Baidu (BIDU) 16 May 242024 Q1 Earnings call transcript

Hello, and thank you for standing by for Baidu's First Quarter 2024 Earnings Conference Call. [Operator Instructions] Today's conference is being recorded.

If you have any objection, you may disconnect at this time.

I would now like to turn the meeting over to your host for the conference, Juan Lin, Baidu's Director of Investor Relations.

Hello, everyone, and welcome to Baidu's First Quarter 2024 Earnings Conference Call. distributed our and as IR services. was a can on release on copy website earnings earlier you today, find newswire Baidu's as well

the our EVP our charge of Robin Luo, On Cloud CFO; in we Group AI CEO; ACG. our Shen, and Rong Li, Dou and Baidu today, call Co-Founder have

U.S. discussion Reform safe we forward-looking statements contain of remarks, prepared Q&A harbor our will of hold a Litigation session. under today the Credit Please XXXX. After will note Act provisions the that Securities the made

Forward-looking statements risks to expectations. from subject cause actual and may are our uncertainties results materially that to differ current

For detailed the of risks with Stock SEC report annual refer and uncertainties, Kong discussions our latest to filed these other and documents and Hong please Exchange.

to required except forward-looking obligation update Baidu any under does as undertake statements, optical. not any

press release certain call measures. include this financial non-GAAP and discussions unaudited Our of earnings

a ir.baidu.com. unaudited most available Our of at side directly press GAAP measures unaudited on measures comparable reconciliation release and is to IR the contains the our non-GAAP

is As a reminder, this being recorded. conference

In available on call addition, the will be this webcast website. a conference of

turn now Robin. our CEO, over to will the call I

everyone. Hello,

to from the on a basis. March XXXX non-GAAP path. and Baidu continued a ago. total and XX% year-over-year In year-over-year second to on non-GAAP quarter. billion, profit Baidu's of reached RMB in in increased from AI by business Our X% the accelerated revenue our to growth margin AI revenue to operating continuing is the cloud an XX.X grow operating improvement the XX.X% first core particular, quarter deliver while Gen year year the

an are Internet-centric we foundation business. from our business company . As leadership AI-first we solidify in the models, position to an transforming

experience. models. both and customers business we in the is more provide enable efficient agents envelope effective pushing powerful enjoy better To earnings to and to applications XB increase Given advertiser user to to business write more developers the XC ROI and that China, to are our aggressively adopt ERNIE most for let

legacy AI in term, and profit experienced our that revenue the growth business operate growth bring run. the in a in and environment will sustained lower While we confident near in us remain long challenging we revenue

cloud ] loss of business the the the We rest year. our to bins accelerate of our for narrow expect [ to robotaxi and

be We mobile existing near Gen recover the to becomes business core our next start soft the term in year. expect when and new products AI of to

benefit to AI like Looking developers and out individual a is to the missing bring is applications. technology influencing prepared build We important a Gen be ERNIE for XX Baidu foundation well in key models tremendous AI our will one from growth the opportunity driver to of transitioned beyond have model future. us near the foundation will opportunities, revenue AI long-term innovation most the models term, opportunities transformation. greatly believe which fear cycle. leveraging Enterprise swiftly new from offering this on and of this

life of ecosystem by developed developers To handled around in especially applications, around enterprise to various and This a ERNIE XXX ERNIE range points April, revenue diverse vibrant jump of millions contain in of inferencing. ERNIE of a individual accelerate from potential it. meeting daily, growth in needs building time, people's are API of million considerable calls significant work. indicates increasing the about believe industries and healthy future we last over well year. agents XX adoption wide adoption ERNIE, strong In the million a model and December a community from ecosystem We across everyday

our is ERNIE X.X. will of to The desktop base user flagship agents models, ERNIE family the of X.X large apps models, whoever in us enable distribute and to including and anchor these appropriate. ecosystem needs it mobile Our whenever this and ERNIE

proprietary unique architecture continued we introduced as Throughout models leveraging lightweight end-to-end improving well in QX. quarter, our strong the ability and AI the ERNIE's X-layer optimization. our in we As efficiency

For we about on March and compared mass sets only inference cost offer efficiency to is ERNIE of has and platform. increasingly its now trading the our increased accessible to X XXXX its affordable, version. tools X% X.Xx example, ERNIE To make

we for an introduced we app and important agents. platform This easily encompassing quarter, forms took the by one most builder In and of by for models. and AI develop Last AI agent model apps to tools individual applications creating envision of Builder, foundation AI April, model. a and is enterprise become builder will we step powered further a because build Gen Agent developers introducing

able line the of travel write are including wide be to the and developers of new programming code. per build they together, ability millions natural range day, without language, distributed a to platform BXB, as need every created more. agents ERNIE our day, Currently, on language verticals AI will serving education, the to and legal, With use single times a of agents

assisting I'm native experiences note using across in ERNIE. such quarter, as All API. Samsung used these ERNIE's built believe enhancing that running through smartphone will only is insights true proudly it and renowned realized Last numerous initiatives them app to applications. ERNIE drive when by native as experience ERNIE, devices well developing pleased and applications AI we top as that of widely customers. partnerships brands announced our smart are be users from influence extensive in and We on value building their China and expanding with its Honor,

This Vivo with to experience. quarter, excited leveraged we to APIs user makers who more extend ERNIE leading and such Xiaomi, elevate Oppo, are as collaboration smartphone our

for its AI. enabled assistant point This to the become smart up way user began vehicles. for empower [ large-scale enhance browser adoption, using of to our of devices smart the incoming China's now to into brand reach for ERNIE of experiences API manufacturer EV electric its utilized [ into opportunities PCs leading ERNIE the ample ERNIE partnerships now beyond AI the broadening world applications opens ] paving our top the PC Moreover, generated entry PCs. Neil extends ], default its include Lenovo, the are APIs and smartphones in vehicles. to by

just Tourism brands to also have new I [indiscernible] [ many the mentioned, such we customers Trip.com, notable as Board. acquired and addition Soto ], In Singapore

our is business. consumer-facing in opportunity Another long-term

of our have consumer-facing been reconstructing We products. all

Our goal AI so, be opportunities. proprietary should apps potentially to the for doing new killer ERNIE able to we our growth is native ecosystem. generate build By applications,

expected of the are and quarter. took to year-over-year for of Baidu generally Gen order base is after For the our document time. and cases paying and sensitive search Baidu with in feed than example, earnings users double-digit of [ because one-stop in ], AI and hundreds millions, these response very experienced first increase LLM longer user rebuilding cost shop in for the creation Penetration

models strategy. for best optimized for sizes of After for scenarios performance trial are ratio. quarters, range error different wide a few in firming need ERNIE our a different up we price We and

plan of more accelerate search new ERNIE adoption Going as recently, to of forward, results, features the and agents. such and product multimodel distribution generating launch we interaction in multi-round search

forefront territory. definition, the in technological we are change, in this are are we operating at We of confident to innovate. uncharted and globally unique By our abilities

want day-to-day users always, make adjustments to consumer we -- their incorporate needs and flexible features to we be in how product want life. As timely new evolving with

continued mainly generate reached by driven review models. In and the RMB each me key AI quarter. and to basis. profit business quarter, revenue Gen The foundation the operating a continue In XX% first of year-over-year will this AI AI the but model Gen Now cloud model majority revenue. on revenue first X.X AI for the quickly. growth let revenue to billion, quarter, from was non-GAAP as up for adoption highlights the been growing from inferencing for first improves. total revenue foundation X.X% and is accounted customer cloud training, revenue from Currently, such believe has of We rise revenue

internal instance, ERNIE. leveraging as the For and power of mobile [indiscernible] revenue within driving groups group intelligent such cloud by other groups, our business increasingly ecosystem

self-developed train because a Compared Baidu to peers, AI and we the believe and for model. and Cloud of in the their cloud Gen powerful AI layer, primarily payments China. model our inferencing reducing we expertise. infrastructure two, XX% efficient in efficiency. to and have As us number continuous also This capabilities are technical to superior model X cloud to allocated AI trigributable host their models our is while inferencing each and allowed innovate our have model [ enabling optimize AI has ] insights train we all on AI Group one, to cost. leveraging training And X-layer and result, sizes enterprises to they reasons. at Number TPU foundation our most Enterprises management, a to architecture help AI to cluster

computing vendors We integrate cluster to and from GPUs can ] unified Twin [ LLS. a into now various

of thousands Our with ] cluster important that breakthrough GPU this platform [ of entree of the demonstrated efficiency setup of availability very ] on an [ because This is GPUs. GPUs. is limited imported a even post has high

is GPU cross-selling Cloud customers. to AI cloud cloud Another our growth for services driver of our CPU

among recognition high With and new the cloud GPU customers. existing of our

usage CPU on switch open-sourced models seen customers ERNIE have We cloud to of we increasingly side, to their more Baidu. more increasingly models. efficient Earlier the affordable the than many and make initiatives took family mass and of

this some quarter. for highlights So

model ERNIE enhanced builder. model lightweight LLM performance enterprises in portfolio, model balanced development. cost model to X total our and and expanded LLM for have with developers, and ERNIE a We audience a X offering broader professional reach of This test-specific helps model

most X.X handle complex In a use MOE assigning can each performance more or tasks, addition, inferencing faster to distinct similar to the level our models. responses [indiscernible] most models experts tasks. This only into and approach or for for and maintaining query X.X while user cost mixture ERNIE the lower of approach suitable using allows task

introduced the quarter, App steps developers for [indiscernible] AI [indiscernible] refining we App reaching Builder to platform. quarter, tools continued our Builder X the just throughout in develop create in easily on Last and to we enabling

just of operating confident a April, to AI basis. continue on the and can non-GAAP AI the we remain Agent strong in we create for an in on aim anyone generating sentences With our agent Builder launch few cloud Baidu. Overall, a revenue, profit with

growth quarter, a deliver the ecosystem cash from and flow. have monetization margins by pushing to the conversion the free ongoing experience. reconstruct experience ERNIE challenging by a started. strong mobile continued first yet has macro need traditional is leverage healthy impacted to environment. to AI and Mobile better for hard gain. not our product system monetization At user our Revenue We This efficiency X% to to has same marketing transition and online generated year-over-year. revenue transform all grew been we was time, In

stand quarter, scaled improvement in opportunity we During revenue. ERNIE in agents ad generated to. targeting efforts These for our and up incremental further long-term conversion the upgrade generation. and real-time capabilities ad a resulted [indiscernible] for an enhanced

but we Recently, SMEs not adopting only agents. brand gradually seen also advertisers, have earning

designed can us agents productivity improve the of serving as for CPS step and expand transform the CPC We meanwhile, and clock. important and use significantly on sell-through We to users. agent storefront enhance that among this be around their enhancing more This virtual will for ease service their a consumers model. efficient to desk, SMEs our have Baidu. reach And experience traditional rate, believe an model user

opportunities greater Gen believe last QX I total for will Baidu arise year, addressable forward, AI expanding ERNIE. have reconstructing native from [indiscernible]. Since Going search, complements Search traditional we AI been with apps, AI market. the Gen particularly

and continue experience real usage Search. more variety and a of to more point the the [indiscernible]. is which we text, directly search formats questions by like are results results of growing in increasing search Now are users' rally and Baidu have problems. in links, time to for These will so, By generated ] improved crucial ERNIE produced interest third-party of address [ doing and image, enhance

user While with in is we the on has are this reconstructing stages early important feature still of It been product note Search feedback to ERNIE. that and renovations, Baidu encouraging.

the native to and gradually. mentioned I Baidu killer key build evolve and will enhancing is and opportunity years for the to AI ultimately Gen our creators, ERNIE behavior experience as builder, in that and This capitalize increasingly potential. this to It given that process an on has offerings XX be time introduced Overall, will Search trajectory AI important on With user service it are search content provide most we Baidu's on agents likely history will platform. an take era on a I likely one Baidu. believe app over user publishers easy newly providers agent monetization. find will right

In XXX,XXX April, Go Moving provider stands autonomous the number public, on year-over-year provided first the a heating Apollo ride X public. surpassed right million. total measured the the In to We about by service of offered to Go to Apollo quarter, believe largest the XX% increase. marking rights Intelligent as globally, ride Driving. the

then the reach unit replicate is continuing breakeven the regions to -- towards up To Apollo largest the service scaling are a make towards are part move In each progressing breakeven point, this is We and Apollo to driverless utilization city's gradually strategy transportation heating Go our ] Go. in for operations forward we move ride this integral goal. an focusing other becoming Go's enhance of fully happen, is field. economics [ key achieving the regional in and regions. Apollo To Wuhan, breakeven on regional reach success Wuhan, of network.

driverless than more of ago, X serving March, further of operate as and a by quarter the improving quarter its its the growth of Go's led total population Apollo XX% crossing remarkable All XX/X area [indiscernible]. year. accounted achieving part is Go over QX, for ride our in up from progresses milestone rights fully vehicles [ ] Wuhan, XX% over in Wuhan of Apollo In the provided million the driverless Yandi have to River fully with from of last and rapid to in early vehicles in fourth started fully doubled utilization. which operational expanding expansion. reach vehicle vehicles Moreover, the these a the

the The quarters. continues of expectations rapid in exceeding XXX% and in to coming figure rise, growth sustained April, with ahead reaching XX%

plan our we our ahead, hardware reduce this year will which to Go Wuhan operation generation Apollo significantly robotaxi costs. Looking deploy depreciation in RTX,

in of scaling future. cost the Go UE Apollo will improvement operational and structure, With Wuhan believe driverless breakeven continuous operations achieve near the in of we

and As overall closely operation monitor Go the progress, Apollo in to we driving our will optimizing business. continues efficiency intelligent of persist

ASD call technology Apollo in I the in to models is use global of our Apollo that evolve. mentioned earnings a On Auto autonomous last in Solutions, our visual foundation pioneer driving. for self-driving continue

XXX,XXX environments Now hundreds in ASP on China, coming cities made to vehicles. to high-end to of as and proof leadership. us over across solely reliant solution economy make driving to cities technology urban across to our expand in our of can broad state-of-the-art available From XXX priced autonomous as [ spectrum ] allows a with driving vision passenger in as is navigate it This plans low OEMs. serves autonomous models effectively months. attainable advanced of RMB the complex another

me that, results. call to turn over financial the go let to With through the Rong

Thank you, Robin.

X% the were year-over-year. billion, results. revenue was online of RMB Online Baidu Baidu X.X me XX.X X% revenues marketing year-over-year. quarter was was year-over-year, up of the driven and our billion, let Now through partially ] costs costs by X.X X% content traffic year-over-year. an by year-over-year, AI [indiscernible] walk related RMB from the billion, in AI Revenue first acquisition increasing to increasing XX.X Total XX.X revenue Cloud billion, billion, [ RMB Cloud primarily XX.X financial costs. offset business. Revenue marketing year-over-year. was details RMB mainly [indiscernible] was Cost for Core RMB [indiscernible] billion, increasing X% to increasing X% in business due Baidu X% RMB decrease decreasing increase revenue

accounting X.X billion, for called were the compared year-over-year. Baidu the partially Baidu quarter X.X in costs X% year-over-year, X% compared SG&A last due billion, XX%, decrease development billion, offset other for Core Operating same in decreasing revenue RMB research server decreasing RMB in expenses XX% expenditures, year. were and by costs accounting R&D depreciation year-over-year. XX% X-X expenses same a R&D period support [indiscernible] and decreasing XX.X custody expenses period this in expenses and costs expenses which personnel-related in year-over-year. were Baidu increase Baidu expenses tools. fees, last to this SG&A operating X% decreasing X.X Baidu XX% of RMB billion, of RMB year. in R&D X% revenue were primarily to the to quarter solar

billion, and XX%, operating billion, long-term Baidu was margin was due other fair the [indiscernible] gain X.X X.X income And billion. decreasing Value was Baidu margin to billion. income non-GAAP per income year. was Total attributable primarily operating decrease income to X.X operating income by income a Income XX% foreign was billion. and Baidu non-GAAP partially operating RMB investments, was Net billion XX.X%. Core was in in offset net billion, cost year-over-year, the net X.X was RMB was cost value XXX from RMB diluted margin expenses to income was Core Baidu billion. attributable was RMB Non-GAAP X-X operating to Operating RMB gain. X.X X attributable net Baidu Baidu increase earnings Baidu XX.X. to period net Net income RMB and same million and last ADS exchange for in Core was X.X tax RMB X.X [indiscernible]. were Core RMB Non-GAAP billion, RMB compared

net for ADS to was was XX%. XX.XX. net per RMB Baidu non-GAAP Core income were Non-GAAP RMB attributable Core margin billion, earnings X.X Baidu diluted and Non-GAAP

were cash, cash cash cash Free billion. flow, and RMB billion. restricted of billion. in XXX.X XXX.X restricted RMB and investments RMB IT, excluding excluding XX, was were ] cash short [ cash As March ], X.X Free short-term investments, IT credence XXXX, X.X equivalents, was RMB [ cash flow billion

Finally, as With that, the questions. call now ]. XXXX Baidu to Core of had employees operator, approximately XX,XXX open let's [ XX,

Your Instructions] comes Alicia first Yap [Operator Citigroup. with from question

AI be wanted to improve the who adopted advertisers monetization enhanced? some to improvement? those have areas share are on technology any helping what of been feedback there ask, they further do management Baidu largest be ad I areas would you kind Can quantify system? of that the can So able see to And In rate.

Robin. is this Alicia,

improvement million know, Gen increase RMB As Baidu. to This system, with the spending monetization sales advertisers. improving utilizing creative seen the [ more our been strategy to have has to motivated have we better system year, add various AI, system. generating generation of conversion ad was of in technology. This formation aspects their and first benefit seen Advertisers for last refining includes and leads. quarter automating advertisers second capabilities, half our ] auction ERNIE Since has per incremental It revenue. our monetization several you from upgrade in hundred the

broader advertising and quarter, weakness incremental grow the a our time basis, us brought revenue on product some incremental us revenue helped ERNIE. and in user AI-related we also continue. expect this quarter-over-quarter with the has trend to first So mitigate The reconstruct to macro

customers [indiscernible] language Agents opening offers As long-term presence of conversations. their customers Builder, online domains, using multiround and emphasized natural potential online interact function believe whether with can establish can help ERNIE Agent When improve also our can The more improving advertisers our their press experience goals, marketing easily can agent, Agents effectively service a intentions business. their introduction potential we enrich virtual agents. customer create ERNIE in offering significant Baidu. I my to their remarks, for achieve or this quality. advertisers helping content products customized it's as they or on user understand and through advertisers

have still agents with the some While in advertisers ERNIE ROIs. stage, achieve better helped ERNIE

service agent by adoption consultations. enhanced example, for education. conversion subject This For in providing lead AI expertise, like the we customer high-quality key Builder company. online increase ERNIE its around-the-clock to XX for a matter create Agent education has offering introduction product refinement. customer have it agent online online rate used significantly insights print The injecting to and company's will with They ad feedback of agent, this continually in

the agent. I continue beginning. AI believe just will ERNIE the era, form the new think We in and capabilities this of a of to is agent improve services content will be major we

both help conversion our us into over shift CPS should but for will user time, model. Agents not elevate a generated experience, traditional increase to more on CPC transactions model transform and platform. directly advertisers, This our also in foster only ROI efficient

Morgan Yu from Stanley. Gary with comes Your next question

a also, question sustainable as price cloud of think cloud we Cloud outlook regarding growth hits profitability the should business? And have should be by the level the AI key going about how your what of And we should Cloud how AI revenue initiated forward? up? some the affected companies by competition business. How I has -- cut

Yes, Gary, this is Dou.

and computing mentioned, purpose have Gen the from computing. models cloud to a industry AI we already foundation or As transforming AI general

a So in industry opportunity this establish is the leader reshipping shift the valuable cloud. cloud ourselves the with AI landscape competitive to us as within presenting

platform most advanced we China's mass we model AI our believe for offered inference. So and the infrastructure and efficient training most

fine-tuning training, has cloud. revenue. significantly are development AI for This As us increase our AI and demand our model public more result, native a in boosted and cloud more app enterprises choosing on our

further second accelerate half to a this last increase cloud of in the in started Actually, since year-over-year the the year. to to an revenue the third second first has year, year, accelerated quarter of XX% -- fourth the from XX% quarter growth our last in quarter quarter and AI so decline then

models brought supported So we the new Gen acceleration incremental factors. our and generated foundation directly revenue X legacy was by the revenue main cloud by The AI business. to also and opportunities

AI the our cloud in X.X% first of Gen were year, foundation factors from our AI So account both our Cloud. on businesses traditional drivers models this of for foundation models, as presented then and revenue. quarter capitalizing by of the just AI CPU or Robin Cloud important mentioned, growth And revenue AI Gen and opportunities

Cloud a transportation QX subdued, note, quarters. previous remained while was in substantially business separate smart impact on its On business the AI than overall smaller

So our continue trend, upcoming the benefiting from in to revenue AI overall, growth quarters. momentum we maintaining make expect strong the Cloud

AI Baidu operating previous profit have Cloud in side, GAAP seen continued profit, as we to the quarters. generate On

to health growth. committed are revenue and We sustainable a

maintained the quarter, business. on revenue growth During by our down achieving low-margin scaling high-quality focus we

stage. of business models, the is market still very early the a Gen and AI the at foundation Regarding

broadening footprint remains our market enterprises. So and educating our across more on the focus

we improve and business AI to be foundation business. and run, so higher our the to expect long the for traditional into Looking models normalized Gen than margin cloud further related

common Regarding cloud our policies certain pricing to it the of pretty of their products. actually change right, mentioned, for pricing address you is competitors just vendors some for

we a past. is trend in the this times multi Now have observed

minimum impact AI cloud Given expanded high-value the pricing traditional our has Cloud. products of the a to industry our on changes development AI CPU CPU cloud so and services that has offerings cloud the a on beyond

approximately the the daily. million year. lowered cloud May, March optimization, handles only our X-layer our XXX last or AI ability of This XXX ERNIE's have in unique end-to-end architecture cost we by in billion about tokens our inference strong API proprietary platform calls services leveraging X% for ERNIE vary Actually, [ and to ]

reduce to It's rather that adoption confident continue boosting different then models at performance, ERNIE's some costs. different evaluate to prices. for focusing the present enhance its further and just on important expanding ratio efficiency superficial than performance will also are We workloads

So [indiscernible] the customers. our infrastructure advanced we value platform believe best and mass our AI delivers to

leveraging spending AI we to increase capabilities, At ones continuously Baidu same strong their non-GAAP and we positive forward, to on aim our AI operating to Looking generate Cloud. new existing attract the aim customers time, profit. encouraging consistently

Your next comes JPMorgan. Yao question Alex from with

differentiate shortage Given maintain leading the help in the chief while to delivery commitment us in value to its China, China. enhancing by advantage in [indiscernible] technology

differently taking Alex, think application-driven we do AI our very in an for aspect. approach this We're push.

mass the most the For not solving time. all at priority be advanced example, the high using school problems LLM may this highest

more users convincing it's generating While important lot a for products, the of applications. buy to in reasons right

So ERNIE AI App to and Agent X-layer customers taking can models. and optimize applications Model Builder, Builder. And like with stack we tools sure Builder of of unique tech cost using are build mind, advantage that we make our the performance in developers and easily

layer, [ a to ] factor capability infrastructure in GPU contributing efficiency in management. our France AI model superior the and high For train in key our cluster is

integrating from training for and cluster, vendors effective large-scale use We highly less us one made breakthrough into have by different advanced a inference. chips unified recently allowing to model computing GPUs

community constant and has grown XX the compatible on framework with has to chips, inferencing different as deep learning developer training helped PaddlePaddle over million. cost XX a innovation and model continuous many of is domestically designed enhancement Our PaddlePaddle through and basis. reduce

flagship and gained lightweight our well We're experience ERNIE tasks. making around have ERNIE to be insights available these introducing vibrant ERNIE products, to ERNIE. we affordable toolkits mindset, all more With and healthy With and capabilities in making and used ecosystem a applications extensively as for X.X, and all LLMs remain We're early model development using applying to AI are training accessible better the lower performance our and our sophisticated fostering it. and efforts, on app have MOE renovate developing models mono for by developers. these approach customers we application-driven we and With development, cost. own ERNIE as launching for X.X inference for and

of holistic different are actually approach some Gen which can from You that very is see a LLM developing and we taking competitors. AI to our

reserve model run, most the chips should on an efficient ample form There's in but its be for market ecosystem homegrown to think software for access and And with framework quite application will With in ecosystem AI well the China's leader to developed long I team, our strong AI and application-driven building layer ecosystem AI the probably future. layer, that I'm our the Our ERNIE, room long an as in of layer. around in the applications stack. less to of infrastructure support self innovation us dedication a powerful R&D approach as and AI our the X-layer China as in own millions our by stand confident a will chips sufficient in term.

Lincoln from Sachs. Kong next comes Goldman question Your with

is business. the about enterprising question my So

with and drag trends we have the major peers? advertising compare growth growth May? for advertiser the seen for if we XXXX? April the and what for the quarter, our second So normalized So the So into should sentiment, advertising we budgets especially are about what's think how [indiscernible]

Yes.

X% by revenue in Well, quarter. the marketing search is maturing. traditional year-over-year online first grew Our

renovate filled user We do users could of answers now XX% Right our hard some with the about organized enable questions, experience not with cases, are result AI. they in before. better and to These working things pages and to search more Gen results generated accurate, do provide users' are bright results. to

have not the AI Gen those monetization of started results. We

macro revenue business. of to contributed the for some weak softness And So to catch the time up. take it our ad also will

range Our come wide a being advertisers of from industries with the SMEs. majority

to but weak, muted to verticals estate to both estate, spending specifically developers extended sentiment in and the advertiser off-line sensitive economy. macroenvironment, remained our real the and some impact revenue the particularly in first upstream advertising also the downstream not industry, sectors. In only agencies as This was ad quarter, takes highly real the and such from franchising

to example, for years. our platform. in materials constrained more in on of energy have furniture, many kind sectors at For few past working time Also chemicals, machinery, building recover SMEs been also the and hard as they renovation offline downstream home spending the upstream this need

As we seen second improvements segment. we quarter, not enter in have the advertiser

for fundamentally visibility But the QX, should limited quarters. the the few with remain revenue solid. our Given online comp marketing from sentiment and next tough paired growth improvement, perspective, in soft over a

apps near-term to to offering foreseeable in be the Baidu a likely one one the with of age, future. enable developers to in innovation engine will expect Search better search directly apps for the the Baidu natural is marketing will and more Beyond in us million Internet China the popular engage most challenges, users' killer we in most product relevant close remains remain users merchants, connect of ways. of for and XXX MAUs. and with the service [indiscernible] [indiscernible] intention Technology largest the AI. with age Gen online

Your next with from Jefferies. Thomas comes Chong question

about we look question lag your Previously, room between more XXXX, My AI how margin there revenue is AI in you expand offerings? do cost contribution. at see mentioned we we investment will a How in to intend and much cutting. should be if trends

your much you thank question. so for Thomas,

the that for operational do us efficiencies. are I optimizing marketing still our on be confident we winning businesses. there still to will But macro challenges think are to ways continue some

still our And we ecosystem, think flexibility. take resources the the prioritize to expenses. and into We we as structure strategic areas. expense will If we [indiscernible], strategy strangely agility support manage mobile to organization we we the necessary for for further steps to streamline will some cost can the will costs the manage including try also businesses our and issues take I key relocate enhance and

remain So profitable flow a the positive. Mobile Group to strongly and Systems continue cash

and out Dou we For phase businesses products. just mentioned, AI will continue the Cloud, as to low-margin

into businesses So we margins than higher for business. Cloud on term, a profit can AI-related normalized generating the longer basis. If we the the should look operating legacy non-GAAP be margin the the Gen continue and cloud

and For investors to many our in particularly than intelligent operational we we other past our about how ] ask impairment businesses, aim loss And efficiency investments in reduce with business OE our [ margins. robotaxi. driving gains as the talked what X for improvements more the ERNIE quarters, me

mainly related training for CapEx investments inference. In are our to and fact, modern

times and the batch which startbase. in In different has have time XXXX, different depreciation made purchase a we arrived at large

the is expenses inference use and impact annualized model are While expenses reviews and and computing see depreciation revenue will computing booked that the Baidu the R&D for business manageable. the depreciation XXXX, power available both training on expansions. can expenses cost our quarter, Cost, whole The of non-GAAP R&D quarterly under power first cost, for cost In we year very be the and predicable the of expenses and the overall in for quite slightly. earnings increased revenue

margin spending such operating on expanding actually SG&A. due non-GAAP disciplined to XX.X% our other While we're items as to

have in to generated we the be of for revenue we because sufficient will profit X and and from years. high-performance We believe are have X our AI growth more XXXX, a margins. in should availability conclusion, ERNIE's CapEx support manageable of for year ecosystem investment large and with Overall, the monetization In chips And year. AI both mobile more businesses last our this already of Cloud. -- from trends And versus high-quality the hand imaging [indiscernible] we the China expect next off, in investments will expect balanced. that to kind in be smaller earnings and the or of both near-term well impact that believe limited be taking of we that models that on Gen

For with business. and have on solid to many line cost the years of a this top futures disciplines, resolute very we growth we building our record now, track continue intend kind delivered of the

UBS. next Your [indiscernible] from comes question with

My shareholder question return. about is

X current pace the of should improve program, how shareholder execution current other in approach So to return? buyback diversified about expect program, buyback think to we billion we addition about RMB

And much our the questions. so have shareholder Thanks returns. I value for making been to Yes. increase highly efforts we we shareholders, think and your

around repurchased flow consistently market over shares our towards years, the past buyback programs. X -- our averaging the total, billion have have from share In allocated free cash $X XX% of our We annually. the we

and shares significant in and And going comp we to outstanding I buy we the the our we the growth economies shareholders' total have dilution returns. to opportunities current of repurchase committed very prevent think during any utilized Baidu. shares more stated increase in forward, this program addition, share our the continue in aiming potential in reduce to to believe as will shareholder we are our period from back loan market

XXXX was outstanding year-over-year the XXXX, increase the you flat a decrease. a that shares shares total our year to the of year X.X% this XXXX. see X.X% compared in year In number and outstanding quarter, total can to In in the increase began

declining You it been has can that see compared the prior quarter. by X.X% to as

We of market. time, ahead open us. are take we opportunities adopting programs a same consideration strategy of at sustainable and from recurring the share into And buyback

foundation AI it. a opportunity to plan facing huge are Gen and have we capitalize in structured a models, concrete we Now

shareholders that to of way best in to values have believe building we the consider the to the Furthermore, as create shareholders. invest business want most for interest to necessary So fundamentals. we we value is by strong flexibility long-term the and the effective

business marketing steady growth that will remains Thank another so and much we question. help to build core us you believe AI the for Our engine.

BofA comes Your next from Securities. question [indiscernible] with

question My about robotaxi. is

updates I in I want breakeven the continue impact on of take? it you What's mentioned what's efforts So operating coverage vehicle size will this how projected to previously, understand management initiatives can to behind the think more Go potentially for EE fleet UE. the future. Wuhan And the geographical year? will in And improve near cost and robotaxi this achieve the Apollo share to logic the the year. that for

Marina. Sure,

Let the business. cities. robotaxi Go me economics key more unit color some significant regional improving made give the XXXX, on In Apollo progress in you in

to Since operation of operations can XXXX. scale expanding be how to operation attributed kicked me Apollo then, largest UE driverless witnessed bank vehicle. off and in operational we Go's that. use commercial the we've Wuhan, explain consistent Go Apollo a enhancement decreasing per achieved which a cost Let We in in

growing. been Compared the hours drivetrain Thanks driving government. in fully area operation for ago, scale the driving grown service Regarding vehicles reaching both steadily [ our increasing fold has XXX been recognition to our about our vehicle local operational technology by ] consistently X have Wuhan, expansion, of straight autonomous fully to expanding. year has the speed fleet today. Meanwhile, and

per XX/X consistent terms rides the is eightfold a been resulted and operational The of the of March Go its a the ago, peak add revenue in X this coverage area people gradually have operational expanded year. in in to in by labor to for also hours from cost, fully over The in beginning hardware vehicle expansion year now operation million majority cost [indiscernible] hours expenditures. Wuhan. growing. from both service covering extending UE to to ride daily hours eventually and Apollo in off-peak distant When and improvement only steel increased of it comes

track operations, of increase [indiscernible] the to help consistent fully safe which demonstrating been have We record us heading operation. of deployment

the April, of XX% XX%. of XXXX QX last proportion up In XX% to only year. fully-driverless from August rose and That's orders in

minimize figure this officers. to the in to quarters, us expect safety to We the thereby related enabling coming XXX% reach cost

lowering adopting the we generation below are mass costs. track down USD veteran, In of driving price addition hardware to mass RTX, sixth steadfast our in robotaxi the solution, remains labor excluding swapping line battery The production RTX, cost, production on XX,XXX. a for is time

We reduce improving to for will to hardware fleet us future the the vehicle and closer it the in cost help use and significantly RTX primary each depreciation further as should bring profitability. expansion, vehicle and

by of end tripling year. year, of of driver operation this X,XXX into last the fleet Wuhan more vehicles the we our expand fully to plan than end quarter, Looking from the the to

we operational Thank in Our once Go the will Apollo UE With that's we Go Wuhan losses scale improvement focus narrow operations regional continued and operational the quickly. and for Apollo can breakeven you. achieved, in remains business. efficiency on improving up achieve believe UE And first. cost reduction,

and today. you conclude for for our gentlemen, participating. does Ladies Thank conference that

You disconnect. may now