저자: Alex · 외국인
요약 보기
엔비디아 실적(매출 351억달러, 데이터센터 매출 전년비 112%↑, 분기 37억달러 중 네트워킹 37억 언급 등)을 근거로 AI 시대 독점적 지위와 블랙웰·후속 칩 로드맵을 강조. 2025년 시총 4조달러 도달 전망, 데이터센터 매출 연간 1750억달러 가능성 제시. 경쟁자 등장 가능성은 낮다고 주장하며 달러코스트평균 매수를 지속한다고 밝힘. 중간중간 스폰서(수면 장비) 광고 포함.
다듬은 글
I'm convinced that Wall Street still doesn't understand Nvidia whether you're a longtime shareholder or you're thinking about buying Nvidia stock for the first time you should know that this earnings call wasn't about revenues or profit margins Nvidia just showed us why it's the king of the entire AI era so in this video I'll break down everything you need to know about nvidia's latest earnings and what that means for Nvidia stock in 2025 and beyond your time is valuable so let's get right into it as always I'm not here to waste your time so here's everything I'll cover in this video I'll highlight nvidia's earnings results focusing on data centers I'll walk you through a few things that most investors still don't understand about nvidia's data center gpus I'll share my thoughts on Nvidia stock and whether I'd still buy it now that it's the biggest company on Earth by market cap and of course where Nvidia will end the year really well all year long there's a ton to talk about so let's Dive Right into Nvidia earnings Nvidia reported record revenues of $35.1 billion for the quarter which is up 177% quarter over quarter and a massive 94% year-over-year Nvidia also posted earnings per share of 78 cents which is up 11% from last year after accounting for their 10 for one stock split this past summer nvidia's data center segment is their biggest business unit by far accounting for about 88% of their total revenues today another reason it's the main main focus of my videos nvidia's data center revenues came in at $3.8 billion which was up 177% quarter over quarter and a whopping 112% year-over-year just to be clear that means that nvidia's data center revenues more than doubled in one year after they became a trillion dollar company and they're on Pace to make well over $100 billion in AI accelerator Revenue this year alone and if you still think the biggest companies can't get bigger remember that Warren Buffett didn't invest in Apple until 2016 10 years after the iPhone came out Apple was the world's biggest company by market cap for 5 years when Warren Buffett bought it and if you've been following this channel for a while then you know that the key to finding great stocks is understanding a company's products not just their profits and the best investments are in companies with platforms that other businesses pay to build on top of today Nvidia is the biggest company on Earth not Apple exactly because every major tech company relies on their accelerated Computing platforms nvidia's hardware and software ecosystems are the picks and shovels of this AI Gold Rush and how big this Gold Rush will ultimately get depends on how fast Nvidia can deploy their AI chips and at what price on this past earnings call nvidia's CFO coet Crest said that Hopper is seeing continued demand and Blackwell's demand greatly exceeds Supply so Nvidia is on track to exceed their previous Blackwell Revenue estimates of several billion dollars for quarter four at this growth rate we could see nvidia's data center revenues reach 175 billion by the end of next year and their total revenues cross the $200 billion Mark as a result I really believe that Nvidia could reach a market cap of $4 trillion in 2025 remember Nvidia currently holds well over a 90% share of the data center GPU market and that's before sales from their Blackwell gpus fully ramp up a ramp that we now know is growing faster than Nvidia previously expected that's because the global artificial intelligence Market is expected to almost 11x in size over the next 9 years which is a compound annual growth rate of over 30% through 2033 and that strong demand is why Nvidia has so much pricing power even as a hardware company they reported gross margins of 74.6% which makes Nvidia as profitable as most software companies that's why I think Nvidia is still a great investment even though it's the biggest company by market cap but there's an even better investment that you need to know about and it comes with huge returns right away and that's eight sleep the sponsor of this video I'm a really light sleeper I always struggle to fall asleep and stay asleep I either get too cold and need an extra blanket or I overheat and get up to crank the AC either way I'm way less productive the next day only to sleep even worse the next night but I've been using the eight sleep pod for a few weeks now and my Sleep Quality has skyrocketed the Pod is a mattress cover packed with advanced technology to track your sleep patterns as well as cool heat and even Elevate you just slip it onto your bed and it adjusts the temperature optimizes each phase of your sleep and shows you how your sleep score is improving over time it's kind of like having a sleep coach a personal thermostat and a fitness tracker allinone it also has Precision temperature and time controls for each side of the bed so my fiance and I can fine-tune our settings separately she absolutely loves our eight sleep pod and I can't recommend it enough because I'm finally getting a good night's sleep so if you're tired of always being tired you can get up to $600 off your pod 4 ultra with my code tsy by using the link in the description that's their best offer of the year and easily one of the best investments I've ever made all right so nidia's earnings per share and data center revenues grew by triple digits year-over-year on top of that Blackwell is already on track to exceed previous Revenue estimates and on top of that Jensen Wong already announced nvidia's next three sets of data center chips after Blackwell earlier this year the Blackwell Ultras which are set to ship in 2025 the new Reuben gpus shipping in 2026 and the Reuben Ultras scheduled for 2027 let me point out a few things that most investors still don't understand about nvidia's data centers systems when most people think about gpus they think about an expensive chip that goes into one computer like a gaming PC or a professional workstation but when Nvidia announces a new architecture like Hopper Blackwell or Reuben it actually includes six or seven chips depending on how you count them there's the GPU itself a CPU a data processing unit or dpu multiple configurations of nvlink chipto Chip switches and infin band or Spectrum X networking Solutions at the rack level and Beyond in fact $3.7 billion of nvidia's Revenue came from networking products in quarter 2 which is still more than all of amd's data center revenues today all right so if each architecture really includes six or seven chips we should walk through how they all fit together I've gained almost 100,000 subscribers since the last time I walked through this so it'll be great to get everybody on the same page first this Blackwell GPU actually combines two dieses into a single chip and connects them with a 10 terb pers second link so they act like one GPU this is a really clever chip design but two dies an ultra high-speed link and a bigger total area for advanced packaging means that Blackwell costs much more to make than Hopper higher unit costs mean lower margins and higher part counts means lower times between failures on average simply because there are more parts that can break personally I think the trade-off is worth it at least in the short term because Blackwell's insane performance is what gives Nvidia a lot of its pricing power compared to other GPU makers and that flexibility is what lets them get even more businesses onto the Kuda platform at a time where every Data Center and cloud service provider is scrambling to buy as many gpus as they can during a gold rush getting everyone on your pix and shovels is what matters most the next level up is the gb200 super chip which pairs two Blackwell b200 gpus with one Gray server CPU they're connected via NV link which is a chipto chip connection with 900 GB per second of bandwidth that's fast enough to move 150 featurelength 4K movies between chips every single second two of those Grace Blackwell chips go into a Blackwell compute node which is one tray in an AI data center rack and the two Super Chips in each tray are also connected via NV link Nvidia also makes these Blue Field dpus or data processing units dpus have a ton of cores to handle highly paralleled workloads like coordinating securing analyzing and transferring data between many different chips inside a data center that way the CPUs and gpus don't have to do that work all right so 18 of these compute trays go into one gb200 nvl 72 system it's called that because the gb200 super chip has two blackwall gpus on it connected by NV link there are two Super Chips per compute tray also connected by NV link and there are 18 trays per system so 2 * 2 * 18 equals 72 Blackwell gpus connected by NV link that's why the whole system is called the gb200 nvl 72 but that's just the compute side of the system so stick with me as I go through the networking side next it's important for investors to understand that nvlink is not just a chipto chip connection the at which we're advancing Computing is insane and it's still not fast enough so we built another chip we call it the MV link switch it's 50 billion transistors it's almost the size of Hopper all by itself this switch ship has four MV links in it each 1.8 terabytes per second what is this chip for if we were to build such a chip we can have every single GPU talk to every other GP at full speed at the same time that's insane sandwiched between the compute nodes we just talked about are nine nvlink switch trays with two nvlink chips per tray and each nvlink chip connects four Blackwell gpus so four ports per nvlink chip times two chips per tray time nine trays is 72 ports supporting all toall GPU Communications at a mindboggling bandwidth of 130 terab per second that's the back that's the that's the back the dgx MV link spine 130 terabytes per second goes through the back of that chassis that is more than the aggregate bandwidth of the internet we could basically send everything to everybody within a second more bandwidth than the entire internet the first time I heard that my brain melted a little all right the 18 compute tray and the nine nvlink switch trays come together to form the gb200 nvl 72 system which is one Rack in an AI data center this one rack has 1.4 exf flops of AI training performance and just to be clear there are only a handful of exaflop supercomputers on Earth today and Nvidia built one in a single 120 KW rack that they sell right off the shelf and at the top of each of these single rack supercomputers exists a single infin band or ethernet X switch tray to connect them to other racks all of which are liquid cool room temperature water flows in hot tub temperature water flows out and the resulting system can be connected to others to make 1 32,000 GPU 645 xof flop AI Factory for the new Industrial Revolution this is more than one giant GPU it's one accelerated Computing platform for the AI era the only other thing I want to point out is that these systems are architecturally compatible which means a hopper tray can be replaced with a Blackwell or a Reuben tray down the road that means when a data center buys Nvidia systems today they're also building out the infrastructure to support the new Chips down the road that's how Nvidia gets in the door this also means that older chips benefit whenever Nvidia writes a new acceleration library or software application so the whole Nvidia Hardware ecosystem is getting better over time which leads to more adop option which leads to more developers and the cycle keeps repeating full circle let me give you a real example last fall Nvidia came out with an open-source software package called tensor RT llm which literally doubled the inference performance for large language models running on nvidia's gpus that means the hundreds of thousands of h100s bought by companies like Amazon Google Microsoft meta open Ai and X got twice as good at running large language models for their users over for night but tensor RT llm also works on nvidia's love La gpus their Amper A1 100s which predate the h100s and it will work on blackw chips too so this software compatibility between GPU Generations helps make nvidia's ecosystem very sticky and that's just the bottom layer of nvidia's Stack nvidia's Cuda ecosystem and acceleration libraries sit on top of that part of the acceleration process is rewriting different software functions and operations that run sequentially on a CPU so that they can work in parallel on many gpus so at a high level this part of the stack enables accelerated Computing across many different markets from Chip design and Gene sequencing to physics simulations Graphics processing and of course generative Ai and the top layer of this stack is nvidia's AI software Suites and applications like Ace their Avatar Cloud engine which lets game developers use generative AI models to power in game characters these applications stitch together many different microservices called Nims to do things like understand and produce speech match a characters lips and facial expressions to what they're saying follow a specific set of rules and guard rails and so on and that's just for digital avatars Nvidia also has software for training and controlling robots called Isaac or self-driving cars via Nvidia drive or modeling and simulating weather patterns in Earth 2 and nvidia's Omniverse is the real Real Time physics-based 3D environment where many of these applications run including where different AI systems go to get trained up before getting deployed in different real world applications like the robots and self-driving cars I just mentioned I've only scratched the surface of nvidia's software stack since nvidia's software revenues are a tiny fraction of their Hardware revenues today and that's a whole different growth area for NVIDIA in the future that really deserves its own videos I spent most of this video covering nvidia's data center ecosystem because that's the real science behind this stock and it has a lot of moving parts that aren't easy to understand until you see how they all fit together the different chips the server racks the infrastructure that connects them and the software that sits on top of that but now that you understand it at least at a high level we can talk about the future of Nvidia stock put out here's what I think is in store for Nvidia stock in 2025 and Beyond like I said earlier I think Nvidia will be the world's first $4 trillion company their revenue and earnings are still growing by triple digits every year and the demand for their AI systems is going faster than even they expected on top of that there's still no competition in sight at least not with enough scale to really make a dent in nvidia's market share based on everything we just watch through I strongly suspect that Nvidia will always be the king of gpus the same way that Google will always be the king of search in Google's case the world is just now starting to move away from search altogether and towards answer engines like chat GPT or perplexity since these answer engines don't make you sift through tons of ads and links to find what you're looking for but think about how long Google's Monopoly on search has lasted likewise I think the company that will disrupt Nvidia won't build a better GPU at all they'll build a fundamentally different kind of chip that will change the way that companies invest in Computing in the first place just like Nvidia is doing right now by disrupting the CPU Market with their gpus on top of that Nvidia stock is still relatively cheap with a forward price to earnings ratio lower than their two main competitors broadcom and AMD as fast as their stock price has gone up their earnings have gone up much faster so long story short I've been dollar cost averaging into Nvidia for years now and that's not about to change but what did change is where Nvidia goe Believe It or Not Nvidia started the year in the number five spot on this list below the NASDAQ 100 Microsoft Google and Amazon since I thought these companies would be the first to integrate AI into their massive software Suites and online applications which means they were going to profit the most from this AI boom especially Microsoft which owns 49% of open Ai and Google which has Deep Mind and their Gemini models they're also the only companies on Earth with the resources to make their own AI chips at scale potentially eating into nvidia's market share but with each earnings call and keynote presentation I saw that more and more value was accruing to Nvidia instead of two nvidia's clients and even though Microsoft Google and Amazon do all make their own AI chips none of them even came close to disrupting Nvidia if anything they're buying even more Nvidia chips today that's why I'm moving Nvidia up one final spot on my list second to only the index that it's already at the top of anyway like I said at the start of the video I think Nvidia will be the king of the entire AI era the same way that Apple was the king of the mobile era and Microsoft was the king of the PC era nd if you want to see what else I've been investing in check out this video
Comments 0