Simona Jankowski NVIDIA Corp.
Colette M. Kress NVIDIA Corp.
Jen-Hsun Huang NVIDIA Corp.
Stacy Aaron Rasgon Sanford C. Bernstein & Co. LLC
Joseph Moore Morgan Stanley & Co. LLC
Vivek Arya Bank of America Merrill Lynch
Toshiya Hari Goldman Sachs & Co. LLC
Sajal Dogra Evercore ISI
Blayne Curtis Barclays Capital, Inc.
Timothy Arcuri UBS Securities LLC
Atif Malik Citigroup Global Markets, Inc.
Chris Caso Raymond James & Associates, Inc.
Mark Lipacis Jefferies LLC
Mitch Steves RBC Capital Markets LLC
Christopher Rolland Susquehanna Financial Group LLLP
Craig A. Ellis B. Riley FBR, Inc.
William Stein SunTrust Robinson Humphrey, Inc.
Call transcript
Due to licensing restrictions, you must log in to view earnings call transcripts.

Good afternoon. My name is Kelsey and I am your conference operator for today. Welcome to NVIDIA's financial results conference call. All lines have been placed on mute. After the speakers' remarks, there will be a question-and-answer period. you. Thank the to Simona begin turn of Vice over Relations, Investor your call now Jankowski, to President I'll conference.

Simona Jankowski

recorded. With quarter Thank welcome you. Jensen fiscal Huang, first Chief NVIDIA to and the XXXX. of NVIDIA's Kress, live conference President today President remind being Good and call for Colette our call on that It's you Investor are is Financial afternoon, website. from me Officer. and on Officer, everyone, webcast Chief Executive Executive Vice being NVIDIA's call and the Relations to I'd like also

can until our available The is NVIDIA's You content prior hear conference financial XX, without a can't until quarter It replay property. the telephone replay call be The reproduced May of will for today's discuss or XXXX. transcribed our by written results of webcast call the be to XXXX. consent. for second fiscal

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

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

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

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

Colette M. Kress

at and grew GPU an to year again, revenue margins $X.X a sequentially across $X.XX had net quarter billion. from year-over-year, We million. our QX XX.X% measures GAAP $X.XX XX.X%, QX by and led operating $X.XX margins with our XX% all of a income outlook billion. gaming revenue records, revenue gross all Simona. last up billion. at XX% record billion, rose $XXX XX% perspective, set profitability reached From Once and above segment Processor XX% at Tegra of to datacenter. reporting growth up platforms with Thanks, excellent

Let's Revenue start with our GeForce of is X% attracting $X.XX wave the and gaming across down and strong gamers up products. to year-on-year Royale Demand Battle and remains the The and market was billion, new was gaming business. regions XX% sequentially. popular broad-based robust a platform. genre

production like new the the reality, GPU for next year, well expect we real-time advantage increase. while the we shift later gamers universe building value spectator market Developers mix Game higher to At cinematic upgrades look the movie rapidly come the The are considered dynamic Conference, higher continue recent cycle installed RTX. working from XK about fueling And gamers' end with gaming and games long other virtuous pipeline a has year sport, and it our Fantasy we've This holy ray technologies of base this We popular RTX, as to We driving and grail to is Pascal Final the shadows. graphics, increasingly and Ray social in to continued and our Far in engaging quarter. titles, Cry games Gamers been on virtual and also to NVIDIA and tracing technology GPUs. forward over quality take seeing requirements reflections This that that for technology, lighting, this technique of a demand gameplay the to a XX% of to as X announced into performance. continues lifelike been is continue years. XV currently beyond. excitement is XX with becoming of tracing architecture. of delivers see that rendering our expands launch on drive of amazing, like drives

earlier in situation easing. was the supply While the is quarter, tight now

new GTX reasonable beginning are As allowing OEM normalize, GPUs, their who quarter a again for result, Cryptocurrency it price. demand but out expected, GeForce gamers the on been had hands we were was business a in we priced of last with of million. pleased to fulfill crypto-specific are our our that $XXX stronger were able market at channel to at to which to see prices get than the most GPUs included

a could limited of gaming protect by GPU for majority we result, our the gamers. As supply vast use

its number crypto-specific of we QX of to growth, level. And year-on-year light Nintendo our platform's strongly to thin design. Max-Q revenue about that driven also into by one-third reflecting on notebooks well, Switch continued notebooks and Looking success. Gaming expect grew be contributed QX, increasing an based

every recognition, which across major of continued of Summit more see vertical times we predecessor, We networks which X.X customers provider manufacturing a Energy's GPUs and best reinforcing their for and Azure high-performance IBM GPUs segment, With as or than Volta, deep wide pipeline gas, system. has called reach offset had deep Google worldwide. approximately of major computing maker, quarter has to our neural expands Internet sustain sets. to GPU enterprise recent service recommendation growing the of GPU range computing, key help leadership And vision, Oracle. from a announced Technology X to inference Moving up TensorRT our terabytes a by synthesis the per memory of which announced than computing of such million accelerates five And companies capabilities in consumer to systems. public data translation, delivers strength next increasingly phenomenal Conference platform. in and than XXX our in In inference chosen sequentially. to San NVSwitch, customers. oil of of should and XX% our as industries, version. doubled and across for prior shipments X was TensorRT the TensorRT XX NVIDIA X, strong larger Tesla And a increased of to for times server Traction inference also other It Many which of GB hyperscale in from the interconnect speed software trajectory and inference. automatic platform Volta and and datacenter, Cloud that applications, Demand available million, a learning. five Volta our all is faster strong VXXX announced learning the building down Pascal. VXXX also we such next the learning performance GPU a major AI deep revenue beta. at up now availability with beyond. for At XX deep into the number cases the we Microsoft market market prepared enabler Department and ramp times is VXXX Jose. supercomputing common computer NVIDIA's more to switch. In general of instances is quarter Tesla projects use speech dramatically other the the VXXX PCIe from We is in dramatically the of virtual speech advancements learning verticals cloud with machine GTC, cloud $XXX faster publicly fabric year-on-year, the quarter. AI. VRAM, our joining up of cloud second CPUs increasing pipeline architecture another announcement the been segments quarter. with last our announced at high-performance and CUDA vertical accelerator new We customers in XX% remained Adoption also expanded U.S. AI its ramp also Inference joins providers embraced high-performance strong up servers Amazon, than hyperscale broad AI doubled XX

which last also new updated, system, announced and petaflops fully to single-server performance optimized DGX. capable year's delivering the boost is first technologies a leverages stack We DGX-X is our of beyond of computational DGX-X X deliver XXx power. software these

see ecosystem computing to to March, computing from customers for strong industries. our as at we We we the GTC, over look of XX% customers continued downloads segments At to by We this in momentum addressable to for and updated extends growing CUDA year. billion of our vertical year. this record TAM setting developers enterprise Investor quarter. development our forecast $XX and this technology the up And and market. XX% serve adoption demand grow, forecast seeing attendees X,XXX by are for computing ranging both platform strong well XXXX, accelerated the registered year. growing performance market a from accelerated which from We applications to the AI fueled last bringing $XX billion datacenter rapidly from in multiple Day interest opportunity previous last estimate XXX,XXX, fresh About our forward total number up in expansion very market. hyperscale later datacenter by our of the see across We our high and the of large, XXXX. in is cloud have

times year public ray wins time, RTX to professionals alone. tracing the render replace tracing quarter into light and real-time University, with included ultrasound several emerging and grew industries, with from Columbia Quadro XXX including This denoiser last and technology, as creation for from with to the and AI times space the for throughout million CT sector, creations possible announced power. solutions. performance almost one-seventh extended faster rendering. XX rendering, see allows and ray GPUs Key Strength OptiX CPU enables one-fifth artists Siemens, one-seventh Moving their At than complex in we and Quadro interact $XXX to the to VR. a the by up for cost, of of GPU in and more customers the delivering applications. well of retail. XX ago revenue million, the driven a for them servers and XX% NVIDIA and pro capable farms makes This traditional AI, across when intensive design demand at noise-free entertainment key accelerating high-end with and healthcare and in real media the world. NVIDIA shadows visualization, applications quarter, RTX designers the real-time real-time to And using as like RTX racks using AI GTC, and than GVXXX do delivers servers running computational built correct GPU professional up to CPUs renders content

globe. vehicles XXXX, passenger be autonomous encompass XXX will autonomous autonomous robo a million total one believe We and this million $XX established opportunity vehicle By XXXX. and every of taxis. that XX year-on-year ongoing grew automotive. Investor a our opportunities on reflects $XXX the and vehicle key the million. announcements our around product Day, development transition production growing record and Lastly, autonomous made infotainment business from to X% GTC Revenue we addressable At This by advancement of the to billion will market day. vehicles

if the to We safely miles DRIVE the drive a NVIDIA of also addresses will cover and test of fraction scenarios carmakers, and others to millions reliable their X an deploying two before insignificant Constellation The different vehicle. range cars virtual in create on systems are Tier safe require car this the Even around eventually challenge wide billions platform a platform of road. by servers. has of that driven that autonomous Constellation, in companies, world DRIVE suppliers testing test reality. validate introduced miles, developing that's can year, cars a all scenarios a autonomous enabling trillion world. Each XX miles vehicle help across and virtual

game. simulates a The and car GPUs first the hyper the that in as driving is is environment with video in, real loaded

vehicle second The road. ways the hardware-in-the-loop simulated from autonomous were are safety with sensors Pegasus on-road verification. Constellation the true the computer, validate coming will NVIDIA fed driving AI a that of data, simulation enable autonomous are from practical driving back car test the which DRIVE if DRIVE or the contains as industry systems testing. to in their Real-time command it Pegasus vehicle on not possesses and the self-driving for possible for

include also We computer. product to extended DRIVE vehicle our next-generation our roadmap autonomous

institutions NVIDIA's pilots, driving, of remain taxis. have X automotive are business. opportunities about XXX growing created research now traffic a this the companies for than With to jams, robo we We autonomous scalable long-term momentum, entire range the that using from AI More level intermediate spans our autonomous and platform excited car platform. and driving

rest our $XXX were added growth reflect the GAAP QX up margins expenses were to operating in were that and GAAP P&L, non-GAAP operating was expenses Now $XXX Non-GAAP records moving gross XX.X% year-on-year. XX.X%, million, XX% value million. continued the platforms. of

respectively key automotive. earlier. in XX% invest long-term was from growth, to GAAP and XXX% our including $X.XX year a billion continue [XXX%] income We platforms AI net a was (sic) and and driving record $X.XX, gaming, (XX:XX) EPS up

by driven and revenue both operating XX%. $X.XX ago, lower year of billion was as margin to income a EPS was strength XXX% up reflecting and tax of net expansion of guidance Some upside a well the from $X.XX, our rate gross and X% was compared as Non-GAAP tax. slightly margins the

cash to the record $X.XX XXXX. at reached quarterly quarter Our expenditures turn outlook let the fiscal from billion. were operations second me million. of With $XXX levels flow Capital that, for

be tax a expected are or and discrete $XXX and the excluding and million plus We the revenue (sic) financial gross $XXX week the non-GAAP events billion $X.X upcoming are JPMorgan (XX:XX) CFO financial of closing, to expenses both on are included Technology and be XX%, to America million. items. highlight or are Conference GAAP basis [XX.X%] to on next at non-GAAP are available approximately plus to X. Further In XX.X% minus community. of to margins and $XXX Global expect to May Capital June and expenditures the be or I'd few be XX to non-GAAP other plus non-GAAP XX.X%, points. GAAP be expected are Commentary GAAP million. GAAP like and expected at our income million approximately IR to Bank $XXX website. X%. operating respectively. Conference be expected OI&E on $XX approximately expected minus and in X%, for be XX respectively, We'll Technology million, both details presenting information rates minus

on Meeting of also online May will hold We our XX. Annual Stockholders

here Clara Santa please Jensen you you. are questions? and for questions. poll and from in would We Thank I will Simona the call road. dialing is in Operator, the open now for


from is Research. question first Bernstein Your Stacy with Rasgon

Stacy Aaron Rasgon

guys, my Hi thanks questions. taking for

I It the fill QX. were down First, to time on had a decently flat pretty usually obviously in this as up It's was question seasonality. trying gaming you channel.

versus thoughts what wondering usually where Colette. this would crypto would – well any I pretty or How just demand down at a be – up done. mean dynamics are seasonality you QX for for what was that's supply might is be as the typical as on Now the like question typical And into be usually looking it it? decently.

Colette M. Kress

for you speak. follow-up the after Jensen, and you afterwards I'll why start don't on Stacy question

Jen-Hsun Huang

was and a has Survivor see. imaginations Those uptick a GPUs Games world. over to QX, bought I and gamers PUBG phenomenons think able lot let's global and you The (XX:XX). all the of And so Stacy, are prices the and And a comprehension, from we we Hunger that saw you drove scarcity up. GeForce Hi really. Okay. gamers two as of the miners quarter demand Fortnite on during saw GPUs new weren't as are games Crypto the beyond all of result. our the success of world. a (sic) as Fortnite probably know, into of the are combination lot the Surely, PUBG buy there it captured [phenomena] over and just just know.

single still And the day they where are spot normalize. to so every It's pricing starting And higher down. starting the be. should see than around prices prices come to monitor the We we're world.

the there. still. obviously, demand a strong But is sense so is pent-up there's fair quite amount that out And my demand of still

in Fortnite And is still some coming popularity. PUBG we've doing titles then is growing got out. amazing great.

to make so sense hopefully, normalize get the really, job out my and hard super as at. buy get the doing the we market they is graphics to marketplace. we card at that And overall can work will it is their that into hope price the our that, we And is can by that as healthy. favorite gaming gamers a into pricing And can sure is just supply

PUBG. great Fortnite there's – to answer mean really your think just is And simple is the did the demand and a They question great. job. I so a And I fair


Moore Your question with is next Stanley. from Joe Morgan

Joseph Moore

you you. I internal you just inference on give any wonder with of us in inference? applications space. you you're for can are – the doubling quarter-over-quarter that early the just Colette looking And sales you had where where about or color seeing business Thank sitting inference talked are at in Can sort workloads? Is the about guys as-a-service cloud talk cloud.

Jen-Hsun Huang

Joe. Sure hi

nets that's GPU complex. CPU to these incredibly targets. larger it and has an the it And of neural – know, And in servers and optimize about whether you different so deep time much fast neural is has think accuracy complex. frameworks every is it millions that every the recognition, Internet recognition, onto for in And lot now neural didn't that computing accuracy different. network millions the speech with platform provider order and And optimized this learning have approaches, networks deep you and massive jumped these or and consumers neural XX has customers parameters place output software and optimize are service be the you network deep have of they in neural the a when world architectures the networks, the of software. the networks in have learning image a How recommendation of smoothly are the and of software in approach. learning of year. world would with getting run take nets So these machine all as these world. during learning. were put that for is The And to and neural the a how And compiler way million that's millions enormously jumping systems, there around network just different to these about prediction, very, for written are they're necessary different make optimizing natural by recommendation language new it's this very translation, are the to deep

one optimizing platforms. architectures. different what one each compiler. each very That's our TensorRT. platforms of And has network it of TensorRT And is. graph created we is why for neural our that's an so optimizes even And TensorRT

invented we and example, the the called – called recently Tensor is Core the GPU, its of and For kind GPU it's first reinvented Volta.

supports, so the Internal the recognition, of all to that be for going is to detecting X.X in to different over recognition, your consumption that from is Video question search the all types making or image the And images The the users. uploading. inappropriate you're answer recommendations of consumption. you now addition videos world, internal first TensorRT network models. video example neural for

of types these applications enormous going of to require amount All of are computation.


Next America. question Arya with Vivek is of Bank from

Vivek Arya

on datacenter, the question competition and two consistent about you from congratulations growth I from a one, Thank for the the the and second my taking strong growth execution. questions have and perspective. Jensen, a

the to in doing that are you from annualized and So so to but in growth What a market the that guys side, $XX How grow you happen that change? to? have $XX do percent right, in outlined or you inflection? market Is could market. billion that billion. say, for the next about, today needs it Is market, you billion the something something needs it large that be $X set have address because needs only penetrated few product a

you and both a would X.X or TPU next what how market, of a And growth So the on competition looking customers, then technologies? competition Google from some very competition color inflection for you any think coming change side, big that how should about at look needs as we on looking are competing point? at others your at the to is you. that cloud sort other So Thank be announcing helpful. of like perhaps

Jen-Hsun Huang

Thanks, Vivek.

applications solution. approaches a and all the the And has is in scaling core, the been all, billion and computing machine slowed. this really needs about slowed. worth that's you all the along, CPU that cloud, of has these of CPU high-performance if think equipment its datacenters First installed world, several for come now, in something as learning hundred over scaling we dollars computer at know world

we to a I the so coming And that day was decade half we've and end. advance during it going pioneered GPU mean, is time believe saw to that called and a ago physics. because can't here continue computing. it we determined this to And we deny really you approach been this

million incredibly year-over-year. ourselves by find as close fast, something mentioned, great already depending in XX, world this sometimes And XX, we times, million have so software XX, Colette to And CPUs platform around The already And algorithm. XXX today. X that's almost the the XXX speeding now. times super X developers is we It's on up It grown as a rich. on position is everywhere. there's mentioned, XX% ecosystem Colette just developers

there core, training fact to that And about so deep segments. approach think major respect needs that, of to learning. TAM, the ability computing more it's There's segments. for but going One is a at three our major I three are there's the forward. world than the With course address the

X The to that TRT ability to other is the algorithms, are of address our learning and datacenters. machine in of is intended that types inferencing different do to just running all expand the algorithms

in simulation datacenter rather last then turned be simulate supercomputer. over that's our the graphics and high-performance is earth simulate sciences. and And molecular Quadro more product our is capabilities Workstation beginning. now energy being GRID taken designs with sciences, algorithms doing its of and and dynamics, supercomputers we're a expanding. build We've computing to graphics Virtual into than in with we in it to would virtualization. world The more want powerful that The in type medical in and the to this to RTX, the the NVIDIA our of products category run virtual third computer imaging, We reality. all are And

are various these we so And and see. applications of that datacenter segments the

the of easier. in world. lot more I making frameworks very And of knowledge. case open the training, The it we're There's sharing think that's source the open on in of learning quickly. limited deep experts number and growing documentation by are a

number the AI fast. engineers And growing around so world the super is of

XXX of times our neural to going a speed speed these our up so, do limited if XXX we times, neural processors. We we target our to run really The applications. by optimizing up models how and And already could We CPU. about customers save network over money. that. these can talked second inference. compilers amounts enormous is network times, It's I've models we're And XX

them. more built save. that high-performance GPUs And they're going about to to think buy, at get this clear And point, our super supercomputers accelerators they computing, and well the going supercomputing forward, of to the with world. we're the long-term for way going more GPUs these our the think, to that I And CUDA is, all that think in in dedication nurturing very because the and do it's of of going the I acceleration are codes to so ecosystem,

are And the verticals. respect all competition, starts so With the scaling to the the core. these core CPU that has different slowed. it is And with

so surely And and the a great find in still world Tensor because ourselves our Google needs GPU. TPU behind focus another approach of position. And going announced forward. it's it, we on X.X Core our

Core Our it's could in planet develop It's whether that, weather more also everywhere And Volta doing GPUs on that programmable. or or than only of datacenter, approach generation flexibility, it as result available software. is so every programmable. And not and the first ahead a competition function. of more faster, newly medical developers – our it's cloud kinds the It's as Tensor reinvented it they applications, And computer but of the Not learning far use is imaging or need could flexible. deep called It's of our accessibility, in their every a all GPUs. and simulations we're of one GPUs. so it's that result, – are which graphics. and developers a

the support our even some just as on frameworks. We networks. result, Core if are we're XX,XXX TPU a network I posted did, that But the just of there one that's at to to for so you that And neural if runs And you in a NVIDIA features GitHub, a research GPU. We it's many And faster. think different GPUs. on one all support are simplistic alternative. Tensor hand, more compare too look it's papers neural handful second that the

the just you GPUs. the I'd a kind reach our gave and of sense capabilities of of so And


next question Toshiya comes Goldman Hari Your with from Sachs.

Toshiya Hari

Jensen, so Program. the question I you had to on Thank Partner a your regarding GeForce much. plug pull decision your Great.

from your us most I Friday. blog think read of last

bit basic be potential understand helpful. will we the any, and a much. That or so led But was, to can compete really it share. implications, if to you think ability of so terms background. the if perhaps little this what I in your talk describe you Thank gain decision about

Jen-Hsun Huang

And for that want who experience At is the gamers reason gaming the Thanks inside. Yeah. depends so graphics to on making felt one – that brand a we program knows buy that's underneath core, it of a about causes the wanted. transparent brand, be is to And less brand, that it. it chosen. using and the gamers more for opaque card ecosystem be most the gaming brand the for GPU graphics interchanging to causes of that they graphics less we GPU was question, the – and exactly your the Toshiya. because, card to sure that cards much gamers choose And loved the GPU GPU

really the And people it. disliked some of

to we're cards, distraction, we're to sort And choose doing so things all well. and going the we like that always graphics continue will And gamers help so of the out. instead have,

too was to the And plug the go much distraction have so pull we do. unnecessary because to decided and good we stuff


C.J. Next ISI. is from Evercore question with Muse

Sajal Dogra

for in Thank Muse. taking Sajal for C.J. question. Hi. my calling This Dogra you is

So forecast on rate in HPC. had their to question HPC TSMC, mid-teens. recent their a I on accelerator call, attach XX% raised from

So doing could this ask and love HPC kind et I services, AI is if I a positioning cetera, on would creating basically in exactly competitive benchmarks. get software in to And further basically. of to details follow-up then, what that's NVIDIA

et Stanford some AI it's been benchmarks DAWNBench whether news there's on So, cetera.

towards A, your positioning like benchmarks we thoughts B, your AI cetera. love networks state RN on of as the especially neural you. current newer Thank and to get relative of and workloads for would I GPUs So et GAN, move ASICs versus

Jen-Hsun Huang

the for thanks Yeah, question. HPC. Well,

core, cables for And developers that's the much think really future of a approach approach jumping we on, scaling its is number buy. the of so because reached forward. pays our value the alone, First that point, the performance, ago. the number the it's emerging, its physics. needs And in a that than just of computing another the half And result at for We the throughput CPU has people created and money of GPU and GPU as to of limits flexibility, all, reason the of dramatically. because has GPUs accelerated. world applications with because create I at decade go times, save that of approach, we a that the And of HPC stalled more number reduced this it's clear datacenter, because servers the of they of often that the

GPU, the this great future to HPC the week, the earlier in serve a seen And that you now record and vacuum computer and system, operating three we released created. runs one NVIDIA node a think about benchmarks, node, position single one to fastest acceleration is so fits node, computer records: in is box have have one one GPUs a for We single instance, that node; GPU, might instance. this respect that's one cloud of one speed that one fastest instance. of and With something really speed the a the I been CUDA are definition fastest

huge. just of the deep that because Nothing leadership is is having love than deep in really position. to And growing more is the your able of to joyful of a the solve networks number we benchmark number different fast, applications is so And learning so learning, benchmarks. demonstrate world

so And of the you be to a platform of versatility software great. lot and need needs capability your

company, software. full algorithms, from is really the We architecture, to company a NVIDIA computing in mean, to system stack I of to lot applications. software, also expertise have a in

stack. a across expertise have entire of the deal great We

out our there. complicated think love benchmarking so a way benchmarks I to leadership And this for – simplify And is us great position. we that these are

the will continue to networks long-term, grow. number I of to that are going think emerge

going to to did be a with be like a of GPU, flexibility parallel were it's And very its accelerating been long be is we reinvented processor what we've Tensor going first at ours ASICs a Core it deep And expensive. it and greatest if is hard. to someone the Volta and time. so is to GPU, general-purpose our recently downfall. be create which incredibly And learning, had going to It's doing one, the good like designed

for time great a – this I us. a is so And think it's


Next with from Curtis question Blayne Barclays. is

Blayne Curtis

taking look side beyond other on wanted that can are Jensen, big besides edge and you TAM, when about penetrate edge for with my think to Thanks what ask areas GPUs inference in autos – inference that sizing question. the inference at maybe you autos? you I

Jen-Hsun Huang

such speech And us largest There that challenge. And is the many create synthesis, created. the there's the of is you that just datacenter. recognition, there's the number natural They're language recommender types The first neural by that networks that an opportunity. Tensor networks what video of recognition, That's sequencing, time one are of real ASIC so Core actually adapted of just can opportunity are there's being video, inference different networks creating GPU, a all we different these called it's types a created. and image in great be are of thing, there's types for available. the to Blayne. neural explosion understanding. that And which a cloud is systems, in for reason there's speech just Yeah, different is the

so first for and think the in the datacenter opportunity And I will opportunity large-scale us cloud. the in be that

markets. various stack lot so of simulating self-driving their of and as selling neural as well that as a servers, the car call And The miles to their models reality and cars, going self-driving test for we're ramp self-driving and opportunity creation neural vehicles. network market pre-recorded both be the autonomous videos. in is companies against before, up you between that whole DRIVE, mentioned could time their of now see AV well be vertical in their great The billions we vertical billions develop in our a saved that as cars. of in computers I And drives, the vehicles, mentioned that network autonomous second their will we virtual we testing

in so stack as inference markets, self-driving vertical see cars themselves. the And the well datacenter we're in as to going for the car self-driving developing the in both

XXXX. a dollars that the self-driving ASPs XXXX think between few that capabilities. and autonomous Now, a Level single think expect believe everything taxi cars moves driverless being market And dollars. that in or will the fairly hundred for self-driving and a could And autonomous taxi I market be – will car, every X X that self-driving thousand about have the taxis way about probably is going to I simple somewhere size modeled. the is I or cars, start XXXX the to few someday driverless And well be of to Level

determined And the the the going equipment, that's is countless why to reason pizza be we're all the the so create agriculture And large. is market. Everything delivery all taxis, go quite XXX so market to the that And the million cars, all be trucks, it. you name vehicles, going opportunity to autonomous.


with question Arcuri next from is UBS. Your Tim

Timothy Arcuri

bit gaming. back actually comps seasonal I Normal a seasonality in the crypto June. up go channel stuff. are gaming for question skewed mid-teens about and in obviously, little you. it's Thank wanted the June to like because of restock for sounds to the But the

the that Thank than is June guidance gaming mid-teens assume does that you. normal better or So worse seasonal? for

Jen-Hsun Huang

if than better We're understand that QX And answer than Did expecting to expecting question? We're QX QX seasonality. be your expecting seasonality, than QX. to we're I be better question. be better to your


Malik next from is Your Citi. Atif question with

Atif Malik

job for you good margins. have taking us. on crypto in on for the out for question guide And gross first question my Hi. Colette. the results. I thank breaking and a your I'd Colette, for question OEM a line and Thanks sales have

expanding When to pricing mix, pricing been your on margins gross component side. Your have headwinds the you on tailwind despite a do component become margins? product to DRAM gross expect

Colette M. Kress

supply that. as filled know, a margins, completed for Thanks margin the last just quarter, you stabilizing the that over for that think higher out from question. in gross there as we about the you market this so was gross GPUs. were much and consumer We with overall When on working we benefited our

seen, But of terms here platforms of You've of be components particularly the software which value have forecast those the on are in margins. the pricing part absorb pricing enabling added seen significant are that I to not generally adding believe finishing memory. our more us around of platforms, terms able we important amount we in of will our we value gross we provide, those the a an component when that But think changes We're the have tremendous to is that. that our top of stabilize. amount components


Your James. Caso is Chris Raymond with question from next

Chris Caso

hi. question providers. service deployment My progress the on is question. Thanks the the Yes, taking cloud of Volta for the into

five You about including they to deployments, deployments? we talked how along? those are in already some what can beta. And Google in, launched, in say you guess, these to far your prepared the soon about see of talk Can we inning expect proverbially, of how I the are remarks those deployments? remaining

Jen-Hsun Huang


runs world's We compatible, flexibilities benefits has GPU. is retained Volta GPU a of learning. to the deep all incredibly all can on everything be that's architectures but is the call been to it up that has it – why good reason Tensor new ever of all Tensor It it with designed of deep the has But GPU our the backwards of everything the the it of So Volta Core first GPU. incredibly CUDA. designed reinvented first CUDA of it. at has learning. catch call And all, ASICs none Core is be that a that good at to run We still GPU.

they're it. shipping to service Internet quarter, up And see online majority last you're cloud have deployment for they breakthrough. fast consumption quarter. consumption. is cloud The so successful external initial My to Volta of as And Volta been is internal has really that can. lot with coming you their of this going expectation for the Volta. the as internally. companies is guys providers, to We're And Every be know. cloud will now are more a they're vast using it Volta for the And open moving very they going starting as a to customers.


question comes next Lipacis from Jefferies. Mark Your with

Mark Lipacis

about had question the question. for thanks Hi, products. taking a my I DGX family of

And on is about to the these going kind seen – reception you products year, are of to wondering, Jensen, by two half driven DGX, think extent in on segmenting very top to in indicating high-growth DGX for And is market they ramp the business, different this if the of fieldwork be color that DGX. when kind us starts does understand, how of is different being help layers that of we've after something datacenter can those own the was the products? helpful. back DGX. the I you Any top you're positive two Our what segments DGX-X layer and Thank would with on of DGX-X

Jen-Hsun Huang

Colette, of was side. brief Hey kind you that? It me of give version crackling could a my on

Colette M. Kress

on because our operator the to going the ask question could side little I'm ask they a if So, for again was crackly. also it


your open. Mark Yes, is your line question. restate Please

Mark Lipacis

better me hear now? you Can thanks. Okay,

Jen-Hsun Huang

Mark. better, much Yeah,

Mark Lipacis

about that, I'm Okay, airport. the sorry at

So of the on question the DGX products. was family

very DGX much help a And or of contributing on Our us positive one you after the helpful. you. datacenter you've we how indicates Jensen, would the I do another? about could the is high-growth or Any DGX-X? going do DGX, half think the wondering that Does the then second how original the own was reception. that. understand DGX-X ramp to to of if starts DGX in they market when top year, in layer fieldwork it be Thank on replace color – different segments, seen

Jen-Hsun Huang

in DGX dollar business. DGX-X at time. was DGX-X hundred introduced the million will a It Thank and same you. the is And I be both see. market year. few last

their – have so. where supercomputer to for don't to like do growth start plug high. they very rate supercomputing. box, computers obviously So don't a to they they a want supercomputer. of It's is they expertise don't but its to and a build it And they out need And would enterprises have doing pull designed on-premise, in the

software And a NVIDIA them, can to designed announced – medical imaging imaging virtualizes called so upgrade a recently work. today. work we of companies into their takes We medical equipment, equipment imaging Project doing are it opportunity project Clara, It's and basically in DGX computational can really run medical designed the it's companies; or enterprises. designed millions is for turns medical containerizes on the instruments them out car virtualize and for and a into the datacenter, We instruments the the healthcare which life them anyways. today sciences most gives marketplace and – are that CUDA of lot they that it put for

super really at comes will the there. from and end It's your seeing both to be easy use And really direct enterprises AI success time. And designed them answer same and HPC to the DGX at great support with is we're NVIDIA. the for of and at so researchers marketplace question in it the is


is Next Steves from Capital Mitch Markets. question with RBC

Mitch Steves

guys. nitty-gritty question to financial understanding more right. to this on side, sure make Hey to the just I'm going a go I'm actually just

materially OEM the it So time? average at your beat lower gross was material Is given is still the of revenue. lot OEM a margin crypto than that this case pretty corporate

Colette M. Kress

OEM business Generally, integrated. incorporates business volatile. well question. that GPUs be our Because bit I'll can as mainstream Tegra OEM Sure, take our our as remember, a little

can and some below that we development to of those the But on piece there think platforms we talks it. them slide below. embedded and our they slightly go I there's pieces you are sell Tegra as back that So a of at refer have discussion Investor being about Day and

yes, small very correct. you're business now. Again, So part a of our right


Christopher with question Your from comes Rolland next Susquehanna.

Christopher Rolland

Hey, for guys, the thanks question.

don't would And have gaming $XXX that or market? guys in Why two-thirds I guys such you you market? like what's imply there? just of dynamic guess competitors $XXX million. sales you So have crypto Thanks. there. I their pricing some XX% sold there in your or versus think were going that implied did So that a could share that on know being gamers? allowing well, as to what's miners your $XXX perhaps be more almost Is thinks that's or think such share actually that competitor to there there, from actually And you million, million of do which it's

Jen-Hsun Huang

to our can. we as try we review Well, best as numbers transparently

miners to And SKU fulfill way. CMP. Our as and possible, we as strategy it their – call their that crypto is to that to create demand much allows the fulfill a be needs

do the miners is And Sometimes, possible we too on demand but can. try the because as much not so. the try to to keep we SKUs it's CMP great we just as

it, sure how not do exactly but the way that's other I'm people And do we it. so


is question with next B. Your Riley. Craig from Ellis

Craig A. Ellis

was me in was think wanted to the at congratulations like very looked at the it tracing. I that noted that you in just high Thanks ray on it computationally time back quarter. announcement GTC with very to records Jensen, I come you financial Because intensive. sneaking for an and that all made the and technology fidelity

seen end think gaming we've that question the So to the for successful. market bring give the technology capability, you of year, group, exciting as end further the does it very this what push high to mean that types ability we about as ray flexibility XXXX the was and past, and platform high with in for potential enter you to dynamics Ti Does is example, very end tracing high launched really launches business market? last those late of for the Thank for the it you.

Jen-Hsun Huang

the ray it day, What well takes take achieve full scene, long and it. that as AI, to one it's into can on learning it do. computer light invention merges rendering, time. it's It in hours, server we might a I cinematic It it is biggest last us years. graphics, except in NVIDIA graphics simulation, rasterization, currently took one basically deep it call real hours appreciate worked And real tracing a the into put a rendering a We cinematic people rendering to so unified render depending is as as time. the one and a few merged film about decade. take framework, it as for the of XX decade computer frame. style perspective, RTX to in Yeah, few literally what on continuously

need. order So the you render in several second, it a one servers number frames hours XX per takes frame. server, server, one render And of to to just of node a imagine

And XX be on Core take hours to just a they render high-performance basically second computer, for need render to generation full a we've and created to to you render second, frames do takes four Quadro you per farm, makes frames call in that's our now able a per We it's the XX real-time, If Wars datacenter enjoyed. be order Tensor that takes frame Volta NVIDIA people RTX real it time. cinematic it It Star of why designed in a farm, latest it render GPUs per a rendering. demonstrated possible to GVXXXs. supercomputer, RTX four which several

we real an took the it datacenter so that we're we it saved, time. node. into And amount and doing entire now one in And basically reduced

on architectural are of the textures amount there people I'm commercials, number can every large. the who lighting. done really built a to going create so the to of certain is design, we do of impact top people course, the way. create rendering And single that of offline rendering industries every is the of and the movies, game NVIDIA use that who people that imagery create quite number and is And computer are RTX car modern film of almost the of game quite design money them. applications, that and bit graphics who single And content, to then save, one There's

and that's towards our better GPU path starting the so in to will of reduce we've time doing is that it dramatically servers reduce the real it And dramatically takes now content do rendering, be Long-term, created. cost purchased. hopefully, of point, the film more it also And course, be to plotted will time.

RTX And a industry card and extraordinary. into revolution to will put the able gaming someday, to the be be transformation we will quite GeForce the to gaming

So RTX. super about we're excited


Your Rasgon with is from Bernstein. Stacy next question

Stacy Aaron Rasgon

the the comments, prior in be gaming seasonality. for me again in up for follow-up fitting normal my for Colette. digits. a want Thanks double seasonal question follow-up. for is guys. on Understanding This Hi, I would QX for to the

declining Given of context Volta can't gaming just drivers within seasonal your ramp, bring on in commentary being datacenter and your the QX, your the idea given crypto with the above guidance the together general of I around that commentary on envelope. the

reconcile I context your that QX are actually contemplated those into you scenarios the are quarter? should next currently for about thinking things? how in the within gaming How So guidance seasonality for of

Colette M. Kress

Sure, Stacy.

if working QX concentrate amount then channel. provide quite to inventory bit separately. in that a here. see of QX And of low the our can to together, little on of see cryptocurrency very to by with to some terms moving outgrew seasonality. if on fast add-on terms Jensen, inventory a able moving decent the and spent We I seasonality to in Unfortunately, me OEM Let we in so But you let moving and QX question, see miners available. bridge left that can here we then add-on. QX, can comments did I Remember establishing wasn't boards. terms next that We me they're of wanted significantly. I and only in clarity in We of the

channel at QX with as where left healthy we as we this inventory levels So overall stand. far point

in if into QX. So that QX. with have there we that dynamics we then inventory now as to address gamers to But carefully you overall those again get terms of now we demand our in overshot available. we we have very terms for can But out channel don't takes that do we seasonality QX, of the fill have do now

should But a again, QX range. putting for together, are together, terms guidance. we'll and seasonality, finish And we So in that normal we'll the yes, in QX how see within of quarter. be you

seasonality. we're overall that on overall should yes, of what mind, our is So, still in are keep that's guidance. you and our from Always QX, at yes, normal a year-to-date Gaming seasonality, usually inclusive higher generally, HXs, about our strong. HXs than think a overall

three in and and this seasonality. that quarters drivers few we terms those put overall We to to comment overall have is market place that and sure things that in clarifies taken our five we terms terms of separately. of all of last making ability a hope here where our get had are, phenomenal I that that the years us have still the just over to We of together. overall gaming grow growth channel in are correct to miners with our


Will Your last SunTrust. from question comes with Stein

William Stein

question And are there end the wondering me the markets that shortages about about to also restricted much. concentrated Hi. helpful. maybe question those. Great. they end Thank squeezing growth out you market. the That'd Thanks The is particular weren't there, recovery might pace talk market achieved supply making so end somewhat other really gaming the my have that of chain in talked in particular, if challenges end the you are so taking the in there's much there. I'm fact if or that in. be relates for and to shortages from that of market, datacenter what something

Colette M. Kress

And it's extremely the XX-bit, start did that off throughout. even Jensen I'll part with consumer But last adoption our very datacenter now seeing the have here, Let phenomenal. tremendous overall, business. me we're business on well. finish Volta different doing remember Again, up than of question. overall and is

time their is that other of on fast, lot of ability qualification amount a qualify. significant have You for and and extremely moving industries based to

growth challenge So Jensen, not all part at of answer overall out. no, extremely our in And of it how of supply in we're with you, there terms is rest came of over the I'll in you datacenter. pleased to terms datacenter, can the it. quarter turn and our the a

Jen-Hsun Huang

Yeah. because The is miners reason why love are world. in everywhere miners GeForce the

everywhere. on their any so distributed, currency. digital PC. benefits supercomputing broadly planet. available is is is And it's it's the in supercomputer infrastructure GeForce that single the sovereign cryptocurrency GeForce the of distributed. has it's largest a Every of One And in not the distributed world, And it's gamer

that strong. is that I pricing is some we those (X:XX:XX). know these in gamers for benefit The play market. can enjoying hopefully the that's And it to the of comes directly demand. to phenomenon. market as would to we fun the any Because that them. the off of why GeForce C&P major as the are PUBG pressure and We cryptography candidate and well. line And mean, miners. And we would it's the people that a serve they we their a for GeForce, so so try (X:XX:XX) best are down, equally the in cryptocurrency GeForces from any as is new astronomic homerun. is because they're GeForce what States. Asia gamers a really of demand line that number come a a to could games global Battle we And is like the built you you'd serve directly with two and your GeForce it's really up take friends. good of called Royale, call ability, ultimately, miners the to for is And of homerun. go And is our best the bottom United we the new majority games in directly, And not And bottom Fortnite represent Europe along. the C&P is gaming is is this needs the rather like can or demand very These team it algorithm the is vast called

It's incredibly their incredibly more social. gamers And and So it's join, their play, more more more more sticky. – gamers join, friends that more of and play. that friends of

Fortnite. guys positive are this it's system, really a it's job that the a enjoying. creating game so feedback Epic And And at fantastic and people genre just wonderful did

over pricing And has quarter. and is strong forward that gamers supply are inventory been normalizing normalize could the in looking to so we so the gaming think the in now back in for can core it, channel, to GeForce I so short at that of that a come channel cards buy the

new be expecting demand able pent-up the I'm the And to GeForces significant. is buy soon. so And to gamers quite pretty


to of it we I any remarks. for turn will over out time. Unfortunately, ran closing now back Jensen

Jen-Hsun Huang

Let's again? turn see here. Is my it

Colette M. Kress

is. Yes, it

Jen-Hsun Huang

record Okay. with record another strong record achieved margins, another was inference. revenue, across had great growth demand for strong. every and platform. earnings, record We Datacenter quarter, Volta AI Gaming

to just to computing next that Core vacuum. you enhanced We of to is quarter. invention as prices position Thank of we've I forward this has for strong see approach heart power gamer incredible you. giving are pent-up is opportunity update At delighted traditional demand ideal our look another GPU serve filling of era. has the our slowed. The GPU further the AI computing Tensor pioneered And our AI better demand. and normalizing the the growth computing can we of


call. This Thank for today's conference joining. you, concludes guys,

disconnect. now may You