저자: Kevin · 외국인
요약 보기
엔비디아 데이터센터 매출이 급증해 2026년 약 1000억 달러 가능성 제기. 2023년 1월 분기 데이터센터 매출 약 42억 달러, 다음 분기 가이던스 70억 달러(실제 110억 달러). 데이터센터 서버 칩 시장 전환(GPU화)으로 엔비디아 독점 가능성 언급, 주가 3000달러 시나리오 제시. 경쟁자 부재, AI 유행 지속 여부를 주요 리스크로 꼽으며 결론은 유보적.
다듬은 글
nvidia's data center Revenue might double over the next year but not just that I'm not worried about the next year I'm worried about the years after that what's going to happen in 2024 and five and six and who is actually the competition for NVIDIA because I hate to say it but nvidia's Revenue last year ending in January was 15 billion dollars specifically for the server segment the server segment alone data center there it is Data Center Revenue 15 billion dollars that represents out of their 27 billion dollars approximately of total revenue 55 percent of all their money over the last year already came from data centers well next quarter in quarter two of the Year we're expecting uh well a quarter to their Quarter Two we're expecting data center Revenue potentially to come in at 7 billion for the quarter that would be an increase increase from about the 4.3 they did in this quarter well 7 billion in a quarter if we just stay at that rate we're looking at a 28 billion dollar source of Revenue just from data centers alone if you annualize out the next quarter forget gaming for a moment forget professional visualization forget Automotive which is growing already at a rate that's doubling this is their Orient platform this is their Omniverse platform which is like the realistic version of the metaverse that actually makes money and gaming obviously makes a lot of money they're still going to compete in gaming in fact gaming has finally bottomed and we're seeing growth in gaming but the numbers that we got today from Nvidia told us that they're expecting data center revenues to be at around 28 billion dollars annualized for the year that's more than they made all of 2023 2022 oh wait ever so just data centers alone could see a doubling almost a doubling of Revenue just this year and if we get another say 50 growth next year not even a doubling of Revenue you ready for this not even a doubling of data center Revenue anymore forget all of the other data let's just make this so simple 28 billion dollars of data center Revenue in 2023.
now we're going to add 50 that's going to put us at 42 billion in 24. let's add 50 percent growth again that's going to put us at 63 billies at 50 again that'll put us at 90 plus uh one and a half that'd be 94.5 billies right so by 2026 so in other words in three calendar years so called three and a half years or so we could be close to a hundred billies of data center Revenue 100 billies of data center Revenue just roll with me for a moment so you could see the Ridiculousness of what could potentially happen and I don't think anybody saw this coming and I want to be clear about that the stockmart Wall Street was expecting this stock to move four percent in after hours it moved over 20 because nobody expected their forecast of seven billion dollars of Revenue to be a joke and instead their actual forecast to be 11 billion dollars of Revenue that is a massive increase to say we're gonna have seven billies of Revenue and then to beat That by 57 is absolutely insane on top of the fact that you're selling these chips at nearly a 70 gross margin probably around a 50 net margin so let's call it a hundred billies by 2026.
a hundred bill by 2026 and then we'll talk risk factors right that's going to bring you 50 billies to the bottom line there are two and a half billion shares outstanding 50 divided by 2.5 means you're looking at twenty dollars of eps by 2026 per share from data centers alone okay add go ahead and add I don't know 20 billies or whatever for uh for all the other stuff the gaming the automotives uh whatever so 70 divided by 2.5 that's gonna put you at uh 28 of earnings per share so if we now multiply that by let's say because they're growing growing like crazy let's just take a 30pe ratio that's what Apple's trading for now this makes this a 600 stock by 2026 with this sort of growth rate and if we include the other aspects of the business you're actually at close to an 840 stock you're looking at potentially you know what 80 percent return over the next uh three years to potentially a double even at today's levels at 375.
how crazy is that just from going from a double of data center Revenue that would be a hundred percent to get here and then fifty percent fifty percent fifty percent that's assuming Nvidia only grows at these levels now what we heard in the earnings call was actually remarkable that the data center revenue or data center sort of total data center market cap is about a bill now I did a little bit more research after this it's somewhere between a bill to 1.5 Bill and about 50 percent of that is for CPUs and gpus but going forward and this is the big thing so let's go ahead and say it's you know we'll take the midpoint number here 1.25 Bill let's call that the data center Revenue today 50 of that is the chip segment so if 50 of that is the chip segment that puts about 625 million dollars available for chips While most servers are still the old school setups they're just CPUs that's because their data center storage thing servers where you're storing static images pictures video uh you know text files PDFs whatever generic crap Google Drive iCloud kind of stuff text messages right well you now go to generative Ai and let's say you need to rebuild all of these data centers which are now Antiquated as generative AI data centers because everybody wants to start using AI uh assuming AI is not a fad we'll talk about risk factors in a moment well you need to create 725 oh sorry this is a billion actually you need to create 725 billion dollars of of or you're going to spend 700 geez hello you're going to spend 625 billion dollars on chip spend remember we said 50 would be just the chips part the other 50 would be like your fire suppression server racks you know whatever install Services maintenance services whatever the building the land so 625 goes into chips well how much of that 625 is going to be GPU versus CPU well new server systems you would spend about one to two dollars for every about twenty dollars that you spend on a GPU so in other words you're spending about five to ten percent on the CPU which could be like an AMD processor compared to the amount of money you're spending on the gpus okay so let's say the gpus get 90 of that 625 market share we're we're not even going to grow the servers we're not even assuming servers are going to grow because the reality is this one to 1.5 trillion dollar server space could become a three four five ten trillion dollar server space right The Tam could grow but forget about assuming the Tam is going to grow let's just say we're going to take all the old crap and turn it into gpus going forward and not assuming any growth so if we wanted to assume growth maybe we assume these servers stay but we create a new 625 it's not an unrealistic expectation that this would happen within the next three years especially when you consider companies like Google have about 70 percent of their revenue at risk because of AI That's somewhere around 50 billion 48 billion dollars that's at risk per quarter per quarter per year Google has about 200 billion dollars at risk due to Google search advertisements and placement advertisements on websites they've got 200 billion dollars of Revenue at risk if they don't figure out AI out ASAP so Google's got a big incentive and that's just Google it's not Microsoft Amazon Oracle Apple whatever Apple doesn't even do their own servers you think Apple's got chips they use Google servers and some AWS for iCloud crazy right so anyway if you say 90 is going to go to gpus well so 90 goes to gpus that's actually a 562 uh billion dollar spend on gpus well what did we just assume by 2026 we assume just 100 billies of that would be Nvidia well right now you know how much market share Nvidia has of the dedicated graphic chip Market 88 Bros 88 of 562.5 is about 500 billion dollars which is actually 5x this number so now all of a sudden 5x these numbers holy crap 600 times 5x not even considering the other parts of the business this is potentially a three thousand dollar stock now you might think to yourself Kevin this is bubbly this is insane that's true and there's going to be competition and I'm not suggesting this right now is definitely oh yeah that's my price sorry I'm just thinking out loud I'm like oh my God this is a very this is a transformative moment and I want to just speak logically here we don't want to make rash decisions but everybody got caught off guard on this I got caught off guard on this uh even though even though I've been for six to eight months talking about don't invest in software invest in the freaking chips even though I got that right I did not think it would be to this level I thought this was just gonna end up having like you know some growth and then a lot of hype so for me if I drew out sort of my expectation I thought you probably had you know this is the growth part right here and then this purple part is the the recent sort of hype oh the fund manager needs exposure to AI so that's uh you know what you get but what we really got was this was just a fraction you know this was actually like the reality is this right here that whole thing I just drew is like that size and the reality is I like all of this is potentially up for grabs in the server space it's insane so it's a massive massive potential Market at huge margins everybody wants wants the h100 server sector and then you look at tsmc and you're like but Kevin you know they have to be able to manufacture all these chips yeah but what happened during the PC slowdown folks tsmc kept installing new equipment and they ran it at like 50 occupancy in other words they had a very very crumpled up rubber band now Nvidia picks up the phone and says yo we need uh to Triple our orders of h100s yesterday and they're like no problem we have the machines available I don't know why they have that accent but whatever you turn the machines on you start printing checks that's why they think they can supply but they're getting all these it's it's the demand that they're getting that's insane so now you have to ask yourself is is this just a bubble okay so the questions you have to ask yourself are the following number one is AI a fad right I mean like Kevin's got a course featuring AI now how to make more money and get sh9t done we're doing the final release with the final price increase for it on June 1st so if you get in before that you lock in the best price we're going to teach you how to actually be productive with AI how to do video AI how to do image AI how to do daily productive tasks and you get all of the updates long term for free so in other words if in two months there are new ways of doing things better you get an updated course at no cost to you it's a pretty good value proposition I actually think it will be the best AI course that exists now for actual productivity for normal people entrepreneurs and employees alike daily Hardware working Americans that's what this is for so is AI going to be a fad like the metaverse if you think that you probably haven't experienced much AI yet but I'm going to go ahead and conclude that at this point the answer to that is no and the reason for that is that AI is everywhere in practical ways whether it's accounting legal fact checking uh proofreading researching in health care in cancer research I don't care what it is autonomous driving uh you name it AI is there programming right it's in everything sales it doesn't matter it's everywhere unlike the metaverse the metaverse was a joke it still is I mean maybe we'll have virtual reality at some point we'll have the Ready Player one world but that's 10 years out AI is now ai could literally cancel the recession you could see so much spent from businesses that the recession gets canceled because businesses do not want to get left behind by AI so they would actually be more likely to potentially hire or over hire people to make sure that we're not getting left behind in the AI race and I'm saying like AI capable workers right you're not going to hire people who would otherwise get replaced by AI this is why you have to be comfortable with AI and get used to it and add value to AI and realize where its weaknesses are so you still have value all of which will teach but if companies are going to adapt to artificial intelligence and everybody's going to get in on artificial intelligence then companies are going to go through a transition eventually there will probably be more layoffs because artificial intelligence takes your jobs but is now the time to do that weeding probably not now is the time to figure out what's your value proposition how can your company make more Revenue that's the goal right now now it's it's a gold rush folks and Nvidia is selling the pickaxes that's what this is a video is like the only one selling the pickaxes yes AMD does CPUs that'll sit on those server racks but again one Buck maybe two bucks for every 20 you're spending on Nvidia it's insane so tsmc is a winner yes AMD is a winner of course AMD or Nvidia is a winner because they're selling the pickaxes of the gold rush so if you think AI is a fad you're probably not even watching this video so let's get past the fact that it's a fat so what's the next risk to Nvidia well the next risk to Nvidia is competition competition is the next risk who's your competition I'll wait oh uh startups okay well good luck starting a chip a server chip manufacturing company and actually getting the bandwidth to get any of your orders fulfilled at tsmc when Nvidia is just gonna load all that stuff good luck getting into those Supply chains good luck getting designers to make those server chips eventually that'll happen but companies that are relying on AI I don't think are going to gamble trying to save a little bit when they need guaranteed uptime for the AI Services otherwise they're big software businesses fail I don't want to bet on the software businesses it's too speculative to bet on the software businesses I don't know if snowflake's gonna beat our palantir or palantir is going to beat out you know some other service or some other startups gonna be both of them I don't know that but I do know they're all running in video all of them and I don't think a startup can make a better pickaxe can they really make a better pickaxe that's the question and I think the answer that is no so then you're looking at the um Cloud providers basically uh the CPA whatever the cloud provider Services these are your Oracle your Microsoft uh Amazon uh Google right your cloud provider Services well Google came up with their TPU their tensor Processing Unit which they started work on in 2017 but guess what they're still doing they're still buying Nvidia chips because they're better the Nvidia stuff is just better and then you have to ask yourself uh Google spends roughly on just data uh like chip stuff so the way you can figure this out is you look at their whole capex say their capex is seven billion dollars you look at their financial statements you're like okay well of that maybe only about 2.2 billion is uh infrastructure for servers that puts you at about 1 billion for chips because we know about 50 is the chip cost the other 50 is the building the land and the other stuff so if they're spending a billion dollars a quarter on let's say chips uh from Google right what does Google want to do more get a little bit like get those chips for cheaper uh uh basically try to save money on uh save on the one bill or or does Google care about oh wait the 50 billion per quarter we have at risk because of AI well to some extent you could say both but if they de-prioritize actually making the chip and video is just going to create them so why bother that's the problem Nvidia is turning into a monopoly I hate to say it's after today's earnings or a call I'm looking for reasons not to think that Nvidia is a monopoly uh they have an AI chip Monopoly this is insane so I want you to think about some of these these uh tentative things and I have questions for you so number one uh this could cancel uh the recession uh uh number two Nvidia has a monopoly and you might think Nvidia going up uh 20 is you missing out uh or it's a bubble what about when it 10 to 20 X's now people are thinking like oh but Kevin if it's a trillion dollars now could it be five trillion dollars yeah yeah okay maybe not 10x call call it 5x let's be generous 5 to 10x I don't know so okay let's put some of these opinions aside for a moment because we don't need to do these sort of projections right now they don't matter as much what matters is the questions that I have okay here are the real questions so what what growth is the server Market going to see according to asml you're going to see 250 billies of Revenue by 2030 per year that's nominal however that was according to a report that asml put together guess when February of 2023 why is that date important because it was before AI then you have to ask yourself what percent of servers today are GPU based well probably very few because what were gpu-based servers good for before cloud-based gaming who cloud-based games not that many people compared to how many people are going to use AI so in other words the odds are your cloud-based server uh or your gpu-based servers are probably far and few between this is why on the Nvidia call they said most servers are CPU based right now huge Tam opportunity the problem though is anytime you start talking about Tam you get a little bubbly usually right and you kind of lose your foundation in fundamentals and that's scary so you have to be careful about this but then again the growth rate for these chips at Nvidia unless there's a competitor which wears the competitor could be phenomenal now we also don't know if this is going to be like a one and done right what if it's one and done in other words what if everyone buys and orders and h100 and then they have them and then they're done okay now we've translated transitioned the growth slows down you had a big AI hype boom and then it dies and then it's like okay data center revenue is falling again maybe unless of course AI uh explodes and everybody's like Oh I thought everybody was using Ai No actually only one percent of companies were serious about using Ai and now everybody's using it it's very difficult to know but there is a real there are two arguments here and I want to know your thoughts like this is an important one you I need your help in leaving comments down below there are two ways Nvidia I mean there are many ways Nvidia can go here but these these h-100s the these GPU server chips will either get competed away uh or you'll see this one-time purchasing right you'll see this alternatively you could see h100s via software and Via just being the best chips that there are maintain a massive moat the total server segment could 3x with a demand to rebuild all or to build all new as GPU that would be somewhere around two to three trillion in spend on servers that's an insane number by the way trillion on spend that's an insane number uh you know again Google spent like seven on capex last quarter that's like 28 bill in uh in in the year on average annualized even if they doubled they'd be at 50 bill you would need 20 Googles just to get to a trillion you'd need 40 Googles to get to 2 trillion that's assuming they already double so there are definitely some pie in the sky risk factors over oh my gosh let's just extrapolate everything to the Moon so there are definitely risks here and I do not yet have all of the answers and I'm already 22 minutes into this video but my question for you and which I really want you to answer down below is why have you not joined those programs on building your wealth yet where we can brainstorm together and then the second question I have is who's the competition third question what's the growth rate in your opinion going to look like for the server sector and how much needs to be GPU is AI a fad if it's not a fad then what's to stop Nvidia from being the Monopoly of of of this Market I'm blown away I do not have all the answers I was absolutely blindsided by today's report I wholeheartedly admit that I was blindsided by today's report I knew the chips would do well so I was right about that but I was not all the way right because I just didn't realize the magnitude at which this this thesis would be correct and then then we do not know how much longer that will last either so I need your help please leave comments down below thorough details especially if you have experience in this and I will put another video if you leave good comment or like real useful comments I'll put another video together where we do uh some more conclusive uh research but these are just my first impressions and I'm
Comments 0