NVIDIA (NVDA) 20 Aug 202021 Q2 Earnings call transcript

Good afternoon. My name is David and I will be your conference operator today. At this time, I would like to welcome everyone to NVIDIA’s Financial Results Conference Call. [Operator Instructions] Thank you. Simona Jankowski, you may begin your conference.

you. Thank and fiscal conference XXXX. everyone second for to of quarter welcome afternoon, call NVIDIA’s the Good and Vice With NVIDIA President Officer are on Jensen Kress, me the Huang, and call Chief Financial and Chief President today from Colette Officer. Executive Executive to like our that on you webcast call website. Investor Relations remind being live would I is NVIDIA’s The to of results financial third quarter discuss fiscal be for for until replay available will our webcast the call XXXX. conference the property. content today’s is The NVIDIA’s of call our prior It consent. reproduced can’t transcribed without be written or

During of this forward-looking our based and call, materially. we may statements subject and differ on make are These to current actual uncertainties significant expectations. results may a number risks

available For August XX-Q required obligation of on X-K any based on Form and our and that earnings made discussion future release, file with the factors such refer to and our we Securities currently recent may as in us. affect financial Commission. business, XX-K Forms the All today’s are assume XX, a statements. most results no XXXX law, reports statements we and that to to our of Except today, could please update Exchange the information by as disclosure

will call, discuss measures. this financial we During non-GAAP

non-GAAP You Colette. website. measures is can posted the over to a With turn call financial reconciliation on that, CFO commentary, measures these find to financial our our GAAP of in which let me

$X.XX of also leading pandemic of gaming, XX% $X.XX and the through channels. our delivering ahead regions, expectations. while and quarter. of growth another XX% was year-on-year, but a QX outlook. our the combined we agility Mellanox of extraordinary quarter. products the up geographic The was pandemic most Starting was entertainment revenue ahead and The strong Simona. form XX% very dedication, up sequentially and up into COVID-XX of up to sequentially is continue Thanks, upside from highlights team’s across Revenue our significantly emergence continued to world employees a billion, XX% broad-based was the well year-on-year home, and Gaming’s billion, work NVIDIA successfully worldwide. with amid battle

gamers For is example, XX% PC up $XX on for a second PCs reported working, billion. now millions demand at videogames calendar channels. people learning of NPD home. outlets that a are the and levels closed NVIDIA’s online are At leading retail consumer and gaming number U.S. and game of record on outset laptops ideal quarter in shifted pre-pandemic the online grew Steam, XX% the pandemic, to who many daily spending and from the distributor, were the to of

As We ramped the [indiscernible] new XXXX points. from laptop with and over premium with as premium improve RTX the strong picked home. our performance partners segment, the turn the form mainstream price SUPER models In laptop largely delivered unparalleled reopened, to iCafes GeForce-based students systems to demand demand and work, how they game and is stores learn, sales quarter OEM reopened and and XXX we and factors. progressed professionals both continued GPUs online to focused retail XXXX GeForce on thrive. Gaming up, very the

of laptops upgrade RTX Its effects Turing driving Dogs. in while Turing powerful rates which games These to Minecraft XXXX cycle low Cyberpunk In continues that DLSS. the of and XXXX. games ray gamers. support Watch major among STRANDING, support to GTX AI GeForce programmable NVIDIA DEATH DLSS quality. in the other by shading brought and technology architecture, ray and laptop and and points the the AI upcoming Mega-Hit redefined with image mainstream time has at our and graphics. is we games enables RTX as tracing GeForce to for as Xx, crisp gaming $XXX frame blockbuster points anticipated We new price new almost as the also tracing $XXX. brought ray used price to tracing RTX the join the segment, low adds stunning standard highly RTX continues quickly including for for first to generating computer Momentum as release grow, boost visual titles capabilities the

tune We a GeForce. the in with spoil GeForce We the a of to on want day midst don’t in. highlighting special each history September surprise, promoting we campaign you XX-day but are of event in encourage countdown year a X, to the

Games streaming traction with Valve to in access game We NOW second cloud we of devices expanding are NOW game the content service gaming through to Steam, the the Macs quarter PC their richest PC, play Ubisoft its our Android GeForce digital universe GeForce commercial games users Uplay. partnerships enables powerful of leading GFN GeForce. underpowered in offers very or cloud, that of pleased and now with Epic can to the libraries gamers including service reach of any GPUs with with availability. stores,

supported GFN into of now our Chromebooks announced on we millions yesterday, is reach expanding of Just that tens users. further

of in a NVIDIA’s Taiwan and the In own soon Russia, LG in Korea Beeline telecom Japan, or is in available Rostelecom addition to Mobile. world, SoftBank South service, including U+ around number partners coming and GFN to and KDDI

accelerated healthcare, QX Despite including near-term oil in areas quarter, Philips Siemens, business revenue, enterprise Accordingly, declines year-on-year and by sequentially, Sales enable the workstations. as such Electric remote Pro XX% down mobile vGPU by to architectural are closure graphic Vis, the QX in entertainment and during with offices lower and negatively the to General bookings media engineering of workers and desktop enterprises virtual include $XXX we Moving the winning and and in automotive, XX% for new manufacturing, sector. amid year-on-year. challenges, was gas. Initiatives XX% demand both impacted solutions. world. and Industries many hurt around in demand construction, drove cloud-based increasing public and our and were million down

applications RTX-enabled, tracing and including that the work. We will pandemic provide lasting impact NVIDIA The opportunity These over latest with improving are and continue Group, Chaos we to ray performance, expand Foundry, our design design creative release workflows. faster on leading from a accelerated how market and Maxon. creators’ have XX applications

mix this revenue will with that virtual evolution focus as forward on work reflect greater technologies, going enterprise workforce collaboration. enable such workstations virtual a remote likely and trends and NVIDIA laptops in Our

volumes $XXX the pronounced XX% after automotive Moving was down year-over-year a revenue to million, XX% the bottoming with and XX% was of our to than in of auto down than as April. outlook automotive, sequentially. was recover better pandemic sequential This decline slightly expected, production starting less impact

for with centers us computer, to which in to turning software-defined XXXX also Some point quarters. features headwind grows software ownership revenue across intelligent June, landmark an which This will developed. in of vehicles is of a entire In driving integrate future as announced high-performance a Mercedes-Benz, due vehicles starting to launch the software updates, each latest end-to-end model of roll-off on DRIVE the a the easier they autonomous stream. NVIDIA’s NVIDIA’s announcement content will the will over-the-air AI announcement technology. NVIDIA advanced decline making traditional a build upgrade. is for make can recurring software evolving data a of revenue, infotainment industry positioning using driving partnership DRIVE computing spanning features, of the and and remained autonomous legacy the car with it utilize fleet the we in as transformative infrastructure automotive including unifying and from stack, NVIDIA’s wheels. Centralizing core in upgrade updatable and technology AV transformator Mercedes business extending becoming and our It’s receive With the to increasing also the cloud. cockpit full software the majority are value vehicles platforms intelligent AGX

Moving was Mellanox day billion, our of growth. providers, from data of XX, company April center networking up data centers, XXX% QX, with center our a incorporated providers, edge. a record of Both Mellanox $X.XX compute contribution center Ampere year-on-year just set closed data revenue approximately up and a this quarter. full cloud year-on-year center data enterprises, and accelerating the center, first revenue. within The revenue and a was diverse, and XX% supercomputing contributed sequentially. quarter of biggest in which XX% is news data of on telecom quarter record to over QX architecture. launch industrial of consist acquisition, XX% the public In service we cloud the

are with the XX made the execution and launching the this It amid especially ever in the transistors. proud most marvel We technological chip our pandemic. team’s very for The billion workloads. full ramping accelerating largest of stack runs AXXX is software compute-intensive

new greatest Our workloads, powering AI first for new for genomics, analytics, our include ever, CUDA XX libraries of as boosting framework workloads, systems. software such a and It unifying scientific AI and application Merlin AI analytics, training by the predecessor. data services for also major and delivers Jarvis leap The inference its edge AXXX AI deep XXx over conversational leases recommender graphics. and CUDA-X versions and generational NVIDIA’s such over universal accelerator as video XG computing, performance is XX,

servers chart gigaflops than supercomputers offers years efficiencies, energy performance by two-thirds benchmark, with over XX June, first has widely the the to of using Adoption compared NVIDIA’s networking that systems the are all of pack. recognition U.S. service watt. it history. XX the consistently are from than bring XXX records, delivered to which include the the Enterprise launch, TOPXXX providers. latest in the and a on including in AXXX server based and performance outscale In MLPerf TOPXXX is market just the the half for world’s the GPU, to industry’s this GPUs, market Baidu Cloud. of top total performance leading efficiency any availability they is in has and own interconnected also is combined we ranking, also on AI of industry and that AXXX more major Tencent GPUs going it X around Azure ago, is was networking. the based with the list The Google efficient two the more commercially AWS, X.Xx the Mellanox GPU companies which topped years first AXXX inference. released systems for chip debuted vendors InfiniBand with X the server The less incorporated MLPerf the and Cloud, the in average, is categories year. without in is XX on from the incredible in summer top cloud the best only all fastest announced gigaflops and was powerful NVIDIA powers vendors The list They supercomputer, available XX most NVIDIA the into XX Ampere per than customer AI with now offerings both. the leading amplified objective just seven winning as by X.X of of powerful top away sweeping massively cloud XXX-gigabyte-per-second solutions Hewlett-Packard and systems faster number announced On Alibaba adopted list already to and are X the NVIDIA showed of Mellanox AXXX, Cisco, than Selene NVIDIA first makers Microsoft The and Cloud both in benchmark. measured GPU by nearly to watt XXX the AXXX and AXXX by GPUs NVIDIA offerings new leading XX NVIDIA cloud And more clusters, ago. from to the per platform systems end fastest on only Dell, systems cross NVIDIA or use NVIDIA’s morning, been our making this expected Europe. latest training into systems pulling introduced benchmark Since getting for AXXX barrier. per both Lenovo. results the scalable and supercomputers. the record AXXX Cloud training and and come And world, the the prior AXXX. with set

benchmarks is for months Our assemble architecture customers able quickly even than available supercomputer. build will SuperPOD This X a class that weeks it Selene DGX time and in open a to engineers instead our use system to the design that’s years. MLPerf were or using reference the is reference of modular win build we of NVIDIA’s less typical world the to

fastest and more of also We the worldwide. to We time pipeline with Apache of enterprises the enterprise end-to-end Spark GPUs entire enabled acceleration is scientists leading June. acceleration GPU first version to stack analytics released latest And have analytics, than analytics workloads. milestones the in world’s and used in we software and by share. for workload NVIDIA’s the data largest brought data data Spark two XXX,XXX one platform major of the data growing XX,XXX

and total cumulative X over developers just first the of reached It the NVIDIA reach million. a decade a million X We the CUDA number have now of shipped to took total billion GPUs ecosystem the and strength in as fantastic results and NVIDIA. reach less revenue than Board its InfiniBand across has in with Mellanox first quarter accelerated X across second products. years to million. growth of the Mellanox Ethernet part

Edge new critical of several major the high-performance in Major shipments quarter. record. product DGX is build supercomputer quarter for and computing the in networking. part our cloud as the of hyperscale growth new platform. this include DGX our networking AI and was reached introductions a Ethernet SuperPOD Our system, the for a demand AI upside the increased Mellanox EGX drove fueling clusters AI These Selene

a software centers, the data our accelerated, The needs our XXth privately Cumulus data where NVIDIA computing, are the software-defined capabilities generation XX Cumulus launched software with the combination that modern meet the and product exponential networking addition purchased Mellanox of we disaggregated family cloud AI, of the second We held accelerated and is to is of and in growth gigabyte high-performance that network meet designed XXX June. in XX, cloud and We company enables also becoming and the switch NIC, networking data modern and hyperscale computing. center. standard. Networks, expanded to out augments open in of per Cumulus ConnectX-X Mellanox centers building ConnectX acquisition Ethernet Mellanox the

XX% sequentially by automotive a was operating year due EPS non-GAAP Moving XX% a GAAP the quarter $X.XX and QX product Non-GAAP $X.XX year from associated reflecting flow declined let gross expenses was sales. turn of from points to with Non-GAAP increased margin billion with for year GAAP that, up QX a GAAP ago. and to XX% the were operating margins from and acquisition. the GAAP lower up X gross sales billion, was EPS P&L. year-on-year, gross was rest XXXX. operations in costs from QX year-on-year respectively. XX% almost $X.XX of fiscal margin data QX non-GAAP Mellanox was to margin ago, up earlier. cash XX.X% center third and gross With expenses the the $X.XX, billion. were outlook $X.XX, and mix XX%. the me shift higher

expect XX% to We to up with billion, in data be be sequentially. we or X%. expect center gaming plus to digits up just minus over be low revenue mid-single $X.X With sequentially, that, to the

our be to We minus in expect or at excluding be GAAP X%, expected minus GAAP plus auto and and items. both to non-GAAP pro X%, discrete non-GAAP of are the $X.X and OI&E tracking gross year and are in be respectively. CFO to information and are $X.XX line XX.X% highlight financial margins commentary and or let approximately other expense levels points. expected operating GAAP to with billion expenses rates vis website. $XXX OpEx million. details are me $X.XX both Full QX. community. both similar XX Further non-GAAP approximately be $XX expected out are closing, the and million. IR financial expenditures of our included available million expected and is non-GAAP GAAP and respectively. billion to XX.X% be on events upcoming outlook plus billion are In for to expected basis $XXX and respectively to billion in non-GAAP GAAP $X.X approximately be tax Capital

Summit The Mobility, August Evercore’s Conference Technology Memo We X, XX will be on and the September on Technology XXXX Virtual Bank’s BMO Global September on at Future Virtual September Deutsche on of Technology Conference Forum, XX. Citi’s XX, the

on our as financial of part a next Q&A host with GTC. also X will October virtual Jensen analyst We

scheduled quarter’s XX. results November is call for Wednesday, to earnings discuss third Our our

for for open Thank you. Operator, We would will poll you the call questions. now please questions?

first [Operator Vivek the Bank question Certainly. Instructions] Arya Your of comes with America. from line of

Your line is open.

structural rather maybe data into on congratulations data get of center you in of execution. we and Thanks the And how forth? for center secular or into taking gaming it can for because much more even sooner is or much once temporary more Jensen, I normalized customer growth hopefully industry? my of some strength that am and and are the a strong continue pull-ins and is period how in later wondering question seeing so COVID the than more more the that

Vivek, you. Yes, thank

see worlds videogames with and new all, not that from that to of of so brand you itself our of out first your Minecraft, people go made videogames. their thing are People of is are beginning The are friends to hang variety time really So it sharing the but just digitally, that growth around product realize the in can spending than Fortnite, we you ever, the for spending with it’s happening building haven’t the you pull-ins with and in using and we with Fortnite, realized it’s friends, your your time people. center ways techniques largest imaginations. didn’t that’s it data concert industry, to The about are medium Ampere virtual a and vast gaming time using anymore. are spending different majority all create that. in entertainment unfortunate time play, came The cycle people the can whether other it’s ideas for broadcast, in game world in world. – spending with in on the gaming the you More

time an of so videogame. and consumption – then course, the by to way a incredibly video And spend fun of of it’s even just –

return. And different is is time way dynamics you entertainment largest platform that digitally center And over world. steadily. up as about are I pandemic, do things data has coupled game that And games. Video accelerated enjoy the platform gaming in GeForce just different been – with I With that��s just just mentioned, PC of a you result at is videogames structural it’s and time. gaming. this that enjoying been data I so to And platform. think think change could the so are pretty entertainment the largest as now the with but the many in largest has could now. There’s the – single not it’s is that happening gaming of so the And the now center happening going there much going don’t adoption that same ways

applications way east-west center is cloud the accelerate a whole is of of could way are course, multiple it’s to that be single computers. servers models. the application a center of on of it intelligence bunch GPU about artificial A this The for large hyperscalers first network which it. cloud really, And versus And data the generates or – that cluster disaggregated. dynamic, data computers important basically the we service of is the which platform driving to a And acceleration then architecture of really center the of perfect is the called hyper know service applications hosting the movement center from perfect different. at to original of neuro and services the same this that’s the is lot artificial an time, disaggregation the architectural a acceleration. built And And applications of providers the really type last a hyperconverged is to new fundamentally cloud, the and today converged. change name ton type of But hosting traffic, is NVIDIA’s reason offer, today ability is running would data intelligence. data really to It’s a have built like that wants enterprise is most a beneficial that medium, software is that everything. one of

architecture, really two this so, are future technologies type the two cloud. of important of to because components, And of of types

bunch of you east-west yourself data GPU and low-latency ideal transition networking. run. fantastic the most for the could is up send is that And parts to now microservice them, as fractionalized of are across answer into is is the I mentioned, And needs buy the called high-speed pieces high-speed, disaggregated, that traffic. – is running is for is server our for was for really the an that’s And And other the server possibly of reason small at. is And the the – application this acceleration, center. it. one is That to and another networking. do in then One thing important a because application what whenever to to Mellanox broken

this where be more transactions And of and successful. where more And really and find be a – future microservice lastly, world comes find benefit that’s that’s and GeForce future is that’s going then, digital, And to autonomous we GPUs. perfect networking, the the benefit to going the cloud condition so, reason so we in ourselves our high-speed in really that’s virtual, AI-driven. why why then ourselves is more the is where Mellanox in. more And

think that this the – am it’s I happening everything there’s accelerated I the because – future, going just accelerated. of present, no us. that to permanent, are is – back. so, It’s describing dynamics And and and that’s But this to everything been is not we the just found it’s and

Arcuri comes with of line Timothy Your next from question UBS. the

Your line is open.

or lot. time. on a question also architecture CPU made light for had that think of you you approach, moving the But radical the of more I in that. And you chiplet I guess past approach. that manufacturing. have on on Jensen, a been guys sort about and when been side, manufacturing of the some clearly Thanks have And both a picking asked to I tiled are

then And question how side see the make the demand we is outstripping And why you on this, about then guess, the compute similar architecture, I have was So, a computing you theme power? do week Chips talk of and to move? this really won’t is of Hot ARM. far think you

maybe future Thanks. the and guess and to the opportunity So, whether I integrate GPU? CPU GPU talk you broader can is about

enjoying The longer center-scale or push push even system. cluster. start is Vivek an from bunch of And is we just But are the a data I Yes. computing unit into up we to center. data a running architecture hard. server The that broken those of the that the center computer way that really hundreds it future believe the explaining as a we we are And even a that is is a microservices billions are microprocessor now. entire unit a no We now architecture across a company of was transactions data microservice or day, company. entire computing whole

data the running center an the – several this application. from is mean, We happened pretty a running years. in center computing and And we is and that entire center were the we scale we trend, architect data a that, of I remarkable company, starting. data recognized that’s last thing. as so, And ahead that have to be company

X invented or, one our could the in together. of GPUDirect then Storage, that you this all architecture, If inspired of that the the idea holistic data Mellanox. and technology why with Ampere high-speed of last Ampere, invention in was And to node, it’s that’s make that way. world’s to recently, importance the GPU combined, we GPUDirect that think the first simultaneously bought NVlink, at current the networking the – it with you whether we we being invented multi-instance processors DGX. amount networking, world’s GPU, along recognize and or, which of the the for multi-instance synchronized our with working center and NVLink, way the all concept world be the low-latency software the of look have GPU. And were first we low-latency by the in generation one communications, in computing we across about fully possible invented first to means the We create created GPUs eight We an GPU

working instance, could need of think could a X powerful working fractionalize Volta the them time. each call being so together GPU lap of are a And performant GPUs one our GPU titled, instances you we the those little that MIG, GPUs itself, you So each your into if harmoniously of entire workload, are instance. fractionalize one don’t and tiny as much more than on multi-GPU or tiny those more

or Mellanox, would an GPU the way networking so, you architect DGX we larger high-speed, like. oftentimes, of the And low-latency NVLink, together GPU, using which even create any POD, create fractionalize could GPU, you a to connected whether larger it happens with with size like

ARM. comment about You question made about a a asked – you

ARM, ARM a driving then, And long-term application, and to bunch computing applications. to partner We have we autonomous to we bring or of it’s a ARM a robotics whether brought ARM. business And in accelerated that been Switch, of console the we and are recently, whole use Nintendo CUDA in.

the They very so, are worked you know very the And architecture the we levels it’s well one with that that great specials to into the closely. headroom energy-efficient, And guys. time. ARM incredibly team because And scale ARM really has is it about energy-efficient. it’s of very high-performance over

And so, anyways, ARM the we love guys. working with

Your from Rakers Aaron line with of the next question comes Fargo. Wells

is open. Your line

Yes. question. Congratulations Thanks quarter. on for taking the the

questions. prior on some building Just

views are look back the give at changed Turing what The as kind gaming over kind the first year? half to could we curious generations, has the the was was And base me through of a GPU of one, upgrade cycle, What last see I be could Colette, months? it or just then how on business, How Pascal relative I that X wondering, like just look you if that where secondly, does center still do business? before? help data on you prior – visibility us appreciate we of commentary of the in that the the calendar or you Thank updated installed you.

lot, Yes. Thanks Aaron. a

and it’s ray two technology we new we the technology ray a problem. GPU. is done to to still in, generation. super, the computer pixels independently be first it acceleration traversal the that want We We concurrently. tracing ray scene ray shading problem. tracing RTX technologies: that and added – in that’s of between – technologies. scene fast. core One objects are a we three ramping of looking current intersections that it world’s programmable Turing, tracing super complicated the – and the graphics, for concurrently introduced fuses the ago are fundamental the shader the complicated the and a makes be is of And the a super in and revolutionized fuses the Turing could that long rays generation running shading – that It’s we time the are And so RTX

at we examples, DLSS, which second that technology take bring when high-resolution artificial images beautiful. And called the learning it’s graphics. than you capability really, even to it The of to Super introduced native rendered this and is to is a deep its AI, doing so it’s example Learning from we the a using invented beautiful that. benefit the from essentially synthesize computer RTX the essentially neural doubled by of examples and what performance really it, is remembering thing Deep that as only previous allows commented it remembering network, of images learning algorithm you image through the have people run like type also one intelligence, more synthesizes essentially low-resolution that’s result nearly of a And super this not is us algorithm of beautiful, images beautiful new look called image, fast. previous And We learning native resolution. it Sampling, deep

and of ray as So, very tracing high well very speed. you can have as the benefit high resolution

our installed the installed not of probably as which close, of single-largest the in the so gaming RTX. you base Turing is, even is all GeForce that’s of not total GPUS, And called one-third know, even And platforms base world. of

new size of ray be it’s us. single platform, be the developers the to And now, great going which to RTX. with to going installed generation much, bring play, console bringing tracing, And is content game planet so, that large coming, important richer the future in these multi-platform, base, and with doing of cross-platform this we and creating are generation them the on they game the for the are we support developer PC And PCs, and process gaming of latest because platform, to of because much the really gaming is every content.

And the on then center visibility? data

Yes.

It’s represents Right ramping. cycle. of into center new talking Let computing have very portfolio that see answer we existing our ramping our accelerated QX early can in this now, overall probably product can me installed adoption keep markets. quarter nicely. but have into our But also as visibility than still you. you mind, has we less data in seen quite one see are center. AXXX of a Yes, strong is the if of I data really for in our NVIDIA the AXXX you our about bases And results, revenues. been

grow. still So to have a we lot

have into good looking with our We QX hyperscales. visibility

We a market little vertical and and of strong more because the have terms bit remember, in QX. given and of overall in COVID. uncertainty mixed to industries, to in challenged the outlook a a are a continuing overall, industries expect the lot we of the move overall economy. are On-premises of the our But But of cloud. do very quickly adopt overall terms

Your line next question comes ISI. from C.J. of Evercore with the Muse

is line open. Your

two your on have focused taking headway I questions. or questions. Spark know revenues. I Thank as get – there’s you Jensen, would you XX% Yes, hi. made driving for kind of platforms visibility Thank and obligations, X If I where that platforms excited, gaming the sequentially. particular Is year? if you comment grew I last question, second September to And inventory Jarvis, you whether real at launch? very half prepare into guess you. you love look of for on the Merlin, whatever. hear X the the any the over are outstanding months can purchase the back or then And you recurring

much, Yes. C.J. Thanks so

really strong for second expecting gaming. a are half We

is that well PC is really, sharing or create on requires you because cameras Fortnite using, of with they now, things other systems now, PCs game their friends. has And it and and just content. out to could I of even may games They captures, art. amazing really are large a I the it system has reason for e-sport one – create with gaming are and the keyboards very Minecraft the using They but the gaming The broadcast hanging community. multitasking. combination an FX It’s it best are because of medium. be ways different The described, that ever. is number of They that become And it works exploded. the think require game a such open streaming seasons people because all or and game format. gaming this are e-sporting, like an their way

We and so since the become run. platform the beautiful, large are been programmable graphics, we PC has just introduced years It’s gaming. really it’s second games And for ago. And the so, a not the amazing such that it’s a bar and XX really, home next is RTX, thing the about raised really this with level. computer shaders

because leap And shaders been better. tracing as never RTX a the programmable and it and getting so, well the like artificial then launch. giant gaming and is to think for better this. third be brought is most – gearing the as been But right gaming. are many been making – PC gaming how there’s And better, There’s season console for going home, And market vibrant a developers the And gaming last it’s XX both ray factor years, ever. we big I now intelligence people has have better at people around the depending leap. are – is amazing how world up and of the really to game because the the on really

on numbers seeing Switch all partner, they make the to time. mean We are Nintendo, – about gaming all their Super platform was Switch our mean, all time. – from successful sell way are amazing consoles console best than the of I Famicom. more already Nintendo. one the to of I than more gaming most of which has

them. And excited so, super I am for

so, a half be And going huge of I think year. second it’s quite to the

Goldman Sachs. the from of Hari comes Toshiya of question next line Your

open. is line Your

I jump Colette, I second – and I answer didn’t we felt question. on C.J.’s it? Can missed like

referring I was – that the regarding part that inventory are you Is purchases the to? think your question that think our I piece. on

Yes. the Yes, that’s one.

mind, full long as Yeah. for between and are the of gaming. now think our systems produce well that extremely what for have that center, C.J., of data we both we lead about overall do, you you complexity are the times, we between when the seeing what the Keep building, QX to growth in sequential terms in we need QX products that as

second in for in are it Nothing guidance. that to about those that So, of our half. have is the revenue got unusual QX preparation hit all other numbers we than, yes,

Okay.

the from comes Sachs. and question of next line Hari Goldman Toshiya with Your

open. is line Your

you following-up on I had data Colette. Hi. Good the afternoon and just another so one question. Jensen Jensen, for the center business. much for taking and one for thank

think potential As quite your of next your of the a have that talking the data And the Or enough in breadth hyperscale over about that digestion you like do I drivers worry have growth data are of call the few built that your X business customers data know, idiosyncratic business? Curious, are probably one guess over you trend ramp? center part months? months. the XX is capacity center center October you enough within it, that for about basis XX peers on been something X to to you it about, AXXX networking, margins have you the quarter next positive. on segments, sequentially. across you second margins, for Colette, then color like mix the looks provided gross And gross remains the and on within those individual compute Based the XX points buck just to guiding pretty for down you your

Thank in decline marginal the quarter? you. driving the margins what’s gross in curious October just So

Thank Yes. you.

in terrific. to the from by computational deep our So center I services to are and over seeing several years, One XXx is artificial more really using that. number requirement – artificial the And in cloud – that and providers. all of that service factors. a the intelligence are has we thanks say of more the is than last the proliferation breakthroughs fair think – the it’s learning anywhere And question. to are year data tied each been few in really intelligence for The trend

X,XXXx And the models, last AI seen have somewhere so, these create years, in in computational models between requirement the necessary of increase deploy to X,XXXx AI to and size three we the to these models.

– and are the with along probably answer of and – the I’ll are using indexed And how I and And are is so, breakthroughs of AI that. one usefulness and to of people question we the.X remember C.J.’s the AI it. trend that now, we this

One is are of based of for big these – demand. gigantic the things what that and how that on should on look are ones, three And one. AI services deep learning all for and kind you based there we look breakthroughs just

there breakthrough, a complicated to second text as to Of which the vector system, the that way of ability is reduction very called use major text them to somebody. on on and a essentially deep use and is of teach create how Recommendation reduction, recommender trillions it’s course, recommendation the ads a take are embedding, body videos and make predict of web. dimension could language deep system, that so learning recommender videos and one understanding. you a dimension that to natural The on a to

news to form in – is ways and the just both require going true instantaneous. that of recommenders need to amount information You them, to that dynamic these the recommend is be

answer predict the that first I going And the going recommender one is you’re years. predict so, models what breakthroughs just there with happening fast. to understanding, that ask. I automate. can clever. started I the And we’re right one natural just the have the haven’t mean, breakthroughs robotics, conversational that And gigantic And in three about last second make that are are conversational engines they in to described. there the the talking need for system, the cars, factories you, to improving based AI. all self-driving seeing really super is And They’re recommendations to language are the the even pillars you are is on third we breakthroughs that about to several

that indexed And you, kind first are so, similar, answer and to the C.J. the to we’re Toshiya, AI. one, of on

indexed second, of we’re The breakthroughs to AI.

possible of do to the graphics can is and workstations to the cloud, that So virtual thing more workstations to remote that is rendering in cloud. indexed it single NVIDIA we’re technology. in cloud. And movement the is capability third workloads in It continue more the And to now consume the cloud. every more and then

few and where do these like only other for the applications AI the virtualize allows I to of big like have training, conversational applications types microservices data computing could computing, is cloud. you And up accelerated NVIDIA architecture analytics applications to that you workstations. applications applications do high-performance to scale AI given in scientific have big fundamentally all analytics architectures. applications, can computing cloud. data in a different These do You you just the

are just virtualized, universal. is microservices-based, that And scale-up-based. facts these platform three Our complex, supremely I described and

all the of The that three And reasons they because Mellanox the scale-up. early and of we are that. they complicated, just I bought described intersection one of that. of dynamics the the and very, powerful the at are storage, very And and of so, are these – the networking, bare-metal virtualization, intersection these are security, it’s at at are core why yet. And the at the all stages

somebody through TCO. money, as adopts a we well scale their GPUs. And we TCO Their out, things lastly, saves is in amount multi-instance of ever really indexed we the application. the roof. GPU created. such those as Ampere It are reduced when is up for And people is – – then product to. scale gigantic that up multiyear scale Ampere, value And both fractionalization, the a as multi who universal – when just money of able use first platform it’s tremendous their it so goes out up scale It are beginning in reduces breakthrough. it. It’s speeds GPUs, new It cycle, we’re a like introduce it the to

been multiyear and fastest enthusiasm has This this ever we’ve cycle is the the of ramp had. fantastic. beginning the in we’re so, And

so, second to on going through the racing keep half. we are And

QX guidance, little mix, Okay. is have. of ramping And, new our small we products we our within from architecture a associated a but really with in our a gross regarding decline margin guidance regarding gross you QX. of Toshiya, asked terms of also have that bit our And forward question that going just Most margin. in Ampere the

and our are margin, below of included, company which also strongly expect we that will total our growth, to overall quarter shipments. be be up in quarter average gross Within expected consoles continue keep mind, is over a to overall totals gaming and be console data will in the QX our that for keep mind, percentage that strong growth gaming QX. given lower So, likely revenue, between center

when to GPUs going the mature, new are have too. as to architectures We over time gross ramping be Ampere expand our those we margin ability

with the Rasgon from comes question next Bernstein Stacy of Your line Research.

is line open. Your

Hi, is question question. into wanted little my for taking dig This I a a bit. for data to guys. Thanks center Colette.

QX. to similar So, what’s little be the driving that center was center, trajectory? limited you What data in does on supply bit you or about you maybe do to this have guide more a the X%. demand- Like this at capacity your lead have core up, like look times roughly data supply The Mellanox, point? like? point? right And Can products the a X%, like have you to the for stronger more quarter, us ex than into talk looks you’d data more AXXX trajectory if situation especially look at now? Are are what think center you available,

Yes. Stacy, so thanks for the question.

seeing. on complex, Let are quarters and procuring lot terms is QX And and of of products we demand start very supply, supply time spent we’re what previously. over first and about we in in every are our our outlook me mind, our have. multiple a Keep when we with of our think that aspect quite supply our comfortable the that

we not we center. with board are the we that data confident supply in AXXX. very just how that work. Keep that’s in the have But it’s overall across mind So

market. to to overall continuing coming are bringing VXXX, of We And are we versions new sell the AXXX also TX.

you hope in are of terms on QX statements that at our I where understand So, helps the we guidance.

We bit to wants that. will little see if add Jensen more a to

are complicated. plan. Colette the is love executing we complicated is this to are more have as sooner. mentioned. but It our we It sure ramping, we and is and when very most the Well, product, And – plan, a to

Got about growth the Got Mellanox? center just data core sequential it. follow-up, do think the versus a like you guidance, within how data quick And it. center

Yes.

that. terms we that largest growth driver from likely – the that So, most we will will to QX moving mid-single-digit QX, that single of NVIDIA low growth be to single-digits actual from of receive the stem will believe compute in in of

comes Stanley. question from with next Moore Your the Morgan Joseph of line

Your open. line is

well. that lot ask seven-nanometer, Great. a you the been how the you you the guys of on question the have in around – Thank There’s need I about you. CPU really could and very a see discussion done domain. importance technology. felt be haven’t I wonder first guys But process longer-term of if to

do of NVIDIA? important to does innovation process the early you into that is at be node? And cycle new how Just in the think how it factor

Yes.

technology First lot thanks, is more of Joe. all, a process The a than number. complex

architecture, we team. that I really a that with we us foundries, to engineer with we process we to level, make alright? are think and absolutely it paid rest the us. second for chip, at And procurement engineer a transistors that architecture engineering technology a complicated final process recognize it process, technology it awesome it’s closely, is have work metallization transistors. ideal will the thing, is and the NVIDIA Then simplified we part do that And process is part. for architecture, people of the not ideal Everybody almost the for, ridiculous It’s World-class. TSMC of have high work the so, of where and where analysis, down what of design really systems the sure we and world-class. the

a incredible. We our is are architecture architectures. rest for vast and paid between architecture. second-best difference the And and there’s architecture difference of the The the

– easily the the of side. in efficiency all twice are energy the time, the We irrespective number the of transistor

that. it so, be than more And complicated must

forward, a that really it’s that. lot then into we center-scale And going so, last of would computing. And the thing is say put data I about energy

at I fact home sitting data runs center right write entire a if this on a will you the engineer, of that Going software and data the because is And you’re forward, would at a know be you cloud. why optimize the in center you reason software for the scale. piece now

it, you no do have idea what’s You care. underneath nor

that, center And There throughput want sure to you lot in are of data that high really is is what as there. possible. make so, as code

our And is to game a so, to the new NVIDIA take decided has level. to over years what do

we Of very the best use the companies to partner engineer and to best start closely and We world’s the them the process best foundries, we course, create world’s best processors, with world’s packaging. we us. the with packaging partnered building for best

X.XD user And by whether are XD the packaging. and of world’s cobots. highest think I’m we first volume we it’s sure We’re are– the – I pretty far of still

start a from from there. beginning for end we but just don’t chip, And That’s us. a start chip. We great great so, the we

took the four the way take work is in record through one the the the we It’s networking, in systems, shocking. that difference way the in And the center. centers. – the energy-efficient MLPerf. is absolutely broke for supercomputer four the the is And this thing up world. the single the future it time. It Selene in and put about every future. data complexity algorithms, up it weeks’ this built that we center, And the most the fastest in Now and of AI It’s our that of system all We we data world, work weeks. the us all our software, supercomputers to data we Selene, entire supercomputer – shows is about of world, the something and scale do. seventh-fastest kind you

call questions this time. to have Jensen turn back over We further no I at you. Huang, the

On computing NVIDIA Adopting accelerating. model leveraging transformed The computing accelerated foundation three pioneered in dimensions. this the our we of is and have you. we company clearly passed one Thank point. tipping has architecture,

to most center architect, full-stack build is out capability, create like data partners, and modern computing operate are architectures to maker pressing is industry. computing the industries, their then data data center offering unit. and With the NVIDIA world’s is most libraries computer new second, dynamic NVIDIA that the the challenges. capabilities this data advanced a – tackle NVIDIA the company, The First, world’s we centers. scale can software NVIDIA company and a platform systems, to AI with most center-scale chips

virtual a NOW, to In will model. GeForce with Drive business the coming worlds play NVIDIA increasingly machines, become will and rich Third, through will NVIDIA-accelerated NVIDIA revenue today, advancements VR company software the recurring will integrated NVIDIA computing cloud, will and Industry role. AI AR. in our revolutionize is software-defined years, software Robotics the virtual add that content software. workstation an like and and important automate physical AI and will accelerate, NVIDIA

everyone. kitchen. me. Our Thanks, my October next I be will share you. to with some coming X, have exciting on from developments GTC Join again

call. conference today’s concludes This

disconnect. may You now