4:30:40 PM ETAfter the speaker

4:30:42 PM ETpresentation, there will be a question and

4:30:44 PM ETanswer session.

4:30:44 PM ETafter the speaker presentation there will be a question and answer

4:30:46 PM ETTo ask a question during the session, you will

4:30:48 PM ETneed to press star one on your

4:30:50 PM ETtelephone. I would now like

4:30:52 PM ETto hand the conference over to your speaker

4:30:54 PM ETtoday, Jim Friedland, head

4:30:54 PM ETi would now like to hand the conference over to your speaker today

4:30:56 PM ETof Investor Relations.

4:30:58 PM ETPlease go ahead.

4:30:58 PM ETjim friedland head of investor relations please go ahead

4:31:00 PM ETThank you. Good afternoon,

4:31:02 PM ETeveryone, and welcome to Alphabet's third quarter

4:31:04 PM ET2026 Earnings Conference

4:31:04 PM ETthank you good afternoon everyone and

4:31:06 PM ETCall. With us today are Senator Pachai,

4:31:06 PM ETto alphabet's third quarter twenty twenty six earnings conference call

4:31:08 PM ETPhilip Schindler, Thomas Curian, and

4:31:10 PM ETAnat Ashkenazi. Now I'll

4:31:10 PM ETwith us today are sundar pichai philip schindler

4:31:12 PM ETquickly cover the safe harbor. Some

4:31:12 PM ETthomas curian and anat ashkenazi now i'll quickly cover the safe harbor

4:31:14 PM ETof the statements that we make today regarding our

4:31:16 PM ETbusiness, operations, and financial

4:31:18 PM ETperformance may be considered

4:31:18 PM ETsome of the statements that we make today regarding our business operations and financial performance

4:31:20 PM ETforward-looking. Such statements are based on current

4:31:22 PM ETexpectations and assumptions that are subject to

4:31:22 PM ETmay be considered forward looking such statements are based on

4:31:24 PM ETa number of risks and

4:31:26 PM ETuncertainties. Actual results could differ

4:31:26 PM ETcurrent expectations and assumptions that are subject to a number of risks and uncertainties

4:31:28 PM ETmaterially. Please refer to our

4:31:30 PM ETforms 10K and 10Q, including the

4:31:30 PM ETactual results could differ materially please refer to our forms ten ks

4:31:32 PM ETrisk factors. We undertake

4:31:32 PM ETand ten q including the risk factors

4:31:34 PM ETno obligation to update any

4:31:36 PM ETforward-looking statement. During this

4:31:36 PM ETwe undertake no obligation to update any forward looking statement

4:31:38 PM ETcall, we will present both GAAP and non-GAAP

4:31:40 PM ETfinancial measures, a

4:31:40 PM ETduring this call we will present both gaap and non gaap financial measures

4:31:42 PM ETreconciliation of non-GAP to GAAP measures is

4:31:44 PM ETincluded in today's earnings press release,

4:31:44 PM ETa reconciliation of non gaap to gaap measures is included in today's earnings press release

4:31:46 PM ETwhich is distributed and available to the

4:31:48 PM ETpublic through our Investor Relations

4:31:50 PM ETwebsite located at

4:31:50 PM ETwhich is distributed and available to the public through our investor relations website located at

4:31:52 PM ETabc.xyz forward slash

4:31:54 PM ETinvestor. Our comments will be on

4:31:54 PM ETabc xyzinvestor

4:31:56 PM ETyear-over-year comparisons unless we

4:31:58 PM ETstate otherwise. And now I'll

4:31:58 PM ETour comments will be on year over year comparisons unless we state otherwise

4:32:00 PM ETturn the call over to

4:32:00 PM ETand now i'll turn the call over to sundar

4:32:02 PM ETSundar. Thanks,

4:32:04 PM ETJim, and hello, everyone. Q3

4:32:04 PM ETthanks jim and hello everyone

4:32:06 PM ETwas a strong quota for our

4:32:08 PM ETalphabet. Our deep full stack approach

4:32:10 PM ETis driving momentum across the

4:32:10 PM ETq3 was a strong quarter for alphabet our deep full stack up

4:32:12 PM ETbusiness with many opportunities

4:32:12 PM ETapproach is driving momentum across the business with many opportunities ahead

4:32:14 PM ETahead. Let me give you a quick snapshot

4:32:16 PM ETof the quarter.

4:32:16 PM ETlet me give you a quick snapshot of the quarter

4:32:18 PM ETAlphabet revenues grew 25% year

4:32:20 PM ETover year. We are now at a

4:32:22 PM EThalf a trillion dollar annual run

4:32:22 PM ETalphabet revenues grew twenty five percent year over year

4:32:24 PM ETrate. That's remarkable growth.

4:32:24 PM ETwe are now at a half a trillion dollar annual run rate that's remarkable growth

4:32:26 PM ETSearch has strong

4:32:28 PM ETmomentum. Our AI investments are driving

4:32:30 PM ETincreased usage and overall query

4:32:30 PM ETsearch has strong momentum our ai investments are driving increased usage

4:32:30 PM ETand overall query growth

4:32:32 PM ETgrowth, including commercial

4:32:34 PM ETqueries. In fact, our

4:32:36 PM ETyear-over-year query growth rate is the

4:32:36 PM ETincluding commercial queries in fact our year over year query growth rate

4:32:38 PM EThighest it's been in the last five

4:32:38 PM ETis the highest it's been in the last five years

4:32:40 PM ETyears. Google Cloud revenues grew

4:32:42 PM ET103%.

4:32:42 PM ETgoogle cloud revenues grew one hundred and three percent

4:32:44 PM ETOur AI infrastructure and

4:32:46 PM ETsolutions are the major drivers for cloud

4:32:48 PM ETgrowth, and our customers see us as

4:32:48 PM ETour ai infrastructure and solutions are the major drivers for cloud growth

4:32:50 PM ETthe leading provider here so

4:32:52 PM ETthere's huge demand.

4:32:52 PM ETand our customers see us as the leading provider here so there's huge demand

4:32:54 PM ETWe are working hard to bring more

4:32:56 PM ETcapacity online to meet this long-term

4:32:58 PM ETdemand. At the same

4:32:58 PM ETwe are working hard to bring more capacity online to meet this long term

4:33:00 PM ETtime, we are seeing strong demand for

4:33:02 PM ETour TPU systems

4:33:02 PM ETat the same time we are seeing strong demand for our tpu systems

4:33:04 PM ETacross a wide range of

4:33:06 PM ETcustomers, and we expect that to continue as

4:33:08 PM ETwe scale. I'm

4:33:08 PM ETacross a wide range of customers and we expect

4:33:08 PM ETthat to continue as we scale

4:33:10 PM ETexcited to welcome Thomas to the

4:33:12 PM ETcall and he'll share more about clouds

4:33:12 PM ETi'm excited to welcome thomas to the call

4:33:14 PM ETMomentum. On our

4:33:16 PM ETAI models, customers are

4:33:16 PM ETand he'll share more about cloud's momentum on our ai models

4:33:18 PM ETusing our models more than ever.

4:33:18 PM ETcustomers are using our models more than ever

4:33:20 PM ETA first-party model APIs are now

4:33:22 PM ETprocessing nearly 32 billion tokens per

4:33:24 PM ETminute.

4:33:24 PM ETour first party model apis are now processing nearly thirty two billion tokens per minute

4:33:26 PM ET40% more than last quarter, and

4:33:28 PM ETmore than four times what we

4:33:28 PM ETforty percent more than last quarter

4:33:30 PM ETprocessed a year ago.

4:33:32 PM ETAnd with Gemini 4 argon, and now

4:33:32 PM ETand more than four times what we processed a year ago

4:33:34 PM ETcarbon, which we announced today, we

4:33:36 PM ETare excited to be leading at the

4:33:36 PM ETand with gemini iv argon and now carbon

4:33:38 PM ETfrontier. The Gemini app is

4:33:38 PM ETwhich we announced today we are excited to be leading at the frontier

4:33:40 PM ETnow at 1 billion users and is the

4:33:42 PM ETfastest growing product in Google's

4:33:42 PM ETthe gemini app is now at one billion users and is the fastest growing product

4:33:44 PM EThistory. Lastly, we have surpassed

4:33:46 PM ET150 million YouTube

4:33:46 PM ETin google's history lastly we have surpassed

4:33:48 PM ETmusic and premium subscribers

4:33:50 PM ETglobally, including

4:33:50 PM ETone hundred and fifty million youtube music and premium subscribers globally

4:33:50 PM ETincluding trials

4:33:52 PM ETtrials. Today I'll talk more about

4:33:54 PM ETour AI progress, our

4:33:54 PM ETtoday i'll talk more about our ai progress

4:33:56 PM ETglobal consumer platforms led by search

4:33:58 PM ETand YouTube, then our long-term

4:33:58 PM ETour global consumer platforms led by search and youtube

4:34:00 PM ETbets. Starting

4:34:00 PM ETthan our long term bets

4:34:02 PM ETwith AI, highlighted by

4:34:04 PM ETour Gemini 4 series of frontier

4:34:06 PM ETmodels, we are seeing

4:34:06 PM ETstarting with ai highlighted by our gemini four series of frontier models

4:34:08 PM ETstrong demand and usage for Argon, which we

4:34:10 PM ETlaunched two weeks ago. In

4:34:10 PM ETwe are seeing strong demand and usage for argonne which we launched two weeks ago

4:34:12 PM ETfact, on our developer platforms, usage in

4:34:14 PM ETthe first week ramped up three and a half

4:34:16 PM ETtimes faster than Gemini

4:34:16 PM ETin fact on our developer platforms usage in the first week ramped up

4:34:18 PM ET3, even with capacity

4:34:20 PM ETconstraints. Oregon is proving to do

4:34:20 PM ET3.5x faster than gemini three even with capacity constraints

4:34:22 PM ETreally well in areas like finance,

4:34:24 PM ETlegal, cybersecurity and

4:34:24 PM ETargonne is proving to do really well in areas like

4:34:26 PM ETscience. Developers and cloud

4:34:28 PM ETcustomers are already using it for things

4:34:28 PM ETfinance legal cybersecurity and science

4:34:30 PM ETlike untangling legacy code

4:34:30 PM ETdevelopers and cloud customers are already using it for things like untangling legacy

4:34:32 PM ETbases, accelerating scientific

4:34:34 PM ETresearch, robotics,

4:34:34 PM ETcodebases accelerating scientific research robotics

4:34:36 PM ETdiscovering product improvements,

4:34:36 PM ETdiscovering product improvements

4:34:38 PM ETand domain-specific tasks

4:34:40 PM ETrequiring in-depth reasoning.

4:34:40 PM ETand domain specific tasks requiring in-depth reasoning

4:34:42 PM ETAnd at Google, just in the last

4:34:44 PM ETfew weeks, teams

4:34:44 PM ETand at google just in the last few weeks

4:34:46 PM ETacross Chrome, Ads and Search have used

4:34:48 PM ETArgon to write and optimize code much

4:34:48 PM ETteams across chrome ads and search have used oregon to write and optimize code much

4:34:50 PM ETfaster, shipping

4:34:52 PM ET35% more code with 30%

4:34:54 PM ETfewer errors.

4:34:54 PM ETfaster shipping thirty five percent more code with thirty percent fewer errors

4:34:56 PM ETOregon is just the start of the Gemini 4

4:34:58 PM ETseries. Today we

4:34:58 PM ETargon is just the start of the gemini four series

4:35:00 PM ETannounced Carbon, it's our strongest frontier

4:35:02 PM ETmodel with state-of-the-art

4:35:02 PM ETtoday announced carbon it's our strongest frontier model with

4:35:04 PM ETcybersecurity and long horizon

4:35:06 PM ETcapabilities. It ranks number one on the

4:35:06 PM ETstate of the art cybersecurity and long horizon capabilities

4:35:08 PM ETrecently updated artificial

4:35:10 PM ETAnalysis Intelligence Index at

4:35:10 PM ETit ranks number one on the recently updated

4:35:12 PM ET59.

4:35:14 PM ETCarbon outperforms argon across

4:35:14 PM ETartificial analysis intelligence index at fifty nine

4:35:16 PM ETalmost every domain, like full stack

4:35:18 PM ETweb development, multimodal

4:35:18 PM ETcarbon outperforms argon across almost every domain like full stack

4:35:20 PM ETunderstanding, complex knowledge

4:35:22 PM ETwork, and scientific reasoning.

4:35:22 PM ETweb development multimodal understanding complex knowledge work

4:35:22 PM ETand scientific reasoning

4:35:24 PM ETWe are rolling carbon out first to

4:35:26 PM ETa subset of trusted cyber defenders

4:35:28 PM ETbefore we expand access and it's

4:35:28 PM ETwe are rolling carbon out first to a subset of trusted cyber defenders

4:35:30 PM ETbeing tested by the US

4:35:32 PM ETgovernment. Beyond our Gemini

4:35:32 PM ETbefore we expand access and it's being tested by the u s government

4:35:34 PM ET4 series, yesterday we

4:35:34 PM ETbeyond our gemini four series

4:35:36 PM ETannounced Genie 4, our most advanced world

4:35:38 PM ETmodel yet that simulates safe,

4:35:38 PM ETyesterday we announced genie four our most advanced world model yet that simulates

4:35:40 PM ETrealistic, interactive worlds

4:35:42 PM ETwith native audio in real

4:35:42 PM ETsafe realistic interactive worlds with native audio in real time

4:35:44 PM ETtime. This is an important building block

4:35:46 PM ETfor training robots and

4:35:48 PM ETagents. This rapid progress

4:35:48 PM ETit is an important building block for training robots and agents

4:35:50 PM ETunderscores our incredible model

4:35:52 PM ETmomentum. We are at the point

4:35:52 PM ETthis rapid progress underscores our incredible model momentum

4:35:54 PM ETwhere AI is helping us increase

4:35:56 PM ETour training and launch velocity.

4:35:56 PM ETwe are at the point where ai is helping us increase our training and launch

4:35:58 PM ETI would expect the next 12 months

4:36:00 PM ETto see continued acceleration

4:36:00 PM ETi would expect the next twelve months to see continued

4:36:02 PM ETbecause of RSI.

4:36:02 PM ETacceleration because of rsi

4:36:04 PM ETWe have more models coming soon,

4:36:06 PM ETincluding more efficient versions to

4:36:08 PM EThelp developers optimize for speed and

4:36:08 PM EThave more models coming soon including more efficient

4:36:10 PM ETcost and to

4:36:10 PM ETversions to help developers optimize for speed and cost

4:36:12 PM ETdeploy across our billions of

4:36:12 PM ETand to deploy across our billions of users

4:36:14 PM ETusers. Let's turn to

4:36:16 PM ETsearch, which is bringing the benefits of AI

4:36:18 PM ETto more people than any other product in the

4:36:18 PM ETlet's turn to search which is bringing the benefits

4:36:20 PM ETworld. Just like our shift to

4:36:22 PM ETmobile over a decade ago, we are focused

4:36:22 PM ETof ai to more people than any other product in the world

4:36:24 PM ETon building the best search user experience

4:36:24 PM ETjust like our shift to mobile over a decade ago we are focused on building the best

4:36:26 PM ETfollowed by

4:36:28 PM ETmonetization. We are building one

4:36:28 PM ETsearch user experience followed by monetization

4:36:30 PM ETseamless experience across AI

4:36:32 PM EToverviews, AI mode, and

4:36:32 PM ETwe are building one seamless experience across ai overviews

4:36:32 PM ETai mode

4:36:34 PM ETtraditional search. And people are

4:36:36 PM ETusing our AI features more, searching

4:36:38 PM ETin entirely new ways.

4:36:38 PM ETand traditional search and people are using our ai features more

4:36:40 PM ETTheir queries are longer,

4:36:40 PM ETsearching in entirely new ways their queries are longer

4:36:42 PM ETpersonalized, and

4:36:44 PM ETmultimodal. Queries in AI mode are on

4:36:46 PM ETaverage three times longer than

4:36:46 PM ETpersonalized and multimodal queries in ai mode are on average

4:36:48 PM ETtraditional searches, and one in

4:36:50 PM ETfive AI mode searches globally

4:36:50 PM ETthree times longer than traditional searches and one in five ai mode searches

4:36:52 PM ETuse voice, images or

4:36:54 PM ETdocuments. We have now

4:36:56 PM ETcrossed over 50 billion visual

4:36:56 PM ETglobally use voice images or documents

4:36:58 PM ETsearches per month. One

4:36:58 PM ETwe have now crossed over fifty billion visual searches per month

4:37:00 PM ETarea I'm excited about is helping

4:37:02 PM ETpeople get things done, be it

4:37:04 PM ETbooking reservations, filling in

4:37:04 PM ETone area i'm excited about is helping people get things done

4:37:06 PM ETforms, scheduling appointments, or

4:37:06 PM ETbe it booking reservations filling in forms scheduling appointments

4:37:08 PM ETmaking purchases. As AI

4:37:10 PM ETbecomes more personal and starts taking

4:37:12 PM ETaction for people and

4:37:12 PM ETor making purchases as ai becomes more personal and starts

4:37:14 PM ETbusinesses, trust and security matter more than

4:37:16 PM ETever. People

4:37:16 PM ETtaking action for people and businesses trust and security matter more than ever

4:37:18 PM ETalready put their trust in Google to get

4:37:20 PM ETthings done in their daily lives and

4:37:20 PM ETpeople already put their trust in google to get

4:37:22 PM ETour leadership here remains an

4:37:22 PM ETthings done in their daily lives and their leadership here

4:37:24 PM ETimportant differentiator.

4:37:24 PM ETremains an important differentiator

4:37:26 PM ETWe focus first on getting

4:37:28 PM ETthese agentic experiences right with the

4:37:30 PM ETGemini Agent for Enterprise because

4:37:30 PM ETwe focus first on getting these agentic experiences right with the gemini agent

4:37:30 PM ETenterprise

4:37:32 PM ETthe bar for security, safety,

4:37:34 PM ETand governance is extremely

4:37:36 PM EThigh. As a next step, we are bringing these

4:37:36 PM ETbecause the bar for security safety and governance is extremely high

4:37:38 PM ETfeatures to search and

4:37:40 PM ETalso, of course, to our billion users on

4:37:40 PM ETas a next step we are bringing these features to search

4:37:42 PM ETthe Gemini app. People are

4:37:42 PM ETand also of course to our billion users on the gemini app

4:37:44 PM ETalready using the Gemini app to do more

4:37:46 PM ETthan ask questions. They're

4:37:46 PM ETpeople are already using the gemini app to do more than ask questions

4:37:48 PM ETusing it to do things including

4:37:50 PM ETcomplex multi-step agentic

4:37:52 PM ETqueries. Our Gemini 4 series has

4:37:52 PM ETthey're using it to do things including complex multi step agentic queries

4:37:54 PM ETbeen a really big step forward

4:37:56 PM EThere and has already been handling

4:37:56 PM ETour gemini four series has been a really big step forward here

4:37:58 PM ETthese complex tasks really

4:38:00 PM ETwell. Agents will operate

4:38:00 PM ETand has already been handling these complex tasks really well

4:38:02 PM ETas a unifying layer across our

4:38:04 PM ETproducts, so you will see us extend

4:38:04 PM ETagents will operate as a unifying layer across our products

4:38:06 PM ETthese agentic experiences beyond

4:38:08 PM ETsearch in the Gemini app to Chrome.

4:38:08 PM ETso you will see us extend these agentic experiences beyond search in the gemini app

4:38:10 PM ETAndroid and our workspace products as

4:38:12 PM ETwell.

4:38:12 PM ETto chrome android and our workspace products as well

4:38:14 PM ETOn to YouTube, which continues to lead in

4:38:16 PM ETthe living room as the number one streaming platform

4:38:16 PM ETon to youtube which continues to lead in the living room as the number one

4:38:18 PM ETin the US in terms of watch

4:38:20 PM ETtime. It's also seeing good subscriber

4:38:22 PM ETgrowth, as I mentioned.

4:38:22 PM ETstreaming platform in the u s in terms of watch time

4:38:24 PM ETAt Made on YouTube last month, we announced more

4:38:24 PM ETit's also seeing good subscriber growth as i mentioned

4:38:26 PM ETthan 30 products and features for

4:38:28 PM ETcreators and viewers. One great

4:38:28 PM ETat made on youtube last month we announced more than thirty products and features

4:38:28 PM ETfor creators and viewers

4:38:30 PM ETexample is custom feeds, where you can simply

4:38:32 PM ETpick a topic you love and get a

4:38:34 PM ETtailored video home feed based on that

4:38:34 PM ETone great example is custom feeds where you can simply pick a topic you love

4:38:36 PM ETtopic. Philip will say more on

4:38:38 PM ETYouTube. On to our

4:38:38 PM ETand get a tailored video home feed based on that topic

4:38:38 PM ETphilip will say more on youtube

4:38:40 PM ETlong-term bets. Waymo is on a great

4:38:40 PM ETonto our long term bets

4:38:42 PM ETtrajectory. We have now driven more than

4:38:44 PM ET300 million fully autonomous

4:38:44 PM ETwaymo is on a great trajectory we have now driven more than

4:38:46 PM ETmiles. Woemo now operates in

4:38:48 PM ET15 major U.S. cities and just

4:38:48 PM ETthree hundred million fully autonomous miles waymo now

4:38:50 PM ETrecently opened up service to the public in San

4:38:52 PM ETAntonio, San Diego, and

4:38:52 PM EToperates in fifty major u s cities and just recently opened up service to the

4:38:52 PM ETpublic in san antonio

4:38:54 PM ETTampa with planned

4:38:56 PM ETexpansions into Canada, Germany,

4:38:56 PM ETsan diego and tampa with planned expansions

4:38:58 PM ETSingapore, and Spain.

4:38:58 PM ETinto canada germany singapore and spain

4:39:00 PM ETWe are also seeing strong progress

4:39:02 PM ETacross our other long-term

4:39:04 PM ETinitiatives. Isomorphic is

4:39:04 PM ETare also seeing strong progress across our other long term initiatives

4:39:06 PM ETusing Oregon to support drug design

4:39:08 PM ETresearch. We demonstrated

4:39:08 PM ETisomorphic is using oregon to support drug design research

4:39:10 PM ETmagic state cultivation for the first time

4:39:12 PM ETin quantum computing and

4:39:12 PM ETwe demonstrated magic state cultivation for the first time in quantum computing

4:39:14 PM ETlaunched our TPUs into space with

4:39:16 PM ETProject Suncatcher. So it's

4:39:18 PM ETbeen a busy quarter, and there's much more to

4:39:18 PM ETand launched our tpus into space with project suncatcher

4:39:20 PM ETcome. I want to thank our employees

4:39:20 PM ETso it's been a busy quarter and there's much more to come

4:39:22 PM ETfor their work this quarter

4:39:24 PM ETand over to you, Philip.

4:39:24 PM ETi want to thank our employees for their work this quarter and over to you philip

4:39:26 PM ETThanks, Ondar, and hello,

4:39:28 PM ETeveryone. I'll highlight our

4:39:28 PM ETthanks sundar and hello everyone

4:39:30 PM ETmomentum across search, YouTube and

4:39:32 PM ETpartnerships. We continue to

4:39:32 PM ETi'll highlight our momentum across search youtube and partnerships

4:39:34 PM ETaccelerate the deployment of Gemini models across

4:39:36 PM ETour entire ads infrastructure

4:39:36 PM ETwe continue to accelerate the deployment of gemini models across our

4:39:38 PM ETto boost performance in the three areas I've

4:39:40 PM EThighlighted before: ads

4:39:40 PM ETentire ads infrastructure to boost performance in the three areas i've highlighted before

4:39:42 PM ETquality, advertiser tools, and AI user

4:39:44 PM ETexperiences. First, ads

4:39:44 PM ETads quality advertiser tools and ai user experiences

4:39:46 PM ETquality. The integration

4:39:48 PM ETof Gemini models into our ads

4:39:48 PM ETfirst ads quality the integration of gemini models into our ads

4:39:50 PM ETinfrastructure is improving how we decode

4:39:52 PM ETuser intent and drive ads

4:39:54 PM ETrelevance. For example, in

4:39:54 PM ETinfrastructure is improving how we decode user intent and drive ads relevance

4:39:56 PM ETQ3, we made the biggest update to

4:39:58 PM ETproduct listing ads in a decade, adding

4:39:58 PM ETfor example in q3 we made the biggest update to product

4:40:00 PM ETnew AI capabilities to help users

4:40:02 PM ETmake more confident purchase

4:40:02 PM ETlisting ads in a decade adding new ai capabilities

4:40:04 PM ETdecisions. Already, we've seen

4:40:06 PM ETsatisfaction increase by 19%

4:40:06 PM ETto help users make more confident purchase decisions

4:40:08 PM ETand dissatisfaction drop more than

4:40:08 PM ETalready we've seen satisfaction increase by nineteen percent

4:40:10 PM EThalf. We're also seeing AI

4:40:12 PM ETmodels enable better prediction of

4:40:12 PM ETand dissatisfaction dropped more than half we're also seeing ai models

4:40:14 PM ETclicks and conversions, stronger

4:40:16 PM EToverall performance, and deeper personalization

4:40:16 PM ETenable better prediction of clicks and conversions

4:40:18 PM ETof ads. For

4:40:18 PM ETstronger overall performance and deeper personalization of ads

4:40:20 PM ETinstance, through several innovations and

4:40:22 PM ETpersonalization modeling, we delivered

4:40:22 PM ETfor instance through several innovations and personalization modeling

4:40:24 PM ET18% incremental advertiser

4:40:26 PM ETvalue and an 8% increase in

4:40:26 PM ETwe delivered eighteen percent incremental advertiser value

4:40:28 PM ETuser sentiment on Discover ads in

4:40:30 PM ET2026.

4:40:30 PM ETand an eight percent increase in user sentiment on discover ads in twenty twenty six

4:40:32 PM ETOverall, AI is accelerating our ads

4:40:34 PM ETdevelopment and drives the majority of our

4:40:34 PM EToverall ai is accelerating our ads development and drives the

4:40:36 PM ETproduct launches.

4:40:38 PM ETSecond, advertiser tools. We're

4:40:38 PM ETmajority of our product launches second advertiser tools

4:40:40 PM ETensuring every advertiser from global

4:40:42 PM ETbrands to small businesses, can

4:40:44 PM ETseamlessly participate in our AI

4:40:44 PM ETwe're ensuring every advertiser from global brands

4:40:46 PM ETservices. AI Max and

4:40:46 PM ETto small businesses can seamlessly participate in our ai services

4:40:48 PM ETPerformance Max continue to see strong

4:40:50 PM ETadoption to give marketers even more

4:40:50 PM ETai max and performance max continue to see strong adoption

4:40:52 PM ETcontrol, we're now introducing a way

4:40:54 PM ETto test different budgets and ROI

4:40:54 PM ETto give marketers even more control we're now introducing a way to test

4:40:56 PM ETtargets across multiple search

4:40:58 PM ETcampaigns in a single AI

4:40:58 PM ETdifferent budgets and roi targets across multiple search campaigns

4:41:00 PM ETA-B test.

4:41:00 PM ETin a single ai max ab test

4:41:02 PM ETTo drive better performance, we

4:41:04 PM ETintegrated our AI creative tools

4:41:06 PM ETdirectly into AI Max.

4:41:06 PM ETto drive better performance we integrated our ai creative tools

4:41:08 PM ETIn Q3 alone, Gemini

4:41:10 PM ETmodels helped advertisers generate over

4:41:10 PM ETinto ai max in q three alone

4:41:12 PM ET1.5 billion campaign

4:41:12 PM ETgemini models helped advertisers generate over one point five billion campaign assets

4:41:14 PM ETassets, up over 7x

4:41:16 PM ETyear over year, for

4:41:18 PM ETexample, by using AI Max to capture

4:41:18 PM ETup over seven x year over year for example

4:41:20 PM ETlong-form and multilingual

4:41:22 PM ETqueries, Telstra saw a 52%

4:41:22 PM ETby using ai maps to capture long form and multilingual queries

4:41:24 PM ETincrease in conversions and a

4:41:26 PM ET36% reduction in cost per

4:41:26 PM ETtelstra saw a fifty two percent increase in conversions

4:41:28 PM ETorder showing up in key moments, however,

4:41:28 PM ETand a thirty six percent reduction in cost per order showing up in

4:41:30 PM ETthe customer search.

4:41:30 PM ETkey moments however the customer search

4:41:32 PM ETThird, AI user

4:41:34 PM ETexperiences. Sonder mentioned the

4:41:36 PM ETdistinct behavioral changes we are seeing as

4:41:36 PM ETthird ai user experiences

4:41:38 PM ETusers ask entirely new categories of

4:41:38 PM ETsundar mentioned the distinct behavioral changes we're seeing as users ask and

4:41:40 PM ETquestions. We think of search

4:41:42 PM ETholistically, a unified

4:41:42 PM ETentirely new categories of questions we think of search holistically

4:41:44 PM ETproduct spanning traditional results and

4:41:46 PM ETAI-powered search. On

4:41:46 PM ETa unified product spending traditional results and ai powered search

4:41:48 PM ETAI mode, many of these queries are upper

4:41:50 PM ETfunnel queries, allowing us to capture

4:41:50 PM ETon ai mode many of these queries are upper funnel

4:41:52 PM ETnew high-value commercial

4:41:54 PM ETintent. In addition, the

4:41:54 PM ETqueries allowing us to capture new high value commercial intent

4:41:56 PM ETmajority of AI mode queries are

4:41:58 PM ETincremental, creating more monetization

4:41:58 PM ETin addition the majority of ai mode queries are incremental

4:42:00 PM ETopportunities over time.

4:42:02 PM ETAdditionally, our user research shows that

4:42:02 PM ETcreating more monetization opportunities over time

4:42:04 PM ETpeople are comfortable with ads on these

4:42:06 PM ETsurfaces and find them highly

4:42:06 PM ETadditionally our user research shows that people are comfortable with ads

4:42:08 PM ETrelevant. Click-through rates are already

4:42:10 PM ETsimilar between sponsored and organic

4:42:10 PM ETon these surfaces and find them highly relevant click through rates are already

4:42:12 PM ETproduct list items within AI mode.

4:42:14 PM ETBased on

4:42:14 PM ETsimilar between sponsored and organic product list items within ai mode

4:42:16 PM ETthese collective results, we're now

4:42:18 PM ETscaling high-quality ads within AI

4:42:18 PM ETbased on these collective results we are now skating

4:42:20 PM EToverviews, in AI mode, to more

4:42:22 PM ETusers and queries. This

4:42:22 PM EThigh quality ads within ai overviews in ai mode to more users and queries

4:42:24 PM ETincludes taking our monetization

4:42:26 PM ETfootprint global for the first time.

4:42:28 PM ETToday we're announcing the

4:42:28 PM ETthis includes taking our monetization footprint global for the first time

4:42:30 PM ETexpansion of ads in AI mode into

4:42:32 PM ETinternational markets alongside

4:42:32 PM ETtoday we are announcing the expansion of ads in ai mode into international markets

4:42:34 PM ETmulti-turn ads in AI mode in the

4:42:36 PM ETUS. As

4:42:36 PM ETalongside multi turn ads in ai mode in the u s

4:42:38 PM ETalways, our approach is deliberate.

4:42:40 PM ETprioritizing the user experience and engagement

4:42:40 PM ETas always our approach is deliberate prioritizing the user experience and engaging

4:42:42 PM ETfirst, and then scaling the formats

4:42:44 PM ETthat drive proven RI for

4:42:46 PM ETadvertisers. That discipline is what

4:42:46 PM ETfirst and then scaling the formats that drive proven roi for advertisers

4:42:48 PM ETgives us a strong, durable

4:42:50 PM ETfoundation for the future.

4:42:50 PM ETthat discipline is what gives us a strong durable foundation for the future

4:42:52 PM ETAs Sondar mentioned, we're in the

4:42:54 PM ETearly stages of agentic experiences

4:42:56 PM ETacross our surfaces and excited about the

4:42:56 PM ETas sundar mentioned we are in the early stages of agentic experiences across our

4:42:58 PM ETcommercial opportunities

4:43:00 PM ETahead. The Universal Commerce Protocol,

4:43:00 PM ETservices and excited about the commercial opportunities ahead

4:43:02 PM ETis being rapidly adopted

4:43:04 PM ETby retailers. Today,

4:43:04 PM ETthe universal commerce protocol ucp is being rapidly adopted by retailers

4:43:06 PM ETUCP enables direct checkout

4:43:08 PM ETfor hundreds of thousands of brands across

4:43:08 PM ETtoday ucp enables direct checkout

4:43:10 PM ETGoogle including Chewy, Nike,

4:43:12 PM ETTapestry, and many more.

4:43:12 PM ETfor hundreds of thousands of brands across google including chewy nike tapestry

4:43:14 PM ETWe're combining this with the Agent Payments

4:43:16 PM ETProtocol as we build for the future of

4:43:16 PM ETand many more we're combining this with the

4:43:18 PM ETcommerce. Our focus

4:43:20 PM ETspans two fronts: unlocking

4:43:20 PM ETagent payments protocol as we build for the future of commerce

4:43:22 PM ETagentic capabilities that act on behalf of

4:43:22 PM ETour focus spans two fronts unlocking agentic capabilities that act

4:43:24 PM ETthe consumer and designing

4:43:26 PM ETpractical tools that give shoppers

4:43:28 PM ETconfidence and control.

4:43:28 PM ETon behalf of the consumer and designing practical tools that gives shoppers confidence

4:43:28 PM ETand control

4:43:30 PM ETWe launched Universal Card this

4:43:32 PM ETquarter to support that second

4:43:34 PM ETtrack, mirroring how people actually shop

4:43:34 PM ETwe launched universal card this quarter to support that second track

4:43:36 PM ETtoday across Google. This allows

4:43:38 PM ETretailers to maintain their relationships with

4:43:38 PM ETmirroring how people actually shop today across google

4:43:40 PM ETcustomers while helping to capture

4:43:40 PM ETthis allows retailers to maintain their relationships with customers

4:43:42 PM ETsales and avoid abandoned

4:43:44 PM ETcards. Moving to

4:43:44 PM ETwhile helping to capture sales and avoid abandoned carts

4:43:46 PM ETYouTube. We view YouTube as

4:43:48 PM ETa flywheel of creators, viewers,

4:43:48 PM ETmoving to youtube we view youtube as a

4:43:50 PM ETand advertisers. First,

4:43:52 PM ETcreators. Every day, more than

4:43:52 PM ETflywheel of creators viewers and advertisers first creators

4:43:54 PM ET20 million videos are being

4:43:56 PM ETuploaded. We are making

4:43:56 PM ETevery day more than twenty million videos are being uploaded

4:43:58 PM ETcreation easier with new generative AI

4:44:00 PM ETtools in YouTube Studio, plus

4:44:00 PM ETwe are making creation easier with new generative ai tools in youtube studio

4:44:02 PM ETa conversational editing assistant

4:44:04 PM ETin Shorts and YouTube

4:44:04 PM ETplus a conversational editing assistant in shorts and youtube create

4:44:06 PM ETCrate. Looking at

4:44:08 PM ETviewership. At Made on YouTube, we

4:44:10 PM ETintroduced short series where creators can

4:44:10 PM ETlooking at viewership it made on youtube

4:44:12 PM ETstructure their shorts into seasons and

4:44:14 PM ETepisodes. Sequential playback makes

4:44:14 PM ETwe introduced short series where creators can structure their shorts into seasons and episodes

4:44:16 PM ETit seamless for viewers to discover

4:44:18 PM ETnew shorts and keep

4:44:18 PM ETsequential playback makes it seamless for viewers to discover new shorts

4:44:18 PM ETand keep watching

4:44:20 PM ETwatching. The third part is

4:44:22 PM ETmonetization, starting with ads.

4:44:24 PM ETOur focus remains on delivering

4:44:24 PM ETthe third part is monetization starting with ads

4:44:26 PM ETvalue across the entire funnel.

4:44:28 PM ETIn Q3, living room accounted

4:44:28 PM ETour focus remains on delivering value across the entire funnel

4:44:30 PM ETfor over one-third of total

4:44:32 PM ETYouTube ads revenues.

4:44:32 PM ETin q3 living room accounted for over one third of total youtube ads

4:44:34 PM ETSmall and medium advertisers continue to adopt

4:44:36 PM ETdemand gen driving direct

4:44:36 PM ETrevenues small and medium advertisers continue to adopt

4:44:38 PM ETresponse on

4:44:38 PM ETmain gen driving direct response

4:44:40 PM ETYouTube, Wayfair's demand gen campaigns

4:44:42 PM ETdrove a 2.3 times higher return

4:44:42 PM ETon youtube wayfair's demand gen campaigns drove it

4:44:44 PM ETon ad spend compared with traditional video

4:44:46 PM ETreach campaigns. Our

4:44:46 PM ETtwo point three times higher return on ad spend compared with traditional video reach campaigns

4:44:48 PM ETYouTube Creator Partnerships

4:44:50 PM ETprogram connects brands with over three

4:44:50 PM ETour youtube creator partnerships program

4:44:52 PM ETmillion creators to drive highly

4:44:54 PM ETeffective campaigns. Brands like

4:44:54 PM ETconnects brands with over three million creators to drive highly effective campaigns

4:44:56 PM ETLexus continue to partner with creators

4:44:58 PM ETlike Colin and Samir, driving the

4:44:58 PM ETbrands like lexus continue to partner with creators like colin and samir

4:45:00 PM ETeleven percent lift in brand

4:45:02 PM ETconsideration. More than

4:45:02 PM ETdriving eleven percent lift in brand consideration

4:45:04 PM ET1.3 million channels are

4:45:06 PM ETenrolled in YouTube shopping

4:45:06 PM ETmore than one point three million channels are enrolled in youtube shopping

4:45:08 PM ETglobally. Building on this momentum, we

4:45:10 PM ETrecently announced that Amazon is joining YouTube

4:45:12 PM ETShopping. U.S. creators can now

4:45:12 PM ETbuilding on this momentum we recently announced that amazon is joining youtube shopping

4:45:14 PM ETtag Amazon products directly in their

4:45:16 PM ETvideos, shorts, and live streams,

4:45:16 PM ETus creators can now tag amazon products directly in their videos

4:45:18 PM ETdriving direct conversions.

4:45:20 PM ETImportantly,

4:45:20 PM ETshorts and live streams driving direct conversions

4:45:22 PM ETour subscription ecosystem continues

4:45:24 PM ETto complement our advertising

4:45:24 PM ETimportantly our subscription ecosystem continues to complement our advertising

4:45:26 PM ETbusiness. Subscriptions continue to grow

4:45:28 PM ETfaster than ads, led by YouTube Music

4:45:30 PM ETPremium. YouTube Premium

4:45:30 PM ETbusiness subscriptions continue to grow faster than ads

4:45:32 PM EThas more than 150 million

4:45:32 PM ETled by youtube music premium youtube premium

4:45:34 PM ETsubscribers with Premium and Premium

4:45:36 PM ETLight, viewers have

4:45:36 PM EThas more than a hundred fifty million subscribers with premium and premium lite

4:45:38 PM ETchoice. As more people opt into premium

4:45:40 PM ETexperiences, it's a win for

4:45:40 PM ETviewers have choice as more people opt into premium experiences

4:45:42 PM ETviewers, creators and our

4:45:44 PM ETbusiness. Subscriptions help fund the

4:45:44 PM ETit's a win for viewers creators and our business

4:45:46 PM ETcreator ecosystem and drive the

4:45:48 PM ETplatform forward.

4:45:48 PM ETsubscriptions help fund the creator ecosystem and drive the platform forward

4:45:50 PM ETFinally, strategic

4:45:52 PM ETpartnerships. Businesses are choosing Google

4:45:54 PM ETbecause we bring together cloud,

4:45:54 PM ETfinally strategic partnerships businesses are choosing google

4:45:56 PM ETAI, ads, and consumer platforms

4:45:58 PM ETto accelerate their transformation.

4:45:58 PM ETbecause we bring together cloud ai ads and consumer platforms

4:46:00 PM ETWe expanded our partnership with American

4:46:02 PM ETExpress to support the ongoing

4:46:02 PM ETto accelerate their transformation we expanded our partnership with american express to support

4:46:04 PM ETmodernization of their contact centers

4:46:06 PM ETand collaborate on agent

4:46:06 PM ETthe ongoing modernization of their contact centers and collaborate

4:46:08 PM ETe-commerce across search and wallet.

4:46:10 PM ETAnd Unity signed an expanded

4:46:10 PM ETon agentic commerce across search and wallet

4:46:12 PM ETmulti-year cloud deal to use our full

4:46:14 PM ETAI stack to power the next generation

4:46:14 PM ETand unity signed an expanded multiyear cloud deal to use our full ai stack to

4:46:16 PM ETgame engine. Our

4:46:18 PM ETcommercial priorities are clear. Expanding advertising.

4:46:18 PM ETpower the next generation game engine our commercial priorities are clear

4:46:20 PM ETRI, systematically

4:46:22 PM ETadvancing our AI monetization

4:46:22 PM ETexpanding advertiser roi systematically advancing our ai

4:46:24 PM ETefficiency and scaling our ecosystem

4:46:26 PM ETpartnerships to support the next era of

4:46:26 PM ETmonetization efficiency and scaling our ecosystem partnerships to support the next area of

4:46:28 PM ETdiscovery. In closing, I want to

4:46:30 PM ETthank Googlers Everywhere for the

4:46:32 PM ETincredible execution and our customers

4:46:32 PM ETdiscovery in closing i want to thank googlers everywhere for the

4:46:34 PM ETand partners for their continued

4:46:36 PM ETtrust. Thomas, over to

4:46:36 PM ETincredible execution and our customers and partners for their continued trust

4:46:38 PM ETyou.

4:46:38 PM ETthomas over to you

4:46:40 PM ETOur strong momentum is driven by our

4:46:42 PM ETintegrated AI

4:46:42 PM ETour strong momentum is driven by our integrated ai

4:46:44 PM ETportfolio of chips,

4:46:46 PM ETmodels, data.

4:46:46 PM ETportfolio of chips models

4:46:48 PM ETsecurity, and agent

4:46:50 PM ETplatforms, all designed

4:46:50 PM ETdata security and agent platforms

4:46:52 PM ETto work together. We

4:46:52 PM ETall designed to work together

4:46:54 PM EThave strong

4:46:56 PM ETdiversification across our product

4:46:56 PM ETwe have strong diversification across our product lines

4:46:58 PM ETlines, industries and

4:47:00 PM ETcustomers. Google Cloud

4:47:00 PM ETindustries and customers

4:47:02 PM ETnow has eighteen product

4:47:04 PM ETlines with more than one

4:47:04 PM ETgoogle cloud now has eighteen product lines

4:47:06 PM ETbillion dollars each in

4:47:08 PM ETannual revenue and our

4:47:08 PM ETwith more than one billion dollars each in annual revenue

4:47:10 PM ETgrowth this quarter was driven by

4:47:12 PM ETstrong growth in Google

4:47:12 PM ETand our growth this quarter was driven by strong growth

4:47:14 PM ETCloud Platform with

4:47:16 PM ETstrength in both AI and core

4:47:16 PM ETin google cloud platform with strength in both ai and

4:47:18 PM ETGCP. TPU

4:47:18 PM ETcore gcp

4:47:20 PM ETsystem sales were also a

4:47:22 PM ETmeaningful contributor. Over

4:47:22 PM ETtpu system sales were also a meaningful contributor

4:47:24 PM ETthe past 12 months, more than

4:47:26 PM ETtwo thousand six hundred

4:47:26 PM ETover the past twelve months more than two thousand six hundred customers

4:47:28 PM ETcustomers have used at least

4:47:30 PM ET100 billion

4:47:32 PM ETtokens and nearly seventy

4:47:32 PM EThave used at least one hundred billion tokens

4:47:34 PM ETat least ten trillion

4:47:36 PM ETtokens on Gemini

4:47:36 PM ETand nearly seventy five at least ten trillion tokens

4:47:38 PM ETEnterprise.

4:47:38 PM ETon gemini enterprise

4:47:40 PM ETNearly 90% of the Fortune

4:47:42 PM ET100 use Gemini

4:47:42 PM ETnearly ninety percent of the fortune one hundred

4:47:44 PM ETEnterprise, and

4:47:46 PM ETcustomers who use Gemini

4:47:46 PM ETuse gemini enterprise and customers

4:47:48 PM ETEnterprise in Q3 twenty twenty

4:47:50 PM ETmore than

4:47:50 PM ETwho use gemini enterprise in q3 twenty twenty five

4:47:52 PM ETquadruple their spend with us year

4:47:54 PM ETover year. One of the

4:47:54 PM ETmore than quadrupled their spend with us year over year

4:47:56 PM ETimportant ways in which

4:47:58 PM ETcustomers are using

4:47:58 PM ETone of the important ways in which customers are using gemini

4:48:00 PM ETGemini is for knowledge work.

4:48:02 PM ETAcross all our AI

4:48:04 PM ETservices, we have

4:48:04 PM ETis for knowledge work across

4:48:06 PM ETsold more than one hundred

4:48:06 PM ETall our ai services we have sold more than

4:48:08 PM ETmillion seats for knowledge

4:48:10 PM ETwork. Gemini

4:48:10 PM ETone hundred million seats for knowledge work

4:48:12 PM ETEnterprise now

4:48:14 PM ETincludes the Gemini agent. That

4:48:14 PM ETgemini enterprise now includes

4:48:16 PM ETis deeply integrated across

4:48:18 PM ETour entire cloud

4:48:18 PM ETthe gemini agent that is deeply integrated

4:48:18 PM ETacross our entire cloud portfolio

4:48:20 PM ETportfolio. It can

4:48:22 PM ETresearch and answer questions.

4:48:24 PM ETwrite and run

4:48:24 PM ETit can research and answer questions

4:48:26 PM ETcode, and automate process

4:48:28 PM ETflows with one

4:48:28 PM ETwrite and run code and automate process flows

4:48:30 PM ETagent, one API, and a

4:48:32 PM ETsingle prompt window. It

4:48:32 PM ETwith one agent one api and a single prompt window

4:48:34 PM ETgives companies strong

4:48:36 PM ETsecurity and cost

4:48:36 PM ETit gives companies strong security and cost controls

4:48:38 PM ETcontrols, and the widest choice

4:48:40 PM ETof frontier models.

4:48:40 PM ETand the widest choice of frontier models

4:48:42 PM ETCustomers include BNP

4:48:44 PM ETPariba, who is deploying Gemini

4:48:46 PM ETEnterprise to more than

4:48:46 PM ETcustomers include bnp paribas

4:48:48 PM ET65,000 employees to

4:48:48 PM ETwho is deploying gemini enterprise to more than

4:48:50 PM ETstreamline workflows

4:48:52 PM ETwhile maintaining strict

4:48:52 PM ETsixty five thousand employees to streamline workflows

4:48:54 PM ETenterprise-grade security and data

4:48:56 PM ETgovernance, Tata

4:48:56 PM ETwhile maintaining strict enterprise grade security and data governance

4:48:58 PM ETSteel, who deployed more than three

4:49:00 PM EThundred specialized AI

4:49:00 PM ETtata steel who deployed more than

4:49:02 PM ETagents in just nine

4:49:04 PM ETmonths to enhance

4:49:04 PM ETthree hundred specialized ai agents in just nine months

4:49:06 PM ETmanufacturing safety,

4:49:08 PM EToptimize supply chains, and cut

4:49:08 PM ETto enhance manufacturing safety optimize supply chains

4:49:10 PM ETservice times for customers by

4:49:12 PM ET50%.

4:49:12 PM ETand cut service times for customers by fifty percent

4:49:14 PM ETAlta Beauty, who built

4:49:14 PM ETulta beauty

4:49:16 PM ETthe Alta AI

4:49:18 PM ETshopping assistant,

4:49:18 PM ETwho built the ulta ai shopping assistant

4:49:20 PM ETresulting in a 3x

4:49:22 PM EThigher conversion rate than

4:49:24 PM ETtraditional browsing on their site.

4:49:24 PM ETresulting in a 3x higher conversion rate

4:49:24 PM ETthan traditional browsing on their site

4:49:26 PM ETGemini is also

4:49:28 PM ETdriving acceleration for our

4:49:30 PM ETdatabase, analytics, and

4:49:30 PM ETgemini is also driving acceleration

4:49:32 PM ETcybersecurity products.

4:49:34 PM ETGemini can now help

4:49:34 PM ETfor our database analytics and cybersecurity products

4:49:36 PM ETusers find,

4:49:38 PM ETanalyze, and visualize

4:49:38 PM ETgemini can now help users

4:49:40 PM ETinformation from structured and unstructured

4:49:40 PM ETfind analyze and visualize information from

4:49:42 PM ETdata. Gemini

4:49:44 PM ETuses our knowledge catalog

4:49:44 PM ETstructured and unstructured data gemini uses our knowledge catalog

4:49:46 PM ETto find the right sources and

4:49:48 PM ETdefinitions of

4:49:50 PM ETinformation, improving

4:49:50 PM ETto find the right sources and definitions of information

4:49:52 PM ETaccuracy, lowering token cost,

4:49:54 PM ETand making responses more

4:49:54 PM ETimproving accuracy lowering token cost

4:49:56 PM ETreliable. Gemini uses

4:49:58 PM ETour lake house to

4:49:58 PM ETand making responses more reliable

4:50:00 PM ETaccess information from multiple

4:50:00 PM ETgemini uses our lakehouse to access

4:50:02 PM ETsystems with zero

4:50:04 PM ETcopy. To respond

4:50:04 PM ETinformation from multiple systems with zero copy

4:50:06 PM ETto agentic analysis,

4:50:08 PM ETBigQuery improved query

4:50:08 PM ETto respond to agentic analysis

4:50:10 PM ETspeed over forty

4:50:12 PM ETyear over year while

4:50:12 PM ETbigquery improved query speed

4:50:14 PM ETreducing query processing

4:50:14 PM ETover forty percent year over year while reducing

4:50:16 PM ETcost by over fifty

4:50:18 PM ETyear over year.

4:50:18 PM ETquery processing cost by over fifty percent year over year

4:50:20 PM ETCustomers include Deutsche

4:50:22 PM ETTelekom, which uses

4:50:22 PM ETcustomers include deutsche telekom

4:50:24 PM ETBigQuery, cloud storage, and knowledge

4:50:26 PM ETcatalog to build a

4:50:26 PM ETwhich uses bigquery cloud storage and knowledge catalog

4:50:28 PM ETunified data lake house

4:50:30 PM ETacross 40 plus legacy

4:50:30 PM ETto build a unified data lakehouse across

4:50:32 PM ETon-premise systems.

4:50:32 PM ETforty plus legacy on premise systems

4:50:34 PM ETEtsy, which

4:50:34 PM ETetsy

4:50:36 PM ETmigrated three petabytes of

4:50:38 PM ETdata to Unified

4:50:38 PM ETwhich migrated three petabytes of data to unified lakehouse

4:50:40 PM ETLake House. By using BigQuery and

4:50:42 PM ETIceberg, they now

4:50:44 PM ETperform critical data joins up to

4:50:44 PM ETby using bigquery and iceberg they now

4:50:46 PM ET60% faster.

4:50:46 PM ETperform critical data joins up to sixty percent faster

4:50:48 PM ETThe growing risk of cybersecurity

4:50:50 PM ETthreats is also

4:50:52 PM ETdriving strong demand for

4:50:52 PM ETthe growing risk of cybersecurity threats

4:50:54 PM ETour AI-powered security

4:50:54 PM ETis also driving strong demand for our ai powered

4:50:56 PM ETplatform. Our

4:50:56 PM ETsecurity platform

4:50:58 PM ETGemini flash cyber

4:51:00 PM ETmodels are frontier models that

4:51:00 PM ETour gemini flash cyber models

4:51:02 PM ETautomate vulnerability

4:51:04 PM ETdiscovery across twenty programming

4:51:04 PM ETare frontier models that automate

4:51:06 PM ETlanguages and

4:51:06 PM ETvulnerability discovery across twenty programming languages

4:51:08 PM ETpatch 2.6 times more

4:51:10 PM ETreliably than other

4:51:10 PM ETand patch two point six times more reliably

4:51:12 PM ETmodels. And there's even better

4:51:14 PM ETperformance to come as

4:51:14 PM ETthan other models and there's even better performance to come

4:51:16 PM ETwe roll out Argon.

4:51:16 PM ETas we roll out argonne

4:51:18 PM ETand carbon.

4:51:18 PM ETand carbon

4:51:20 PM ETOur security platform

4:51:22 PM ETintegrates threat

4:51:24 PM ETintelligence, cyber response

4:51:24 PM ETour security platform integrates

4:51:24 PM ETthreat intelligence

4:51:26 PM ETprioritization, automated

4:51:26 PM ETcyber response prioritization

4:51:28 PM ETscanning, code remediation

4:51:28 PM ETautomated scanning

4:51:30 PM ETand monitoring.

4:51:32 PM ETGemini and Google Threat

4:51:32 PM ETcode remediation and monitoring gemini in google threat intelligence

4:51:34 PM ETIntelligence autonomously

4:51:36 PM ETanalyzes millions of external

4:51:38 PM ETevents every day with ninety-eight

4:51:38 PM ETautonomously analyzes millions of

4:51:40 PM ETaccuracy.

4:51:40 PM ETexternal events every day with ninety eight percent accuracy

4:51:42 PM ETAnd more than 95% of

4:51:44 PM ETWISC customers are

4:51:46 PM ETadopting AI-powered

4:51:46 PM ETand more than ninety five percent of wiz customers

4:51:48 PM ETfeatures, and growth

4:51:50 PM ETcontinues. These security

4:51:50 PM ETare adopting ai powered features

4:51:52 PM ETtools protect critical

4:51:52 PM ETand growth continues these security tools protect critical

4:51:54 PM ETinfrastructure, including financial

4:51:56 PM ETinstitutions like Van Norte

4:51:58 PM ETand Jack Hendry and

4:51:58 PM ETinfrastructure including financial institutions like van norte

4:52:00 PM ETtelecommunication providers such as

4:52:02 PM ETATT.

4:52:02 PM ETand jack henry and telecommunication providers such as at and t

4:52:04 PM ETAll of this is

4:52:06 PM ETpowered by leading AI

4:52:06 PM ETall of this is powered

4:52:08 PM ETinfrastructure, spanning our TPU

4:52:10 PM ETsystems, Axion

4:52:10 PM ETby our leading ai infrastructure spanning our tpu systems

4:52:12 PM ETCPUs, the latest

4:52:14 PM ETNIDA GPUs, and

4:52:14 PM ETaxione cpus the latest nvidia gpus

4:52:16 PM ETour agent compute

4:52:18 PM ETengine, which is specifically designed

4:52:20 PM ETfor AI agents.

4:52:20 PM ETand our agent compute engine which is specifically designed

4:52:20 PM ETfor ai agents

4:52:22 PM ETTPU systems are

4:52:24 PM ETspecialized systems built

4:52:26 PM ETfor AI, combining

4:52:26 PM ETtpu systems are specialized systems

4:52:28 PM EThardware, compute storage,

4:52:28 PM ETbuilt for ai combining hardware

4:52:30 PM ETand networking with

4:52:32 PM ETsystems components and

4:52:32 PM ETcompute storage and networking with systems components

4:52:34 PM EToptimized software for training and

4:52:36 PM ETinference. TPU

4:52:36 PM ETand optimized software for training and inference

4:52:38 PM ET8T and

4:52:40 PM ET8I offer up to

4:52:40 PM ETtpu 8t and 8i

4:52:42 PM ET2.7 times better trading

4:52:44 PM ETperformance and

4:52:44 PM EToffer up to two point seven times better trading performance

4:52:46 PM ET80% better inference price

4:52:48 PM ETperformance,

4:52:48 PM ETand eighty percent better inference price performance

4:52:50 PM ETrespectively, versus our previous generation

4:52:52 PM ETof TPUs. To make

4:52:52 PM ETrespectively versus our previous generation of tpus

4:52:54 PM ETit easy for developers to

4:52:56 PM ETport workloads to

4:52:56 PM ETto make it easy for developers to port workloads to tpus

4:52:58 PM ETTPUs. We recently

4:53:00 PM ETworked closely with Meta

4:53:02 PM ETto deliver Torch

4:53:02 PM ETwe recently worked closely with meta

4:53:02 PM ETto deliver torch tpu

4:53:04 PM ETTPU, which allows users to

4:53:06 PM ETtrain, serve, and

4:53:08 PM ETdeploy PyTorch models on

4:53:08 PM ETwhich allows users to train serve

4:53:10 PM ETTPUs using the

4:53:10 PM ETand deploy pytorch models on tpus

4:53:12 PM ETtools they're already familiar

4:53:14 PM ETwith and rely on.

4:53:14 PM ETusing the tools they're already familiar with and rely on

4:53:16 PM ETNearly a hundred

4:53:18 PM ETcustomers use TPUs

4:53:18 PM ETnearly one hundred customers use tpus today

4:53:20 PM ETtoday, and we're seeing strong

4:53:22 PM ETand growing demand from many

4:53:24 PM ETothers. In addition to our

4:53:24 PM ETand we're seeing strong and growing demand from many others

4:53:26 PM ETwork with Anthropic,

4:53:28 PM ETboth directly and

4:53:28 PM ETin addition to our work with anthropic both directly

4:53:30 PM ETvia Broadcom and

4:53:30 PM ETand via broadcom

4:53:32 PM ETpublic reference customers like

4:53:34 PM ETApple, who are scaling

4:53:34 PM ETand public reference customers like apple

4:53:36 PM ETpre-training of their foundation

4:53:38 PM ETmodels on the latest generation of

4:53:38 PM ETwho are scaling pre training of their foundation models

4:53:40 PM ETTPUs. We're

4:53:42 PM ETalso working with new

4:53:42 PM ETon the latest generation of tpus

4:53:44 PM ETfrontier model companies,

4:53:46 PM ETincluding Mirandal

4:53:46 PM ETwe're also working with new frontier model companies

4:53:48 PM ETand Physical Intelligence,

4:53:48 PM ETincluding mirandel and physical intelligence

4:53:50 PM ETcapital markets firms

4:53:52 PM ETincluding Citadel Securities and Jump

4:53:54 PM ETTrading,

4:53:54 PM ETcapital markets firms including citadel securities and jump trading

4:53:56 PM ETuniversities like Caltech

4:53:58 PM ETand public sector service

4:53:58 PM ETuniversities like caltech

4:54:00 PM ETproviders including

4:54:02 PM ETCassava and Cirascale.

4:54:02 PM ETand public sector service providers

4:54:02 PM ETincluding kosovo and cirrascale

4:54:04 PM ETGiven continued

4:54:06 PM ETstrong and exhilarating

4:54:08 PM ETdemand, we've recently entered

4:54:08 PM ETgiven continued strong and accelerating demand

4:54:10 PM ETinto partnerships to

4:54:12 PM ETdeliver TPU systems

4:54:12 PM ETwe've recently entered into partnerships

4:54:14 PM ETwith both Crux AI and

4:54:16 PM ETOracle, with more in the

4:54:16 PM ETto deliver tpu systems with both crux ai

4:54:16 PM ETand oracle

4:54:16 PM ETwith more in the works

4:54:18 PM ETworks. Over to you,

4:54:20 PM ETAnat. Thank

4:54:20 PM ETover to you anat

4:54:22 PM ETyou, Thomas. My comments will focus

4:54:24 PM ETon year-over-year comparisons for the third

4:54:24 PM ETthank you thomas my comments will focus

4:54:26 PM ETquarter unless I state

4:54:28 PM ETotherwise. I will start with the results at

4:54:28 PM ETon year over year comparisons for the third quarter unless i state otherwise

4:54:30 PM ETthe alphabet level and will then

4:54:32 PM ETcover our segment results. I'll end with

4:54:32 PM ETi will start with the results at the alphabet level and will then cover our segment results

4:54:34 PM ETsome commentary on our outlook for the

4:54:36 PM ETfourth quarter and full year

4:54:38 PM ET2026. We had a

4:54:38 PM ETi'll end with some commentary in our outlook for the fourth quarter

4:54:38 PM ETand full year twenty twenty six

4:54:40 PM ETstrong third quarter

4:54:40 PM ETwe had a strong third quarter

4:54:42 PM ETdelivering our thirteenth consecutive

4:54:44 PM ETquarter of double-digit revenue

4:54:46 PM ETgrowth. Consolidated revenues were

4:54:46 PM ETdelivering our thirteenth consecutive quarter of double digit revenue growth

4:54:48 PM ET$127.6

4:54:50 PM ETbillion dollars, up twenty five

4:54:50 PM ETconsolidated revenues were one hundred and twenty seven point six billion dollars

4:54:52 PM ETin both reported and constant

4:54:54 PM ETcurrency. Total costs

4:54:54 PM ETup twenty five percent in both reported and constant currency

4:54:56 PM ETof revenue was $51.4

4:54:58 PM ETbillion, up

4:54:58 PM ETtotal cost of revenue was fifty one point four billion dollars

4:55:00 PM ET24%. TAC was

4:55:02 PM ET$16.4 billion

4:55:02 PM ETup twenty four percent tac was sixteen point four billion dollars

4:55:04 PM ETup 10%.

4:55:06 PM ETOther cost of revenues was

4:55:08 PM ET$35.1 billion.

4:55:08 PM ETup ten percent other cost of revenues

4:55:10 PM ETup 32%, driven

4:55:12 PM ETby hard work costs primarily

4:55:12 PM ETwas thirty five point one billion dollars up thirty two percent

4:55:14 PM ETrelated to the sales of TPU

4:55:16 PM ETsystems, increasing

4:55:16 PM ETdriven by hardware costs primarily related

4:55:18 PM ETappreciation, content acquisition

4:55:18 PM ETto the sales of tpu systems increase in depreciation

4:55:20 PM ETcosts, and technical infrastructure

4:55:22 PM EToperations costs.

4:55:22 PM ETcontent acquisition costs and technical infrastructure operations costs

4:55:24 PM ETTotal operating expenses were up

4:55:26 PM ET20% to

4:55:28 PM ET$35.6 billion.

4:55:28 PM ETtotal operating expenses were up twenty percent

4:55:28 PM ETto thirty five point six billion dollars

4:55:30 PM ETRD expenses increased by

4:55:32 PM ET36 percent, driven by

4:55:34 PM ETcompensation, primarily for investment in

4:55:34 PM ETr and d expenses increased by thirty six percent

4:55:36 PM ETAI talent, as well as

4:55:36 PM ETdriven by compensation primarily for investment in ai talent

4:55:38 PM ETdepreciation. Sales and

4:55:40 PM ETmarketing expenses were up twenty-two

4:55:40 PM ETas well as depreciation sales and marketing

4:55:42 PM ETdue to investments in

4:55:44 PM ETGemini and

4:55:44 PM ETexpenses were up twenty two percent due to investments in gemini and

4:55:44 PM ETcompensation

4:55:46 PM ETcompensation. And GMA expenses

4:55:48 PM ETdecreased 17%,

4:55:50 PM ETprimarily due to lapping of higher legal

4:55:50 PM ETand g and a expenses decreased seventeen percent

4:55:52 PM ETexpenses that occurred in Q3 last

4:55:54 PM ETyear, partially offset by

4:55:54 PM ETprimarily due to lapping of higher legal expenses that

4:55:56 PM ETcurrent-year legal

4:55:56 PM EToccurred in q3 last year partially offset by current year legal charges

4:55:58 PM ETcharges. Operating income increased

4:56:00 PM ET30% to $40.6

4:56:02 PM ETbillion, and operating

4:56:02 PM EToperating income increased thirty percent

4:56:04 PM ETmargin was

4:56:04 PM ETto forty point six billion dollars and operating margin

4:56:06 PM ET31.8%. Other

4:56:06 PM ETwas thirty one point eight percent

4:56:08 PM ETincome and expenses was negative

4:56:10 PM ET$8.2 billion due primarily

4:56:10 PM ETother income and expenses was negative eight point two billion dollars

4:56:12 PM ETto an unrealized loss from our investment

4:56:14 PM ETin SpaceX, from a decline in

4:56:16 PM ETits stock price in the third

4:56:16 PM ETdue primarily to unrealized loss from our investment in spacex

4:56:18 PM ETquarter. Net income and earnings per

4:56:18 PM ETfrom a decline in its stock price in the third quarter

4:56:20 PM ETshare decreased due to the

4:56:22 PM ETdecline in OINE.

4:56:22 PM ETnet income and earnings per share decreased due to the

4:56:22 PM ETdecline in oi and e

4:56:24 PM ETWe generated strong operating cash

4:56:26 PM ETflow of $42.3

4:56:26 PM ETwe generated strong operating cash flow

4:56:28 PM ETbillion in the third quarter.

4:56:30 PM ETand $179.6

4:56:30 PM ETof forty two point three billion dollars in the third quarter

4:56:32 PM ETbillion for the trailing 12

4:56:34 PM ETmonths. Operating cash

4:56:34 PM ETand one hundred and seventy nine point six billion dollars for the trailing twelve months

4:56:36 PM ETflow was negatively impacted by the

4:56:38 PM ETbuildup of inventory for TPU

4:56:38 PM EToperating cash flow was negatively impacted by the buildup of inventory

4:56:40 PM ETsystem sales, as well as

4:56:42 PM ETa $5.2 billion cash

4:56:44 PM ETpayment for a

4:56:44 PM ETfor tpu system sales as well as a five point two billion dollars

4:56:46 PM ET2018 EC fine following the conclusion

4:56:48 PM ETof the appeal process.

4:56:48 PM ETcash payment for a twenty eighteen ec fine

4:56:50 PM ETCapEx was $51.4

4:56:52 PM ETbillion in the third quarter.

4:56:52 PM ETfollowing the conclusion of the appeal process capex

4:56:54 PM ETwith the vast majority of the spend in

4:56:54 PM ETwas fifty one point four billion dollars in the third quarter

4:56:56 PM ETtechnical infrastructure to support

4:56:58 PM ETour investments in AI.

4:56:58 PM ETwith the vast majority of the spend in technical infrastructure

4:57:00 PM ETApproximately

4:57:02 PM ET65% of our investments in technical

4:57:02 PM ETto support our investments in ai approximately

4:57:04 PM ETinfrastructure in Q3 was

4:57:04 PM ETsixty five percent of our investments in technical infrastructure in q3

4:57:06 PM ETrelated to servers and networking

4:57:06 PM ETwas related to servers

4:57:08 PM ETequipment, which generally

4:57:10 PM EThave useful lives of six

4:57:10 PM ETand networking equipment which generally have useful lives of six years

4:57:12 PM ETyears. The remaining thirty-five

4:57:14 PM ETwas for data centers, which

4:57:16 PM EThave useful lives of 7 to 40

4:57:16 PM ETthe remaining thirty five percent was for data centers

4:57:18 PM ETyears. We

4:57:18 PM ETwhich have useful lives of seven to forty years

4:57:20 PM EThad negative free cash flow of

4:57:22 PM ET$9.1 billion in the third

4:57:22 PM ETwe had negative free cash flow of nine point one billion dollars in the third

4:57:24 PM ETquarter, driven by our investments in

4:57:24 PM ETquarter driven by our investments in capex

4:57:26 PM ETCapEx. Free cash flow in the third quarter

4:57:28 PM ETwas also affected by the aforementioned

4:57:30 PM ETinventory bill to support TPU

4:57:30 PM ETfree cash flow in the third quarter was also affected by the aforementioned inventory

4:57:32 PM ETsystem sales and payment of

4:57:34 PM ETthe EC fine. Free

4:57:34 PM ETbuild to support tpu system sales and payment of the ec fine

4:57:36 PM ETcash flow was $19.7

4:57:38 PM ETbillion for the trail in 12

4:57:40 PM ETmonths. We ended the

4:57:40 PM ETfree cash flow was nineteen point seven billion dollars

4:57:40 PM ETfor the trailing twelve months

4:57:42 PM ETquarter with $265.1

4:57:44 PM ETbillion in cash and marketable

4:57:44 PM ETwe ended the quarter with two hundred and sixty five

4:57:46 PM ETsecurities, which

4:57:48 PM ETincludes $90.4

4:57:48 PM ETpoint one billion dollars in cash and marketable securities which includes

4:57:50 PM ETbillion of marketable equity

4:57:52 PM ETsecurities. Long-term debt was

4:57:52 PM ETninety point four billion dollars of marketable equity securities

4:57:54 PM ET$125.1

4:57:56 PM ETbillion.

4:57:56 PM ETlong term debt was one hundred and twenty five point one billion dollars

4:57:58 PM ETWe have also entered into multi-year

4:58:00 PM ETagreements to secure capacity for

4:58:00 PM ETwe have also entered into multi year agreements to

4:58:02 PM ETtechnical infrastructure and

4:58:04 PM ETinventory, as well as leases and energy.

4:58:04 PM ETsecure capacity for technical infrastructure and inventory as well as leases and energy

4:58:06 PM ETAt the end of the third quarter, we

4:58:08 PM EThad future purchase commitments

4:58:10 PM ETof $924

4:58:10 PM ETat the end of the third quarter we had future purchase commitments

4:58:12 PM ETbillion and pre-commencement leases

4:58:14 PM ETand other obligations of

4:58:14 PM ETof nine twenty four billion dollars and pre commencement lease

4:58:16 PM ETone hundred fifty two billion dollars.

4:58:16 PM ETand other obligations of one hundred and fifty two billion dollars

4:58:18 PM ETThese dollar amounts reflect the

4:58:20 PM ETmaximum contractual obligations.

4:58:22 PM ETHowever, a number of these agreements have variable

4:58:22 PM ETthese dollar amounts reflect the maximum contractual obligations

4:58:24 PM ETpricing and volume

4:58:26 PM ETflexibility, which could result in lowering

4:58:26 PM EThowever a number of these agreements have variable pricing and volume flexibility

4:58:28 PM ETspend. And as mentioned in our press

4:58:30 PM ETrelease, our board of directors declared a

4:58:30 PM ETwhich could result in lower spend and as mentioned in our press release

4:58:32 PM ETquarterly cash dividend on our common

4:58:34 PM ETstock of twenty two cents per

4:58:34 PM ETour board of directors declared a quarterly cash dividend on our common stock

4:58:36 PM ETshare, which is payable in

4:58:36 PM ETof zero point two two dollars per share which is payable in december

4:58:38 PM ETDecember. Turning to segment

4:58:40 PM ETresults, Google Services revenues

4:58:42 PM ETincreased 11% to

4:58:42 PM ETturning to segment results google services revenues increased eleven percent

4:58:44 PM ET$96.3 billion,

4:58:46 PM ETreflecting strong growth in search

4:58:46 PM ETto ninety six point three billion dollars reflecting strong growth

4:58:48 PM ETand subscriptions.

4:58:50 PM ETTotal advertising revenues were up

4:58:50 PM ETin search and subscriptions total ad

4:58:52 PM ETeleven percent, with growth across

4:58:54 PM ETall major verticals.

4:58:54 PM ETadvertising revenues were up eleven percent with growth across all major verticals

4:58:56 PM ETAs we stated in our outlook last

4:58:58 PM ETquarter, FX was a slight headwind

4:59:00 PM ETon year-over-year growth

4:59:00 PM ETas we stated in our outlook last quarter

4:59:02 PM ETrates and had a more significant impact

4:59:02 PM ETfx was a slight headwind on year over year growth rates

4:59:04 PM ETon a sequential basis.

4:59:04 PM ETand had a more significant impact on sequential basis

4:59:06 PM ETGoogle search and other advertising

4:59:08 PM ETrevenues increased by twelve percent

4:59:10 PM ETto $63.3

4:59:10 PM ETgoogle search and other advertising revenues increased by twelve percent

4:59:12 PM ETbillion, with retail driving the

4:59:14 PM ETlargest contribution, followed by

4:59:14 PM ETto sixty three point three billion dollars with retail driving the largest

4:59:16 PM ETfinance. YouTube advertising

4:59:18 PM ETrevenues increased nine

4:59:18 PM ETfollowed by finance youtube advertising revenues

4:59:20 PM ETto $11.2 billion.

4:59:22 PM ETwith brand driving the largest

4:59:22 PM ETincreased nine percent to eleven point two billion dollars

4:59:24 PM ETcontribution and notable strength in

4:59:26 PM ETliving room.

4:59:26 PM ETwith brand driving the largest contribution and notable strength in living room

4:59:28 PM ETNetwork advertising revenues of

4:59:30 PM ET$7.5 billion were up 1%.

4:59:30 PM ETnetwork advertising revenues of seven point five billion dollars were up one

4:59:32 PM ETSubscription

4:59:34 PM ETplatforms, and devices revenues increased

4:59:36 PM ET11% this quarter to

4:59:36 PM ETpercent subscription platforms and devices revenues increased eleven

4:59:38 PM ET$14.3 billion.

4:59:40 PM ETdriven by strong growth in YouTube

4:59:40 PM ETpercent this quarter to fourteen point three billion dollars

4:59:42 PM ETsubscriptions,

4:59:44 PM ETparticularly in YouTube Music and

4:59:44 PM ETdriven by strong growth in youtube subscriptions

4:59:46 PM ETpremium, and strong demand for AI

4:59:46 PM ETparticularly in youtube music and premium and strong demand

4:59:48 PM ETplans in Google One.

4:59:48 PM ETfor ai plans in google one

4:59:50 PM ETGoogle Services operating income

4:59:52 PM ETincreased 17% to

4:59:52 PM ETgoogle services operating income increased seventeen percent

4:59:54 PM ET$39.2 billion and operating

4:59:54 PM ETto thirty nine point two billion dollars

4:59:56 PM ETmargin was $40.2

4:59:58 PM ETand operating margin was forty point seven percent

5:00:00 PM ETThe Google Cloud segment

5:00:02 PM ETdelivered record growth in the third

5:00:04 PM ETquarter. Cloud revenues were

5:00:04 PM ETthe google cloud segment delivered record growth in the third quarter

5:00:06 PM ETup 103% to

5:00:08 PM ET$30.7 billion, driven by

5:00:08 PM ETcloud revenues were up one hundred and three percent to thirty point seven billion dollars

5:00:10 PM ETstrong growth in GCP, was

5:00:12 PM ETwith strength in AI, including both

5:00:12 PM ETdriven by strong growth in gcp with strength in ai

5:00:14 PM ETAI infrastructure and AI

5:00:16 PM ETsolutions, and core

5:00:16 PM ETincluding both ai infrastructure and ai solutions

5:00:16 PM ETand core gcp

5:00:18 PM ETGCP. TPU system cells were

5:00:20 PM ETalso a meaningful contributor.

5:00:22 PM ETGCP accounted for the vast majority of

5:00:22 PM ETtpu system sales were also a meaningful contributor

5:00:24 PM ETthe revenue generated in

5:00:26 PM ETQ3. The cloud revenue growth rate, excluding

5:00:26 PM ETgcp accounted for the vast majority of the revenue generated in q3

5:00:28 PM ETthe impact of TPU

5:00:30 PM ETsystems, was again incredibly

5:00:30 PM ETthe cloud revenue growth rate excluding the impact of tpu systems

5:00:30 PM ETwas again incredibly strong

5:00:32 PM ETstrong. Cloud

5:00:34 PM EToperating income was $11.1

5:00:36 PM ETbillion, more than tripling

5:00:36 PM ETcloud operating income was eleven point one billion dollars

5:00:38 PM ETyear over year. Operating

5:00:38 PM ETmore than tripling year over year

5:00:40 PM ETmargin increased from

5:00:42 PM ET23.7% in the third quarter last

5:00:42 PM EToperating margin increased from twenty three point seven percent in the third quarter

5:00:44 PM ETyear to

5:00:46 PM ET36.2 percent. The third

5:00:48 PM ETbridging capacity, which we

5:00:48 PM ETlast year to thirty six point two percent

5:00:50 PM ETdiscussed on our prior call, came

5:00:50 PM ETthe third party bridging capacity which we discussed on our prior call

5:00:52 PM ETonline only at the very end of

5:00:54 PM ETQ3 and therefore had

5:00:56 PM ETlimited negative impact to our margin in the

5:00:56 PM ETcame online only at the very end of q3

5:00:58 PM ETquarter. Google

5:00:58 PM ETand therefore had limited negative impact to our margin in the quarter

5:01:00 PM ETCloud's backlog was

5:01:02 PM ET$561 billion in the third

5:01:02 PM ETgoogle cloud's backlog was five

5:01:04 PM ETquarter, with the majority of the backlog related to

5:01:06 PM ETtypical GCP

5:01:06 PM ETone billion dollars in the third quarter with the majority of the backlog

5:01:08 PM ETcontracts, we expect to recognize

5:01:10 PM ETjust over 55% of the

5:01:10 PM ETrelated to typical gcp contracts

5:01:12 PM ETtotal backlog as revenues over the

5:01:12 PM ETwe expect to recognize just over fifty five

5:01:14 PM ETnext 24 months.

5:01:14 PM ETpercent of the total backlog as revenues over the next twenty four months

5:01:16 PM ETIn other bets,

5:01:18 PM ETrevenues were $372

5:01:20 PM ETmillion, and the operating loss

5:01:20 PM ETin other bets revenues were three seventy two million dollars

5:01:22 PM ETwas $2 billion as we continue to

5:01:24 PM ETexpand Waymo's business and invest

5:01:24 PM ETand the operating loss was two billion dollars as we continue to expand waymo's business

5:01:26 PM ETin key other bets.

5:01:28 PM ETIn alphabet level activities,

5:01:30 PM ETthe operating loss was $7.7

5:01:30 PM ETand invest in key other bets in alpha bet level activities

5:01:32 PM ETbillion, driven by shared

5:01:34 PM ETAI RD expenses.

5:01:34 PM ETthe operating loss was seven point seven billion dollars driven by shared ai r and d

5:01:34 PM ETexpenses

5:01:36 PM ETI'll now turn to

5:01:38 PM ETour outlook for the remainder of

5:01:40 PM ET2026. First, in terms of

5:01:40 PM ETi'll now turn to our outlook for the remainder of twenty twenty six

5:01:42 PM ETrevenues, at current spot

5:01:44 PM ETrates, we would expect a slightly larger

5:01:44 PM ETfirst in terms of revenues at current spot rates

5:01:46 PM ETFX headwind to our consolidated revenues in

5:01:48 PM ETQ4 relative to the

5:01:48 PM ETwe would expect a slightly larger fx headwinds to our consolidated revenues in q4

5:01:50 PM ETprior quarter. This impact

5:01:52 PM ETwill be seen primarily in search

5:01:52 PM ETrelative to the prior quarter this impact will be seen primarily in search

5:01:54 PM ETand YouTube ads. In

5:01:56 PM ETGoogle Services, we will continue to lap

5:01:58 PM ETthe search performance acceleration that

5:01:58 PM ETand youtube ads in google services we will continue to lap the

5:02:00 PM ETbegan in the third quarter last year.

5:02:00 PM ETsearch performance acceleration that began in the third quarter last year

5:02:02 PM ETwhich will pressure revenue growth in

5:02:04 PM ETQ4.

5:02:06 PM ETIn Google Cloud, we expect TPU

5:02:06 PM ETwill pressure revenue growth in q4

5:02:08 PM ETsystem sales to continue to scale in

5:02:10 PM ETQ4, followed by meaningful

5:02:10 PM ETin google cloud we expect tpu system sales to continue to scale in q4

5:02:12 PM ETrevenue ramp in

5:02:12 PM ETfollowed by meaningful revenue ramp in twenty twenty seven

5:02:14 PM ET2027. With respect to cloud

5:02:16 PM ETmargins, given the supply constraint

5:02:18 PM ETenvironment, we started utilizing

5:02:18 PM ETwith respect to cloud margins given the

5:02:20 PM ETthird-party compute capacity at the end of

5:02:22 PM ETQ3 and will ramp

5:02:22 PM ETsupply constrained environment we started utilizing third party compute capacity

5:02:22 PM ETat the end of q3

5:02:24 PM ETusage in Q4. We

5:02:26 PM ETexpect margin headwinds in the near

5:02:28 PM ETterm as we utilize this

5:02:28 PM ETwill ramp usage in q4 we expect margin headwinds

5:02:30 PM ETcapacity. Moving to investments, our

5:02:30 PM ETin the near term as we utilize this capacity

5:02:32 PM ETfull year 2026 CapEx guidance

5:02:34 PM ETrange of $195 to

5:02:34 PM ETmoving to investments our full year twenty twenty six capex guidance range

5:02:34 PM ETof one hundred and ninety five million

5:02:36 PM ET$205 billion

5:02:36 PM ETdollars to two zero five billion dollars

5:02:38 PM ETremain unchanged. We

5:02:40 PM ETcontinue to expect our capex to increase

5:02:42 PM ETsignificantly in

5:02:42 PM ETremain unchanged we continue to expect our capex to increase significantly in

5:02:44 PM ET2027, and we'll provide further details on our

5:02:46 PM ETQ4 call. In terms

5:02:46 PM ETtwenty twenty seven and we'll provide further details on our q4 call

5:02:48 PM ETof expenses, the significant increase in

5:02:50 PM ETour investment in technical

5:02:52 PM ETinfrastructure will continue to weigh on the PL and

5:02:52 PM ETterms of expenses the significant increase in our investment in technical infrastructure

5:02:54 PM ETin the form of higher depreciation

5:02:56 PM ETexpenses and related data centers

5:02:56 PM ETwill continue to weigh on the p and l in the form of higher depreciation expenses

5:02:58 PM EToperation costs such as

5:03:00 PM ETenergy. Before we move to

5:03:00 PM ETand related data centers operation costs such as energy

5:03:02 PM ETQA, let me say a few words on how

5:03:04 PM ETwe think about return on invested

5:03:04 PM ETbefore we move to q and a let me say a few words on how we

5:03:06 PM ETcapital and free cash

5:03:06 PM ETthink about return on invested capital and free cash flow

5:03:08 PM ETflow. We're in a unique position to be able

5:03:10 PM ETto monetize our AI

5:03:12 PM ETinvestments across many products that reach billions

5:03:12 PM ETwe're in a unique position to be able to monetize our ai investments

5:03:14 PM ETof users and

5:03:16 PM ETenterprises. Our AI investments in search

5:03:16 PM ETacross many products that reach billions of users and enterprises

5:03:18 PM ETare driving our highest query growth

5:03:20 PM ETin half a decade.

5:03:20 PM ETour ai investments in search are driving our highest

5:03:22 PM ETGoogle Cloud has generated tens of

5:03:24 PM ETbillions of dollars in AI revenues year to

5:03:24 PM ETquery growth in half a decade google

5:03:26 PM ETdate, and Google One is our

5:03:26 PM ETcloud has generated tens of billions of dollars in ai revenues year to date

5:03:28 PM ETfastest-growing subscription

5:03:30 PM ETproducts driven by demand for AI

5:03:30 PM ETand google one is our fastest growing subscription products

5:03:32 PM ETplans. Throughout this

5:03:34 PM ETinvestment cycle, we expect to generate an

5:03:34 PM ETdriven by demand for ai plans throughout this investment cycle we

5:03:36 PM ETattractive ROIC that

5:03:38 PM ETremains meaningfully above our cost of

5:03:38 PM ETexpect to generate an attractive roic that remains meaningfully above our cost of

5:03:40 PM ETcapital and anticipate generating

5:03:42 PM ETstrong cash flow from

5:03:42 PM ETcapital and anticipate generating strong cash flow from operations

5:03:44 PM EToperations. Our sizable investments in

5:03:46 PM ETtechnical infrastructure and

5:03:48 PM ETinventory build of TPU systems as

5:03:48 PM ETour sizable investments in technical infrastructure

5:03:50 PM ETwe scale that business will put

5:03:50 PM ETand inventory build of tpu systems as we scale that business

5:03:52 PM ETpressure on free cash

5:03:52 PM ETwill put pressure on free cash flows

5:03:54 PM ETflows. Q3 was a strong

5:03:56 PM ETquarter, highlighted by our

5:03:58 PM ETinnovation and disciplined

5:03:58 PM ETq3 was a strong quarter highlighted by our innovation

5:04:00 PM ETexecution across all areas of the

5:04:02 PM ETbusiness, including the launch of Gemini

5:04:02 PM ETand disciplined execution across all areas of the business

5:04:04 PM ET4. I want to take this

5:04:06 PM ETopportunity to thank our employees for their

5:04:06 PM ETincluding the launch of gemini four

5:04:08 PM ETcontributions to our performance.

5:04:08 PM ETwant to take this opportunity to thank our employees for their contributions to our performance

5:04:10 PM ETWe will now open the call for

5:04:10 PM ETwe will now open the call for q and a

5:04:12 PM ETQA.

5:04:16 PM ETThank you. As a reminder, to ask a question,

5:04:16 PM ETthank you

5:04:18 PM ETyou will need to press star one on your

5:04:20 PM ETtelephone. To prevent

5:04:22 PM ETany background noise, we ask that

5:04:24 PM ETyou please mute your line once your

5:04:26 PM ETquestion has been stated.

5:04:28 PM ETAnd our first

5:04:30 PM ETquestion comes from Doug

5:04:32 PM ETand our first question comes from doug anmuth

5:04:34 PM ETwith jpmorgan your line is now open

5:04:40 PM ETthanks for taking the questions one for sundar and one for philip

5:04:42 PM ETsundar on a recent podcast you said that

5:04:48 PM ETtwenty twenty seven could be the golden age of productivity for google because of the new cloud

5:04:52 PM ETof models and how they'll integrate with your other products

5:04:56 PM ETi was hoping you could elaborate on that and just share more how you think about the opportunity set

5:04:56 PM ETthere ahead

5:05:02 PM ETand then philip just on query growth the fastest in five years

5:05:04 PM ETbut obviously some deceleration in search growth

5:05:10 PM ETyou just help us understand the disconnect there

5:05:12 PM ETis it simply that you haven't been showing as many ads

5:05:16 PM ETbasically in ai overviews and ai mode and that's

5:05:20 PM ETnow becoming a bigger percentage of the overall queries thanks

5:05:26 PM ETthanks doug

5:05:28 PM ETwe definitely started seeing

5:05:32 PM ETstep change i would say with three point eight flash

5:05:38 PM ETinternally as our models got better at

5:05:40 PM ETlong horizon tasks agent decoding

5:05:44 PM ETetcetera but i'm definitely

5:05:48 PM ETseeing a huge point of inflection

5:05:52 PM ETonce we have started deploying oregon internally

5:05:54 PM ETit is one of the most meaningful internal

5:06:00 PM ETproductivity levers we have seen on software engineering

5:06:02 PM ETdirectly tracking high quality code that's getting checked in the rate at which

5:06:08 PM ETgetting checked in reducing the number of rollbacks

5:06:10 PM ETbut more importantly people are

5:06:16 PM ETtackling massive problems

5:06:20 PM ETas well as we are beginning to see innovation on the

5:06:24 PM ETproduct and the consumer experience side from deploying those models but

5:06:28 PM ETall this happened only about four weeks ago we are now

5:06:32 PM ETdeploying carbon internally and carbon is a meaningful step up from oregon

5:06:36 PM ETso as we roll it out and make it all widely available to googlers with more

5:06:42 PM ETcompute i expect to that's what i was referring to and i

5:06:46 PM ETi'm even more excited since i said that in the podcast and what i would view as

5:06:50 PM ETa golden year ahead in terms of us being able to be more

5:06:54 PM ETproductive and so excited for the innovation ahead philip yes doug on the second part of your

5:06:58 PM ETquestion look in q3 search grew twelve percent year over year to

5:07:02 PM ETsixty three billion euros and when you look at the change in growth rate from q2 to q3

5:07:06 PM ETthere were a few drivers here so first there was fx it was the single largest driver of

5:07:12 PM ETsearch deceleration relative to q2

5:07:14 PM ETsecond there were vertical specific dynamics in

5:07:18 PM ETretail some of our apac advertisers were impacted by the eu eliminating its

5:07:22 PM ETde minimis exemption in media and entertainment spend

5:07:26 PM ETnormalized after an unseasonably strong q2 driven

5:07:30 PM ETby the fifa world cup which had more matches and prefer more

5:07:34 PM ETadvertising budgets allocated in june than actually in july

5:07:38 PM ETit's also to the next part of your question it's worth noting that the strong growth in ai mode

5:07:44 PM ETqueries had an impact because of the lower monetization we are seeing in ai mode

5:07:48 PM ETto be clear we feel very good about our trajectory here our

5:07:52 PM ETad experiments have shown promising results and we're now scaling ads

5:07:56 PM ETwithin both ai overuse and ai mode to more users and more queries

5:07:58 PM ETincluding as i mentioned international markets and multi turnouts in the u s

5:08:02 PM ETthank you

5:08:10 PM ETour next question comes from the line of brian nowak with morgan

5:08:10 PM ETstanley

5:08:14 PM ETyour line is now open thanks for taking my questions i have two maybe

5:08:18 PM ETlet me start with agenteq you rolled out

5:08:24 PM ETspark to some of the users throughout the year

5:08:26 PM ETi know sundar you've talked a lot about shipping more agentic products to the broader base of users

5:08:28 PM ETcan you just help us understand a little bit some of the

5:08:30 PM ETthe key hurdles

5:08:34 PM EThave to overcome technologically to ship out

5:08:38 PM ETagentic products to the full base of users is it model efficiency is it

5:08:42 PM ETcapacity is it the technologies that ready like what do we have to do technologically to do that

5:08:48 PM ETand then the second one just on to doug's question on search

5:08:50 PM ETanant can you help us better understand the search performance comparison

5:08:56 PM ETfactor that you talked about impacting search and how do we think about

5:09:00 PM ETthat as we kind of go into 4q 1q just kind of looking at the compares thanks

5:09:04 PM ETthanks brian first of all on agents

5:09:08 PM ETwhat an exciting and dynamic area i think

5:09:12 PM ETwhen i look at our new models

5:09:16 PM ETand we see the capabilities of

5:09:20 PM ETthe kind of complex tasks they can take in

5:09:22 PM ETin an automated way and execute with very very high quality

5:09:28 PM ETit's clear that the agentic capabilities we have all

5:09:32 PM ETenvisioned are going to be real and so

5:09:34 PM ETi couldn't be more excited to bring the next generation agent experience

5:09:40 PM ETto our users to your question i would say a few things

5:09:42 PM ETi think the gemini four model series has been important we are

5:09:48 PM ETworking on distill versions of our carbon models

5:09:50 PM ETso that we can deploy it across a billion user

5:09:54 PM ETscale efficiently both from a compute from a latency

5:09:58 PM ETstandpoint and from a quality standpoint so that's been

5:10:02 PM ETan important area of investment definitely making sure you need more than

5:10:08 PM ETcapable model at scale you need a robust harness

5:10:10 PM ETyou need a highly secure runtime and this is where gcp comes in

5:10:14 PM ETbeing able to give each and every one of our users

5:10:18 PM ETa secure runtime with twenty fourseven availability in the cloud

5:10:22 PM ETthe right ecosystem of skills and connectors so all of that is the foundation

5:10:28 PM ETwe've been hard at work and we decided to roll it out to enterprise

5:10:32 PM ETfirst which we announced with our gemini agent for enterprise

5:10:36 PM ETessentially because they have the highest bar for security governance and

5:10:40 PM ETcontrols and it's been well received and we are in the process of

5:10:46 PM ETexpanding that later this year we will bring that

5:10:50 PM ETsame experience to consumers obviously starting with the

5:10:52 PM ETone billion users in our gemini app we already see people in the app

5:10:58 PM ETmoving beyond chat to what our task and agent like

5:11:02 PM ETexperiences and so the trust

5:11:06 PM ETusers have in our products the fact that all their context is right there i

5:11:12 PM ETthink sets us up well but we are really focused as a company on

5:11:16 PM ETdelivering that unified experience across our products

5:11:20 PM ETand you should expect to see a lot of progress later this

5:11:20 PM ETyear

5:11:24 PM ETthanks and on the

5:11:28 PM ETsecond question with regards to my comments i believe you're referring to the outlook

5:11:32 PM ETwith regards to search a couple of dynamics that i've mentioned one is on sequential

5:11:38 PM ETrelated to impact of fx headwinds which we saw this

5:11:42 PM ETthis quarter as well versus the previous quarter so

5:11:44 PM ETwhere we are today with current spot rates we expect that this will be

5:11:50 PM ETsomewhat of a headwind to q4 results and then when we look at the year over year

5:11:54 PM ETcomparisons this quarter we started lapping the acceleration we

5:11:58 PM ETsaw in search growth in twenty twenty five the back end of twenty twenty five

5:12:02 PM ETwe left about three points of growth

5:12:06 PM ETversus q3 and then we expect to see that

5:12:10 PM ETcontinue in q4 a couple of points of

5:12:12 PM ETlapping that growth acceleration that we saw at the end of

5:12:16 PM ETat the end of twenty twenty five

5:12:24 PM ETour next question comes from the line of eric sheridan with goldman sachs

5:12:26 PM ETyour line is now open

5:12:30 PM ETthanks so much for taking the question based on some of

5:12:34 PM ETcomments you've made recently sundar on compute constraints can you refresh

5:12:40 PM ETfor us your key priorities for

5:12:42 PM ETthe application of compute as it becomes available across the organization

5:12:48 PM ETand the second part maybe either for you or thomas would be within

5:12:52 PM ETgoogle cloud specifically how does your view continue to evolve with

5:12:56 PM ETrespect to the application compute to cloud

5:12:58 PM ETneeds relative to external tpu thanks so much

5:13:04 PM ETthanks eric i'll start and then thomas will

5:13:06 PM ETanswer the second half

5:13:12 PM ETdefinitely compute allocation is an area where

5:13:16 PM ETwe are spending a lot of time on because of

5:13:18 PM ETthe forward planning that's needed i would start with saying we feel

5:13:24 PM ETfortunate to have an extraordinary set of opportunities ahead of us

5:13:26 PM ETwe are operating a scale consumer business that reaches billions of people

5:13:32 PM ETa hyperscale cloud business with a lot of momentum

5:13:34 PM ETand of course a range of other highly important areas across google and alphabet

5:13:38 PM ETwe always take a long range planning and so

5:13:44 PM ETeach of our business areas submit long range

5:13:48 PM ETplans with a very disciplined roic framework

5:13:50 PM ETand so i would broadly think of all of this as a long range

5:13:54 PM ETplanning process not a short term allocation exercise

5:14:00 PM ETobviously our ai models underlie everything we do from search and youtube to

5:14:04 PM ETcloud to waymo and so on so it's the foundation and

5:14:08 PM ETwe view it as a highly leveraged

5:14:12 PM ETapproach the full stack approach

5:14:16 PM ETdrives a lot of efficiency as well as gives us a lot of

5:14:20 PM EToptionality across a wide variety of opportunities

5:14:22 PM ETso obviously that is the foundation and we plan for that for the long term too

5:14:28 PM ETand about that as i said we really look at our businesses

5:14:32 PM ETand really take a disciplined long range planning approach

5:14:34 PM ETto us obviously

5:14:38 PM ETwe are seeing very very robust demand signals across a wide

5:14:42 PM ETvariety of areas and so we view it as an extraordinary

5:14:46 PM ETset of opportunities and we are investing for that moment thomas

5:14:50 PM ETeric on the demand in cloud

5:14:54 PM ETwe are always managing our portfolio to ensure we have a breadth of

5:15:00 PM ETcustomers and we have different types of customers from digital natives

5:15:04 PM ETai labs traditional enterprises

5:15:08 PM ETfinancial and capital markets customers so it's a broad set

5:15:12 PM ETof customers many of them have either started with

5:15:16 PM ETusing tpu systems for example in our cloud

5:15:22 PM ETwe have over one hundred significant tpu customers

5:15:26 PM ETwho start in our cloud they learn to use tpus

5:15:30 PM ETand remember tpu system is a combination of hardware

5:15:34 PM ETsystem software compilers

5:15:38 PM ETthe whole thing assembled as a system including optical networking and other things

5:15:44 PM ETwe've taken a very deliberate approach on how we're bringing tpus to

5:15:48 PM ETthe first one is to build the right silicon so we have eight

5:15:52 PM ETfor training ai for inference they both offer

5:15:56 PM ETsubstantial price performance and performance

5:16:00 PM ETimprovements relative to prior generations

5:16:02 PM ETsecond we've optimized the software

5:16:06 PM ETthe optimization of the software is not just to make

5:16:08 PM ETthe systems run better and to give you great performance but also to

5:16:14 PM ETmake portability seamless recently we introduced

5:16:18 PM ETfor example pytorch which opens up a much larger market for us

5:16:20 PM ETby optimizing pytorch on the tpu

5:16:24 PM ETthird we are broadening distribution so in the past

5:16:30 PM ETyou could use tpus in our cloud now we make for customers who want

5:16:34 PM ETtpu systems in their data centers we're making tpu systems available

5:16:38 PM ETin their data centers we're also broadening availability to

5:16:42 PM ETother providers for example we announced a partnership with oracle

5:16:46 PM ETand we've launched along with blackstone an agreement

5:16:50 PM ETfor crux ai a joint venture to offer customers

5:16:52 PM ETtpu systems we are seeing extraordinary demand

5:16:56 PM ETand strong growth from both

5:17:02 PM ETai labs capital markets financial

5:17:06 PM ETservice institutions and enterprises and if you imagine these one hundred

5:17:10 PM ETplus customers growing to many more and also scaling their use

5:17:14 PM ETin into tpu systems we see strong demand over the future

5:17:24 PM ETour next question comes from ross sandler with barclays

5:17:26 PM ETyour line is now open

5:17:32 PM ETthanks i wanted to stick with the tpu

5:17:34 PM ETresale business topic

5:17:40 PM ETthat you just mentioned so last quarter kind of said that

5:17:44 PM ETthe revenue will be recognized on

5:17:48 PM ETsystem sales i believe if a gigawatt

5:17:48 PM ETof tpu

5:17:52 PM ETfull system costs around thirty five billion

5:17:56 PM ETfor the compute and all the kind of components around it

5:18:00 PM ETwhat percent of that are we actually recognizing as revenue

5:18:04 PM ETand what kind of flow through does that have

5:18:08 PM ETin terms of margin profile relative to

5:18:12 PM ETwhere the google cloud segment currently sits and then

5:18:16 PM ETi guess the second part of the question would be

5:18:20 PM ETas you help companies like anthropic bring on

5:18:22 PM ETmore of their own kind of first party infrastructure

5:18:26 PM ETwhat does that mean for the capex intensity of alphabet

5:18:28 PM ETyou start to kind of hand off some of the

5:18:30 PM ETfuture capex

5:18:32 PM ETto the ai labs and may not require

5:18:34 PM ETas much capex intensity

5:18:40 PM ETon your own in your own data centers thoughts on that would be great thank you

5:18:44 PM ETyeah let me combine these two questions

5:18:48 PM ETfirst obviously don't provide specific financials at the product

5:18:52 PM ETlevel but let me walk through how to think about the financials

5:18:58 PM ETassociated with the tpu system sales and obviously just as a reminder

5:19:02 PM ETand thomas can elaborate on that as well a tpu system is not just

5:19:06 PM ETa chip but it's a specialized system that combines multiple things in

5:19:10 PM EThardware compute storage networking

5:19:14 PM ETso we are selling much more than just a chip

5:19:16 PM ETsoftware training inference there is a

5:19:22 PM ETsignificant added value that's being provided to the customer that's purchasing a t

5:19:26 PM ETsystem and the way this works and this is linked to your second portion of the

5:19:32 PM ETwhen we sign an agreement with the

5:19:34 PM ETcustomer to sell tpu systems

5:19:38 PM ETonce we sign that agreement you'll first see that showing up in our backlog

5:19:42 PM ETand then we start building inventory

5:19:46 PM ETthink about this from a capex versus cash perspective

5:19:48 PM ETwe build the inventory ahead of getting the revenue stream in

5:19:54 PM ETso i think i've alluded to this on the previous call

5:19:56 PM ETyou see inventory balances go up or previously

5:20:00 PM ETinventory balances primarily were some of our hardware products

5:20:04 PM ETnow those tpu systems that are not yet sold will be showing up so you see cash

5:20:08 PM ETgoing out and we're building up the inventory base

5:20:12 PM ETand then as we start

5:20:16 PM ETdelivering and billing customers for these tpu systems

5:20:20 PM ETthen you see cash coming in so it's not the traditional kind of capex

5:20:24 PM ETas you're thinking about this and i think that's the essence of your question

5:20:28 PM ETbut rather a different cycle of cash flow generation

5:20:32 PM ETnow given that we are in the process of building up and ramping up that business

5:20:36 PM ETyou're seeing that cash needs

5:20:40 PM ETin our cash flow statements

5:20:44 PM ETand you see those inventory build

5:20:46 PM ETto date we've had some system sales recorded as revenue but

5:20:52 PM ETexpect more of it to come in through twenty twenty seven

5:21:06 PM ETour next question comes from the line of justin post

5:21:06 PM ETwith bank of america

5:21:08 PM ETyour line is now open

5:21:14 PM ETgreat thank you i have some questions on the agenic

5:21:16 PM ETtransition you did launch the universal shopping cart

5:21:20 PM ETclearly people are liking and using agenic use cases

5:21:24 PM ETas more of that happens how do you plan to manage the

5:21:28 PM ETsearch transition to agentic use do you think that could be a headwind

5:21:32 PM ETand second do you think agentic could actually drive more monetization

5:21:34 PM ETfor your traffic over the long term

5:21:38 PM ETand then i'll go again at tpu margins we're wondering

5:21:42 PM ETdo you think this is a really good business for you and maybe why if you can't give

5:21:48 PM ETus the margins you think this is a good way to use all this capacity thank you

5:21:52 PM ETjustin we're near

5:21:56 PM ETearly days of agendaq but maybe directly to the second part of your question but we view agendaq

5:22:00 PM ETagendaq experiences as clearly additive our focus is build it

5:22:04 PM ETtrusted secure super scale experiences that users love

5:22:08 PM ETanchored frankly on a generic commerce foundational blocks like

5:22:12 PM ETmentioned at universal commerce the card the agents payments protocols and so on

5:22:14 PM ETwhen users shop and that's important there's

5:22:22 PM ETto enhance not just

5:22:24 PM ETautomate away the discovery experience

5:22:26 PM ETthat people frankly love in so many

5:22:26 PM ETto enhance not just automate away the discovery experience that

5:22:28 PM ETcases. Users like

5:22:30 PM ETchoice, they want to consider things way

5:22:30 PM ETpeople frankly love in so many cases users like

5:22:32 PM ETbeyond price like brand

5:22:34 PM ETloyalty, customer service, and so on.

5:22:34 PM ETchoice and they want to consider things way beyond price like brand loyalty loyalty

5:22:36 PM ETwe're really architecting these agentic

5:22:38 PM ETworkflows in a way that we can

5:22:38 PM ETcustomer service and so on so we're really architecting these agentic workflows

5:22:40 PM ETintentionally design the time and

5:22:42 PM ETthe space for users to explore

5:22:42 PM ETa way that we can intentionally design the time

5:22:44 PM ETthose choices. So, simple

5:22:46 PM ETexample: an agent could

5:22:46 PM ETand the space for users to explore those choices

5:22:48 PM ETrespond: here's the insurance product we

5:22:48 PM ETso simple example an agent could respond here's

5:22:50 PM ETfound and here's another strong

5:22:52 PM EToption, one of which is a higher

5:22:52 PM ETinsurance product we found and here's another strong option

5:22:54 PM ETpremium, maybe a little deductible, and they're

5:22:56 PM ETcurrently offering a very special

5:22:56 PM ETone of which is a higher premium maybe a lower deductible and they are currently offering a very

5:22:58 PM ETdiscount. So, given that there's

5:23:00 PM ETalmost never only one perfect option for

5:23:00 PM ETspecial discount so given that there's almost never only one

5:23:02 PM ETa decision, we're actually confident that we

5:23:04 PM ETcan actually provide really

5:23:06 PM ETinteresting, compelling advertising models for

5:23:06 PM ETperfect option for a decision we're actually confident that we can actually provide really interesting

5:23:08 PM ETagenda use cases. And obviously, you can

5:23:10 PM ETimagine going beyond those with

5:23:10 PM ETcompelling advertising models for agentic use cases and obviously you can imagine going beyond the

5:23:12 PM ETsubscriptions and other payment methods.

5:23:14 PM ETBut I'm really actually excited about the advertising.

5:23:14 PM ETwith subscriptions and other payment methods but i'm really actually

5:23:16 PM ETopportunities in an Agentic

5:23:18 PM ETworld. Overall, I would add and

5:23:18 PM ETabout the advertising opportunities in an agentic world overall i would

5:23:20 PM ETsay: look, I would view it

5:23:20 PM ETwould add and say

5:23:22 PM ETas a you know

5:23:24 PM ETexpansion opportunity,

5:23:24 PM ETlook i would view it as

5:23:26 PM ETright? Anytime, just like a few

5:23:28 PM ETyears ago, being able to

5:23:28 PM ETexpansion opportunity right anytime just like a few years ago

5:23:30 PM ETincorporate AI chat use

5:23:32 PM ETcases in search allowed

5:23:32 PM ETbeing able to incorporate ai chat use cases

5:23:34 PM ETpeople to ask more and do more in

5:23:36 PM ETsearch, which is what is playing

5:23:36 PM ETin search allowed people to ask more and do more in search which is what

5:23:38 PM ETout in our usage.

5:23:40 PM ETI think agents would,

5:23:40 PM ETis playing out in our usage

5:23:42 PM ETagain be an opportunity to expand

5:23:44 PM ETthe types of problems we solve for our

5:23:44 PM ETi think agents would again be an opportunity to expand the types

5:23:46 PM ETusers. I think people always

5:23:48 PM ETunderestimate all the

5:23:48 PM ETof problems we solve for our users i think people always underestimate

5:23:50 PM ETcomponents you need to deliver these agent

5:23:52 PM ETthese complex

5:23:52 PM ETall the components you need to deliver these

5:23:54 PM ETexperiences for users.

5:23:56 PM ETAnd search on the surface looks

5:23:56 PM ETcomplex experiences for users and search on the surface look

5:23:58 PM ETsimple, but there's

5:24:00 PM ETextraordinary deep.

5:24:02 PM ETlevels of

5:24:02 PM ETsimple but there's extraordinary deep

5:24:04 PM ETinvestment and things we have built over

5:24:06 PM ETtime, like knowledge graph being an

5:24:06 PM ETlevels of investment and things we have built over time

5:24:08 PM ETexample, and so our understanding of the

5:24:10 PM ETworld, etc. So, I

5:24:10 PM ETlike knowledge graph being an example and so our understanding of the world etcetera

5:24:12 PM ETthink I'm excited about agents, and I

5:24:14 PM ETwould view it as a new frontier, which

5:24:14 PM ETso i think i'm excited about agents and i would view it as a new frontier which

5:24:16 PM ETbrings in a new set

5:24:18 PM ETof use cases to search.

5:24:20 PM ETand so we are excited to work through.

5:24:20 PM ETbrings in a new set of use cases to search

5:24:22 PM ETwork through that.

5:24:24 PM ETOn TPU systems

5:24:24 PM ETand so we are excited to work through that

5:24:26 PM ETand margins, we don't break out

5:24:26 PM ETon tpu systems and margins

5:24:28 PM ETmargins for individual

5:24:30 PM ETproducts or infrastructure

5:24:30 PM ETwe don't break out margins for individual products or infrastructure

5:24:32 PM ETcomponents. However, we

5:24:34 PM ETsee several benefits

5:24:34 PM ETinfrastructure components however we see several benefits

5:24:36 PM ETfrom designing and

5:24:36 PM ETfrom designing and manufacturing

5:24:38 PM ETmanufacturing TPU systems.

5:24:40 PM ETSo, first of all, to remind you, the TPU

5:24:42 PM ETsystem is not just a chip, but it's a

5:24:42 PM ETtpu systems so first of all to remind you the tpu system is not just

5:24:44 PM ETcomplete system.

5:24:46 PM ETGiven the price performance

5:24:46 PM ETa chip but it's a complete system given the

5:24:48 PM ETbenefits of TPU systems.

5:24:50 PM ETrelative to competitors.

5:24:50 PM ETprice performance benefits of tpu systems

5:24:52 PM ETgrowing demand

5:24:54 PM ETfor accelerated

5:24:54 PM ETrelative to competitors growing demand for

5:24:56 PM ETcompute and the scarcity of accelerated

5:24:58 PM ETcompute. We're very focused

5:24:58 PM ETaccelerated compute and the scarcity of accelerated compute

5:25:00 PM ETon continuing to improve margins as we

5:25:02 PM ETscale the business. Two

5:25:02 PM ETwe're very focused on continuing to improve margins as we scale the business

5:25:04 PM ETadditional points. First of

5:25:06 PM ETall, when Anat mentioned

5:25:06 PM ETtwo additional points first of all

5:25:08 PM ETthat we procure

5:25:10 PM ETcomponents to build

5:25:10 PM ETwhen anat mentioned that we procure components

5:25:12 PM ETinventory. We only do that

5:25:14 PM ETafter we have signed contracts. So all

5:25:14 PM ETto build inventory we only do that

5:25:16 PM ETof the inventory

5:25:18 PM ETthat for TPU systems are associated

5:25:18 PM ETafter we have signed contracts so all of the inventory

5:25:20 PM ETwith committed contracts.

5:25:22 PM ETAdditionally, by putting TPU

5:25:22 PM ETthat for tpu systems are associated with committed contracts

5:25:24 PM ETsystems in third-party

5:25:26 PM ETfacilities and in customer data

5:25:26 PM ETadditionally by putting tpu systems in

5:25:28 PM ETcenters. It allows us, one, to

5:25:28 PM ETthird party facilities and in customer data centers

5:25:30 PM ETbroaden our addressable markets.

5:25:32 PM ETFor some industries, they have

5:25:32 PM ETit allows us one to broaden our addressable markets some

5:25:34 PM ETdata in their premise that they cannot

5:25:36 PM ETmove to a cloud, and it allows us to

5:25:36 PM ETfor some industries they have data in their premise

5:25:38 PM ETbring accelerated compute to where

5:25:40 PM ETthe data sets.

5:25:40 PM ETthat they cannot move to a cloud and it allows us to bring accelerated compute

5:25:42 PM ETAdditionally, it also offsets the need for

5:25:44 PM ETus to have to

5:25:44 PM ETto where the data sets additionally it also offsets the need for us

5:25:46 PM ETpre-build facilities to host

5:25:48 PM ETthese TPU systems.

5:25:48 PM ETto have to pre build facilities to host

5:25:50 PM ETthese tpu systems

5:25:54 PM ETOur

5:25:56 PM ETnext question comes from the line of

5:25:58 PM ETMark Schmullek with Alliance

5:25:58 PM ETour next question comes from the line of mark

5:26:00 PM ETBernstein. Your line is now

5:26:02 PM ETopen. Yes, thanks

5:26:02 PM ETschmullich with alliance bernstein your line is now open

5:26:04 PM ETfor taking the questions.

5:26:06 PM ETI'm going to be on the same couple topics, but Philip,

5:26:06 PM ETyes thanks for taking the questions

5:26:08 PM ETjust on Agentic, you

5:26:10 PM ETknow, if you've kind of just taken a look at some of the early

5:26:10 PM ETi'm going be on the same couple of topics but philip just on agenteq

5:26:12 PM ETmore agentic-style use cases on

5:26:14 PM ETGemini AI mode and kind of observing some

5:26:14 PM ETas you've kind of just taken a look at some of the early more agentic style

5:26:16 PM ETof the 3P integrations with

5:26:16 PM ETuse cases on gemini ai mode and and kind of observing some of the

5:26:18 PM ETGoogle properties. Anything that's

5:26:20 PM ETsurprised you or changed your view? I know we've talked a lot

5:26:20 PM ETthree p integrations with google properties anything that's

5:26:22 PM ETabout commerce, but perhaps on like consumer

5:26:24 PM ETproductivity as well.

5:26:24 PM ETyou or changed your view i know we've talked a lot about commerce but perhaps on like consumer product

5:26:26 PM ETAnd then either Thomas or not, just on

5:26:28 PM ETuncomfortable cloud growth, is there anything you

5:26:28 PM ETproductivity as well and then either thomas or not just on kind of cloud growth

5:26:30 PM ETcould share just around the rate of growth

5:26:32 PM ETof Google Cloud excluding the TPU

5:26:32 PM ETis there anything you could share just around the rate of growth of google cloud

5:26:34 PM ETsystem sales, and just how much of that's been kind

5:26:36 PM ETof constrained by capacity

5:26:36 PM ETexcluding the tpu system sales and just how much of that's been kind of constrained by capacity

5:26:38 PM ETand as some of that 3P

5:26:40 PM ETcapacity comes online in Q4, is it fair

5:26:42 PM ETto kind of assume an

5:26:42 PM ETcapacity and as some of that 3p capacity comes online in q4 is it fair to kind of

5:26:44 PM ETacceleration of that Google Cloud business? Thank

5:26:46 PM ETyou. Look, you know, maybe I'll take

5:26:46 PM ETassume an acceleration of that google cloud business thank you

5:26:48 PM ETthe first part on

5:26:50 PM ETw what we are observing on these early

5:26:50 PM ETlook maybe i'll take the first part on what

5:26:52 PM ETagentic style use cases.

5:26:54 PM ETWe are definitely seeing it. Just

5:26:54 PM ETare observing on these early agent style use cases

5:26:56 PM ETrecently in the Gemini app, we

5:26:58 PM EThave added a task tab and where

5:26:58 PM ETwe are definitely seeing it just recently in the gemini app

5:27:00 PM ETpeople are actually typing.

5:27:02 PM ETin agentic queries and so we can observe

5:27:02 PM ETwe have added a task tab and where people are actually typing and

5:27:04 PM ETthat. Look,

5:27:04 PM ETagentic queries and so we can observe that look

5:27:06 PM ETin all of this,

5:27:08 PM ETusers really want, and they want to

5:27:10 PM ETbe delighted, so they effectively

5:27:10 PM ETin all of this users really want

5:27:12 PM ETare, you know, looking for, they have a

5:27:14 PM EThigh bar. What in

5:27:14 PM ETthey want to be delighted so they effectively are looking for

5:27:16 PM ETrepeat use

5:27:18 PM ETis your ability

5:27:18 PM ETthey have a high bar what in courageous repeat use is

5:27:20 PM ETto more handle

5:27:22 PM ETthese longer tasks and

5:27:22 PM ETyour ability to more handle these

5:27:24 PM ETactually deliver a

5:27:26 PM ETsatisfactory outcome at the end.

5:27:26 PM ETlonger tasks and actually deliver a satisfactory outcome at the end

5:27:28 PM ETYou know, just like having seen this in

5:27:30 PM ETother areas, including coding, et

5:27:32 PM ETcetera, when you're doing it at

5:27:32 PM ETjust like having seen this in other areas including coding etcetera

5:27:34 PM ET70%, it looks like it doesn't

5:27:34 PM ETwhen you're doing it at seventy percent

5:27:36 PM ETwork until you cross a certain

5:27:38 PM ETthreshold and it really takes

5:27:38 PM ETlooks like it doesn't work until you cross a certain threshold

5:27:40 PM EToff. And so

5:27:42 PM ETwe are closely.

5:27:42 PM ETand it really takes off and so we are

5:27:44 PM ETlearning from all of it.

5:27:46 PM ETyou know, identifying gaps,

5:27:46 PM ETclosely learning from all of it

5:27:48 PM ETlaying all the foundations I talked about

5:27:50 PM ETearlier, and I'm

5:27:50 PM ETidentifying gaps laying all the foundations i talked about earlier

5:27:52 PM ETconfident it's an area where you will see a lot of

5:27:54 PM ETprogress across our

5:27:54 PM ETand i'm confident it's an area where you will see a lot

5:27:56 PM ETsurfaces as we go

5:27:58 PM ETthrough later this year in 27.

5:27:58 PM ETprogress across our surfaces as we go through later

5:28:00 PM ETThe one thing as we

5:28:02 PM ETtransition to TPUs, one of the

5:28:02 PM ETyear in 'twenty seven the one thing as we transition to tpus

5:28:04 PM ETcomment I would make on the TPU systems

5:28:06 PM ETis: look, in the semiconductor industry,

5:28:08 PM ETeconomies of scale matter.

5:28:08 PM ETone other comment i would make on the tpu systems is look in the semiconductor industry economies

5:28:10 PM ETSo I would look at our ability to

5:28:12 PM ETdo, you know,

5:28:12 PM ETof scale matter so i would look at our ability to do

5:28:14 PM ETuh TPUs

5:28:16 PM ETboth for

5:28:16 PM ETtpus both for

5:28:18 PM ETdeploying

5:28:18 PM ETdeploying

5:28:20 PM ETin our data centers and TPU systems and

5:28:22 PM ETothers' data centers as an

5:28:24 PM ETability to scale up

5:28:24 PM ETin our data centers and tpu systems and others data centers as an

5:28:26 PM ETthe product, which directly translates

5:28:28 PM ETinto better

5:28:28 PM ETability to scale up the product which

5:28:30 PM ETeconomics.

5:28:30 PM ETdirectly translates into better economics

5:28:32 PM ETTo address the question on the

5:28:34 PM ETrate of growth of

5:28:34 PM ETto address the question on the rate of growth of cloud

5:28:36 PM ETcloud, you know, a Google Cloud

5:28:38 PM ETsegment delivered record growth in the

5:28:40 PM ETthird quarter with

5:28:40 PM ETour google cloud segment delivered record growth

5:28:40 PM ETin the third quarter

5:28:42 PM ETcloud revenues up one hundred and three

5:28:44 PM ETpercent to 30 point seven

5:28:44 PM ETwith cloud revenues up one hundred and three percent to thirty point seven billion

5:28:46 PM ETbillion. The strongest growth

5:28:48 PM ETcontributor was Google Cloud

5:28:50 PM ETPlatform, with strength in both

5:28:50 PM ETthe strongest growth contributor was google cloud platform with strength in both

5:28:52 PM ETAI and core

5:28:54 PM ETGCP. So GCP

5:28:54 PM ETai and core gcp

5:28:56 PM ETaccounted for the vast majority of

5:28:58 PM ETrevenue generated in Q3.

5:28:58 PM ETso gcp accounted for the vast majority of revenue generated in

5:29:00 PM ETand cloud revenue

5:29:02 PM ETgrowth, excluding the impact of TPU

5:29:02 PM ETq3 and cloud revenue growth

5:29:04 PM ETsystem sales, was again very, very

5:29:06 PM ETstrong. Demand

5:29:06 PM ETexcluding the impact of tpu system sales was again very very strong

5:29:08 PM ETwas also strong across

5:29:10 PM ETour product lines. So

5:29:10 PM ETdemand was also strong across our product lines

5:29:12 PM ETwe see three things going

5:29:14 PM ETon. One, as people

5:29:14 PM ETso we see three things going on one

5:29:16 PM ETuse Gemini, we see revenue growth from

5:29:18 PM ETGemini itself, but

5:29:18 PM ETas people use gemini we see revenue growth from gemini itself

5:29:20 PM ETwe also see that as people use

5:29:22 PM ETGemini, it drives growth in other

5:29:24 PM ETparts of our portfolio.

5:29:24 PM ETbut we also see that as people use gemini it drives growth in

5:29:26 PM ETFor instance, as they build agents, they

5:29:28 PM ETuse compute to perform

5:29:28 PM ETother parts of our portfolio for instance as they build agents

5:29:30 PM ETa task, and so they

5:29:32 PM ETbuy compute, they use

5:29:32 PM ETthey use compute to perform a task

5:29:34 PM ETstorage to store memories of

5:29:34 PM ETand so they buy compute they use storage

5:29:36 PM ETthe agent, database

5:29:38 PM ETto serve up what's called context for

5:29:38 PM ETto store memories of the agent database to serve up what's

5:29:40 PM ETthe agent to reason on.

5:29:42 PM ETAnd obviously, given all this

5:29:42 PM ETcalled context for the agent to reason on and obviously given all this

5:29:44 PM ETcybersecurity threats, we're seeing strong growth in

5:29:46 PM ETour cybersecurity

5:29:48 PM ETportfolio. So we are confident

5:29:48 PM ETcybersecurity threats we are seeing strong growth in our cybersecurity portfolio

5:29:50 PM ETabout our growth and

5:29:52 PM ETthe diversity in our

5:29:52 PM ETso we are confident about our growth and the

5:29:54 PM ETgrowth. We're also seeing

5:29:56 PM ETnew customers buy into

5:29:56 PM ETdiversity in our growth we're also seeing new customers

5:29:58 PM ETGCP and accelerating

5:30:00 PM ETgrowth in new customer

5:30:00 PM ETbuy in to gcp and accelerating growth in new customer

5:30:02 PM ETadoption. So we also see that as a growth vector for

5:30:04 PM ETus, and we're

5:30:06 PM ETare always at any point in

5:30:06 PM ETso we also see that as a growth vector for us

5:30:08 PM ETtime, we could obviously serve a lot

5:30:08 PM ETand we are always at any point in time

5:30:10 PM ETmore demand with capacity, but we have

5:30:12 PM ETare managing all of these parameters and

5:30:12 PM ETwe could obviously serve a lot more demand with capacity but we are managing

5:30:14 PM ETgrowth carefully.

5:30:14 PM ETall of these parameters and growth carefully

5:30:22 PM ETAnd our last question comes from

5:30:24 PM ETMichael Nathanson with

5:30:26 PM ETMoffitt Nathanson. Your line is

5:30:26 PM ETand our last question comes from michael

5:30:28 PM ETnow open.

5:30:28 PM ETwith moffettnathanson your line is now open

5:30:30 PM ETThanks. I have two for

5:30:30 PM ETthanks i have two for sundar

5:30:32 PM ETSundar. In an early answer

5:30:34 PM ETto Eric's question

5:30:36 PM ETabout CPU and

5:30:36 PM ETin an early answer to eric's question about

5:30:36 PM ETtpu and cloud

5:30:38 PM ETcloud. We heard a

5:30:40 PM ETwillingness to work with some of your

5:30:42 PM ETcompetitors or companies that we may view as

5:30:42 PM ETwe heard a willingness to work with some of your

5:30:44 PM ETcompetitors in models and

5:30:44 PM ETcompetitors or companies that we may view as competitors

5:30:46 PM ETcloud. So, I wonder how you think

5:30:46 PM ETin models and cloud

5:30:46 PM ETwonder how you think about

5:30:48 PM ETabout the philosophy here about work with people that

5:30:50 PM ETmaybe are competing with

5:30:50 PM ETthe philosophy here about work with people that maybe

5:30:52 PM ETyou and why this strategy is an

5:30:54 PM ETopportunity, not a risk, longer term that makes

5:30:54 PM ETcompeting with you and why the strategy is an opportunity not a risk

5:30:56 PM ETsense. And then, secondly, on

5:30:56 PM ETlonger term that makes sense

5:30:58 PM ETejectic consumer behavior. One of

5:31:00 PM ETthe debates is around

5:31:00 PM ETand then secondly on agentic consumer behavior one of the debates

5:31:02 PM ETsafety and trust. If you share with us

5:31:04 PM ETwhat you've learned so far

5:31:04 PM ETis around safety and trust can you share with us what you've learned so far

5:31:04 PM ETabout

5:31:06 PM ETabout safety and trust? Are those

5:31:08 PM ETvariables that matter to consumers?

5:31:10 PM ETand how do you think if those variables

5:31:10 PM ETsafety and trust are those variables that matter to consumers

5:31:12 PM ETmatter, what are you doing to

5:31:12 PM ETand how do you think if those variables matter

5:31:12 PM ETwhat are you doing to

5:31:14 PM ETprotect consumer trust and privacy as

5:31:16 PM ETyou build out Agenti Consumer

5:31:18 PM ETTools. Thanks.

5:31:18 PM ETprotect consumer trust and privacy as you build out

5:31:20 PM ETThanks, Michael. Great questions.

5:31:20 PM ETthe consumer tools thanks thanks michael great

5:31:22 PM ETOn the first one, look, you

5:31:24 PM ETknow, a long

5:31:24 PM ETquestions on the first one look

5:31:26 PM ETviewed uh what we built as you

5:31:28 PM ETcategories of things which we build as

5:31:28 PM ETi've long viewed what we build as

5:31:30 PM ETproducts, and then there are

5:31:32 PM ETcategories of things which we build as platforms,

5:31:32 PM ETcategories of things which we build as products and then

5:31:34 PM ETright? And you need to have a clear

5:31:36 PM ETview on which is which.

5:31:36 PM ETthere are categories of things which we build as platforms right and you need to have a

5:31:38 PM ETAnd when you're building things as

5:31:40 PM ETplatforms you really work

5:31:40 PM ETclear view on which is which and when you're building things as platforms

5:31:42 PM EThard to make the platform great and make

5:31:44 PM ETsure it works for everyone.

5:31:44 PM ETyou really work hard to make the platform great and make sure it works

5:31:46 PM ETwe have a long history of operating many

5:31:48 PM ETplatforms, be it a browser, be it

5:31:48 PM ETfor everyone we have a long history of operating many

5:31:50 PM ETan operating system, be it a

5:31:52 PM ETextraordinarily successful

5:31:52 PM ETplatforms be it a browser be it operating system be it

5:31:54 PM ETcloud platform, and in

5:31:56 PM ETall of that,

5:31:56 PM ETextraordinarily successful cloud platform

5:31:58 PM ETyou know, we worked hard to provide it

5:32:00 PM ETto all players.

5:32:00 PM ETand in all of that we worked hard to provide it to

5:32:02 PM ETand be a great

5:32:04 PM ETcustomer. You've seen us

5:32:04 PM ETall players and be a great customer

5:32:06 PM ETwork on our models

5:32:08 PM ETwith other companies, including

5:32:08 PM ETyou've seen us work on our models with other

5:32:10 PM ETApple, with whom we compete in many, many

5:32:12 PM ETareas.

5:32:12 PM ETcompanies including apple with whom we compete in many many areas

5:32:14 PM ETSimilarly, we view our TPU systems as

5:32:16 PM ETan important platform we are

5:32:16 PM ETsimilarly we view our tpu systems as an

5:32:18 PM ETbuilding. We want to be

5:32:20 PM ETworld-class in how we build it. That

5:32:20 PM ETimportant platform we are building we want to be world class

5:32:22 PM ETmeans you're putting it

5:32:24 PM ETagainst market forces.

5:32:24 PM ETin how we build it that means your

5:32:26 PM ETtesting it, and competing

5:32:28 PM ETfor people who would have the

5:32:28 PM ETputting it against market forces testing it

5:32:30 PM ETmost demanding workloads on top of

5:32:30 PM ETand competing for people who would have the most demanding workloads on top

5:32:32 PM ETit. And I think that's

5:32:34 PM ETwhat makes sure.

5:32:36 PM ETthose teams are

5:32:36 PM ETof it and i think that's what makes sure

5:32:38 PM ETcontinuing to be at the

5:32:40 PM ETfrontier. So that's how we think about it.

5:32:40 PM ETthose teams are continuing to be at the frontier

5:32:42 PM ETAnd over time, I think that

5:32:44 PM ETalways ends up in

5:32:44 PM ETso that's how we think about it and over time i think that

5:32:46 PM ETexpanding opportunities

5:32:48 PM ETand we've been able to do well

5:32:48 PM ETalways ends up in expanding opportunities

5:32:50 PM ETtaking that approach over a very long

5:32:52 PM ETperiod of time.

5:32:52 PM ETand we've been able to do well taking that

5:32:54 PM ETSecond,

5:32:54 PM ETapproach over a very long period of time

5:32:56 PM ETon agentic behavior.

5:32:58 PM ETI mean, to the earlier

5:32:58 PM ETsecond on agentic behavior

5:33:00 PM ETquestion, and I think I should have

5:33:00 PM ETi mean to the earlier question and i think i should have emphasized

5:33:02 PM ETemphasized, people really, when people are

5:33:04 PM ETapproaching these agentic

5:33:06 PM ETexperiences, we clearly

5:33:06 PM ETpeople really when people are approaching these agentic experiences

5:33:08 PM ETsee people are concerned

5:33:08 PM ETwe clearly see people are concerned about

5:33:10 PM ETabout safety and

5:33:12 PM ETtrust. I mean, it is a very,

5:33:14 PM ETvery high

5:33:14 PM ETsafety and trust i mean it is a very very

5:33:16 PM ETfactor in in how they want to use

5:33:18 PM ETthese things

5:33:18 PM EThigh factor in how they want to use these things

5:33:20 PM ETAnd over

5:33:20 PM ETand over

5:33:22 PM ETdecades, the trust we have

5:33:24 PM ETbuilt by delivering these

5:33:26 PM ETexperiences in a safe way

5:33:26 PM ETdecades the trust we have built by delivering these experiences

5:33:28 PM ETacross many consumer products with

5:33:30 PM ETscale in billions of

5:33:30 PM ETin a safe way across many consumer products with scale

5:33:33 PM ETusers, as well as this putting

5:33:35 PM ETin

5:33:35 PM ETin billions of users as well as this product putting in

5:33:38 PM ETend-to-end as we are building our models and

5:33:40 PM ETagents, deploying them both with

5:33:40 PM ETend to end as we are building our models and agents

5:33:42 PM ETextensive testing, the kind

5:33:44 PM ETof real-time monitoring we do as

5:33:44 PM ETdeploying them both with extensive testing

5:33:46 PM ETwe deploy these at scale.

5:33:46 PM ETthe kind of real time monitoring we do as we deploy these at scale

5:33:48 PM ETI think all of that will end up being a huge

5:33:50 PM ETdifferentiator, and we already

5:33:52 PM EThear about it from our users.

5:33:52 PM ETi think all of that will end up being a huge differentiator

5:33:54 PM ETand I think they would vote with their

5:33:54 PM ETand we already hear about it from our users and i think they would

5:33:56 PM ETfeet as this frontier goes

5:33:58 PM ETthrough.

5:33:58 PM ETwork with their feet as this frontier goes through

5:34:00 PM ETThank

5:34:00 PM ETthank you

5:34:02 PM ETyou.

5:34:04 PM ETThank you.

5:34:04 PM ETthank you

5:34:06 PM ETAnd that concludes our question and answer

5:34:08 PM ETsession for today. I'd like to turn the

5:34:10 PM ETconference back over to Jim

5:34:10 PM ETand that concludes our question and answer session for today

5:34:12 PM ETFriedman for any further

5:34:12 PM ETi'd like to turn the conference back over to jim friedland for any further remarks

5:34:14 PM ETremarks. Thanks, everyone, for joining us

5:34:16 PM ETtoday. We look forward to speaking with you

5:34:18 PM ETagain on our fourth quarter 2026

5:34:18 PM ETthanks everyone for joining us today we look forward to speaking with you again

5:34:20 PM ETcall. Thank you and have a good

5:34:22 PM ETevening.

5:34:22 PM ETon our fourth quarter twenty twenty six call thank you and have a good evening

5:34:26 PM ETThank you. Everyone, this concludes today's

5:34:28 PM ETconference call. Thank you for

5:34:28 PM ETthank you everyone this concludes today's conference call

5:34:30 PM ETparticipating. You may now

5:34:32 PM ETdisconnect.

5:34:32 PM ETthank you for participating you may now disconnect

5:35:24 PM ETHad some system

5:35:24 PM ETsales use cases early days of agenda, but maybe

5:35:25 PM ETdirect way the discovery experience that so given that there's almost in search allowed people

5:35:25 PM EThad some system sales genic

5:35:25 PM ETto investment and things we have built over time, like GPU systems. So, we only do that after you have to pre-build

5:35:25 PM ETuse cases as days of agendic but maybe direct way the discovery

5:35:25 PM ETour style use cases, we definitely deliver a satisfactory end of this year in 2027.

5:35:25 PM ETexperience that so given that there's almost

5:35:25 PM ETTo address the question on the rate sales was again very, very serve up what's called context for

5:35:25 PM ETall of these parameters and growth in models and

5:35:25 PM ETin search allow people and things we have built over time like if you use systems

5:35:25 PM ETcloud. Thanks.

5:35:25 PM ETso if we only do that after

5:35:25 PM ETThanks, Mark. You know, extraordinarily successful cloud market forces.

5:35:25 PM ETtest to the earlier question, as well as this putting

5:35:25 PM EThave to prebuild facilities anything that's surprise style use cases

5:35:25 PM ETin end-to-end as we are building our models and agents, deploying them both with

5:35:25 PM ETwe definitely deliver a satisfactory this year in twenty seven

5:35:25 PM ETextensive testing, the kind of real-time monitoring we do as we do.

5:35:26 PM ETAny further remarks?

5:35:26 PM ETto address the question on the rave sales was again very very

5:35:26 PM ETwhat's called context for the all of these parameters and growth

5:35:26 PM ETin models and cloud thanks

5:35:26 PM ETthanks mark extraordinarily successful cloud market forces

5:35:26 PM ETtest i mean to the earlier question and as well as this putting in

5:35:26 PM ETend to end as we are building our models and agents

5:35:26 PM ETdeploying them both with extensive testing

5:35:27 PM ETthe kind of real time monitoring we do as we do any further remarks