저자: JoseNajarro · 외국인
요약 보기
엔비디아 관련 반법무부 조사 루머는 가짜뉴스 성격으로 우려 불필요, 사카나AI 투자 및 xAI 슈퍼컴퓨터 등 AI 인프라 투자 확대는 엔비디아에 우호적. TSMC CoWoS 증설로 엔비디아 공급 확대 전망. 브로드컴은 AI 매출 비중이 엔비디아 대비 작고 커스텀 AI 가속기는 마진을 낮춰 엔비디아의 실질적 경쟁자로 보기 어렵다는 입장. 브로드컴 시간외 하락 언급하나 1~3개월 방향 예측은 없음.
다듬은 글
welcome back to another episode today we're going to take a closer look at Nvidia and some other semiconductor news impacting the overall AI space for example broom just reported earnings and the stock is down after hours so we're going to take a small look at what's happening there and what they had to share more importantly in the AI space a market where they kind of compete with Nvidia so let's get to it with today's episode I want to thank the mle fool for sponsoring this video and check out full.com hose for the 10 best stocks to buy now with that link you get a promotional offer for the subscription service now let's continue with today's episode if you're new here my name is Jose Naro Masters in electrical engineer semiconductor nerd and check out the links below for a free newsletter and my earnings ebook for the semiconductor market now let's look at Nvidia so the first news I want to cover is we did hear from Nvidia that it they mentioned that they did not receive any Anti-Trust files or subpoena from the Department of Justice and this is pretty important where I believe it was Tuesday we kind of heard those news after hours and it did create kind of a lot of bearish sentiment for the stock and now we're seeing that at the moment this seems like fake news it's almost a little bit like the um like the black well delays where there were some truth behind it but not a complete truth so at the moment I don't think this is anything investors should worry about and I've talked about this in my previous episode and like I mentioned I kind of talked about some of the points that that were made against Nvidia and why I personally wouldn't be worried about it as well now the next thing we have for NVIDIA is we did see reports that Nvidia is investing in a Japanese AI company that just did their $100 million funding round this AI company is called Sakana Ai and will work with Nvidia on artificial intelligence research data centers and AI community building so it seems to be a big AI company in Japan that's focusing a a lot of things like research and development and also kind of sovereign AI they do get funding from the overall Japanese government so I believe Nvidia is actually being really smart and making investments in these AI companies for a few reasons obviously if these companies do well in forms of returns um that could be a nice boost in forms of nvidia's um numbers but the other thing is NVIDIA is getting closer and closer to those that are using its products so if you want to see where the next iteration of AI is going to go if you need to see where the next type of AI workload is going to go what's the next type of AI solution the best place are going to be the developers and the engineers working in that so I think by Nvidia making investments in these AI companies it allows them to see what's going on in the AI community in the very kind of first view point first point of view where they're able to design the chips uh the chips that are specialized for that market and I think that's very very important right because you can make the best chip uh in the world but if that best chip doesn't have any type of use cases then there's no point in customers buying it but now if you make the best chip but also the best chip with the best Ed cases and kind of optimize for those types of workloads that are coming up in the future then that definitely gives companies like Nvidia a head start over the weekend Elon musks did mention that xai brought their coloss supercomputer up which is their 100,000 h100 training cluster online and it took them about 122 days to do it from start to finish Elon Musk also mentions that Colossus is the most powerful AI training system in the world more however it will double in size to 200,000 by adding 50,000 more h20s in a few months um so this is going to be a very very powerful system and the reason I believe this is bullish for NVIDIA is just showcases the amount of investment needed for these AI players so obviously if x AI is going to be investing like this you have Claude which is anthropic you have chat GPT which is open AI you have companies like Sakana um that are going to be investing in infrastructure as well um stability AI who else do you have you have um companies for like uh mid Journey for Runway and list goes on and on so there's a lot of different AI use cases right now and the training infrastructure that they need are probably going to be very very huge so I think that's just a bullish sentiment for the AI space in general now there were some reports on X daily news that some kind of competitors of xai are kind of a bit worried uh more importantly it reports that open AI CEO samman for exam for instance has told some Microsoft ex Executives that he is concerned that X AI could soon have more access to computing power than open AI does and I think that's huge right because now this is open AI CEO if true is going to be forcing Microsoft to buy and buy more computing power which obviously bolts well for a company like Nvidia but we can see this is a very competitive space then you have obviously Google which they do their training on their tpus but again you can continue to see the use cases for AI accelerators now staying here in the AI Market there's no shortage in the AI space right now in forms of money one of open ai's co-founder who left the companies to start their own company just raised $1 billion at I believe at a 5 billion valuation for this company um that he's starting SSI so again you're seeing another big company with a very big valuation coming out and most likely are going to need a lot and a lot of infrastructure power um finally the last kind of Ai No I have two more AI news before we look at brockham and some other stuff the CEO of open AI Japan just said that AI called GPT next is is coming which is going to be like GPT 5 and it's expected to be a 100 times better than GPT 4 obviously more computational power means more AI accelerators now the last type of AI news I want to cover before looking at broadcom is Tesla Tesla was up roughly 5% main reason is Tesla share gains on plans to launch full self-driving in Europe and China uh so obviously we don't know Elon can say one thing in forms of when the actual time frame is going to be for that but regardless on when it's going to happen I do believe it's more going to be of it's no longer an if it's more of a when and things like that obviously are going to need strong computational power um so again we see a lot of more use cases in AI continue to evolve and evolve and kind of give more of this bullish sentiment that the AI Market is definitely heading in the right direction now for those that are familiar right Nvidia right now doesn't have a de it has a demand issue but also has Supply issue they have way wait actually they don't have a demand issue the demand issue is that they don't have enough Supply to cover it and one of the main reasons that they don't have enough Supply to cover it is because of coas capacity um and coas is a form of advanced packaging at a recent event tsmc mentions that they plan rapid Coos expansion through 2026 in response to client demand they see between 2022 to 2026 a 50% annual compounded growth rate in coas capacity and this high-speed expansion is continuing at least until 2026 so obviously that means that Nvidia will be able to provide more chips and more chips means more Revenue again this is kind of already known but it's good to kind of get reinform that this is where the market is going that tsmc is increasing that coas capacity because of high demand because it means Nvidia is most likely making a lot and a lot of orders now the final news I want to take a closer look at is brockham right brockham on after hours is down roughly 6.7% the stock is sitting at about $142 so later today or tomorrow I will drop more of a broadcom earnings um but this is probably a great segue check out my ebook um I have a eBook that talks about quarter earnings and this will be coming up in the next update within the next few days special code YouTube will get you $10 off the link is down below and also you there's my free substack newsletter so make sure to check that out well anyways broadcom right now is down roughly $ 140 it's down to $142 we can see year to date this kind of did Peak at roughly $182 so these 140 levels were kind of those bottom levels that we saw when we kind of had that dip early in August so from that Peak the stock is down roughly 20% um so I think broadcom which we're going to see has a mixture of two of three markets for those that are not familiar Amar two main markets it has their software side and their semiconductor side their semiconductor side is pretty much broken down into two markets non aai and AI so very similar to Marvel every kind of Market outside of AI is kind of either pretty low versus year-over-year basis so right now brockham is mainly having the benefactor of its acquisition of VMware and the huge growth in AI accelerators so I thought I I wanted to share some points of what the CEO and some Q&A sessions that kind of correlate with Nvidia so hwan which is the CEO of Brom mentioned that AI Revenue continues to grow strongly with Brom we can see another proof of the point that AI Market continues to grow now another thing that they mentioned is that the main growth is this was again driven by strong demand from hyperscalers for both AI networking and oncustom AI accelerators as you know our hyperscale customers continue to scale up and scale out their AI clusters custom AI accelerators grew 3.5% year-over-year so a lot of these big cap um be big cloud server providers these hyperscalers are designing their own AI chips and they're going to um to broadcom to be able to do that but the kind of number is nowhere near and a lot of people kind of talk about broadcom as this huge huge competitor to Nvidia and we know Nvidia gpus are used for various workloads not just one type of AI accelerator load and we know that Nvidia is making multi-billion dollars right over $30 billion in Revenue $30 billion in Revenue this most recent quarter and over 20 of that comes from Data Center Market for here per quarter for broadcom though the AI demand they had roughly for this quarter was about $3.1 billion and that's not just AI accelerators that's AI accelerators and their networking side now they mentioned for quarter 4 that's expected to grow 10% to over $3.5 billion and we're going to see in a bit that they believe this is about 23 AI accelerator 1/3 networking so out of that $3.5 billion they're only really seeing about 2 something billion in AI accelerators you can see how this is not much of a competition to Nvidia in my opinion now for them because of this AI Revenue they're able to guide their AI revenue for the full year from 11 billions up to billion the other thing that they did mention though is their gross margins for their semiconductor solution segment was down 270 basis points year on-ear driven primarily by higher mix of custom AI accelerators so it seems for them unlike for NVIDIA Nvidia sells their gpus at a very very strong margins and um broadcom unfortunately their AI accelerators are actually hurting their margins a nice amount versus a year ago uh so again very interesting there now some other things that a hawan mentions is this might be a little bit worrisome for the Nvidia space as he just mentions that hey look right now ai compute is needed so much that these hyperscale players are continuing to invest money into as6 solution or how they call it xus which is custom silicon um so that's something that could be seen as a bearish flag for NVIDIA but based on the numbers that we're seeing it seems like they're just trying to offset some of that computational power at the end of the day nvidia's gpus are still need it because obviously they're used for a variety of workload and these hyperscale players need a variety of use cases for their Investments so obviously broadcom can can definitely grow in this space but I don't don't believe it's a true competitor to Nvidia just yet obviously that may move in the numbers but $12 billion in total revenue for AI per year and here I actually have the numbers I think I I'm going to talk about it in my other video um out of that two 12 billion dollar they mentioned roughly um where do I have it here roughly 23 in computes and one3 in networking so out of $12 billion per year that means about 8 billion is coming from AI accelerator that's obviously better than AMD but am uh Nvidia is doing what over a hundred billion dollar in data sensor Revenue alone for the next in for for this fiscal year expected uh so we can see the big difference there the other thing that I thought was pretty interesting is where Nvidia and this is one that really kind of I believe maybe why the stock is down right now is NVIDIA is right now Supply constraint Unfortunately they can't keep up with the demand um and then the analyst kind of ask the CEO what is your main issue right now are you supply constraint could you make more money if you saw if you're um if you had more Supply and Hogan the CEO mentions that hey look right now we can meet those upsides so it seems like this company is not supply constraint they have demand that they're able to provide um which is obviously good you always want to be able to do that but investors would like it more as in the Nvidia case where the issue is actually the supply side because it means you probably have a very very long Viewpoint of where the market is going to go so again I'll probably do more in depth in Broad come later today or tomorrow but I hope that shared a lot of great information and like always thank you for the support take care have a good day and see you next time
Comments 0